annotate mercurial/parsers.c @ 16955:92e1c64ba0d4

parsers: add a C function to pack the dirstate This is about 9 times faster than the Python dirstate packing code. The relatively small speedup is due to the poor locality and memory access patterns caused by traversing dicts and other boxed Python values.
author Bryan O'Sullivan <bryano@fb.com>
date Wed, 30 May 2012 12:55:33 -0700
parents bbedef66c6f3
children 249cc4ec4327
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1 /*
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
2 parsers.c - efficient content parsing
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
3
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
4 Copyright 2008 Matt Mackall <mpm@selenic.com> and others
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
5
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
6 This software may be used and distributed according to the terms of
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
7 the GNU General Public License, incorporated herein by reference.
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
8 */
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
9
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
10 #include <Python.h>
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
11 #include <ctype.h>
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
12 #include <string.h>
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
13
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
14 #include "util.h"
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
15
16617
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
16 static inline int hexdigit(const char *p, Py_ssize_t off)
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
17 {
16617
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
18 char c = p[off];
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
19
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
20 if (c >= '0' && c <= '9')
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
21 return c - '0';
7092
fb3fc27617a2 parsers: speed up hex decoding for manifests
Matt Mackall <mpm@selenic.com>
parents: 7091
diff changeset
22 if (c >= 'a' && c <= 'f')
fb3fc27617a2 parsers: speed up hex decoding for manifests
Matt Mackall <mpm@selenic.com>
parents: 7091
diff changeset
23 return c - 'a' + 10;
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
24 if (c >= 'A' && c <= 'F')
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
25 return c - 'A' + 10;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
26
7092
fb3fc27617a2 parsers: speed up hex decoding for manifests
Matt Mackall <mpm@selenic.com>
parents: 7091
diff changeset
27 PyErr_SetString(PyExc_ValueError, "input contains non-hex character");
fb3fc27617a2 parsers: speed up hex decoding for manifests
Matt Mackall <mpm@selenic.com>
parents: 7091
diff changeset
28 return 0;
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
29 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
30
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
31 /*
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
32 * Turn a hex-encoded string into binary.
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
33 */
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
34 static PyObject *unhexlify(const char *str, int len)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
35 {
7092
fb3fc27617a2 parsers: speed up hex decoding for manifests
Matt Mackall <mpm@selenic.com>
parents: 7091
diff changeset
36 PyObject *ret;
6395
3f0294536b24 fix const annotation warning
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 6389
diff changeset
37 char *d;
16617
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
38 int i;
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
39
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
40 ret = PyBytes_FromStringAndSize(NULL, len / 2);
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
41
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
42 if (!ret)
7092
fb3fc27617a2 parsers: speed up hex decoding for manifests
Matt Mackall <mpm@selenic.com>
parents: 7091
diff changeset
43 return NULL;
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
44
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
45 d = PyBytes_AsString(ret);
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
46
16617
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
47 for (i = 0; i < len;) {
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
48 int hi = hexdigit(str, i++);
4fb16743049d parsers: change the type signature of hexdigit
Bryan O'Sullivan <bryano@fb.com>
parents: 16616
diff changeset
49 int lo = hexdigit(str, i++);
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
50 *d++ = (hi << 4) | lo;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
51 }
7091
12b35ae03365 parsers: clean up whitespace
Matt Mackall <mpm@selenic.com>
parents: 6395
diff changeset
52
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
53 return ret;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
54 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
55
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
56 /*
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
57 * This code assumes that a manifest is stitched together with newline
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
58 * ('\n') characters.
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
59 */
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
60 static PyObject *parse_manifest(PyObject *self, PyObject *args)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
61 {
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
62 PyObject *mfdict, *fdict;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
63 char *str, *cur, *start, *zero;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
64 int len;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
65
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
66 if (!PyArg_ParseTuple(args, "O!O!s#:parse_manifest",
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
67 &PyDict_Type, &mfdict,
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
68 &PyDict_Type, &fdict,
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
69 &str, &len))
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
70 goto quit;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
71
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
72 for (start = cur = str, zero = NULL; cur < str + len; cur++) {
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
73 PyObject *file = NULL, *node = NULL;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
74 PyObject *flags = NULL;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
75 int nlen;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
76
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
77 if (!*cur) {
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
78 zero = cur;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
79 continue;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
80 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
81 else if (*cur != '\n')
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
82 continue;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
83
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
84 if (!zero) {
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
85 PyErr_SetString(PyExc_ValueError,
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
86 "manifest entry has no separator");
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
87 goto quit;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
88 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
89
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
90 file = PyBytes_FromStringAndSize(start, zero - start);
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
91
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
92 if (!file)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
93 goto bail;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
94
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
95 nlen = cur - zero - 1;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
96
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
97 node = unhexlify(zero + 1, nlen > 40 ? 40 : nlen);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
98 if (!node)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
99 goto bail;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
100
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
101 if (nlen > 40) {
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
102 flags = PyBytes_FromStringAndSize(zero + 41,
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
103 nlen - 40);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
104 if (!flags)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
105 goto bail;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
106
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
107 if (PyDict_SetItem(fdict, file, flags) == -1)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
108 goto bail;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
109 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
110
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
111 if (PyDict_SetItem(mfdict, file, node) == -1)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
112 goto bail;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
113
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
114 start = cur + 1;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
115 zero = NULL;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
116
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
117 Py_XDECREF(flags);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
118 Py_XDECREF(node);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
119 Py_XDECREF(file);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
120 continue;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
121 bail:
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
122 Py_XDECREF(flags);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
123 Py_XDECREF(node);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
124 Py_XDECREF(file);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
125 goto quit;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
126 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
127
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
128 if (len > 0 && *(cur - 1) != '\n') {
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
129 PyErr_SetString(PyExc_ValueError,
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
130 "manifest contains trailing garbage");
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
131 goto quit;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
132 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
133
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
134 Py_INCREF(Py_None);
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
135 return Py_None;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
136 quit:
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
137 return NULL;
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
138 }
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
139
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
140 static PyObject *parse_dirstate(PyObject *self, PyObject *args)
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
141 {
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
142 PyObject *dmap, *cmap, *parents = NULL, *ret = NULL;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
143 PyObject *fname = NULL, *cname = NULL, *entry = NULL;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
144 char *str, *cur, *end, *cpos;
7174
4da87407b845 parsers.c: fix integer overflows
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7168
diff changeset
145 int state, mode, size, mtime;
4da87407b845 parsers.c: fix integer overflows
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7168
diff changeset
146 unsigned int flen;
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
147 int len;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
148
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
149 if (!PyArg_ParseTuple(args, "O!O!s#:parse_dirstate",
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
150 &PyDict_Type, &dmap,
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
151 &PyDict_Type, &cmap,
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
152 &str, &len))
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
153 goto quit;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
154
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
155 /* read parents */
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
156 if (len < 40)
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
157 goto quit;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
158
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
159 parents = Py_BuildValue("s#s#", str, 20, str + 20, 20);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
160 if (!parents)
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
161 goto quit;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
162
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
163 /* read filenames */
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
164 cur = str + 40;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
165 end = str + len;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
166
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
167 while (cur < end - 17) {
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
168 /* unpack header */
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
169 state = *cur;
16437
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
170 mode = getbe32(cur + 1);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
171 size = getbe32(cur + 5);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
172 mtime = getbe32(cur + 9);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
173 flen = getbe32(cur + 13);
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
174 cur += 17;
10449
7c8266c1d15a parsers: fix some signed comparison issues
Matt Mackall <mpm@selenic.com>
parents: 10282
diff changeset
175 if (cur + flen > end || cur + flen < cur) {
7174
4da87407b845 parsers.c: fix integer overflows
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7168
diff changeset
176 PyErr_SetString(PyExc_ValueError, "overflow in dirstate");
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
177 goto quit;
7174
4da87407b845 parsers.c: fix integer overflows
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7168
diff changeset
178 }
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
179
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
180 entry = Py_BuildValue("ciii", state, mode, size, mtime);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
181 if (!entry)
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
182 goto quit;
7175
5d8626b2c1db parsers.c: do not try to untrack after a failure
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7174
diff changeset
183 PyObject_GC_UnTrack(entry); /* don't waste time with this */
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
184
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
185 cpos = memchr(cur, 0, flen);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
186 if (cpos) {
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
187 fname = PyBytes_FromStringAndSize(cur, cpos - cur);
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
188 cname = PyBytes_FromStringAndSize(cpos + 1,
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
189 flen - (cpos - cur) - 1);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
190 if (!fname || !cname ||
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
191 PyDict_SetItem(cmap, fname, cname) == -1 ||
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
192 PyDict_SetItem(dmap, fname, entry) == -1)
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
193 goto quit;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
194 Py_DECREF(cname);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
195 } else {
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
196 fname = PyBytes_FromStringAndSize(cur, flen);
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
197 if (!fname ||
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
198 PyDict_SetItem(dmap, fname, entry) == -1)
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
199 goto quit;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
200 }
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
201 cur += flen;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
202 Py_DECREF(fname);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
203 Py_DECREF(entry);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
204 fname = cname = entry = NULL;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
205 }
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
206
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
207 ret = parents;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
208 Py_INCREF(ret);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
209 quit:
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
210 Py_XDECREF(fname);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
211 Py_XDECREF(cname);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
212 Py_XDECREF(entry);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
213 Py_XDECREF(parents);
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
214 return ret;
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
215 }
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
216
16955
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
217 static inline int getintat(PyObject *tuple, int off, uint32_t *v)
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
218 {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
219 PyObject *o = PyTuple_GET_ITEM(tuple, off);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
220 long val;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
221
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
222 if (PyInt_Check(o))
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
223 val = PyInt_AS_LONG(o);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
224 else if (PyLong_Check(o)) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
225 val = PyLong_AsLong(o);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
226 if (val == -1 && PyErr_Occurred())
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
227 return -1;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
228 } else {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
229 PyErr_SetString(PyExc_TypeError, "expected an int or long");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
230 return -1;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
231 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
232 if (LONG_MAX > INT_MAX && (val > INT_MAX || val < INT_MIN)) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
233 PyErr_SetString(PyExc_OverflowError,
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
234 "Python value to large to convert to uint32_t");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
235 return -1;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
236 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
237 *v = (uint32_t)val;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
238 return 0;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
239 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
240
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
241 static PyObject *dirstate_unset;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
242
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
243 /*
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
244 * Efficiently pack a dirstate object into its on-disk format.
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
245 */
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
246 static PyObject *pack_dirstate(PyObject *self, PyObject *args)
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
247 {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
248 PyObject *packobj = NULL;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
249 PyObject *map, *copymap, *pl;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
250 Py_ssize_t nbytes, pos, l;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
251 PyObject *k, *v, *pn;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
252 char *p, *s;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
253 double now;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
254
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
255 if (!PyArg_ParseTuple(args, "O!O!Od:pack_dirstate",
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
256 &PyDict_Type, &map, &PyDict_Type, &copymap,
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
257 &pl, &now))
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
258 return NULL;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
259
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
260 if (!PySequence_Check(pl) || PySequence_Size(pl) != 2) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
261 PyErr_SetString(PyExc_TypeError, "expected 2-element sequence");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
262 return NULL;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
263 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
264
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
265 /* Figure out how much we need to allocate. */
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
266 for (nbytes = 40, pos = 0; PyDict_Next(map, &pos, &k, &v);) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
267 PyObject *c;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
268 if (!PyString_Check(k)) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
269 PyErr_SetString(PyExc_TypeError, "expected string key");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
270 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
271 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
272 nbytes += PyString_GET_SIZE(k) + 17;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
273 c = PyDict_GetItem(copymap, k);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
274 if (c) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
275 if (!PyString_Check(c)) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
276 PyErr_SetString(PyExc_TypeError,
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
277 "expected string key");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
278 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
279 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
280 nbytes += PyString_GET_SIZE(c) + 1;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
281 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
282 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
283
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
284 packobj = PyString_FromStringAndSize(NULL, nbytes);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
285 if (packobj == NULL)
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
286 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
287
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
288 p = PyString_AS_STRING(packobj);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
289
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
290 pn = PySequence_ITEM(pl, 0);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
291 if (PyString_AsStringAndSize(pn, &s, &l) == -1 || l != 20) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
292 PyErr_SetString(PyExc_TypeError, "expected a 20-byte hash");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
293 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
294 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
295 memcpy(p, s, l);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
296 p += 20;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
297 pn = PySequence_ITEM(pl, 1);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
298 if (PyString_AsStringAndSize(pn, &s, &l) == -1 || l != 20) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
299 PyErr_SetString(PyExc_TypeError, "expected a 20-byte hash");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
300 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
301 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
302 memcpy(p, s, l);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
303 p += 20;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
304
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
305 for (pos = 0; PyDict_Next(map, &pos, &k, &v); ) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
306 uint32_t mode, size, mtime;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
307 Py_ssize_t len, l;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
308 PyObject *o;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
309 char *s, *t;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
310 int err;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
311
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
312 if (!PyTuple_Check(v) || PyTuple_GET_SIZE(v) != 4) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
313 PyErr_SetString(PyExc_TypeError, "expected a 4-tuple");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
314 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
315 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
316 o = PyTuple_GET_ITEM(v, 0);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
317 if (PyString_AsStringAndSize(o, &s, &l) == -1 || l != 1) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
318 PyErr_SetString(PyExc_TypeError, "expected one byte");
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
319 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
320 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
321 *p++ = *s;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
322 err = getintat(v, 1, &mode);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
323 err |= getintat(v, 2, &size);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
324 err |= getintat(v, 3, &mtime);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
325 if (err)
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
326 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
327 if (*s == 'n' && mtime == (uint32_t)now) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
328 /* See dirstate.py:write for why we do this. */
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
329 if (PyDict_SetItem(map, k, dirstate_unset) == -1)
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
330 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
331 mode = 0, size = -1, mtime = -1;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
332 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
333 putbe32(mode, p);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
334 putbe32(size, p + 4);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
335 putbe32(mtime, p + 8);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
336 t = p + 12;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
337 p += 16;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
338 len = PyString_GET_SIZE(k);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
339 memcpy(p, PyString_AS_STRING(k), len);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
340 p += len;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
341 o = PyDict_GetItem(copymap, k);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
342 if (o) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
343 *p++ = '\0';
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
344 l = PyString_GET_SIZE(o);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
345 memcpy(p, PyString_AS_STRING(o), l);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
346 p += l;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
347 len += l + 1;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
348 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
349 putbe32((uint32_t)len, t);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
350 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
351
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
352 pos = p - PyString_AS_STRING(packobj);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
353 if (pos != nbytes) {
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
354 PyErr_Format(PyExc_SystemError, "bad dirstate size: %ld != %ld",
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
355 (long)pos, (long)nbytes);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
356 goto bail;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
357 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
358
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
359 return packobj;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
360 bail:
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
361 Py_XDECREF(packobj);
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
362 return NULL;
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
363 }
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
364
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
365 /*
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
366 * A base-16 trie for fast node->rev mapping.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
367 *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
368 * Positive value is index of the next node in the trie
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
369 * Negative value is a leaf: -(rev + 1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
370 * Zero is empty
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
371 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
372 typedef struct {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
373 int children[16];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
374 } nodetree;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
375
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
376 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
377 * This class has two behaviours.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
378 *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
379 * When used in a list-like way (with integer keys), we decode an
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
380 * entry in a RevlogNG index file on demand. Our last entry is a
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
381 * sentinel, always a nullid. We have limited support for
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
382 * integer-keyed insert and delete, only at elements right before the
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
383 * sentinel.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
384 *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
385 * With string keys, we lazily perform a reverse mapping from node to
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
386 * rev, using a base-16 trie.
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
387 */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
388 typedef struct {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
389 PyObject_HEAD
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
390 /* Type-specific fields go here. */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
391 PyObject *data; /* raw bytes of index */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
392 PyObject **cache; /* cached tuples */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
393 const char **offsets; /* populated on demand */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
394 Py_ssize_t raw_length; /* original number of elements */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
395 Py_ssize_t length; /* current number of elements */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
396 PyObject *added; /* populated on demand */
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
397 PyObject *headrevs; /* cache, invalidated on changes */
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
398 nodetree *nt; /* base-16 trie */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
399 int ntlength; /* # nodes in use */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
400 int ntcapacity; /* # nodes allocated */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
401 int ntdepth; /* maximum depth of tree */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
402 int ntsplits; /* # splits performed */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
403 int ntrev; /* last rev scanned */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
404 int ntlookups; /* # lookups */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
405 int ntmisses; /* # lookups that miss the cache */
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
406 int inlined;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
407 } indexObject;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
408
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
409 static Py_ssize_t index_length(const indexObject *self)
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
410 {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
411 if (self->added == NULL)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
412 return self->length;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
413 return self->length + PyList_GET_SIZE(self->added);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
414 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
415
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
416 static PyObject *nullentry;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
417 static const char nullid[20];
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
418
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
419 static long inline_scan(indexObject *self, const char **offsets);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
420
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
421 #if LONG_MAX == 0x7fffffffL
16393
ee163a9cf37c util.h: more Python 2.4 fixes
Matt Mackall <mpm@selenic.com>
parents: 16385
diff changeset
422 static char *tuple_format = "Kiiiiiis#";
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
423 #else
16393
ee163a9cf37c util.h: more Python 2.4 fixes
Matt Mackall <mpm@selenic.com>
parents: 16385
diff changeset
424 static char *tuple_format = "kiiiiiis#";
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
425 #endif
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
426
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
427 /* A RevlogNG v1 index entry is 64 bytes long. */
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
428 static const long v1_hdrsize = 64;
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
429
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
430 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
431 * Return a pointer to the beginning of a RevlogNG record.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
432 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
433 static const char *index_deref(indexObject *self, Py_ssize_t pos)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
434 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
435 if (self->inlined && pos > 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
436 if (self->offsets == NULL) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
437 self->offsets = malloc(self->raw_length *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
438 sizeof(*self->offsets));
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
439 if (self->offsets == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
440 return (const char *)PyErr_NoMemory();
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
441 inline_scan(self, self->offsets);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
442 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
443 return self->offsets[pos];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
444 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
445
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
446 return PyString_AS_STRING(self->data) + pos * v1_hdrsize;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
447 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
448
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
449 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
450 * RevlogNG format (all in big endian, data may be inlined):
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
451 * 6 bytes: offset
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
452 * 2 bytes: flags
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
453 * 4 bytes: compressed length
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
454 * 4 bytes: uncompressed length
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
455 * 4 bytes: base revision
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
456 * 4 bytes: link revision
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
457 * 4 bytes: parent 1 revision
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
458 * 4 bytes: parent 2 revision
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
459 * 32 bytes: nodeid (only 20 bytes used)
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
460 */
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
461 static PyObject *index_get(indexObject *self, Py_ssize_t pos)
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
462 {
7154
7fdf7a0a41b7 index parser: fix refcounting in case of errors, refactor
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7135
diff changeset
463 uint64_t offset_flags;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
464 int comp_len, uncomp_len, base_rev, link_rev, parent_1, parent_2;
7154
7fdf7a0a41b7 index parser: fix refcounting in case of errors, refactor
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7135
diff changeset
465 const char *c_node_id;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
466 const char *data;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
467 Py_ssize_t length = index_length(self);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
468 PyObject *entry;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
469
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
470 if (pos < 0)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
471 pos += length;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
472
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
473 if (pos < 0 || pos >= length) {
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
474 PyErr_SetString(PyExc_IndexError, "revlog index out of range");
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
475 return NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
476 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
477
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
478 if (pos == length - 1) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
479 Py_INCREF(nullentry);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
480 return nullentry;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
481 }
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
482
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
483 if (pos >= self->length - 1) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
484 PyObject *obj;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
485 obj = PyList_GET_ITEM(self->added, pos - self->length + 1);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
486 Py_INCREF(obj);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
487 return obj;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
488 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
489
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
490 if (self->cache) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
491 if (self->cache[pos]) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
492 Py_INCREF(self->cache[pos]);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
493 return self->cache[pos];
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
494 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
495 } else {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
496 self->cache = calloc(self->raw_length, sizeof(PyObject *));
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
497 if (self->cache == NULL)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
498 return PyErr_NoMemory();
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
499 }
7190
aecea6934fdd Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents: 7186
diff changeset
500
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
501 data = index_deref(self, pos);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
502 if (data == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
503 return NULL;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
504
16437
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
505 offset_flags = getbe32(data + 4);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
506 if (pos == 0) /* mask out version number for the first entry */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
507 offset_flags &= 0xFFFF;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
508 else {
16437
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
509 uint32_t offset_high = getbe32(data);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
510 offset_flags |= ((uint64_t)offset_high) << 32;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
511 }
7154
7fdf7a0a41b7 index parser: fix refcounting in case of errors, refactor
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents: 7135
diff changeset
512
16437
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
513 comp_len = getbe32(data + 8);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
514 uncomp_len = getbe32(data + 12);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
515 base_rev = getbe32(data + 16);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
516 link_rev = getbe32(data + 20);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
517 parent_1 = getbe32(data + 24);
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
518 parent_2 = getbe32(data + 28);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
519 c_node_id = data + 32;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
520
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
521 entry = Py_BuildValue(tuple_format, offset_flags, comp_len,
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
522 uncomp_len, base_rev, link_rev,
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
523 parent_1, parent_2, c_node_id, 20);
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
524
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
525 if (entry)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
526 PyObject_GC_UnTrack(entry);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
527
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
528 self->cache[pos] = entry;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
529 Py_INCREF(entry);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
530
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
531 return entry;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
532 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
533
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
534 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
535 * Return the 20-byte SHA of the node corresponding to the given rev.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
536 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
537 static const char *index_node(indexObject *self, Py_ssize_t pos)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
538 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
539 Py_ssize_t length = index_length(self);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
540 const char *data;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
541
16664
5bc6edf71b39 parsers: ensure that nullid is always present in the radix tree
Bryan O'Sullivan <bryano@fb.com>
parents: 16663
diff changeset
542 if (pos == length - 1 || pos == INT_MAX)
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
543 return nullid;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
544
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
545 if (pos >= length)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
546 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
547
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
548 if (pos >= self->length - 1) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
549 PyObject *tuple, *str;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
550 tuple = PyList_GET_ITEM(self->added, pos - self->length + 1);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
551 str = PyTuple_GetItem(tuple, 7);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
552 return str ? PyString_AS_STRING(str) : NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
553 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
554
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
555 data = index_deref(self, pos);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
556 return data ? data + 32 : NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
557 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
558
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
559 static int nt_insert(indexObject *self, const char *node, int rev);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
560
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
561 static int node_check(PyObject *obj, char **node, Py_ssize_t *nodelen)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
562 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
563 if (PyString_AsStringAndSize(obj, node, nodelen) == -1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
564 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
565 if (*nodelen == 20)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
566 return 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
567 PyErr_SetString(PyExc_ValueError, "20-byte hash required");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
568 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
569 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
570
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
571 static PyObject *index_insert(indexObject *self, PyObject *args)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
572 {
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
573 PyObject *obj;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
574 char *node;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
575 long offset;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
576 Py_ssize_t len, nodelen;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
577
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
578 if (!PyArg_ParseTuple(args, "lO", &offset, &obj))
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
579 return NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
580
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
581 if (!PyTuple_Check(obj) || PyTuple_GET_SIZE(obj) != 8) {
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
582 PyErr_SetString(PyExc_TypeError, "8-tuple required");
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
583 return NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
584 }
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
585
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
586 if (node_check(PyTuple_GET_ITEM(obj, 7), &node, &nodelen) == -1)
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
587 return NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
588
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
589 len = index_length(self);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
590
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
591 if (offset < 0)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
592 offset += len;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
593
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
594 if (offset != len - 1) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
595 PyErr_SetString(PyExc_IndexError,
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
596 "insert only supported at index -1");
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
597 return NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
598 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
599
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
600 if (offset > INT_MAX) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
601 PyErr_SetString(PyExc_ValueError,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
602 "currently only 2**31 revs supported");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
603 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
604 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
605
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
606 if (self->added == NULL) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
607 self->added = PyList_New(0);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
608 if (self->added == NULL)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
609 return NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
610 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
611
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
612 if (PyList_Append(self->added, obj) == -1)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
613 return NULL;
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
614
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
615 if (self->nt)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
616 nt_insert(self, node, (int)offset);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
617
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
618 Py_CLEAR(self->headrevs);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
619 Py_RETURN_NONE;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
620 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
621
16370
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
622 static void _index_clearcaches(indexObject *self)
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
623 {
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
624 if (self->cache) {
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
625 Py_ssize_t i;
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
626
16732
277e2acb7e5c parsers: use Py_CLEAR where appropriate
Bryan O'Sullivan <bryano@fb.com>
parents: 16699
diff changeset
627 for (i = 0; i < self->raw_length; i++)
277e2acb7e5c parsers: use Py_CLEAR where appropriate
Bryan O'Sullivan <bryano@fb.com>
parents: 16699
diff changeset
628 Py_CLEAR(self->cache[i]);
16370
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
629 free(self->cache);
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
630 self->cache = NULL;
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
631 }
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
632 if (self->offsets) {
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
633 free(self->offsets);
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
634 self->offsets = NULL;
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
635 }
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
636 if (self->nt) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
637 free(self->nt);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
638 self->nt = NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
639 }
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
640 Py_CLEAR(self->headrevs);
16370
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
641 }
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
642
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
643 static PyObject *index_clearcaches(indexObject *self)
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
644 {
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
645 _index_clearcaches(self);
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
646 self->ntlength = self->ntcapacity = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
647 self->ntdepth = self->ntsplits = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
648 self->ntrev = -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
649 self->ntlookups = self->ntmisses = 0;
16370
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
650 Py_RETURN_NONE;
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
651 }
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
652
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
653 static PyObject *index_stats(indexObject *self)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
654 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
655 PyObject *obj = PyDict_New();
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
656
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
657 if (obj == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
658 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
659
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
660 #define istat(__n, __d) \
16629
1435866c1937 parser: use PyInt_FromSsize_t in index_stats
Adrian Buehlmann <adrian@cadifra.com>
parents: 16621
diff changeset
661 if (PyDict_SetItemString(obj, __d, PyInt_FromSsize_t(self->__n)) == -1) \
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
662 goto bail;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
663
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
664 if (self->added) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
665 Py_ssize_t len = PyList_GET_SIZE(self->added);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
666 if (PyDict_SetItemString(obj, "index entries added",
16629
1435866c1937 parser: use PyInt_FromSsize_t in index_stats
Adrian Buehlmann <adrian@cadifra.com>
parents: 16621
diff changeset
667 PyInt_FromSsize_t(len)) == -1)
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
668 goto bail;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
669 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
670
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
671 if (self->raw_length != self->length - 1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
672 istat(raw_length, "revs on disk");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
673 istat(length, "revs in memory");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
674 istat(ntcapacity, "node trie capacity");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
675 istat(ntdepth, "node trie depth");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
676 istat(ntlength, "node trie count");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
677 istat(ntlookups, "node trie lookups");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
678 istat(ntmisses, "node trie misses");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
679 istat(ntrev, "node trie last rev scanned");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
680 istat(ntsplits, "node trie splits");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
681
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
682 #undef istat
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
683
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
684 return obj;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
685
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
686 bail:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
687 Py_XDECREF(obj);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
688 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
689 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
690
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
691 /*
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
692 * When we cache a list, we want to be sure the caller can't mutate
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
693 * the cached copy.
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
694 */
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
695 static PyObject *list_copy(PyObject *list)
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
696 {
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
697 Py_ssize_t len = PyList_GET_SIZE(list);
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
698 PyObject *newlist = PyList_New(len);
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
699 Py_ssize_t i;
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
700
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
701 if (newlist == NULL)
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
702 return NULL;
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
703
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
704 for (i = 0; i < len; i++) {
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
705 PyObject *obj = PyList_GET_ITEM(list, i);
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
706 Py_INCREF(obj);
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
707 PyList_SET_ITEM(newlist, i, obj);
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
708 }
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
709
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
710 return newlist;
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
711 }
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
712
16786
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
713 static PyObject *index_headrevs(indexObject *self)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
714 {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
715 Py_ssize_t i, len, addlen;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
716 char *nothead = NULL;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
717 PyObject *heads;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
718
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
719 if (self->headrevs)
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
720 return list_copy(self->headrevs);
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
721
16786
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
722 len = index_length(self) - 1;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
723 heads = PyList_New(0);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
724 if (heads == NULL)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
725 goto bail;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
726 if (len == 0) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
727 PyObject *nullid = PyInt_FromLong(-1);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
728 if (nullid == NULL || PyList_Append(heads, nullid) == -1) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
729 Py_XDECREF(nullid);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
730 goto bail;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
731 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
732 goto done;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
733 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
734
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
735 nothead = calloc(len, 1);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
736 if (nothead == NULL)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
737 goto bail;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
738
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
739 for (i = 0; i < self->raw_length; i++) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
740 const char *data = index_deref(self, i);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
741 int parent_1 = getbe32(data + 24);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
742 int parent_2 = getbe32(data + 28);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
743 if (parent_1 >= 0)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
744 nothead[parent_1] = 1;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
745 if (parent_2 >= 0)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
746 nothead[parent_2] = 1;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
747 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
748
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
749 addlen = self->added ? PyList_GET_SIZE(self->added) : 0;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
750
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
751 for (i = 0; i < addlen; i++) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
752 PyObject *rev = PyList_GET_ITEM(self->added, i);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
753 PyObject *p1 = PyTuple_GET_ITEM(rev, 5);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
754 PyObject *p2 = PyTuple_GET_ITEM(rev, 6);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
755 long parent_1, parent_2;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
756
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
757 if (!PyInt_Check(p1) || !PyInt_Check(p2)) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
758 PyErr_SetString(PyExc_TypeError,
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
759 "revlog parents are invalid");
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
760 goto bail;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
761 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
762 parent_1 = PyInt_AS_LONG(p1);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
763 parent_2 = PyInt_AS_LONG(p2);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
764 if (parent_1 >= 0)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
765 nothead[parent_1] = 1;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
766 if (parent_2 >= 0)
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
767 nothead[parent_2] = 1;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
768 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
769
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
770 for (i = 0; i < len; i++) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
771 PyObject *head;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
772
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
773 if (nothead[i])
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
774 continue;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
775 head = PyInt_FromLong(i);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
776 if (head == NULL || PyList_Append(heads, head) == -1) {
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
777 Py_XDECREF(head);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
778 goto bail;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
779 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
780 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
781
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
782 done:
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
783 self->headrevs = heads;
16786
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
784 free(nothead);
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
785 return list_copy(self->headrevs);
16786
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
786 bail:
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
787 Py_XDECREF(heads);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
788 free(nothead);
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
789 return NULL;
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
790 }
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
791
16618
6bae941b58ad parsers: change the type of nt_level
Bryan O'Sullivan <bryano@fb.com>
parents: 16617
diff changeset
792 static inline int nt_level(const char *node, Py_ssize_t level)
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
793 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
794 int v = node[level>>1];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
795 if (!(level & 1))
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
796 v >>= 4;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
797 return v & 0xf;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
798 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
799
16616
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
800 /*
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
801 * Return values:
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
802 *
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
803 * -4: match is ambiguous (multiple candidates)
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
804 * -2: not found
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
805 * rest: valid rev
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
806 */
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
807 static int nt_find(indexObject *self, const char *node, Py_ssize_t nodelen,
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
808 int hex)
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
809 {
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
810 int (*getnybble)(const char *, Py_ssize_t) = hex ? hexdigit : nt_level;
16641
e6dfbc5df76f parsers: use the correct maximum radix tree depth
Bryan O'Sullivan <bryano@fb.com>
parents: 16604
diff changeset
811 int level, maxlevel, off;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
812
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
813 if (nodelen == 20 && node[0] == '\0' && memcmp(node, nullid, 20) == 0)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
814 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
815
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
816 if (self->nt == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
817 return -2;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
818
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
819 if (hex)
16665
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
820 maxlevel = nodelen > 40 ? 40 : (int)nodelen;
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
821 else
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
822 maxlevel = nodelen > 20 ? 40 : ((int)nodelen * 2);
16641
e6dfbc5df76f parsers: use the correct maximum radix tree depth
Bryan O'Sullivan <bryano@fb.com>
parents: 16604
diff changeset
823
e6dfbc5df76f parsers: use the correct maximum radix tree depth
Bryan O'Sullivan <bryano@fb.com>
parents: 16604
diff changeset
824 for (level = off = 0; level < maxlevel; level++) {
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
825 int k = getnybble(node, level);
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
826 nodetree *n = &self->nt[off];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
827 int v = n->children[k];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
828
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
829 if (v < 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
830 const char *n;
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
831 Py_ssize_t i;
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
832
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
833 v = -v - 1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
834 n = index_node(self, v);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
835 if (n == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
836 return -2;
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
837 for (i = level; i < maxlevel; i++)
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
838 if (getnybble(node, i) != nt_level(n, i))
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
839 return -2;
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
840 return v;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
841 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
842 if (v == 0)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
843 return -2;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
844 off = v;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
845 }
16616
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
846 /* multiple matches against an ambiguous prefix */
8f79aabd96f6 parsers: allow nt_find to signal an ambiguous match
Bryan O'Sullivan <bryano@fb.com>
parents: 16615
diff changeset
847 return -4;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
848 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
849
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
850 static int nt_new(indexObject *self)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
851 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
852 if (self->ntlength == self->ntcapacity) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
853 self->ntcapacity *= 2;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
854 self->nt = realloc(self->nt,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
855 self->ntcapacity * sizeof(nodetree));
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
856 if (self->nt == NULL) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
857 PyErr_SetString(PyExc_MemoryError, "out of memory");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
858 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
859 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
860 memset(&self->nt[self->ntlength], 0,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
861 sizeof(nodetree) * (self->ntcapacity - self->ntlength));
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
862 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
863 return self->ntlength++;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
864 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
865
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
866 static int nt_insert(indexObject *self, const char *node, int rev)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
867 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
868 int level = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
869 int off = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
870
16641
e6dfbc5df76f parsers: use the correct maximum radix tree depth
Bryan O'Sullivan <bryano@fb.com>
parents: 16604
diff changeset
871 while (level < 40) {
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
872 int k = nt_level(node, level);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
873 nodetree *n;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
874 int v;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
875
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
876 n = &self->nt[off];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
877 v = n->children[k];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
878
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
879 if (v == 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
880 n->children[k] = -rev - 1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
881 return 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
882 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
883 if (v < 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
884 const char *oldnode = index_node(self, -v - 1);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
885 int noff;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
886
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
887 if (!oldnode || !memcmp(oldnode, node, 20)) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
888 n->children[k] = -rev - 1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
889 return 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
890 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
891 noff = nt_new(self);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
892 if (noff == -1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
893 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
894 /* self->nt may have been changed by realloc */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
895 self->nt[off].children[k] = noff;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
896 off = noff;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
897 n = &self->nt[off];
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
898 n->children[nt_level(oldnode, ++level)] = v;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
899 if (level > self->ntdepth)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
900 self->ntdepth = level;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
901 self->ntsplits += 1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
902 } else {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
903 level += 1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
904 off = v;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
905 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
906 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
907
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
908 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
909 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
910
16615
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
911 static int nt_init(indexObject *self)
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
912 {
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
913 if (self->nt == NULL) {
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
914 self->ntcapacity = self->raw_length < 4
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
915 ? 4 : self->raw_length / 2;
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
916 self->nt = calloc(self->ntcapacity, sizeof(nodetree));
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
917 if (self->nt == NULL) {
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
918 PyErr_NoMemory();
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
919 return -1;
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
920 }
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
921 self->ntlength = 1;
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
922 self->ntrev = (int)index_length(self) - 1;
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
923 self->ntlookups = 1;
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
924 self->ntmisses = 0;
16664
5bc6edf71b39 parsers: ensure that nullid is always present in the radix tree
Bryan O'Sullivan <bryano@fb.com>
parents: 16663
diff changeset
925 if (nt_insert(self, nullid, INT_MAX) == -1)
5bc6edf71b39 parsers: ensure that nullid is always present in the radix tree
Bryan O'Sullivan <bryano@fb.com>
parents: 16663
diff changeset
926 return -1;
16615
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
927 }
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
928 return 0;
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
929 }
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
930
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
931 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
932 * Return values:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
933 *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
934 * -3: error (exception set)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
935 * -2: not found (no exception set)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
936 * rest: valid rev
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
937 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
938 static int index_find_node(indexObject *self,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
939 const char *node, Py_ssize_t nodelen)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
940 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
941 int rev;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
942
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
943 self->ntlookups++;
16663
a955e05dd7a0 parsers: allow hex keys
Bryan O'Sullivan <bryano@fb.com>
parents: 16642
diff changeset
944 rev = nt_find(self, node, nodelen, 0);
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
945 if (rev >= -1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
946 return rev;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
947
16615
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
948 if (nt_init(self) == -1)
96fa9dd1db38 parsers: factor out radix tree initialization
Bryan O'Sullivan <bryano@fb.com>
parents: 16614
diff changeset
949 return -3;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
950
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
951 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
952 * For the first handful of lookups, we scan the entire index,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
953 * and cache only the matching nodes. This optimizes for cases
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
954 * like "hg tip", where only a few nodes are accessed.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
955 *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
956 * After that, we cache every node we visit, using a single
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
957 * scan amortized over multiple lookups. This gives the best
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
958 * bulk performance, e.g. for "hg log".
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
959 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
960 if (self->ntmisses++ < 4) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
961 for (rev = self->ntrev - 1; rev >= 0; rev--) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
962 const char *n = index_node(self, rev);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
963 if (n == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
964 return -2;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
965 if (memcmp(node, n, nodelen > 20 ? 20 : nodelen) == 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
966 if (nt_insert(self, n, rev) == -1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
967 return -3;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
968 break;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
969 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
970 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
971 } else {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
972 for (rev = self->ntrev - 1; rev >= 0; rev--) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
973 const char *n = index_node(self, rev);
16614
1d800eb9ba52 parsers: update ntrev when we stop scanning
Bryan O'Sullivan <bryano@fb.com>
parents: 16597
diff changeset
974 if (n == NULL) {
1d800eb9ba52 parsers: update ntrev when we stop scanning
Bryan O'Sullivan <bryano@fb.com>
parents: 16597
diff changeset
975 self->ntrev = rev + 1;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
976 return -2;
16614
1d800eb9ba52 parsers: update ntrev when we stop scanning
Bryan O'Sullivan <bryano@fb.com>
parents: 16597
diff changeset
977 }
1d800eb9ba52 parsers: update ntrev when we stop scanning
Bryan O'Sullivan <bryano@fb.com>
parents: 16597
diff changeset
978 if (nt_insert(self, n, rev) == -1) {
1d800eb9ba52 parsers: update ntrev when we stop scanning
Bryan O'Sullivan <bryano@fb.com>
parents: 16597
diff changeset
979 self->ntrev = rev + 1;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
980 return -3;
16614
1d800eb9ba52 parsers: update ntrev when we stop scanning
Bryan O'Sullivan <bryano@fb.com>
parents: 16597
diff changeset
981 }
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
982 if (memcmp(node, n, nodelen > 20 ? 20 : nodelen) == 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
983 break;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
984 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
985 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
986 self->ntrev = rev;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
987 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
988
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
989 if (rev >= 0)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
990 return rev;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
991 return -2;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
992 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
993
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
994 static PyObject *raise_revlog_error(void)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
995 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
996 static PyObject *errclass;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
997 PyObject *mod = NULL, *errobj;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
998
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
999 if (errclass == NULL) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1000 PyObject *dict;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1001
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1002 mod = PyImport_ImportModule("mercurial.error");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1003 if (mod == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1004 goto classfail;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1005
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1006 dict = PyModule_GetDict(mod);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1007 if (dict == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1008 goto classfail;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1009
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1010 errclass = PyDict_GetItemString(dict, "RevlogError");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1011 if (errclass == NULL) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1012 PyErr_SetString(PyExc_SystemError,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1013 "could not find RevlogError");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1014 goto classfail;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1015 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1016 Py_INCREF(errclass);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1017 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1018
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1019 errobj = PyObject_CallFunction(errclass, NULL);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1020 if (errobj == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1021 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1022 PyErr_SetObject(errclass, errobj);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1023 return errobj;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1024
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1025 classfail:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1026 Py_XDECREF(mod);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1027 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1028 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1029
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1030 static PyObject *index_getitem(indexObject *self, PyObject *value)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1031 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1032 char *node;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1033 Py_ssize_t nodelen;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1034 int rev;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1035
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1036 if (PyInt_Check(value))
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1037 return index_get(self, PyInt_AS_LONG(value));
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1038
16679
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1039 if (node_check(value, &node, &nodelen) == -1)
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1040 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1041 rev = index_find_node(self, node, nodelen);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1042 if (rev >= -1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1043 return PyInt_FromLong(rev);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1044 if (rev == -2)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1045 raise_revlog_error();
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1046 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1047 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1048
16665
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1049 static int nt_partialmatch(indexObject *self, const char *node,
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1050 Py_ssize_t nodelen)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1051 {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1052 int rev;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1053
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1054 if (nt_init(self) == -1)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1055 return -3;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1056
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1057 if (self->ntrev > 0) {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1058 /* ensure that the radix tree is fully populated */
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1059 for (rev = self->ntrev - 1; rev >= 0; rev--) {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1060 const char *n = index_node(self, rev);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1061 if (n == NULL)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1062 return -2;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1063 if (nt_insert(self, n, rev) == -1)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1064 return -3;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1065 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1066 self->ntrev = rev;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1067 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1068
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1069 return nt_find(self, node, nodelen, 1);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1070 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1071
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1072 static PyObject *index_partialmatch(indexObject *self, PyObject *args)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1073 {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1074 const char *fullnode;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1075 int nodelen;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1076 char *node;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1077 int rev, i;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1078
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1079 if (!PyArg_ParseTuple(args, "s#", &node, &nodelen))
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1080 return NULL;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1081
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1082 if (nodelen < 4) {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1083 PyErr_SetString(PyExc_ValueError, "key too short");
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1084 return NULL;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1085 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1086
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1087 if (nodelen > 40)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1088 nodelen = 40;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1089
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1090 for (i = 0; i < nodelen; i++)
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1091 hexdigit(node, i);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1092 if (PyErr_Occurred()) {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1093 /* input contains non-hex characters */
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1094 PyErr_Clear();
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1095 Py_RETURN_NONE;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1096 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1097
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1098 rev = nt_partialmatch(self, node, nodelen);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1099
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1100 switch (rev) {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1101 case -4:
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1102 raise_revlog_error();
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1103 case -3:
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1104 return NULL;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1105 case -2:
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1106 Py_RETURN_NONE;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1107 case -1:
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1108 return PyString_FromStringAndSize(nullid, 20);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1109 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1110
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1111 fullnode = index_node(self, rev);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1112 if (fullnode == NULL) {
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1113 PyErr_Format(PyExc_IndexError,
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1114 "could not access rev %d", rev);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1115 return NULL;
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1116 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1117 return PyString_FromStringAndSize(fullnode, 20);
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1118 }
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1119
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1120 static PyObject *index_m_get(indexObject *self, PyObject *args)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1121 {
16679
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1122 Py_ssize_t nodelen;
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1123 PyObject *val;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1124 char *node;
16679
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1125 int rev;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1126
16679
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1127 if (!PyArg_ParseTuple(args, "O", &val))
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1128 return NULL;
16679
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1129 if (node_check(val, &node, &nodelen) == -1)
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1130 return NULL;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1131 rev = index_find_node(self, node, nodelen);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1132 if (rev == -3)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1133 return NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1134 if (rev == -2)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1135 Py_RETURN_NONE;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1136 return PyInt_FromLong(rev);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1137 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1138
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1139 static int index_contains(indexObject *self, PyObject *value)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1140 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1141 char *node;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1142 Py_ssize_t nodelen;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1143
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1144 if (PyInt_Check(value)) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1145 long rev = PyInt_AS_LONG(value);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1146 return rev >= -1 && rev < index_length(self);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1147 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1148
16679
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1149 if (node_check(value, &node, &nodelen) == -1)
2950d186a927 parsers: strictly check for 20-byte hashes where they're required
Bryan O'Sullivan <bryano@fb.com>
parents: 16641
diff changeset
1150 return -1;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1151
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1152 switch (index_find_node(self, node, nodelen)) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1153 case -3:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1154 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1155 case -2:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1156 return 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1157 default:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1158 return 1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1159 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1160 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1161
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1162 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1163 * Invalidate any trie entries introduced by added revs.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1164 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1165 static void nt_invalidate_added(indexObject *self, Py_ssize_t start)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1166 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1167 Py_ssize_t i, len = PyList_GET_SIZE(self->added);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1168
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1169 for (i = start; i < len; i++) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1170 PyObject *tuple = PyList_GET_ITEM(self->added, i);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1171 PyObject *node = PyTuple_GET_ITEM(tuple, 7);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1172
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1173 nt_insert(self, PyString_AS_STRING(node), -1);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1174 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1175
16732
277e2acb7e5c parsers: use Py_CLEAR where appropriate
Bryan O'Sullivan <bryano@fb.com>
parents: 16699
diff changeset
1176 if (start == 0)
277e2acb7e5c parsers: use Py_CLEAR where appropriate
Bryan O'Sullivan <bryano@fb.com>
parents: 16699
diff changeset
1177 Py_CLEAR(self->added);
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1178 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1179
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1180 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1181 * Delete a numeric range of revs, which must be at the end of the
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1182 * range, but exclude the sentinel nullid entry.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1183 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1184 static int index_slice_del(indexObject *self, PyObject *item)
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1185 {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1186 Py_ssize_t start, stop, step, slicelength;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1187 Py_ssize_t length = index_length(self);
16784
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1188 int ret = 0;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1189
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1190 if (PySlice_GetIndicesEx((PySliceObject*)item, length,
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1191 &start, &stop, &step, &slicelength) < 0)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1192 return -1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1193
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1194 if (slicelength <= 0)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1195 return 0;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1196
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1197 if ((step < 0 && start < stop) || (step > 0 && start > stop))
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1198 stop = start;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1199
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1200 if (step < 0) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1201 stop = start + 1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1202 start = stop + step*(slicelength - 1) - 1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1203 step = -step;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1204 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1205
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1206 if (step != 1) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1207 PyErr_SetString(PyExc_ValueError,
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1208 "revlog index delete requires step size of 1");
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1209 return -1;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1210 }
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1211
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1212 if (stop != length - 1) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1213 PyErr_SetString(PyExc_IndexError,
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1214 "revlog index deletion indices are invalid");
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1215 return -1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1216 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1217
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1218 if (start < self->length - 1) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1219 if (self->nt) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1220 Py_ssize_t i;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1221
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1222 for (i = start + 1; i < self->length - 1; i++) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1223 const char *node = index_node(self, i);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1224
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1225 if (node)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1226 nt_insert(self, node, -1);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1227 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1228 if (self->added)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1229 nt_invalidate_added(self, 0);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1230 if (self->ntrev > start)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1231 self->ntrev = (int)start;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1232 }
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1233 self->length = start + 1;
16784
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1234 if (start < self->raw_length)
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1235 self->raw_length = start;
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1236 goto done;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1237 }
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1238
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1239 if (self->nt) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1240 nt_invalidate_added(self, start - self->length + 1);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1241 if (self->ntrev > start)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1242 self->ntrev = (int)start;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1243 }
16784
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1244 if (self->added)
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1245 ret = PyList_SetSlice(self->added, start - self->length + 1,
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1246 PyList_GET_SIZE(self->added), NULL);
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1247 done:
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
1248 Py_CLEAR(self->headrevs);
16784
12a852c7c763 parsers: reduce raw_length when truncating
Bryan O'Sullivan <bryano@fb.com>
parents: 16732
diff changeset
1249 return ret;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1250 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1251
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1252 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1253 * Supported ops:
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1254 *
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1255 * slice deletion
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1256 * string assignment (extend node->rev mapping)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1257 * string deletion (shrink node->rev mapping)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1258 */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1259 static int index_assign_subscript(indexObject *self, PyObject *item,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1260 PyObject *value)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1261 {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1262 char *node;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1263 Py_ssize_t nodelen;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1264 long rev;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1265
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1266 if (PySlice_Check(item) && value == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1267 return index_slice_del(self, item);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1268
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1269 if (node_check(item, &node, &nodelen) == -1)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1270 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1271
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1272 if (value == NULL)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1273 return self->nt ? nt_insert(self, node, -1) : 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1274 rev = PyInt_AsLong(value);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1275 if (rev > INT_MAX || rev < 0) {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1276 if (!PyErr_Occurred())
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1277 PyErr_SetString(PyExc_ValueError, "rev out of range");
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1278 return -1;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1279 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1280 return nt_insert(self, node, (int)rev);
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1281 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1282
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1283 /*
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1284 * Find all RevlogNG entries in an index that has inline data. Update
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1285 * the optional "offsets" table with those entries.
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1286 */
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1287 static long inline_scan(indexObject *self, const char **offsets)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1288 {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1289 const char *data = PyString_AS_STRING(self->data);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1290 const char *end = data + PyString_GET_SIZE(self->data);
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1291 long incr = v1_hdrsize;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1292 Py_ssize_t len = 0;
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
1293
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1294 while (data + v1_hdrsize <= end) {
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1295 uint32_t comp_len;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1296 const char *old_data;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1297 /* 3rd element of header is length of compressed inline data */
16437
d126a0d16856 util.h: replace ntohl/htonl with get/putbe32
Matt Mackall <mpm@selenic.com>
parents: 16414
diff changeset
1298 comp_len = getbe32(data + 8);
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1299 incr = v1_hdrsize + comp_len;
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1300 if (incr < v1_hdrsize)
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1301 break;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1302 if (offsets)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1303 offsets[len] = data;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1304 len++;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1305 old_data = data;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1306 data += incr;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1307 if (data <= old_data)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1308 break;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1309 }
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
1310
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1311 if (data != end && data + v1_hdrsize != end) {
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1312 if (!PyErr_Occurred())
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1313 PyErr_SetString(PyExc_ValueError, "corrupt index file");
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1314 return -1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1315 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1316
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1317 return len;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1318 }
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
1319
16572
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1320 static int index_init(indexObject *self, PyObject *args)
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1321 {
16572
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1322 PyObject *data_obj, *inlined_obj;
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1323 Py_ssize_t size;
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1324
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1325 if (!PyArg_ParseTuple(args, "OO", &data_obj, &inlined_obj))
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1326 return -1;
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1327 if (!PyString_Check(data_obj)) {
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1328 PyErr_SetString(PyExc_TypeError, "data is not a string");
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1329 return -1;
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1330 }
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1331 size = PyString_GET_SIZE(data_obj);
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1332
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1333 self->inlined = inlined_obj && PyObject_IsTrue(inlined_obj);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1334 self->data = data_obj;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1335 self->cache = NULL;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1336
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1337 self->added = NULL;
16787
bda96ce993f9 parsers: cache the result of index_headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16786
diff changeset
1338 self->headrevs = NULL;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1339 self->offsets = NULL;
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1340 self->nt = NULL;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1341 self->ntlength = self->ntcapacity = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1342 self->ntdepth = self->ntsplits = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1343 self->ntlookups = self->ntmisses = 0;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1344 self->ntrev = -1;
16597
b767382a8675 parsers: fix refcount bug on corrupt index
Matt Mackall <mpm@selenic.com>
parents: 16572
diff changeset
1345 Py_INCREF(self->data);
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1346
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1347 if (self->inlined) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1348 long len = inline_scan(self, NULL);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1349 if (len == -1)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1350 goto bail;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1351 self->raw_length = len;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1352 self->length = len + 1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1353 } else {
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1354 if (size % v1_hdrsize) {
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1355 PyErr_SetString(PyExc_ValueError, "corrupt index file");
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1356 goto bail;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1357 }
16863
bbedef66c6f3 parsers: replace magic number 64 with symbolic constant
Bryan O'Sullivan <bryano@fb.com>
parents: 16787
diff changeset
1358 self->raw_length = size / v1_hdrsize;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1359 self->length = self->raw_length + 1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1360 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1361
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1362 return 0;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1363 bail:
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1364 return -1;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1365 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1366
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1367 static PyObject *index_nodemap(indexObject *self)
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1368 {
16572
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1369 Py_INCREF(self);
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1370 return (PyObject *)self;
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1371 }
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1372
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1373 static void index_dealloc(indexObject *self)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1374 {
16370
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
1375 _index_clearcaches(self);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1376 Py_DECREF(self->data);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1377 Py_XDECREF(self->added);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1378 PyObject_Del(self);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1379 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1380
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1381 static PySequenceMethods index_sequence_methods = {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1382 (lenfunc)index_length, /* sq_length */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1383 0, /* sq_concat */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1384 0, /* sq_repeat */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1385 (ssizeargfunc)index_get, /* sq_item */
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1386 0, /* sq_slice */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1387 0, /* sq_ass_item */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1388 0, /* sq_ass_slice */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1389 (objobjproc)index_contains, /* sq_contains */
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1390 };
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1391
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1392 static PyMappingMethods index_mapping_methods = {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1393 (lenfunc)index_length, /* mp_length */
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1394 (binaryfunc)index_getitem, /* mp_subscript */
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1395 (objobjargproc)index_assign_subscript, /* mp_ass_subscript */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1396 };
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1397
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1398 static PyMethodDef index_methods[] = {
16370
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
1399 {"clearcaches", (PyCFunction)index_clearcaches, METH_NOARGS,
28bb4daf070c parsers: fix a memleak, and add a clearcaches method to the index
Bryan O'Sullivan <bryano@fb.com>
parents: 16363
diff changeset
1400 "clear the index caches"},
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1401 {"get", (PyCFunction)index_m_get, METH_VARARGS,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1402 "get an index entry"},
16786
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
1403 {"headrevs", (PyCFunction)index_headrevs, METH_NOARGS,
2631cd5dd244 revlog: switch to a C version of headrevs
Bryan O'Sullivan <bryano@fb.com>
parents: 16784
diff changeset
1404 "get head revisions"},
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1405 {"insert", (PyCFunction)index_insert, METH_VARARGS,
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1406 "insert an index entry"},
16665
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1407 {"partialmatch", (PyCFunction)index_partialmatch, METH_VARARGS,
e410be860393 revlog: speed up prefix matching against nodes
Bryan O'Sullivan <bryano@fb.com>
parents: 16664
diff changeset
1408 "match a potentially ambiguous node ID"},
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1409 {"stats", (PyCFunction)index_stats, METH_NOARGS,
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1410 "stats for the index"},
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1411 {NULL} /* Sentinel */
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1412 };
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1413
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1414 static PyGetSetDef index_getset[] = {
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1415 {"nodemap", (getter)index_nodemap, NULL, "nodemap", NULL},
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1416 {NULL} /* Sentinel */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1417 };
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1418
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1419 static PyTypeObject indexType = {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1420 PyObject_HEAD_INIT(NULL)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1421 0, /* ob_size */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1422 "parsers.index", /* tp_name */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1423 sizeof(indexObject), /* tp_basicsize */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1424 0, /* tp_itemsize */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1425 (destructor)index_dealloc, /* tp_dealloc */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1426 0, /* tp_print */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1427 0, /* tp_getattr */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1428 0, /* tp_setattr */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1429 0, /* tp_compare */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1430 0, /* tp_repr */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1431 0, /* tp_as_number */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1432 &index_sequence_methods, /* tp_as_sequence */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1433 &index_mapping_methods, /* tp_as_mapping */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1434 0, /* tp_hash */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1435 0, /* tp_call */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1436 0, /* tp_str */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1437 0, /* tp_getattro */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1438 0, /* tp_setattro */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1439 0, /* tp_as_buffer */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1440 Py_TPFLAGS_DEFAULT, /* tp_flags */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1441 "revlog index", /* tp_doc */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1442 0, /* tp_traverse */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1443 0, /* tp_clear */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1444 0, /* tp_richcompare */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1445 0, /* tp_weaklistoffset */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1446 0, /* tp_iter */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1447 0, /* tp_iternext */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1448 index_methods, /* tp_methods */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1449 0, /* tp_members */
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1450 index_getset, /* tp_getset */
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1451 0, /* tp_base */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1452 0, /* tp_dict */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1453 0, /* tp_descr_get */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1454 0, /* tp_descr_set */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1455 0, /* tp_dictoffset */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1456 (initproc)index_init, /* tp_init */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1457 0, /* tp_alloc */
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1458 };
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1459
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1460 /*
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1461 * returns a tuple of the form (index, index, cache) with elements as
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1462 * follows:
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1463 *
16414
e8d37b78acfb parsers: use base-16 trie for faster node->rev mapping
Bryan O'Sullivan <bryano@fb.com>
parents: 16393
diff changeset
1464 * index: an index object that lazily parses RevlogNG records
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1465 * cache: if data is inlined, a tuple (index_file_content, 0), else None
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1466 *
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1467 * added complications are for backwards compatibility
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1468 */
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
1469 static PyObject *parse_index2(PyObject *self, PyObject *args)
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1470 {
16572
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1471 PyObject *tuple = NULL, *cache = NULL;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1472 indexObject *idx;
16572
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1473 int ret;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1474
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1475 idx = PyObject_New(indexObject, &indexType);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1476 if (idx == NULL)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1477 goto bail;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1478
16572
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1479 ret = index_init(idx, args);
8d44b5a2974f parsers: fix refcount leak, simplify init of index (issue3417)
Bryan O'Sullivan <bryano@fb.com>
parents: 16437
diff changeset
1480 if (ret == -1)
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1481 goto bail;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1482
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1483 if (idx->inlined) {
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1484 cache = Py_BuildValue("iO", 0, idx->data);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1485 if (cache == NULL)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1486 goto bail;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1487 } else {
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1488 cache = Py_None;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1489 Py_INCREF(cache);
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1490 }
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1491
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1492 tuple = Py_BuildValue("NN", idx, cache);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1493 if (!tuple)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1494 goto bail;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1495 return tuple;
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1496
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1497 bail:
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1498 Py_XDECREF(idx);
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1499 Py_XDECREF(cache);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1500 Py_XDECREF(tuple);
7108
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1501 return NULL;
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1502 }
1ca878d7b849 C implementation of revlog index parsing
Bernhard Leiner <bleiner@gmail.com>
parents: 7093
diff changeset
1503
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1504 static char parsers_doc[] = "Efficient content parsing.";
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1505
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1506 static PyMethodDef methods[] = {
16955
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
1507 {"pack_dirstate", pack_dirstate, METH_VARARGS, "pack a dirstate\n"},
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1508 {"parse_manifest", parse_manifest, METH_VARARGS, "parse a manifest\n"},
7093
16bafcebd3d1 dirstate: C parsing extension
Matt Mackall <mpm@selenic.com>
parents: 7092
diff changeset
1509 {"parse_dirstate", parse_dirstate, METH_VARARGS, "parse a dirstate\n"},
13254
5ef5eb1f3515 revlog: only build the nodemap on demand
Matt Mackall <mpm@selenic.com>
parents: 11361
diff changeset
1510 {"parse_index2", parse_index2, METH_VARARGS, "parse a revlog index\n"},
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1511 {NULL, NULL}
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1512 };
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1513
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1514 static void module_init(PyObject *mod)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1515 {
16604
48e42f984074 parsers: statically initializing tp_new to PyType_GenericNew is not portable
Adrian Buehlmann <adrian@cadifra.com>
parents: 16597
diff changeset
1516 indexType.tp_new = PyType_GenericNew;
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1517 if (PyType_Ready(&indexType) < 0)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1518 return;
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1519 Py_INCREF(&indexType);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1520
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1521 PyModule_AddObject(mod, "index", (PyObject *)&indexType);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1522
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1523 nullentry = Py_BuildValue("iiiiiiis#", 0, 0, 0,
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1524 -1, -1, -1, -1, nullid, 20);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1525 if (nullentry)
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1526 PyObject_GC_UnTrack(nullentry);
16955
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
1527
92e1c64ba0d4 parsers: add a C function to pack the dirstate
Bryan O'Sullivan <bryano@fb.com>
parents: 16863
diff changeset
1528 dirstate_unset = Py_BuildValue("ciii", 'n', 0, -1, -1);
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1529 }
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1530
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1531 #ifdef IS_PY3K
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1532 static struct PyModuleDef parsers_module = {
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1533 PyModuleDef_HEAD_INIT,
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1534 "parsers",
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1535 parsers_doc,
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1536 -1,
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1537 methods
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1538 };
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1539
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1540 PyMODINIT_FUNC PyInit_parsers(void)
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1541 {
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1542 PyObject *mod = PyModule_Create(&parsers_module);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1543 module_init(mod);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1544 return mod;
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1545 }
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1546 #else
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1547 PyMODINIT_FUNC initparsers(void)
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1548 {
16363
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1549 PyObject *mod = Py_InitModule3("parsers", methods, parsers_doc);
2cdd7e63211b parsers: incrementally parse the revlog index in C
Bryan O'Sullivan <bryano@fb.com>
parents: 15033
diff changeset
1550 module_init(mod);
6389
0231f763ebc8 manifest: improve parsing performance by 8x via a new C extension
Bryan O'Sullivan <bos@serpentine.com>
parents:
diff changeset
1551 }
11361
3de3d670d2b6 parsers.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents: 10449
diff changeset
1552 #endif