Mercurial > public > mercurial-scm > hg
annotate mercurial/base85.c @ 16837:1b9d54c00d50
base85: use Py_ssize_t for string lengths
author | Adrian Buehlmann <adrian@cadifra.com> |
---|---|
date | Thu, 24 May 2012 01:30:12 +0200 |
parents | a8065323c003 |
children | 19a915d43a68 |
rev | line source |
---|---|
3283 | 1 /* |
2 base85 codec | |
3 | |
4 Copyright 2006 Brendan Cully <brendan@kublai.com> | |
5 | |
6 This software may be used and distributed according to the terms of | |
7 the GNU General Public License, incorporated herein by reference. | |
8 | |
9 Largely based on git's implementation | |
10 */ | |
11 | |
16837
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
12 #define PY_SSIZE_T_CLEAN |
3283 | 13 #include <Python.h> |
14 | |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
15 #include "util.h" |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
16 |
3283 | 17 static const char b85chars[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ" |
18 "abcdefghijklmnopqrstuvwxyz!#$%&()*+-;<=>?@^_`{|}~"; | |
19 static char b85dec[256]; | |
20 | |
21 static void | |
22 b85prep(void) | |
23 { | |
24 int i; | |
25 | |
26 memset(b85dec, 0, sizeof(b85dec)); | |
27 for (i = 0; i < sizeof(b85chars); i++) | |
28 b85dec[(int)(b85chars[i])] = i + 1; | |
29 } | |
30 | |
31 static PyObject * | |
32 b85encode(PyObject *self, PyObject *args) | |
33 { | |
34 const unsigned char *text; | |
35 PyObject *out; | |
36 char *dst; | |
16837
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
37 Py_ssize_t len, olen, i; |
3283 | 38 unsigned int acc, val, ch; |
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
39 int pad = 0; |
3283 | 40 |
3369
4bad632913d8
python2.5 PyArg_ParseTuple fix
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents:
3332
diff
changeset
|
41 if (!PyArg_ParseTuple(args, "s#|i", &text, &len, &pad)) |
3283 | 42 return NULL; |
43 | |
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
44 if (pad) |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
45 olen = ((len + 3) / 4 * 5) - 3; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
46 else { |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
47 olen = len % 4; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
48 if (olen) |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
49 olen++; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
50 olen += len / 4 * 5; |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
51 } |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
52 if (!(out = PyBytes_FromStringAndSize(NULL, olen + 3))) |
3283 | 53 return NULL; |
54 | |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
55 dst = PyBytes_AsString(out); |
3283 | 56 |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
57 while (len) { |
3283 | 58 acc = 0; |
59 for (i = 24; i >= 0; i -= 8) { | |
60 ch = *text++; | |
61 acc |= ch << i; | |
62 if (--len == 0) | |
63 break; | |
64 } | |
65 for (i = 4; i >= 0; i--) { | |
66 val = acc % 85; | |
67 acc /= 85; | |
68 dst[i] = b85chars[val]; | |
69 } | |
70 dst += 5; | |
71 } | |
72 | |
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
73 if (!pad) |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
74 _PyBytes_Resize(&out, olen); |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
75 |
3283 | 76 return out; |
77 } | |
78 | |
79 static PyObject * | |
80 b85decode(PyObject *self, PyObject *args) | |
81 { | |
82 PyObject *out; | |
83 const char *text; | |
84 char *dst; | |
16837
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
85 Py_ssize_t len, i, j, olen, cap; |
1b9d54c00d50
base85: use Py_ssize_t for string lengths
Adrian Buehlmann <adrian@cadifra.com>
parents:
16522
diff
changeset
|
86 int c; |
3283 | 87 unsigned int acc; |
88 | |
3369
4bad632913d8
python2.5 PyArg_ParseTuple fix
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents:
3332
diff
changeset
|
89 if (!PyArg_ParseTuple(args, "s#", &text, &len)) |
3283 | 90 return NULL; |
91 | |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
92 olen = len / 5 * 4; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
93 i = len % 5; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
94 if (i) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
95 olen += i - 1; |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
96 if (!(out = PyBytes_FromStringAndSize(NULL, olen))) |
3283 | 97 return NULL; |
98 | |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
99 dst = PyBytes_AsString(out); |
3283 | 100 |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
101 i = 0; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
102 while (i < len) |
3283 | 103 { |
104 acc = 0; | |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
105 cap = len - i - 1; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
106 if (cap > 4) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
107 cap = 4; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
108 for (j = 0; j < cap; i++, j++) |
3283 | 109 { |
110 c = b85dec[(int)*text++] - 1; | |
111 if (c < 0) | |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
112 return PyErr_Format( |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
113 PyExc_ValueError, |
16522
a8065323c003
patch: display a nice error for invalid base85 data
Patrick Mezard <patrick@mezard.eu>
parents:
13302
diff
changeset
|
114 "bad base85 character at position %d", i); |
3283 | 115 acc = acc * 85 + c; |
116 } | |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
117 if (i++ < len) |
3283 | 118 { |
119 c = b85dec[(int)*text++] - 1; | |
120 if (c < 0) | |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
121 return PyErr_Format( |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
122 PyExc_ValueError, |
16522
a8065323c003
patch: display a nice error for invalid base85 data
Patrick Mezard <patrick@mezard.eu>
parents:
13302
diff
changeset
|
123 "bad base85 character at position %d", i); |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
124 /* overflow detection: 0xffffffff == "|NsC0", |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
125 * "|NsC" == 0x03030303 */ |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
126 if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c) |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
127 return PyErr_Format( |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7190
diff
changeset
|
128 PyExc_ValueError, |
16522
a8065323c003
patch: display a nice error for invalid base85 data
Patrick Mezard <patrick@mezard.eu>
parents:
13302
diff
changeset
|
129 "bad base85 sequence at position %d", i); |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
130 acc += c; |
3283 | 131 } |
132 | |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
133 cap = olen < 4 ? olen : 4; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
134 olen -= cap; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
135 for (j = 0; j < 4 - cap; j++) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
136 acc *= 85; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
137 if (cap && cap < 4) |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
138 acc += 0xffffff >> (cap - 1) * 8; |
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
139 for (j = 0; j < cap; j++) |
3283 | 140 { |
141 acc = (acc << 8) | (acc >> 24); | |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
142 *dst++ = acc; |
3283 | 143 } |
144 } | |
145 | |
146 return out; | |
147 } | |
148 | |
149 static char base85_doc[] = "Base85 Data Encoding"; | |
150 | |
151 static PyMethodDef methods[] = { | |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
152 {"b85encode", b85encode, METH_VARARGS, |
7190
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
153 "Encode text in base85.\n\n" |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
154 "If the second parameter is true, pad the result to a multiple of " |
aecea6934fdd
Some additional space/tab cleanups
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3369
diff
changeset
|
155 "five characters.\n"}, |
3288
e93c926e069e
Handle odd-sized base85 input and output
Brendan Cully <brendan@kublai.com>
parents:
3283
diff
changeset
|
156 {"b85decode", b85decode, METH_VARARGS, "Decode base85 text.\n"}, |
3283 | 157 {NULL, NULL} |
158 }; | |
159 | |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
160 #ifdef IS_PY3K |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
161 static struct PyModuleDef base85_module = { |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
162 PyModuleDef_HEAD_INIT, |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
163 "base85", |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
164 base85_doc, |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
165 -1, |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
166 methods |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
167 }; |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
168 |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
169 PyMODINIT_FUNC PyInit_base85(void) |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
170 { |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
171 b85prep(); |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
172 |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
173 return PyModule_Create(&base85_module); |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
174 } |
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
175 #else |
3283 | 176 PyMODINIT_FUNC initbase85(void) |
177 { | |
178 Py_InitModule3("base85", methods, base85_doc); | |
179 | |
180 b85prep(); | |
181 } | |
11362
f42ef9493fa9
base85.c: Added support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
182 #endif |