mercurial/cext/bdiff.c
author Gregory Szorc <gregory.szorc@gmail.com>
Mon, 24 Sep 2018 14:31:31 -0700
changeset 39878 3e896b51aa5d
parent 39421 ad76032d27da
child 41336 763b45bc4483
permissions -rw-r--r--
storageutil: move metadata parsing and packing from revlog (API) Parsing and writing of revision text metadata is likely identical across storage backends. Let's move the code out of revlog so we don't need to import the revlog module in order to use it. Differential Revision: https://phab.mercurial-scm.org/D4754
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     1
/*
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     2
 bdiff.c - efficient binary diff extension for Mercurial
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     3
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     4
 Copyright 2005, 2006 Matt Mackall <mpm@selenic.com>
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     5
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     6
 This software may be used and distributed according to the terms of
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     7
 the GNU General Public License, incorporated herein by reference.
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     8
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
     9
 Based roughly on Python difflib
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    10
*/
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    11
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    12
#define PY_SSIZE_T_CLEAN
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    13
#include <Python.h>
34438
b90e8da190da cext: reorder #include
Gregory Szorc <gregory.szorc@gmail.com>
parents: 32369
diff changeset
    14
#include <limits.h>
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    15
#include <stdlib.h>
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    16
#include <string.h>
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    17
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    18
#include "bdiff.h"
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    19
#include "bitmanipulation.h"
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
    20
#include "thirdparty/xdiff/xdiff.h"
30170
15635d8b17e0 bdiff: include util.h
Gregory Szorc <gregory.szorc@gmail.com>
parents: 29541
diff changeset
    21
#include "util.h"
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    22
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    23
static PyObject *blocks(PyObject *self, PyObject *args)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    24
{
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    25
	PyObject *sa, *sb, *rl = NULL, *m;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    26
	struct bdiff_line *a, *b;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    27
	struct bdiff_hunk l, *h;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    28
	int an, bn, count, pos = 0;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    29
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    30
	l.next = NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    31
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    32
	if (!PyArg_ParseTuple(args, "SS:bdiff", &sa, &sb))
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    33
		return NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    34
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    35
	an = bdiff_splitlines(PyBytes_AsString(sa), PyBytes_Size(sa), &a);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    36
	bn = bdiff_splitlines(PyBytes_AsString(sb), PyBytes_Size(sb), &b);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    37
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    38
	if (!a || !b)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    39
		goto nomem;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    40
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    41
	count = bdiff_diff(a, an, b, bn, &l);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    42
	if (count < 0)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    43
		goto nomem;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    44
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    45
	rl = PyList_New(count);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    46
	if (!rl)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    47
		goto nomem;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    48
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    49
	for (h = l.next; h; h = h->next) {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    50
		m = Py_BuildValue("iiii", h->a1, h->a2, h->b1, h->b2);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    51
		PyList_SetItem(rl, pos, m);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    52
		pos++;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    53
	}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    54
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    55
nomem:
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    56
	free(a);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    57
	free(b);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    58
	bdiff_freehunks(l.next);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    59
	return rl ? rl : PyErr_NoMemory();
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    60
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    61
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    62
static PyObject *bdiff(PyObject *self, PyObject *args)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    63
{
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    64
	Py_buffer ba, bb;
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    65
	char *rb, *ia, *ib;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    66
	PyObject *result = NULL;
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
    67
	struct bdiff_line *al = NULL, *bl = NULL;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    68
	struct bdiff_hunk l, *h;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    69
	int an, bn, count;
30561
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
    70
	Py_ssize_t len = 0, la, lb, li = 0, lcommon = 0, lmax;
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
    71
	PyThreadState *_save = NULL;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    72
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    73
	l.next = NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    74
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    75
	if (!PyArg_ParseTuple(args, PY23("s*s*:bdiff", "y*y*:bdiff"), &ba, &bb))
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    76
		return NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    77
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    78
	if (!PyBuffer_IsContiguous(&ba, 'C') || ba.ndim > 1) {
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    79
		PyErr_SetString(PyExc_ValueError, "bdiff input not contiguous");
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    80
		goto cleanup;
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    81
	}
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    82
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    83
	if (!PyBuffer_IsContiguous(&bb, 'C') || bb.ndim > 1) {
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    84
		PyErr_SetString(PyExc_ValueError, "bdiff input not contiguous");
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    85
		goto cleanup;
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    86
	}
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    87
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    88
	la = ba.len;
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    89
	lb = bb.len;
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    90
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    91
	if (la > UINT_MAX || lb > UINT_MAX) {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    92
		PyErr_SetString(PyExc_ValueError, "bdiff inputs too large");
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    93
		goto cleanup;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    94
	}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    95
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
    96
	_save = PyEval_SaveThread();
30561
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
    97
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
    98
	lmax = la > lb ? lb : la;
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
    99
	for (ia = ba.buf, ib = bb.buf; li < lmax && *ia == *ib;
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
   100
	     ++li, ++ia, ++ib) {
30561
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
   101
		if (*ia == '\n')
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
   102
			lcommon = li + 1;
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
   103
	}
30561
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
   104
	/* we can almost add: if (li == lmax) lcommon = li; */
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
   105
36681
340e4b711df7 bdiff: avoid pointer arithmetic on void*
Matt Harbison <matt_harbison@yahoo.com>
parents: 36675
diff changeset
   106
	an = bdiff_splitlines((char *)ba.buf + lcommon, la - lcommon, &al);
340e4b711df7 bdiff: avoid pointer arithmetic on void*
Matt Harbison <matt_harbison@yahoo.com>
parents: 36675
diff changeset
   107
	bn = bdiff_splitlines((char *)bb.buf + lcommon, lb - lcommon, &bl);
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   108
	if (!al || !bl) {
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   109
		PyErr_NoMemory();
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   110
		goto cleanup;
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   111
	}
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   112
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   113
	count = bdiff_diff(al, an, bl, bn, &l);
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   114
	if (count < 0) {
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   115
		PyErr_NoMemory();
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   116
		goto cleanup;
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   117
	}
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   118
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   119
	/* calculate length of output */
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   120
	la = lb = 0;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   121
	for (h = l.next; h; h = h->next) {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   122
		if (h->a1 != la || h->b1 != lb)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   123
			len += 12 + bl[h->b1].l - bl[lb].l;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   124
		la = h->a2;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   125
		lb = h->b2;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   126
	}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   127
	PyEval_RestoreThread(_save);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   128
	_save = NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   129
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   130
	result = PyBytes_FromStringAndSize(NULL, len);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   131
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   132
	if (!result)
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   133
		goto cleanup;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   134
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   135
	/* build binary patch */
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   136
	rb = PyBytes_AsString(result);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   137
	la = lb = 0;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   138
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   139
	for (h = l.next; h; h = h->next) {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   140
		if (h->a1 != la || h->b1 != lb) {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   141
			len = bl[h->b1].l - bl[lb].l;
30561
7c0c722d568d bdiff: early pruning of common prefix before doing expensive computations
Mads Kiilerich <madski@unity3d.com>
parents: 30170
diff changeset
   142
			putbe32((uint32_t)(al[la].l + lcommon - al->l), rb);
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   143
			putbe32((uint32_t)(al[h->a1].l + lcommon - al->l),
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   144
			        rb + 4);
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   145
			putbe32((uint32_t)len, rb + 8);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   146
			memcpy(rb + 12, bl[lb].l, len);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   147
			rb += 12 + len;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   148
		}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   149
		la = h->a2;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   150
		lb = h->b2;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   151
	}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   152
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   153
cleanup:
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   154
	if (_save)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   155
		PyEval_RestoreThread(_save);
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
   156
	PyBuffer_Release(&ba);
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
   157
	PyBuffer_Release(&bb);
38301
d9e87566f879 cext: stop worrying and love the free(NULL)
Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
parents: 37980
diff changeset
   158
	free(al);
d9e87566f879 cext: stop worrying and love the free(NULL)
Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
parents: 37980
diff changeset
   159
	free(bl);
38309
93b812d5b818 bdiff: one more safe call of bdiff_freehunks(NULL)
Yuya Nishihara <yuya@tcha.org>
parents: 38301
diff changeset
   160
	bdiff_freehunks(l.next);
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   161
	return result;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   162
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   163
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   164
/*
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   165
 * If allws != 0, remove all whitespace (' ', \t and \r). Otherwise,
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   166
 * reduce whitespace sequences to a single space and trim remaining whitespace
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   167
 * from end of lines.
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   168
 */
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   169
static PyObject *fixws(PyObject *self, PyObject *args)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   170
{
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   171
	PyObject *s, *result = NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   172
	char allws, c;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   173
	const char *r;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   174
	Py_ssize_t i, rlen, wlen = 0;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   175
	char *w;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   176
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   177
	if (!PyArg_ParseTuple(args, "Sb:fixws", &s, &allws))
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   178
		return NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   179
	r = PyBytes_AsString(s);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   180
	rlen = PyBytes_Size(s);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   181
31467
08ecec297521 bdiff: use Python memory allocator in fixws
Gregory Szorc <gregory.szorc@gmail.com>
parents: 30561
diff changeset
   182
	w = (char *)PyMem_Malloc(rlen ? rlen : 1);
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   183
	if (!w)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   184
		goto nomem;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   185
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   186
	for (i = 0; i != rlen; i++) {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   187
		c = r[i];
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   188
		if (c == ' ' || c == '\t' || c == '\r') {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   189
			if (!allws && (wlen == 0 || w[wlen - 1] != ' '))
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   190
				w[wlen++] = ' ';
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   191
		} else if (c == '\n' && !allws && wlen > 0 &&
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   192
		           w[wlen - 1] == ' ') {
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   193
			w[wlen - 1] = '\n';
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   194
		} else {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   195
			w[wlen++] = c;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   196
		}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   197
	}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   198
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   199
	result = PyBytes_FromStringAndSize(w, wlen);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   200
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   201
nomem:
31467
08ecec297521 bdiff: use Python memory allocator in fixws
Gregory Szorc <gregory.szorc@gmail.com>
parents: 30561
diff changeset
   202
	PyMem_Free(w);
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   203
	return result ? result : PyErr_NoMemory();
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   204
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   205
36146
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   206
static bool sliceintolist(PyObject *list, Py_ssize_t destidx,
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   207
                          const char *source, Py_ssize_t len)
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   208
{
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   209
	PyObject *sliced = PyBytes_FromStringAndSize(source, len);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   210
	if (sliced == NULL)
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   211
		return false;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   212
	PyList_SET_ITEM(list, destidx, sliced);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   213
	return true;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   214
}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   215
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   216
static PyObject *splitnewlines(PyObject *self, PyObject *args)
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   217
{
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   218
	const char *text;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   219
	Py_ssize_t nelts = 0, size, i, start = 0;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   220
	PyObject *result = NULL;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   221
36620
186c6df3a373 py3: bulk-replace 'const char*' format specifier passed to PyArg_ParseTuple*()
Yuya Nishihara <yuya@tcha.org>
parents: 36146
diff changeset
   222
	if (!PyArg_ParseTuple(args, PY23("s#", "y#"), &text, &size)) {
36146
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   223
		goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   224
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   225
	if (!size) {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   226
		return PyList_New(0);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   227
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   228
	/* This loops to size-1 because if the last byte is a newline,
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   229
	 * we don't want to perform a split there. */
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   230
	for (i = 0; i < size - 1; ++i) {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   231
		if (text[i] == '\n') {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   232
			++nelts;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   233
		}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   234
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   235
	if ((result = PyList_New(nelts + 1)) == NULL)
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   236
		goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   237
	nelts = 0;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   238
	for (i = 0; i < size - 1; ++i) {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   239
		if (text[i] == '\n') {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   240
			if (!sliceintolist(result, nelts++, text + start,
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   241
			                   i - start + 1))
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   242
				goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   243
			start = i + 1;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   244
		}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   245
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   246
	if (!sliceintolist(result, nelts++, text + start, size - start))
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   247
		goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   248
	return result;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   249
abort:
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   250
	Py_XDECREF(result);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   251
	return NULL;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   252
}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   253
36916
1b9f6440506b bdiff: convert more longs to int64_t
Matt Harbison <matt_harbison@yahoo.com>
parents: 36763
diff changeset
   254
static int hunk_consumer(int64_t a1, int64_t a2, int64_t b1, int64_t b2,
1b9f6440506b bdiff: convert more longs to int64_t
Matt Harbison <matt_harbison@yahoo.com>
parents: 36763
diff changeset
   255
                         void *priv)
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   256
{
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   257
	PyObject *rl = (PyObject *)priv;
37980
273ea09f6550 bdiff: fix yet more fallout from xdiff long/int64 conversion (issue5885)
Julien Cristau <jcristau@debian.org>
parents: 36916
diff changeset
   258
	PyObject *m = Py_BuildValue("LLLL", a1, a2, b1, b2);
39421
ad76032d27da xdiff: fix leak in hunk_consumer()
Yuya Nishihara <yuya@tcha.org>
parents: 38309
diff changeset
   259
	int r;
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   260
	if (!m)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   261
		return -1;
39421
ad76032d27da xdiff: fix leak in hunk_consumer()
Yuya Nishihara <yuya@tcha.org>
parents: 38309
diff changeset
   262
	r = PyList_Append(rl, m);
ad76032d27da xdiff: fix leak in hunk_consumer()
Yuya Nishihara <yuya@tcha.org>
parents: 38309
diff changeset
   263
	Py_DECREF(m);
ad76032d27da xdiff: fix leak in hunk_consumer()
Yuya Nishihara <yuya@tcha.org>
parents: 38309
diff changeset
   264
	return r;
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   265
}
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   266
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   267
static PyObject *xdiffblocks(PyObject *self, PyObject *args)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   268
{
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   269
	Py_ssize_t la, lb;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   270
	mmfile_t a, b;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   271
	PyObject *rl;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   272
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   273
	xpparam_t xpp = {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   274
	    XDF_INDENT_HEURISTIC, /* flags */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   275
	};
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   276
	xdemitconf_t xecfg = {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   277
	    XDL_EMIT_BDIFFHUNK, /* flags */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   278
	    hunk_consumer,      /* hunk_consume_func */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   279
	};
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   280
	xdemitcb_t ecb = {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   281
	    NULL, /* priv */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   282
	};
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   283
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   284
	if (!PyArg_ParseTuple(args, PY23("s#s#", "y#y#"), &a.ptr, &la, &b.ptr,
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   285
	                      &lb))
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   286
		return NULL;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   287
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   288
	a.size = la;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   289
	b.size = lb;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   290
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   291
	rl = PyList_New(0);
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   292
	if (!rl)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   293
		return PyErr_NoMemory();
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   294
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   295
	ecb.priv = rl;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   296
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   297
	if (xdl_diff(&a, &b, &xpp, &xecfg, &ecb) != 0) {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   298
		Py_DECREF(rl);
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   299
		return PyErr_NoMemory();
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   300
	}
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   301
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   302
	return rl;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   303
}
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   304
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   305
static char mdiff_doc[] = "Efficient binary diff.";
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   306
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   307
static PyMethodDef methods[] = {
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   308
    {"bdiff", bdiff, METH_VARARGS, "calculate a binary diff\n"},
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   309
    {"blocks", blocks, METH_VARARGS, "find a list of matching lines\n"},
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   310
    {"fixws", fixws, METH_VARARGS, "normalize diff whitespaces\n"},
36146
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   311
    {"splitnewlines", splitnewlines, METH_VARARGS,
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   312
     "like str.splitlines, but only split on newlines\n"},
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   313
    {"xdiffblocks", xdiffblocks, METH_VARARGS,
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   314
     "find a list of matching lines using xdiff algorithm\n"},
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   315
    {NULL, NULL},
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   316
};
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   317
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   318
static const int version = 3;
32355
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   319
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   320
#ifdef IS_PY3K
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   321
static struct PyModuleDef bdiff_module = {
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   322
    PyModuleDef_HEAD_INIT, "bdiff", mdiff_doc, -1, methods,
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   323
};
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   324
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   325
PyMODINIT_FUNC PyInit_bdiff(void)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   326
{
32355
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   327
	PyObject *m;
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   328
	m = PyModule_Create(&bdiff_module);
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   329
	PyModule_AddIntConstant(m, "version", version);
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   330
	return m;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   331
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   332
#else
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   333
PyMODINIT_FUNC initbdiff(void)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   334
{
32355
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   335
	PyObject *m;
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   336
	m = Py_InitModule3("bdiff", methods, mdiff_doc);
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   337
	PyModule_AddIntConstant(m, "version", version);
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   338
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   339
#endif