mercurial/cext/bdiff.c
author Jun Wu <quark@fb.com>
Sat, 03 Mar 2018 12:39:14 -0800
changeset 36675 430fdb717549
parent 36655 68026dd7c4f9
child 36681 340e4b711df7
permissions -rw-r--r--
bdiff: add a xdiffblocks method This is similar to `bdiff.blocks`, but uses xdiff as the backend. The indent heuristic is turned on by default since it has little overhead and improves diff quality significantly. Differential Revision: https://phab.mercurial-scm.org/D2602
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
36655
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
   106
	an = bdiff_splitlines(ba.buf + lcommon, la - lcommon, &al);
68026dd7c4f9 cext: accept arguments as Py_buffer
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36654
diff changeset
   107
	bn = bdiff_splitlines(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);
36654
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   158
	if (al) {
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   159
		free(al);
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   160
	}
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   161
	if (bl) {
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   162
		free(bl);
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   163
	}
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   164
	if (l.next) {
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   165
		bdiff_freehunks(l.next);
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   166
	}
b864f4536ca8 cext: refactor cleanup code in bdiff()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36620
diff changeset
   167
	return result;
29541
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
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
 * 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
   172
 * 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
   173
 * from end of lines.
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   174
 */
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   175
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
   176
{
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   177
	PyObject *s, *result = NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   178
	char allws, c;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   179
	const char *r;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   180
	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
   181
	char *w;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   182
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   183
	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
   184
		return NULL;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   185
	r = PyBytes_AsString(s);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   186
	rlen = PyBytes_Size(s);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   187
31467
08ecec297521 bdiff: use Python memory allocator in fixws
Gregory Szorc <gregory.szorc@gmail.com>
parents: 30561
diff changeset
   188
	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
   189
	if (!w)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   190
		goto nomem;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   191
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   192
	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
   193
		c = r[i];
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   194
		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
   195
			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
   196
				w[wlen++] = ' ';
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   197
		} else if (c == '\n' && !allws && wlen > 0 &&
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   198
		           w[wlen - 1] == ' ') {
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   199
			w[wlen - 1] = '\n';
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   200
		} else {
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   201
			w[wlen++] = c;
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   202
		}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   203
	}
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
	result = PyBytes_FromStringAndSize(w, wlen);
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   206
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   207
nomem:
31467
08ecec297521 bdiff: use Python memory allocator in fixws
Gregory Szorc <gregory.szorc@gmail.com>
parents: 30561
diff changeset
   208
	PyMem_Free(w);
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   209
	return result ? result : PyErr_NoMemory();
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   210
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   211
36146
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   212
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
   213
                          const char *source, Py_ssize_t len)
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
	PyObject *sliced = PyBytes_FromStringAndSize(source, len);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   216
	if (sliced == NULL)
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   217
		return false;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   218
	PyList_SET_ITEM(list, destidx, sliced);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   219
	return true;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   220
}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   221
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   222
static PyObject *splitnewlines(PyObject *self, PyObject *args)
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   223
{
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   224
	const char *text;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   225
	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
   226
	PyObject *result = NULL;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   227
36620
186c6df3a373 py3: bulk-replace 'const char*' format specifier passed to PyArg_ParseTuple*()
Yuya Nishihara <yuya@tcha.org>
parents: 36146
diff changeset
   228
	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
   229
		goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   230
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   231
	if (!size) {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   232
		return PyList_New(0);
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
	/* 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
   235
	 * 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
   236
	for (i = 0; i < size - 1; ++i) {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   237
		if (text[i] == '\n') {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   238
			++nelts;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   239
		}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   240
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   241
	if ((result = PyList_New(nelts + 1)) == NULL)
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
	nelts = 0;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   244
	for (i = 0; i < size - 1; ++i) {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   245
		if (text[i] == '\n') {
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   246
			if (!sliceintolist(result, nelts++, text + start,
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   247
			                   i - start + 1))
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   248
				goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   249
			start = i + 1;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   250
		}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   251
	}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   252
	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
   253
		goto abort;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   254
	return result;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   255
abort:
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   256
	Py_XDECREF(result);
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   257
	return NULL;
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   258
}
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   259
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   260
static int hunk_consumer(long a1, long a2, long b1, long b2, void *priv)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   261
{
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   262
	PyObject *rl = (PyObject *)priv;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   263
	PyObject *m = Py_BuildValue("llll", a1, a2, b1, b2);
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   264
	if (!m)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   265
		return -1;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   266
	if (PyList_Append(rl, m) != 0) {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   267
		Py_DECREF(m);
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   268
		return -1;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   269
	}
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   270
	return 0;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   271
}
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
static PyObject *xdiffblocks(PyObject *self, PyObject *args)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   274
{
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   275
	Py_ssize_t la, lb;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   276
	mmfile_t a, b;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   277
	PyObject *rl;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   278
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   279
	xpparam_t xpp = {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   280
	    XDF_INDENT_HEURISTIC, /* flags */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   281
	    NULL,                 /* anchors */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   282
	    0,                    /* anchors_nr */
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
	xdemitconf_t xecfg = {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   285
	    0,                  /* ctxlen */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   286
	    0,                  /* interhunkctxlen */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   287
	    XDL_EMIT_BDIFFHUNK, /* flags */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   288
	    NULL,               /* find_func */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   289
	    NULL,               /* find_func_priv */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   290
	    hunk_consumer,      /* hunk_consume_func */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   291
	};
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   292
	xdemitcb_t ecb = {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   293
	    NULL, /* priv */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   294
	    NULL, /* outf */
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   295
	};
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 (!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
   298
	                      &lb))
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   299
		return NULL;
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
	a.size = la;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   302
	b.size = lb;
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
	rl = PyList_New(0);
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   305
	if (!rl)
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   306
		return PyErr_NoMemory();
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   307
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   308
	ecb.priv = rl;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   309
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   310
	if (xdl_diff(&a, &b, &xpp, &xecfg, &ecb) != 0) {
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   311
		Py_DECREF(rl);
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   312
		return PyErr_NoMemory();
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   313
	}
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   314
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   315
	return rl;
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   316
}
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   317
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   318
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
   319
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   320
static PyMethodDef methods[] = {
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   321
    {"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
   322
    {"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
   323
    {"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
   324
    {"splitnewlines", splitnewlines, METH_VARARGS,
29dd37a418aa bdiff: write a native version of splitnewlines
Augie Fackler <augie@google.com>
parents: 36055
diff changeset
   325
     "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
   326
    {"xdiffblocks", xdiffblocks, METH_VARARGS,
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   327
     "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
   328
    {NULL, NULL},
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   329
};
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   330
36675
430fdb717549 bdiff: add a xdiffblocks method
Jun Wu <quark@fb.com>
parents: 36655
diff changeset
   331
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
   332
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   333
#ifdef IS_PY3K
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   334
static struct PyModuleDef bdiff_module = {
36055
b4fdc6177b29 bdiff: add to clang-format oversight
Augie Fackler <augie@google.com>
parents: 34438
diff changeset
   335
    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
   336
};
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   337
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   338
PyMODINIT_FUNC PyInit_bdiff(void)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   339
{
32355
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   340
	PyObject *m;
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   341
	m = PyModule_Create(&bdiff_module);
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   342
	PyModule_AddIntConstant(m, "version", version);
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   343
	return m;
29541
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   344
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   345
#else
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   346
PyMODINIT_FUNC initbdiff(void)
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   347
{
32355
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   348
	PyObject *m;
4195b84940e9 bdiff: add version to help detect breaking binary changes
Jun Wu <quark@fb.com>
parents: 31467
diff changeset
   349
	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
   350
	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
   351
}
9631ff5ebbeb bdiff: split bdiff into cpy-aware and cpy-agnostic part
Maciej Fijalkowski <fijall@gmail.com>
parents:
diff changeset
   352
#endif