bdiff: write a native version of splitnewlines

./hg perfunidiff mercurial/manifest.py 0 --count 500 --profile before:
! wall 0.309280 comb 0.350000 user 0.290000 sys 0.060000 (best of 32)

./hg perfunidiff mercurial/manifest.py 0 --count 500 --profile after:
! wall 0.241572 comb 0.260000 user 0.240000 sys 0.020000 (best of 39)

so it's about 20% faster. I hate Python. I wish we could usefully
write this in Rust, but it doesn't look like that's realistic without
using the cpython crate, which I'd still like to avoid.

Differential Revision: https://phab.mercurial-scm.org/D1973

mercurial/cext/bdiff.c

@@ -1,213 +1,263 b''
 /*
  bdiff.c - efficient binary diff extension for Mercurial
 
  Copyright 2005, 2006 Matt Mackall <[e-mail]>
 
  This software may be used and distributed according to the terms of
  the GNU General Public License, incorporated herein by reference.
 
  Based roughly on Python difflib
 */
 
 #define PY_SSIZE_T_CLEAN
 #include <Python.h>
 #include <limits.h>
 #include <stdlib.h>
 #include <string.h>
 
 #include "bdiff.h"
 #include "bitmanipulation.h"
 #include "util.h"
 
 static PyObject *blocks(PyObject *self, PyObject *args)
 {
 	PyObject *sa, *sb, *rl = NULL, *m;
 	struct bdiff_line *a, *b;
 	struct bdiff_hunk l, *h;
 	int an, bn, count, pos = 0;
 
 	l.next = NULL;
 
 	if (!PyArg_ParseTuple(args, "SS:bdiff", &sa, &sb))
 		return NULL;
 
 	an = bdiff_splitlines(PyBytes_AsString(sa), PyBytes_Size(sa), &a);
 	bn = bdiff_splitlines(PyBytes_AsString(sb), PyBytes_Size(sb), &b);
 
 	if (!a || !b)
 		goto nomem;
 
 	count = bdiff_diff(a, an, b, bn, &l);
 	if (count < 0)
 		goto nomem;
 
 	rl = PyList_New(count);
 	if (!rl)
 		goto nomem;
 
 	for (h = l.next; h; h = h->next) {
 		m = Py_BuildValue("iiii", h->a1, h->a2, h->b1, h->b2);
 		PyList_SetItem(rl, pos, m);
 		pos++;
 	}
 
 nomem:
 	free(a);
 	free(b);
 	bdiff_freehunks(l.next);
 	return rl ? rl : PyErr_NoMemory();
 }
 
 static PyObject *bdiff(PyObject *self, PyObject *args)
 {
 	char *sa, *sb, *rb, *ia, *ib;
 	PyObject *result = NULL;
 	struct bdiff_line *al, *bl;
 	struct bdiff_hunk l, *h;
 	int an, bn, count;
 	Py_ssize_t len = 0, la, lb, li = 0, lcommon = 0, lmax;
 	PyThreadState *_save;
 
 	l.next = NULL;
 
 	if (!PyArg_ParseTuple(args, "s#s#:bdiff", &sa, &la, &sb, &lb))
 		return NULL;
 
 	if (la > UINT_MAX || lb > UINT_MAX) {
 		PyErr_SetString(PyExc_ValueError, "bdiff inputs too large");
 		return NULL;
 	}
 
 	_save = PyEval_SaveThread();
 
 	lmax = la > lb ? lb : la;
 	for (ia = sa, ib = sb; li < lmax && *ia == *ib; ++li, ++ia, ++ib)
 		if (*ia == '\n')
 			lcommon = li + 1;
 	/* we can almost add: if (li == lmax) lcommon = li; */
 
 	an = bdiff_splitlines(sa + lcommon, la - lcommon, &al);
 	bn = bdiff_splitlines(sb + lcommon, lb - lcommon, &bl);
 	if (!al || !bl)
 		goto nomem;
 
 	count = bdiff_diff(al, an, bl, bn, &l);
 	if (count < 0)
 		goto nomem;
 
 	/* calculate length of output */
 	la = lb = 0;
 	for (h = l.next; h; h = h->next) {
 		if (h->a1 != la || h->b1 != lb)
 			len += 12 + bl[h->b1].l - bl[lb].l;
 		la = h->a2;
 		lb = h->b2;
 	}
 	PyEval_RestoreThread(_save);
 	_save = NULL;
 
 	result = PyBytes_FromStringAndSize(NULL, len);
 
 	if (!result)
 		goto nomem;
 
 	/* build binary patch */
 	rb = PyBytes_AsString(result);
 	la = lb = 0;
 
 	for (h = l.next; h; h = h->next) {
 		if (h->a1 != la || h->b1 != lb) {
 			len = bl[h->b1].l - bl[lb].l;
 			putbe32((uint32_t)(al[la].l + lcommon - al->l), rb);
 			putbe32((uint32_t)(al[h->a1].l + lcommon - al->l),
 			        rb + 4);
 			putbe32((uint32_t)len, rb + 8);
 			memcpy(rb + 12, bl[lb].l, len);
 			rb += 12 + len;
 		}
 		la = h->a2;
 		lb = h->b2;
 	}
 
 nomem:
 	if (_save)
 		PyEval_RestoreThread(_save);
 	free(al);
 	free(bl);
 	bdiff_freehunks(l.next);
 	return result ? result : PyErr_NoMemory();
 }
 
 /*
  * If allws != 0, remove all whitespace (' ', \t and \r). Otherwise,
  * reduce whitespace sequences to a single space and trim remaining whitespace
  * from end of lines.
  */
 static PyObject *fixws(PyObject *self, PyObject *args)
 {
 	PyObject *s, *result = NULL;
 	char allws, c;
 	const char *r;
 	Py_ssize_t i, rlen, wlen = 0;
 	char *w;
 
 	if (!PyArg_ParseTuple(args, "Sb:fixws", &s, &allws))
 		return NULL;
 	r = PyBytes_AsString(s);
 	rlen = PyBytes_Size(s);
 
 	w = (char *)PyMem_Malloc(rlen ? rlen : 1);
 	if (!w)
 		goto nomem;
 
 	for (i = 0; i != rlen; i++) {
 		c = r[i];
 		if (c == ' ' || c == '\t' || c == '\r') {
 			if (!allws && (wlen == 0 || w[wlen - 1] != ' '))
 				w[wlen++] = ' ';
 		} else if (c == '\n' && !allws && wlen > 0 &&
 		           w[wlen - 1] == ' ') {
 			w[wlen - 1] = '\n';
 		} else {
 			w[wlen++] = c;
 		}
 	}
 
 	result = PyBytes_FromStringAndSize(w, wlen);
 
 nomem:
 	PyMem_Free(w);
 	return result ? result : PyErr_NoMemory();
 }
 
+static bool sliceintolist(PyObject *list, Py_ssize_t destidx,
+                          const char *source, Py_ssize_t len)
+{
+	PyObject *sliced = PyBytes_FromStringAndSize(source, len);
+	if (sliced == NULL)
+		return false;
+	PyList_SET_ITEM(list, destidx, sliced);
+	return true;
+}
+
+static PyObject *splitnewlines(PyObject *self, PyObject *args)
+{
+	const char *text;
+	Py_ssize_t nelts = 0, size, i, start = 0;
+	PyObject *result = NULL;
+
+	if (!PyArg_ParseTuple(args, "s#", &text, &size)) {
+		goto abort;
+	}
+	if (!size) {
+		return PyList_New(0);
+	}
+	/* This loops to size-1 because if the last byte is a newline,
+	 * we don't want to perform a split there. */
+	for (i = 0; i < size - 1; ++i) {
+		if (text[i] == '\n') {
+			++nelts;
+		}
+	}
+	if ((result = PyList_New(nelts + 1)) == NULL)
+		goto abort;
+	nelts = 0;
+	for (i = 0; i < size - 1; ++i) {
+		if (text[i] == '\n') {
+			if (!sliceintolist(result, nelts++, text + start,
+			                   i - start + 1))
+				goto abort;
+			start = i + 1;
+		}
+	}
+	if (!sliceintolist(result, nelts++, text + start, size - start))
+		goto abort;
+	return result;
+abort:
+	Py_XDECREF(result);
+	return NULL;
+}
+
 static char mdiff_doc[] = "Efficient binary diff.";
 
 static PyMethodDef methods[] = {
     {"bdiff", bdiff, METH_VARARGS, "calculate a binary diff\n"},
     {"blocks", blocks, METH_VARARGS, "find a list of matching lines\n"},
     {"fixws", fixws, METH_VARARGS, "normalize diff whitespaces\n"},
+    {"splitnewlines", splitnewlines, METH_VARARGS,
+     "like str.splitlines, but only split on newlines\n"},
     {NULL, NULL},
 };
 
-static const int version = 1;
+static const int version = 2;
 
 #ifdef IS_PY3K
 static struct PyModuleDef bdiff_module = {
     PyModuleDef_HEAD_INIT, "bdiff", mdiff_doc, -1, methods,
 };
 
 PyMODINIT_FUNC PyInit_bdiff(void)
 {
 	PyObject *m;
 	m = PyModule_Create(&bdiff_module);
 	PyModule_AddIntConstant(m, "version", version);
 	return m;
 }
 #else
 PyMODINIT_FUNC initbdiff(void)
 {
 	PyObject *m;
 	m = Py_InitModule3("bdiff", methods, mdiff_doc);
 	PyModule_AddIntConstant(m, "version", version);
 }
 #endif

mercurial/mdiff.py

@@ -1,521 +1,512 b''
 # mdiff.py - diff and patch routines for mercurial
 #
 # Copyright 2005, 2006 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import re
 import struct
 import zlib
 
 from .i18n import _
 from . import (
     error,
     policy,
     pycompat,
     util,
 )
 
 _missing_newline_marker = "\\ No newline at end of file\n"
 
 bdiff = policy.importmod(r'bdiff')
 mpatch = policy.importmod(r'mpatch')
 
 blocks = bdiff.blocks
 fixws = bdiff.fixws
 patches = mpatch.patches
 patchedsize = mpatch.patchedsize
 textdiff = bdiff.bdiff
-
-def splitnewlines(text):
-    '''like str.splitlines, but only split on newlines.'''
-    lines = [l + '\n' for l in text.split('\n')]
-    if lines:
-        if lines[-1] == '\n':
-            lines.pop()
-        else:
-            lines[-1] = lines[-1][:-1]
-    return lines
+splitnewlines = bdiff.splitnewlines
 
 class diffopts(object):
     '''context is the number of context lines
     text treats all files as text
     showfunc enables diff -p output
     git enables the git extended patch format
     nodates removes dates from diff headers
     nobinary ignores binary files
     noprefix disables the 'a/' and 'b/' prefixes (ignored in plain mode)
     ignorews ignores all whitespace changes in the diff
     ignorewsamount ignores changes in the amount of whitespace
     ignoreblanklines ignores changes whose lines are all blank
     upgrade generates git diffs to avoid data loss
     '''
 
     defaults = {
         'context': 3,
         'text': False,
         'showfunc': False,
         'git': False,
         'nodates': False,
         'nobinary': False,
         'noprefix': False,
         'index': 0,
         'ignorews': False,
         'ignorewsamount': False,
         'ignorewseol': False,
         'ignoreblanklines': False,
         'upgrade': False,
         'showsimilarity': False,
         'worddiff': False,
         }
 
     def __init__(self, **opts):
         opts = pycompat.byteskwargs(opts)
         for k in self.defaults.keys():
             v = opts.get(k)
             if v is None:
                 v = self.defaults[k]
             setattr(self, k, v)
 
         try:
             self.context = int(self.context)
         except ValueError:
             raise error.Abort(_('diff context lines count must be '
                                'an integer, not %r') % self.context)
 
     def copy(self, **kwargs):
         opts = dict((k, getattr(self, k)) for k in self.defaults)
         opts = pycompat.strkwargs(opts)
         opts.update(kwargs)
         return diffopts(**opts)
 
 defaultopts = diffopts()
 
 def wsclean(opts, text, blank=True):
     if opts.ignorews:
         text = bdiff.fixws(text, 1)
     elif opts.ignorewsamount:
         text = bdiff.fixws(text, 0)
     if blank and opts.ignoreblanklines:
         text = re.sub('\n+', '\n', text).strip('\n')
     if opts.ignorewseol:
         text = re.sub(br'[ \t\r\f]+\n', r'\n', text)
     return text
 
 def splitblock(base1, lines1, base2, lines2, opts):
     # The input lines matches except for interwoven blank lines. We
     # transform it into a sequence of matching blocks and blank blocks.
     lines1 = [(wsclean(opts, l) and 1 or 0) for l in lines1]
     lines2 = [(wsclean(opts, l) and 1 or 0) for l in lines2]
     s1, e1 = 0, len(lines1)
     s2, e2 = 0, len(lines2)
     while s1 < e1 or s2 < e2:
         i1, i2, btype = s1, s2, '='
         if (i1 >= e1 or lines1[i1] == 0
             or i2 >= e2 or lines2[i2] == 0):
             # Consume the block of blank lines
             btype = '~'
             while i1 < e1 and lines1[i1] == 0:
                 i1 += 1
             while i2 < e2 and lines2[i2] == 0:
                 i2 += 1
         else:
             # Consume the matching lines
             while i1 < e1 and lines1[i1] == 1 and lines2[i2] == 1:
                 i1 += 1
                 i2 += 1
         yield [base1 + s1, base1 + i1, base2 + s2, base2 + i2], btype
         s1 = i1
         s2 = i2
 
 def hunkinrange(hunk, linerange):
     """Return True if `hunk` defined as (start, length) is in `linerange`
     defined as (lowerbound, upperbound).
 
     >>> hunkinrange((5, 10), (2, 7))
     True
     >>> hunkinrange((5, 10), (6, 12))
     True
     >>> hunkinrange((5, 10), (13, 17))
     True
     >>> hunkinrange((5, 10), (3, 17))
     True
     >>> hunkinrange((5, 10), (1, 3))
     False
     >>> hunkinrange((5, 10), (18, 20))
     False
     >>> hunkinrange((5, 10), (1, 5))
     False
     >>> hunkinrange((5, 10), (15, 27))
     False
     """
     start, length = hunk
     lowerbound, upperbound = linerange
     return lowerbound < start + length and start < upperbound
 
 def blocksinrange(blocks, rangeb):
     """filter `blocks` like (a1, a2, b1, b2) from items outside line range
     `rangeb` from ``(b1, b2)`` point of view.
 
     Return `filteredblocks, rangea` where:
 
     * `filteredblocks` is list of ``block = (a1, a2, b1, b2), stype`` items of
       `blocks` that are inside `rangeb` from ``(b1, b2)`` point of view; a
       block ``(b1, b2)`` being inside `rangeb` if
       ``rangeb[0] < b2 and b1 < rangeb[1]``;
     * `rangea` is the line range w.r.t. to ``(a1, a2)`` parts of `blocks`.
     """
     lbb, ubb = rangeb
     lba, uba = None, None
     filteredblocks = []
     for block in blocks:
         (a1, a2, b1, b2), stype = block
         if lbb >= b1 and ubb <= b2 and stype == '=':
             # rangeb is within a single "=" hunk, restrict back linerange1
             # by offsetting rangeb
             lba = lbb - b1 + a1
             uba = ubb - b1 + a1
         else:
             if b1 <= lbb < b2:
                 if stype == '=':
                     lba = a2 - (b2 - lbb)
                 else:
                     lba = a1
             if b1 < ubb <= b2:
                 if stype == '=':
                     uba = a1 + (ubb - b1)
                 else:
                     uba = a2
         if hunkinrange((b1, (b2 - b1)), rangeb):
             filteredblocks.append(block)
     if lba is None or uba is None or uba < lba:
         raise error.Abort(_('line range exceeds file size'))
     return filteredblocks, (lba, uba)
 
 def allblocks(text1, text2, opts=None, lines1=None, lines2=None):
     """Return (block, type) tuples, where block is an mdiff.blocks
     line entry. type is '=' for blocks matching exactly one another
     (bdiff blocks), '!' for non-matching blocks and '~' for blocks
     matching only after having filtered blank lines.
     line1 and line2 are text1 and text2 split with splitnewlines() if
     they are already available.
     """
     if opts is None:
         opts = defaultopts
     if opts.ignorews or opts.ignorewsamount or opts.ignorewseol:
         text1 = wsclean(opts, text1, False)
         text2 = wsclean(opts, text2, False)
     diff = bdiff.blocks(text1, text2)
     for i, s1 in enumerate(diff):
         # The first match is special.
         # we've either found a match starting at line 0 or a match later
         # in the file.  If it starts later, old and new below will both be
         # empty and we'll continue to the next match.
         if i > 0:
             s = diff[i - 1]
         else:
             s = [0, 0, 0, 0]
         s = [s[1], s1[0], s[3], s1[2]]
 
         # bdiff sometimes gives huge matches past eof, this check eats them,
         # and deals with the special first match case described above
         if s[0] != s[1] or s[2] != s[3]:
             type = '!'
             if opts.ignoreblanklines:
                 if lines1 is None:
                     lines1 = splitnewlines(text1)
                 if lines2 is None:
                     lines2 = splitnewlines(text2)
                 old = wsclean(opts, "".join(lines1[s[0]:s[1]]))
                 new = wsclean(opts, "".join(lines2[s[2]:s[3]]))
                 if old == new:
                     type = '~'
             yield s, type
         yield s1, '='
 
 def unidiff(a, ad, b, bd, fn1, fn2, binary, opts=defaultopts):
     """Return a unified diff as a (headers, hunks) tuple.
 
     If the diff is not null, `headers` is a list with unified diff header
     lines "--- <original>" and "+++ <new>" and `hunks` is a generator yielding
     (hunkrange, hunklines) coming from _unidiff().
     Otherwise, `headers` and `hunks` are empty.
 
     Set binary=True if either a or b should be taken as a binary file.
     """
     def datetag(date, fn=None):
         if not opts.git and not opts.nodates:
             return '\t%s' % date
         if fn and ' ' in fn:
             return '\t'
         return ''
 
     sentinel = [], ()
     if not a and not b:
         return sentinel
 
     if opts.noprefix:
         aprefix = bprefix = ''
     else:
         aprefix = 'a/'
         bprefix = 'b/'
 
     epoch = util.datestr((0, 0))
 
     fn1 = util.pconvert(fn1)
     fn2 = util.pconvert(fn2)
 
     if binary:
         if a and b and len(a) == len(b) and a == b:
             return sentinel
         headerlines = []
         hunks = (None, ['Binary file %s has changed\n' % fn1]),
     elif not a:
         without_newline = not b.endswith('\n')
         b = splitnewlines(b)
         if a is None:
             l1 = '--- /dev/null%s' % datetag(epoch)
         else:
             l1 = "--- %s%s%s" % (aprefix, fn1, datetag(ad, fn1))
         l2 = "+++ %s%s" % (bprefix + fn2, datetag(bd, fn2))
         headerlines = [l1, l2]
         size = len(b)
         hunkrange = (0, 0, 1, size)
         hunklines = ["@@ -0,0 +1,%d @@\n" % size] + ["+" + e for e in b]
         if without_newline:
             hunklines[-1] += '\n'
             hunklines.append(_missing_newline_marker)
         hunks = (hunkrange, hunklines),
     elif not b:
         without_newline = not a.endswith('\n')
         a = splitnewlines(a)
         l1 = "--- %s%s%s" % (aprefix, fn1, datetag(ad, fn1))
         if b is None:
             l2 = '+++ /dev/null%s' % datetag(epoch)
         else:
             l2 = "+++ %s%s%s" % (bprefix, fn2, datetag(bd, fn2))
         headerlines = [l1, l2]
         size = len(a)
         hunkrange = (1, size, 0, 0)
         hunklines = ["@@ -1,%d +0,0 @@\n" % size] + ["-" + e for e in a]
         if without_newline:
             hunklines[-1] += '\n'
             hunklines.append(_missing_newline_marker)
         hunks = (hunkrange, hunklines),
     else:
         hunks = _unidiff(a, b, opts=opts)
         if not next(hunks):
             return sentinel
 
         headerlines = [
             "--- %s%s%s" % (aprefix, fn1, datetag(ad, fn1)),
             "+++ %s%s%s" % (bprefix, fn2, datetag(bd, fn2)),
         ]
 
     return headerlines, hunks
 
 def _unidiff(t1, t2, opts=defaultopts):
     """Yield hunks of a headerless unified diff from t1 and t2 texts.
 
     Each hunk consists of a (hunkrange, hunklines) tuple where `hunkrange` is a
     tuple (s1, l1, s2, l2) representing the range information of the hunk to
     form the '@@ -s1,l1 +s2,l2 @@' header and `hunklines` is a list of lines
     of the hunk combining said header followed by line additions and
     deletions.
 
     The hunks are prefixed with a bool.
     """
     l1 = splitnewlines(t1)
     l2 = splitnewlines(t2)
     def contextend(l, len):
         ret = l + opts.context
         if ret > len:
             ret = len
         return ret
 
     def contextstart(l):
         ret = l - opts.context
         if ret < 0:
             return 0
         return ret
 
     lastfunc = [0, '']
     def yieldhunk(hunk):
         (astart, a2, bstart, b2, delta) = hunk
         aend = contextend(a2, len(l1))
         alen = aend - astart
         blen = b2 - bstart + aend - a2
 
         func = ""
         if opts.showfunc:
             lastpos, func = lastfunc
             # walk backwards from the start of the context up to the start of
             # the previous hunk context until we find a line starting with an
             # alphanumeric char.
             for i in xrange(astart - 1, lastpos - 1, -1):
                 if l1[i][0:1].isalnum():
                     func = ' ' + l1[i].rstrip()[:40]
                     lastfunc[1] = func
                     break
             # by recording this hunk's starting point as the next place to
             # start looking for function lines, we avoid reading any line in
             # the file more than once.
             lastfunc[0] = astart
 
         # zero-length hunk ranges report their start line as one less
         if alen:
             astart += 1
         if blen:
             bstart += 1
 
         hunkrange = astart, alen, bstart, blen
         hunklines = (
             ["@@ -%d,%d +%d,%d @@%s\n" % (hunkrange + (func,))]
             + delta
             + [' ' + l1[x] for x in xrange(a2, aend)]
         )
         # If either file ends without a newline and the last line of
         # that file is part of a hunk, a marker is printed. If the
         # last line of both files is identical and neither ends in
         # a newline, print only one marker. That's the only case in
         # which the hunk can end in a shared line without a newline.
         skip = False
         if not t1.endswith('\n') and astart + alen == len(l1) + 1:
             for i in xrange(len(hunklines) - 1, -1, -1):
                 if hunklines[i].startswith(('-', ' ')):
                     if hunklines[i].startswith(' '):
                         skip = True
                     hunklines[i] += '\n'
                     hunklines.insert(i + 1, _missing_newline_marker)
                     break
         if not skip and not t2.endswith('\n') and bstart + blen == len(l2) + 1:
             for i in xrange(len(hunklines) - 1, -1, -1):
                 if hunklines[i].startswith('+'):
                     hunklines[i] += '\n'
                     hunklines.insert(i + 1, _missing_newline_marker)
                     break
         yield hunkrange, hunklines
 
     # bdiff.blocks gives us the matching sequences in the files.  The loop
     # below finds the spaces between those matching sequences and translates
     # them into diff output.
     #
     hunk = None
     ignoredlines = 0
     has_hunks = False
     for s, stype in allblocks(t1, t2, opts, l1, l2):
         a1, a2, b1, b2 = s
         if stype != '!':
             if stype == '~':
                 # The diff context lines are based on t1 content. When
                 # blank lines are ignored, the new lines offsets must
                 # be adjusted as if equivalent blocks ('~') had the
                 # same sizes on both sides.
                 ignoredlines += (b2 - b1) - (a2 - a1)
             continue
         delta = []
         old = l1[a1:a2]
         new = l2[b1:b2]
 
         b1 -= ignoredlines
         b2 -= ignoredlines
         astart = contextstart(a1)
         bstart = contextstart(b1)
         prev = None
         if hunk:
             # join with the previous hunk if it falls inside the context
             if astart < hunk[1] + opts.context + 1:
                 prev = hunk
                 astart = hunk[1]
                 bstart = hunk[3]
             else:
                 if not has_hunks:
                     has_hunks = True
                     yield True
                 for x in yieldhunk(hunk):
                     yield x
         if prev:
             # we've joined the previous hunk, record the new ending points.
             hunk[1] = a2
             hunk[3] = b2
             delta = hunk[4]
         else:
             # create a new hunk
             hunk = [astart, a2, bstart, b2, delta]
 
         delta[len(delta):] = [' ' + x for x in l1[astart:a1]]
         delta[len(delta):] = ['-' + x for x in old]
         delta[len(delta):] = ['+' + x for x in new]
 
     if hunk:
         if not has_hunks:
             has_hunks = True
             yield True
         for x in yieldhunk(hunk):
             yield x
     elif not has_hunks:
         yield False
 
 def b85diff(to, tn):
     '''print base85-encoded binary diff'''
     def fmtline(line):
         l = len(line)
         if l <= 26:
             l = pycompat.bytechr(ord('A') + l - 1)
         else:
             l = pycompat.bytechr(l - 26 + ord('a') - 1)
         return '%c%s\n' % (l, util.b85encode(line, True))
 
     def chunk(text, csize=52):
         l = len(text)
         i = 0
         while i < l:
             yield text[i:i + csize]
             i += csize
 
     if to is None:
         to = ''
     if tn is None:
         tn = ''
 
     if to == tn:
         return ''
 
     # TODO: deltas
     ret = []
     ret.append('GIT binary patch\n')
     ret.append('literal %d\n' % len(tn))
     for l in chunk(zlib.compress(tn)):
         ret.append(fmtline(l))
     ret.append('\n')
 
     return ''.join(ret)
 
 def patchtext(bin):
     pos = 0
     t = []
     while pos < len(bin):
         p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12])
         pos += 12
         t.append(bin[pos:pos + l])
         pos += l
     return "".join(t)
 
 def patch(a, bin):
     if len(a) == 0:
         # skip over trivial delta header
         return util.buffer(bin, 12)
     return mpatch.patches(a, [bin])
 
 # similar to difflib.SequenceMatcher.get_matching_blocks
 def get_matching_blocks(a, b):
     return [(d[0], d[2], d[1] - d[0]) for d in bdiff.blocks(a, b)]
 
 def trivialdiffheader(length):
     return struct.pack(">lll", 0, 0, length) if length else ''
 
 def replacediffheader(oldlen, newlen):
     return struct.pack(">lll", 0, oldlen, newlen)

mercurial/policy.py

@@ -1,116 +1,116 b''
 # policy.py - module policy logic for Mercurial.
 #
 # Copyright 2015 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import os
 import sys
 
 # Rules for how modules can be loaded. Values are:
 #
 #    c - require C extensions
 #    allow - allow pure Python implementation when C loading fails
 #    cffi - required cffi versions (implemented within pure module)
 #    cffi-allow - allow pure Python implementation if cffi version is missing
 #    py - only load pure Python modules
 #
 # By default, fall back to the pure modules so the in-place build can
 # run without recompiling the C extensions. This will be overridden by
 # __modulepolicy__ generated by setup.py.
 policy = b'allow'
 _packageprefs = {
     # policy: (versioned package, pure package)
     b'c': (r'cext', None),
     b'allow': (r'cext', r'pure'),
     b'cffi': (r'cffi', None),
     b'cffi-allow': (r'cffi', r'pure'),
     b'py': (None, r'pure'),
 }
 
 try:
     from . import __modulepolicy__
     policy = __modulepolicy__.modulepolicy
 except ImportError:
     pass
 
 # PyPy doesn't load C extensions.
 #
 # The canonical way to do this is to test platform.python_implementation().
 # But we don't import platform and don't bloat for it here.
 if r'__pypy__' in sys.builtin_module_names:
     policy = b'cffi'
 
 # Our C extensions aren't yet compatible with Python 3. So use pure Python
 # on Python 3 for now.
 if sys.version_info[0] >= 3:
     policy = b'py'
 
 # Environment variable can always force settings.
 if sys.version_info[0] >= 3:
     if r'HGMODULEPOLICY' in os.environ:
         policy = os.environ[r'HGMODULEPOLICY'].encode(r'utf-8')
 else:
     policy = os.environ.get(r'HGMODULEPOLICY', policy)
 
 def _importfrom(pkgname, modname):
     # from .<pkgname> import <modname> (where . is looked through this module)
     fakelocals = {}
     pkg = __import__(pkgname, globals(), fakelocals, [modname], level=1)
     try:
         fakelocals[modname] = mod = getattr(pkg, modname)
     except AttributeError:
         raise ImportError(r'cannot import name %s' % modname)
     # force import; fakelocals[modname] may be replaced with the real module
     getattr(mod, r'__doc__', None)
     return fakelocals[modname]
 
 # keep in sync with "version" in C modules
 _cextversions = {
     (r'cext', r'base85'): 1,
-    (r'cext', r'bdiff'): 1,
+    (r'cext', r'bdiff'): 2,
     (r'cext', r'diffhelpers'): 1,
     (r'cext', r'mpatch'): 1,
     (r'cext', r'osutil'): 3,
     (r'cext', r'parsers'): 4,
 }
 
 # map import request to other package or module
 _modredirects = {
     (r'cext', r'charencode'): (r'cext', r'parsers'),
     (r'cffi', r'base85'): (r'pure', r'base85'),
     (r'cffi', r'charencode'): (r'pure', r'charencode'),
     (r'cffi', r'diffhelpers'): (r'pure', r'diffhelpers'),
     (r'cffi', r'parsers'): (r'pure', r'parsers'),
 }
 
 def _checkmod(pkgname, modname, mod):
     expected = _cextversions.get((pkgname, modname))
     actual = getattr(mod, r'version', None)
     if actual != expected:
         raise ImportError(r'cannot import module %s.%s '
                           r'(expected version: %d, actual: %r)'
                           % (pkgname, modname, expected, actual))
 
 def importmod(modname):
     """Import module according to policy and check API version"""
     try:
         verpkg, purepkg = _packageprefs[policy]
     except KeyError:
         raise ImportError(r'invalid HGMODULEPOLICY %r' % policy)
     assert verpkg or purepkg
     if verpkg:
         pn, mn = _modredirects.get((verpkg, modname), (verpkg, modname))
         try:
             mod = _importfrom(pn, mn)
             if pn == verpkg:
                 _checkmod(pn, mn, mod)
             return mod
         except ImportError:
             if not purepkg:
                 raise
     pn, mn = _modredirects.get((purepkg, modname), (purepkg, modname))
     return _importfrom(pn, mn)

mercurial/pure/bdiff.py

@@ -1,92 +1,102 b''
 # bdiff.py - Python implementation of bdiff.c
 #
 # Copyright 2009 Matt Mackall <[e-mail]> and others
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import difflib
 import re
 import struct
 
 def splitnewlines(text):
     '''like str.splitlines, but only split on newlines.'''
     lines = [l + '\n' for l in text.split('\n')]
     if lines:
         if lines[-1] == '\n':
             lines.pop()
         else:
             lines[-1] = lines[-1][:-1]
     return lines
 
 def _normalizeblocks(a, b, blocks):
     prev = None
     r = []
     for curr in blocks:
         if prev is None:
             prev = curr
             continue
         shift = 0
 
         a1, b1, l1 = prev
         a1end = a1 + l1
         b1end = b1 + l1
 
         a2, b2, l2 = curr
         a2end = a2 + l2
         b2end = b2 + l2
         if a1end == a2:
             while (a1end + shift < a2end and
                    a[a1end + shift] == b[b1end + shift]):
                 shift += 1
         elif b1end == b2:
             while (b1end + shift < b2end and
                    a[a1end + shift] == b[b1end + shift]):
                 shift += 1
         r.append((a1, b1, l1 + shift))
         prev = a2 + shift, b2 + shift, l2 - shift
     r.append(prev)
     return r
 
 def bdiff(a, b):
     a = bytes(a).splitlines(True)
     b = bytes(b).splitlines(True)
 
     if not a:
         s = "".join(b)
         return s and (struct.pack(">lll", 0, 0, len(s)) + s)
 
     bin = []
     p = [0]
     for i in a:
         p.append(p[-1] + len(i))
 
     d = difflib.SequenceMatcher(None, a, b).get_matching_blocks()
     d = _normalizeblocks(a, b, d)
     la = 0
     lb = 0
     for am, bm, size in d:
         s = "".join(b[lb:bm])
         if am > la or s:
             bin.append(struct.pack(">lll", p[la], p[am], len(s)) + s)
         la = am + size
         lb = bm + size
 
     return "".join(bin)
 
 def blocks(a, b):
     an = splitnewlines(a)
     bn = splitnewlines(b)
     d = difflib.SequenceMatcher(None, an, bn).get_matching_blocks()
     d = _normalizeblocks(an, bn, d)
     return [(i, i + n, j, j + n) for (i, j, n) in d]
 
 def fixws(text, allws):
     if allws:
         text = re.sub('[ \t\r]+', '', text)
     else:
         text = re.sub('[ \t\r]+', ' ', text)
         text = text.replace(' \n', '\n')
     return text
+
+def splitnewlines(text):
+    '''like str.splitlines, but only split on newlines.'''
+    lines = [l + '\n' for l in text.split('\n')]
+    if lines:
+        if lines[-1] == '\n':
+            lines.pop()
+        else:
+            lines[-1] = lines[-1][:-1]
+    return lines