diff mercurial/mdiff.py @ 1637:3b1b44b917f4

Add new bdiff based unidiff generation.
author mason@suse.com
date Tue, 24 Jan 2006 15:02:10 +1300
parents 8ca9f5b17257
children fde8fb2cbede
line wrap: on
line diff
--- a/mercurial/mdiff.py
+++ b/mercurial/mdiff.py
@@ -5,9 +5,13 @@
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
-import difflib, struct, bdiff, util, mpatch
+from demandload import demandload
+import struct, bdiff, util, mpatch
+demandload(globals(), "re")
 
-def unidiff(a, ad, b, bd, fn, r=None, text=False):
+
+def unidiff(a, ad, b, bd, fn, r=None, text=False,
+            showfunc=False, ignorews=False):
 
     if not a and not b: return ""
     epoch = util.datestr((0, 0))
@@ -27,9 +31,10 @@ def unidiff(a, ad, b, bd, fn, r=None, te
         l3 = "@@ -1,%d +0,0 @@\n" % len(a)
         l = [l1, l2, l3] + ["-" + e for e in a]
     else:
-        a = a.splitlines(1)
-        b = b.splitlines(1)
-        l = list(difflib.unified_diff(a, b, "a/" + fn, "b/" + fn))
+        al = a.splitlines(1)
+        bl = b.splitlines(1)
+        l = list(bunidiff(a, b, al, bl, "a/" + fn, "b/" + fn,
+                          showfunc=showfunc, ignorews=ignorews))
         if not l: return ""
         # difflib uses a space, rather than a tab
         l[0] = "%s\t%s\n" % (l[0][:-2], ad)
@@ -45,6 +50,128 @@ def unidiff(a, ad, b, bd, fn, r=None, te
 
     return "".join(l)
 
+# somewhat self contained replacement for difflib.unified_diff
+# t1 and t2 are the text to be diffed
+# l1 and l2 are the text broken up into lines
+# header1 and header2 are the filenames for the diff output
+# context is the number of context lines
+# showfunc enables diff -p output
+# ignorews ignores all whitespace changes in the diff
+def bunidiff(t1, t2, l1, l2, header1, header2, context=3, showfunc=False,
+             ignorews=False):
+    def contextend(l, len):
+        ret = l + context
+        if ret > len:
+            ret = len
+        return ret
+
+    def contextstart(l):
+        ret = l - context
+        if ret < 0:
+            return 0
+        return ret
+
+    def yieldhunk(hunk, header):
+        if header:
+            for x in header:
+                yield x
+        (astart, a2, bstart, b2, delta) = hunk
+        aend = contextend(a2, len(l1))
+        alen = aend - astart
+        blen = b2 - bstart + aend - a2
+
+        func = ""
+        if showfunc:
+            # walk backwards from the start of the context
+            # to find a line starting with an alphanumeric char.
+            for x in xrange(astart, -1, -1):
+                t = l1[x].rstrip()
+                if funcre.match(t):
+                    func = ' ' + t[:40]
+                    break
+
+        yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen,
+                                           bstart + 1, blen, func)
+        for x in delta:
+            yield x
+        for x in xrange(a2, aend):
+            yield ' ' + l1[x]
+
+    header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ]
+
+    if showfunc:
+        funcre = re.compile('\w')
+    if ignorews:
+        wsre = re.compile('[ \t]')
+
+    # 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.
+    #
+    diff = bdiff.blocks(t1, t2)
+    hunk = None
+    for i in xrange(len(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]
+        delta = []
+        s1 = diff[i]
+        a1 = s[1]
+        a2 = s1[0]
+        b1 = s[3]
+        b2 = s1[2]
+
+        old = l1[a1:a2]
+        new = l2[b1:b2]
+
+        # bdiff sometimes gives huge matches past eof, this check eats them,
+        # and deals with the special first match case described above
+        if not old and not new:
+            continue
+
+        if ignorews:
+            wsold = wsre.sub('', "".join(old))
+            wsnew = wsre.sub('', "".join(new))
+            if wsold == wsnew:
+                continue
+
+        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] + context + 1:
+                prev = hunk
+                astart = hunk[1]
+                bstart = hunk[3]
+            else:
+                for x in yieldhunk(hunk, header):
+                    yield x
+                # we only want to yield the header if the files differ, and
+                # we only want to yield it once.
+                header = None
+        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:
+        for x in yieldhunk(hunk, header):
+            yield x
+
 def patchtext(bin):
     pos = 0
     t = []