diff-highlight: match multi-line hunks

Currently we only bother highlighting single-line hunks. The
rationale was that the purpose of highlighting is to point
out small changes between two similar lines that are
otherwise hard to see. However, that meant we missed similar
cases where two lines were changed together, like:

   -foo(buf);
   -bar(buf);
   +foo(obj->buf);
   +bar(obj->buf);

Each of those changes is simple, and would benefit from
highlighting (the "obj->" parts in this case).

This patch considers whole hunks at a time. For now, we
consider only the case where the hunk has the same number of
removed and added lines, and assume that the lines from each
segment correspond one-to-one. While this is just a
heuristic, in practice it seems to generate sensible
results (especially because we now omit highlighting on
completely-changed lines, so when our heuristic is wrong, we
tend to avoid highlighting at all).

Based on an original idea and implementation by Michał
Kiedrowicz.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

2 files changed, 52 insertions, 34 deletions

diff --git a/contrib/diff-highlight/README b/contrib/diff-highlight/README
index 1b7b6df8eb..4a58579779 100644
--- a/contrib/diff-highlight/README
+++ b/contrib/diff-highlight/README
@@ -14,13 +14,15 @@ Instead, this script post-processes the line-oriented diff, finds pairs
 of lines, and highlights the differing segments.  It's currently very
 simple and stupid about doing these tasks. In particular:
 
-  1. It will only highlight a pair of lines if they are the only two
-     lines in a hunk.  It could instead try to match up "before" and
-     "after" lines for a given hunk into pairs of similar lines.
-     However, this may end up visually distracting, as the paired
-     lines would have other highlighted lines in between them. And in
-     practice, the lines which most need attention called to their
-     small, hard-to-see changes are touching only a single line.
+  1. It will only highlight hunks in which the number of removed and
+     added lines is the same, and it will pair lines within the hunk by
+     position (so the first removed line is compared to the first added
+     line, and so forth). This is simple and tends to work well in
+     practice. More complex changes don't highlight well, so we tend to
+     exclude them due to the "same number of removed and added lines"
+     restriction. Or even if we do try to highlight them, they end up
+     not highlighting because of our "don't highlight if the whole line
+     would be highlighted" rule.
 
   2. It will find the common prefix and suffix of two lines, and
      consider everything in the middle to be "different". It could
diff --git a/contrib/diff-highlight/diff-highlight b/contrib/diff-highlight/diff-highlight
index 279d21181e..c4404d49c9 100755
--- a/contrib/diff-highlight/diff-highlight
+++ b/contrib/diff-highlight/diff-highlight
@@ -10,23 +10,28 @@ my $UNHIGHLIGHT = "\x1b[27m";
 my $COLOR = qr/\x1b\[[0-9;]*m/;
 my $BORING = qr/$COLOR|\s/;
 
-my @window;
+my @removed;
+my @added;
+my $in_hunk;
 
 while (<>) {
-	# We highlight only single-line changes, so we need
-	# a 4-line window to make a decision on whether
-	# to highlight.
-	push @window, $_;
-	next if @window < 4;
-	if ($window[0] =~ /^$COLOR*(\@| )/ &&
-	    $window[1] =~ /^$COLOR*-/ &&
-	    $window[2] =~ /^$COLOR*\+/ &&
-	    $window[3] !~ /^$COLOR*\+/) {
-		print shift @window;
-		show_hunk(shift @window, shift @window);
+	if (!$in_hunk) {
+		print;
+		$in_hunk = /^$COLOR*\@/;
+	}
+	elsif (/^$COLOR*-/) {
+		push @removed, $_;
+	}
+	elsif (/^$COLOR*\+/) {
+		push @added, $_;
 	}
 	else {
-		print shift @window;
+		show_hunk(\@removed, \@added);
+		@removed = ();
+		@added = ();
+
+		print;
+		$in_hunk = /^$COLOR*[\@ ]/;
 	}
 
 	# Most of the time there is enough output to keep things streaming,
@@ -42,26 +47,37 @@ while (<>) {
 	}
 }
 
-# Special case a single-line hunk at the end of file.
-if (@window == 3 &&
-    $window[0] =~ /^$COLOR*(\@| )/ &&
-    $window[1] =~ /^$COLOR*-/ &&
-    $window[2] =~ /^$COLOR*\+/) {
-	print shift @window;
-	show_hunk(shift @window, shift @window);
-}
-
-# And then flush any remaining lines.
-while (@window) {
-	print shift @window;
-}
+# Flush any queued hunk (this can happen when there is no trailing context in
+# the final diff of the input).
+show_hunk(\@removed, \@added);
 
 exit 0;
 
 sub show_hunk {
 	my ($a, $b) = @_;
 
-	print highlight_pair($a, $b);
+	# If one side is empty, then there is nothing to compare or highlight.
+	if (!@$a || !@$b) {
+		print @$a, @$b;
+		return;
+	}
+
+	# If we have mismatched numbers of lines on each side, we could try to
+	# be clever and match up similar lines. But for now we are simple and
+	# stupid, and only handle multi-line hunks that remove and add the same
+	# number of lines.
+	if (@$a != @$b) {
+		print @$a, @$b;
+		return;
+	}
+
+	my @queue;
+	for (my $i = 0; $i < @$a; $i++) {
+		my ($rm, $add) = highlight_pair($a->[$i], $b->[$i]);
+		print $rm;
+		push @queue, $add;
+	}
+	print @queue;
 }
 
 sub highlight_pair {