Files
git/contrib/diff-highlight/DiffHighlight.pm
Jeff King 6689a6ea49 diff-highlight: fetch all config with one process
When diff-highlight was written, there was no way to fetch multiple
config keys _and_ have them interpreted as colors. So we were stuck
with either invoking git-config once for each config key, or fetching
them all and converting human-readable color names into ANSI codes
ourselves.

I chose the former, but it means that diff-highlight kicks off 6
git-config processes (even if you haven't configured anything, it has to
check each one).

But since Git 2.18.0, we can do:

   git config --type=color --get-regexp=^color\.diff-highlight\.

to get all of them in one shot.

Note that any callers which pass in colors directly to the module via
@OLD_HIGHLIGHT and @NEW_HIGHLIGHT (like diff-so-fancy plans to do) are
unaffected; those colors suppress any config lookup we'd do ourselves.

You can see the effect like:

  # diff-highlight suppresses git-config's stderr, so dump
  # trace through descriptor 3
  git show d1f33c753d | GIT_TRACE=3 diff-highlight 3>&2 >/dev/null

which drops from 6 lines down to 1.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2026-03-23 07:42:27 -07:00

327 lines
8.2 KiB
Perl

package DiffHighlight;
require v5.008;
use warnings FATAL => 'all';
use strict;
# Use the correct value for both UNIX and Windows (/dev/null vs nul)
use File::Spec;
my $NULL = File::Spec->devnull();
# The color theme is initially set to nothing here to allow outside callers
# to set the colors for their application. If nothing is sent in we use
# colors from git config in load_color_config().
our @OLD_HIGHLIGHT = ();
our @NEW_HIGHLIGHT = ();
my $RESET = "\x1b[m";
my $COLOR = qr/\x1b\[[0-9;]*m/;
my $BORING = qr/$COLOR|\s/;
my @removed;
my @added;
my $in_hunk;
my $graph_indent = 0;
our $line_cb = sub { print @_ };
our $flush_cb = sub { local $| = 1 };
# Count the visible width of a string, excluding any terminal color sequences.
sub visible_width {
local $_ = shift;
my $ret = 0;
while (length) {
if (s/^$COLOR//) {
# skip colors
} elsif (s/^.//) {
$ret++;
}
}
return $ret;
}
# Return a substring of $str, omitting $len visible characters from the
# beginning, where terminal color sequences do not count as visible.
sub visible_substr {
my ($str, $len) = @_;
while ($len > 0) {
if ($str =~ s/^$COLOR//) {
next
}
$str =~ s/^.//;
$len--;
}
return $str;
}
sub handle_line {
my $orig = shift;
local $_ = $orig;
# match a graph line that begins a commit
if (/^(?:$COLOR?\|$COLOR?[ ])* # zero or more leading "|" with space
$COLOR?\*$COLOR?[ ] # a "*" with its trailing space
(?:$COLOR?\|$COLOR?[ ])* # zero or more trailing "|"
[ ]* # trailing whitespace for merges
/x) {
my $graph_prefix = $&;
# We must flush before setting graph indent, since the
# new commit may be indented differently from what we
# queued.
flush();
$graph_indent = visible_width($graph_prefix);
} elsif ($graph_indent) {
if (length($_) < $graph_indent) {
$graph_indent = 0;
} else {
$_ = visible_substr($_, $graph_indent);
}
}
if (!$in_hunk) {
$line_cb->($orig);
$in_hunk = /^$COLOR*\@\@ /;
}
elsif (/^$COLOR*-/) {
push @removed, $orig;
}
elsif (/^$COLOR*\+/) {
push @added, $orig;
}
else {
flush();
$line_cb->($orig);
$in_hunk = /^$COLOR*[\@ ]/;
}
# Most of the time there is enough output to keep things streaming,
# but for something like "git log -Sfoo", you can get one early
# commit and then many seconds of nothing. We want to show
# that one commit as soon as possible.
#
# Since we can receive arbitrary input, there's no optimal
# place to flush. Flushing on a blank line is a heuristic that
# happens to match git-log output.
if (/^$/) {
$flush_cb->();
}
}
sub flush {
# Flush any queued hunk (this can happen when there is no trailing
# context in the final diff of the input).
show_hunk(\@removed, \@added);
@removed = ();
@added = ();
}
sub highlight_stdin {
while (<STDIN>) {
handle_line($_);
}
flush();
}
# Ideally we would feed the default as a human-readable color to
# git-config as the fallback value. But diff-highlight does
# not otherwise depend on git at all, and there are reports
# of it being used in other settings. Let's handle our own
# fallback, which means we will work even if git can't be run.
sub color_config {
our $cached_config;
my ($key, $default) = @_;
if (!defined $cached_config) {
$cached_config = {};
my $data = `git config --type=color --get-regexp '^color\.diff-highlight\.' 2>$NULL`;
for my $line (split /\n/, $data) {
my ($key, $color) = split ' ', $line, 2;
$key =~ s/^color\.diff-highlight\.// or next;
$cached_config->{$key} = $color;
}
}
my $s = $cached_config->{$key};
return defined($s) ? $s : $default;
}
sub show_hunk {
my ($a, $b) = @_;
# If one side is empty, then there is nothing to compare or highlight.
if (!@$a || !@$b) {
$line_cb->(@$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) {
$line_cb->(@$a, @$b);
return;
}
my @queue;
for (my $i = 0; $i < @$a; $i++) {
my ($rm, $add) = highlight_pair($a->[$i], $b->[$i]);
$line_cb->($rm);
push @queue, $add;
}
$line_cb->(@queue);
}
sub load_color_config {
# If the colors were NOT set from outside this module we load them on-demand
# from the git config. Note that only one of elements 0 and 2 in each
# array is used (depending on whether you are doing set/unset on an
# attribute, or specifying normal vs highlighted coloring). So we use
# element 1 as our check for whether colors were passed in; it should
# always be set if you want highlighting to do anything.
if (!defined $OLD_HIGHLIGHT[1]) {
@OLD_HIGHLIGHT = (
color_config('oldnormal'),
color_config('oldhighlight', "\x1b[7m"),
color_config('oldreset', "\x1b[27m")
);
}
if (!defined $NEW_HIGHLIGHT[1]) {
@NEW_HIGHLIGHT = (
color_config('newnormal', $OLD_HIGHLIGHT[0]),
color_config('newhighlight', $OLD_HIGHLIGHT[1]),
color_config('newreset', $OLD_HIGHLIGHT[2])
);
};
}
sub highlight_pair {
my @a = split_line(shift);
my @b = split_line(shift);
# Find common prefix, taking care to skip any ansi
# color codes.
my $seen_plusminus;
my ($pa, $pb) = (0, 0);
while ($pa < @a && $pb < @b) {
if ($a[$pa] =~ /$COLOR/) {
$pa++;
}
elsif ($b[$pb] =~ /$COLOR/) {
$pb++;
}
elsif ($a[$pa] eq $b[$pb]) {
$pa++;
$pb++;
}
elsif (!$seen_plusminus && $a[$pa] eq '-' && $b[$pb] eq '+') {
$seen_plusminus = 1;
$pa++;
$pb++;
}
else {
last;
}
}
# Find common suffix, ignoring colors.
my ($sa, $sb) = ($#a, $#b);
while ($sa >= $pa && $sb >= $pb) {
if ($a[$sa] =~ /$COLOR/) {
$sa--;
}
elsif ($b[$sb] =~ /$COLOR/) {
$sb--;
}
elsif ($a[$sa] eq $b[$sb]) {
$sa--;
$sb--;
}
else {
last;
}
}
if (is_pair_interesting(\@a, $pa, $sa, \@b, $pb, $sb)) {
load_color_config();
return highlight_line(\@a, $pa, $sa, \@OLD_HIGHLIGHT),
highlight_line(\@b, $pb, $sb, \@NEW_HIGHLIGHT);
}
else {
return join('', @a),
join('', @b);
}
}
# we split either by $COLOR or by character. This has the side effect of
# leaving in graph cruft. It works because the graph cruft does not contain "-"
# or "+"
sub split_line {
local $_ = shift;
return utf8::decode($_) ?
map { utf8::encode($_); $_ }
map { /$COLOR/ ? $_ : (split //) }
split /($COLOR+)/ :
map { /$COLOR/ ? $_ : (split //) }
split /($COLOR+)/;
}
sub highlight_line {
my ($line, $prefix, $suffix, $theme) = @_;
my $start = join('', @{$line}[0..($prefix-1)]);
my $mid = join('', @{$line}[$prefix..$suffix]);
my $end = join('', @{$line}[($suffix+1)..$#$line]);
# If we have a "normal" color specified, then take over the whole line.
# Otherwise, we try to just manipulate the highlighted bits.
if (defined $theme->[0]) {
s/$COLOR//g for ($start, $mid, $end);
chomp $end;
return join('',
$theme->[0], $start, $RESET,
$theme->[1], $mid, $RESET,
$theme->[0], $end, $RESET,
"\n"
);
} else {
return join('',
$start,
$theme->[1], $mid, $theme->[2],
$end
);
}
}
# Pairs are interesting to highlight only if we are going to end up
# highlighting a subset (i.e., not the whole line). Otherwise, the highlighting
# is just useless noise. We can detect this by finding either a matching prefix
# or suffix (disregarding boring bits like whitespace and colorization).
sub is_pair_interesting {
my ($a, $pa, $sa, $b, $pb, $sb) = @_;
# We hit this case if the prefix consumed the entire line, meaning
# that two lines are identical. This generally shouldn't happen,
# since it implies the diff isn't minimal (you could shrink the hunk by
# making this a context line). But you can see it when the line
# content is the same, but the trailing newline is dropped, like:
#
# -foo
# +foo
# \No newline at end of file
return 0 if $pa == @$a || $pb == @$b;
my $prefix_a = join('', @$a[0..($pa-1)]);
my $prefix_b = join('', @$b[0..($pb-1)]);
my $suffix_a = join('', @$a[($sa+1)..$#$a]);
my $suffix_b = join('', @$b[($sb+1)..$#$b]);
return visible_substr($prefix_a, $graph_indent) !~ /^$COLOR*-$BORING*$/ ||
visible_substr($prefix_b, $graph_indent) !~ /^$COLOR*\+$BORING*$/ ||
$suffix_a !~ /^$BORING*$/ ||
$suffix_b !~ /^$BORING*$/;
}