-sub flush_diff ($$$$) {
- my ($dst, $spfx, $linkify, $diff) = @_;
- my $state = DSTATE_INIT;
- my $dctx = { Q => '' }; # {}, keys: oid_a, oid_b, path_a, path_b
-
- foreach my $s (@$diff) {
- if ($s =~ /^ /) {
- if ($state == DSTATE_HUNK || $state == DSTATE_ADD ||
- $state == DSTATE_DEL || $state == DSTATE_HEAD) {
- $$dst .= "</span><span\nclass=ctx>";
- $state = DSTATE_CTX;
- }
- $$dst .= to_html($linkify, $s);
- } elsif ($s =~ /^-- $/) { # email signature begins
- if ($state != DSTATE_INIT) {
- $state = DSTATE_INIT;
- $$dst .= '</span>';
- }
- $$dst .= $s;
- } elsif ($s =~ m!^diff --git ($PATH_A) ($PATH_B)$!) {
- if ($state != DSTATE_HEAD) {
- my ($pa, $pb) = ($1, $2);
- $$dst .= '</span>' if $state != DSTATE_INIT;
- $$dst .= "<span\nclass=head>";
- $state = DSTATE_HEAD;
- $pa = (split('/', git_unquote($pa), 2))[1];
- $pb = (split('/', git_unquote($pb), 2))[1];
- $dctx = {
- Q => "?b=".uri_escape_utf8($pb, UNSAFE),
- };
- if ($pa ne $pb) {
- $dctx->{Q} .=
- "&a=".uri_escape_utf8($pa, UNSAFE);
+# returns "diff --git" anchor destination, undef otherwise
+sub anchor1 ($$) {
+ my ($ctx, $pb) = @_;
+ my $attr = to_attr($ctx->{-apfx}.$pb) // return;
+
+ my $ok = delete $ctx->{-anchors}->{$attr};
+
+ # unlikely, check the end of long path names we captured,
+ # assume diffstat and diff output follow the same order,
+ # and ignore different ordering (could be malicious input)
+ unless ($ok) {
+ my $fn = shift(@{$ctx->{-long_path}}) // return;
+ $pb =~ /\Q$fn\E\z/s or return;
+ $attr = to_attr($ctx->{-apfx}.$fn) // return;
+ $ok = delete $ctx->{-anchors}->{$attr} // return;
+ }
+ $ok ? "<a\nhref=#i$attr\nid=$attr>diff</a> --git" : undef
+}
+
+sub diff_header ($$$) {
+ my ($x, $ctx, $top) = @_;
+ my (undef, undef, $pa, $pb) = splice(@$top, 0, 4); # ignore oid_{a,b}
+ my $spfx = $ctx->{-spfx};
+ my $dctx = { spfx => $spfx };
+
+ # get rid of leading "a/" or "b/" (or whatever --{src,dst}-prefix are)
+ $pa = (split(m'/', git_unquote($pa), 2))[1] if $pa ne '/dev/null';
+ $pb = (split(m'/', git_unquote($pb), 2))[1] if $pb ne '/dev/null';
+ if ($pa eq $pb && $pb ne '/dev/null') {
+ $dctx->{Q} = '?b='.uri_escape_path($pb);
+ } else {
+ my @q;
+ if ($pb ne '/dev/null') {
+ push @q, 'b='.uri_escape_path($pb);
+ }
+ if ($pa ne '/dev/null') {
+ push @q, 'a='.uri_escape_path($pa);
+ }
+ $dctx->{Q} = '?'.join('&', @q);
+ }
+
+ # linkify early and all at once, since we know the following
+ # subst ops on $$x won't need further escaping:
+ $$x = $ctx->{-linkify}->to_html($$x);
+
+ # no need to capture oid_a and oid_b on add/delete,
+ # we just linkify OIDs directly via s///e in conditional
+ if ($$x =~ s/$NULL_TO_BLOB/$1 . oid($dctx, $spfx, $2)/e) {
+ push @{$ctx->{-qry}->{dfpost}}, $2;
+ } elsif ($$x =~ s/$BLOB_TO_NULL/'index '.oid($dctx, $spfx, $1).$2/e) {
+ push @{$ctx->{-qry}->{dfpre}}, $1;
+ } elsif ($$x =~ $BLOB_TO_BLOB) {
+ # modification-only, not add/delete:
+ # linkify hunk headers later using oid_a and oid_b
+ @$dctx{qw(oid_a oid_b)} = ($1, $2);
+ push @{$ctx->{-qry}->{dfpre}}, $1;
+ push @{$ctx->{-qry}->{dfpost}}, $2;
+ } else {
+ warn "BUG? <$$x> had no ^index line";
+ }
+ $$x =~ s!^diff --git!anchor1($ctx, $pb) // 'diff --git'!ems;
+ print { $ctx->{zfh} } qq(<span\nclass="head">), $$x, '</span>';
+ $dctx;
+}
+
+sub diff_before_or_after ($$) {
+ my ($ctx, $x) = @_;
+ if (exists $ctx->{-anchors} && $$x =~ /\A(.*?) # likely "---\n" # \$1
+ # diffstat lines:
+ ((?:^\x20(?:[^\n]+?)(?:\x20+\|\x20[^\n]*\n))+)
+ (\x20[0-9]+\x20files?\x20)changed,([^\n]+\n)
+ (.*?)\z/msx) { # notes, commit message, etc
+ my @x = ($5, $4, $3, $2, $1);
+ undef $$x;
+ my $lnk = $ctx->{-linkify};
+ my $zfh = $ctx->{zfh};
+ print $zfh $lnk->to_html(pop @x); # $1 uninteresting prefix
+ for my $l (split(/^/m, pop(@x))) { # $2 per-file stat lines
+ $l =~ /^ (.+)( +\| .*\z)/s and
+ anchor0($ctx, $1, $2) and next;
+ print $zfh $lnk->to_html($l);
+ }
+ my $ch = $ctx->{changed_href} // '#related';
+ print $zfh pop(@x), # $3 /^ \d+ files? /
+ qq(<a href="$ch">changed</a>,),
+ ascii_html(pop @x), # insertions/deletions
+ $lnk->to_html(@x); # notes, commit message, etc
+ } else {
+ print { $ctx->{zfh} } $ctx->{-linkify}->to_html($$x);
+ }
+}
+
+# callers must do CRLF => LF conversion before calling this
+sub flush_diff ($$) {
+ my ($ctx, $cur) = @_;
+
+ my @top = split($EXTRACT_DIFFS, $$cur);
+ undef $$cur; # free memory
+
+ my $lnk = $ctx->{-linkify};
+ my $dctx; # {}, keys: Q, oid_a, oid_b
+ my $zfh = $ctx->zfh;
+
+ while (defined(my $x = shift @top)) {
+ if (scalar(@top) >= 4 &&
+ $top[1] =~ $IS_OID &&
+ $top[0] =~ $IS_OID) {
+ $dctx = diff_header(\$x, $ctx, \@top);
+ } elsif ($dctx) {
+ open(my $afh, '>>', \(my $after='')) or die "open: $!";
+
+ # Quiet "Complex regular subexpression recursion limit"
+ # warning. Perl will truncate matches upon hitting
+ # that limit, giving us more (and shorter) scalars than
+ # would be ideal, but otherwise it's harmless.
+ #
+ # We could replace the `+' metacharacter with `{1,100}'
+ # to limit the matches ourselves to 100, but we can
+ # let Perl do it for us, quietly.
+ no warnings 'regexp';
+
+ for my $s (split(/((?:(?:^\+[^\n]*\n)+)|
+ (?:(?:^-[^\n]*\n)+)|
+ (?:^@@ [^\n]+\n))/xsm, $x)) {
+ if (!defined($dctx)) {
+ print $afh $s;
+ } elsif ($s =~ s/\A@@ (\S+) (\S+) @@//) {
+ print $zfh qq(<span\nclass="hunk">),
+ diff_hunk($dctx, $1, $2),
+ $lnk->to_html($s),
+ '</span>';
+ } elsif ($s =~ /\A\+/) { # $s may be huge
+ print $zfh qq(<span\nclass="add">),
+ $lnk->to_html($s),
+ '</span>';
+ } elsif ($s =~ /\A-- $/sm) { # email sig starts
+ $dctx = undef;
+ print $afh $s;
+ } elsif ($s =~ /\A-/) { # $s may be huge
+ print $zfh qq(<span\nclass="del">),
+ $lnk->to_html($s),
+ '</span>';
+ } else { # $s may be huge
+ print $zfh $lnk->to_html($s);