1 # Copyright (C) 2019-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # used by PublicInbox::View
5 # This adds CSS spans for diff highlighting.
6 # It also generates links for ViewVCS + SolverGit to show
7 # (or reconstruct) blobs.
9 package PublicInbox::ViewDiff;
13 use base qw(Exporter);
14 our @EXPORT_OK = qw(flush_diff);
15 use URI::Escape qw(uri_escape_utf8);
16 use PublicInbox::Hval qw(ascii_html to_attr);
17 use PublicInbox::Git qw(git_unquote);
19 sub UNSAFE () { "^A-Za-z0-9\-\._~/" }
21 my $OID_NULL = '0{7,40}';
22 my $OID_BLOB = '[a-f0-9]{7,40}';
24 # cf. git diff.c :: get_compact_summary
25 my $DIFFSTAT_COMMENT = qr/\((?:new|gone|(?:(?:new|mode) [\+\-][lx]))\)/;
27 # link to line numbers in blobs
28 sub diff_hunk ($$$$) {
29 my ($dst, $dctx, $ca, $cb) = @_;
30 my ($oid_a, $oid_b, $spfx) = @$dctx{qw(oid_a oid_b spfx)};
32 if (defined($spfx) && defined($oid_a) && defined($oid_b)) {
33 my ($n) = ($ca =~ /^-([0-9]+)/);
34 $n = defined($n) ? do { ++$n; "#n$n" } : '';
36 $$dst .= qq(@@ <a\nhref="$spfx$oid_a/s/$dctx->{Q}$n">$ca</a>);
38 ($n) = ($cb =~ /^\+([0-9]+)/);
39 $n = defined($n) ? do { ++$n; "#n$n" } : '';
40 $$dst .= qq( <a\nhref="$spfx$oid_b/s/$dctx->{Q}$n">$cb</a> @@);
42 $$dst .= "@@ $ca $cb @@";
47 my ($dctx, $spfx, $oid) = @_;
48 defined($spfx) ? qq(<a\nhref="$spfx$oid/s/$dctx->{Q}">$oid</a>) : $oid;
51 # returns true if diffstat anchor written, false otherwise
53 my ($dst, $ctx, $fn, $rest) = @_;
57 # normal git diffstat output is impossible to parse reliably
58 # without --numstat, and that isn't the default for format-patch.
59 # So only do best-effort handling of renames for common cases;
60 # which works well in practice. If projects put "=>", or trailing
61 # spaces in filenames, oh well :P
62 $fn =~ s/(?: *$DIFFSTAT_COMMENT)? *\z//so;
63 $fn =~ s/{(?:.+) => (.+)}/$1/ or $fn =~ s/.* => (.+)/$1/;
64 $fn = git_unquote($fn);
66 # long filenames will require us to walk backwards in anchor1
67 if ($fn =~ s!\A\.\.\./?!!) {
68 $ctx->{-long_path}->{$fn} = qr/\Q$fn\E\z/s;
71 if (my $attr = to_attr($ctx->{-apfx}.$fn)) {
72 $ctx->{-anchors}->{$attr} = 1;
73 my $spaces = ($orig =~ s/( +)\z//) ? $1 : '';
74 $$dst .= " <a\nid=i$attr\nhref=#$attr>" .
75 ascii_html($orig) . '</a>' . $spaces .
76 $ctx->{-linkify}->to_html($rest);
82 # returns "diff --git" anchor destination, undef otherwise
85 my $attr = to_attr($ctx->{-apfx}.$pb) or return;
87 my $ok = delete $ctx->{-anchors}->{$attr};
89 # unlikely, check the end of all long path names we captured:
91 my $lp = $ctx->{-long_path} or return;
92 foreach my $fn (keys %$lp) {
93 $pb =~ $lp->{$fn} or next;
96 $attr = to_attr($ctx->{-apfx}.$fn) or return;
97 $ok = delete $ctx->{-anchors}->{$attr} or return;
101 $ok ? "<a\nhref=#i$attr\nid=$attr>diff</a> --git" : undef
104 sub diff_header ($$$$) {
105 my ($dst, $x, $ctx, $top) = @_;
106 my (undef, undef, $pa, $pb) = splice(@$top, 0, 4); # ignore oid_{a,b}
107 my $spfx = $ctx->{-spfx};
108 my $dctx = { spfx => $spfx };
110 # get rid of leading "a/" or "b/" (or whatever --{src,dst}-prefix are)
111 $pa = (split('/', git_unquote($pa), 2))[1] if $pa ne '/dev/null';
112 $pb = (split('/', git_unquote($pb), 2))[1] if $pb ne '/dev/null';
113 if ($pa eq $pb && $pb ne '/dev/null') {
114 $dctx->{Q} = "?b=".uri_escape_utf8($pb, UNSAFE);
117 if ($pb ne '/dev/null') {
118 push @q, 'b='.uri_escape_utf8($pb, UNSAFE);
120 if ($pa ne '/dev/null') {
121 push @q, 'a='.uri_escape_utf8($pa, UNSAFE);
123 $dctx->{Q} = '?'.join('&', @q);
126 # linkify early and all at once, since we know the following
127 # subst ops on $$x won't need further escaping:
128 $$x = $ctx->{-linkify}->to_html($$x);
130 # no need to capture oid_a and oid_b on add/delete,
131 # we just linkify OIDs directly via s///e in conditional
132 if (($$x =~ s/^(index $OID_NULL\.\.)($OID_BLOB)\b/
133 $1 . oid($dctx, $spfx, $2)/emos) ||
134 ($$x =~ s/^index ($OID_BLOB)(\.\.$OID_NULL)\b/
135 'index ' . oid($dctx, $spfx, $1) . $2/emos)) {
136 } elsif ($$x =~ /^index ($OID_BLOB)\.\.($OID_BLOB)/mos) {
137 # modification-only, not add/delete:
138 # linkify hunk headers later using oid_a and oid_b
139 @$dctx{qw(oid_a oid_b)} = ($1, $2);
141 warn "BUG? <$$x> had no ^index line";
143 $$x =~ s!^diff --git!anchor1($ctx, $pb) // 'diff --git'!emos;
144 $$dst .= qq(<span\nclass="head">);
150 sub diff_before_or_after ($$$) {
151 my ($dst, $ctx, $x) = @_;
152 my $linkify = $ctx->{-linkify};
153 for my $y (split(/(^---\n)/sm, $$x)) {
154 if ($y =~ /\A---\n\z/s) {
155 $$dst .= "---\n"; # all HTML is "\r\n" => "\n"
156 } elsif ($y =~ /^ [0-9]+ files? changed, /sm) {
157 # ok, looks like a diffstat, go line-by-line:
158 for my $l (split(/^/m, $y)) {
159 if ($l =~ /^ (.+)( +\| .*\z)/s) {
160 anchor0($dst, $ctx, $1, $2) and next;
162 $$dst .= $linkify->to_html($l);
164 } else { # commit message, notes, etc
165 $$dst .= $linkify->to_html($y);
170 # callers must do CRLF => LF conversion before calling this
171 sub flush_diff ($$$) {
172 my ($dst, $ctx, $cur) = @_;
174 state $ANY = qr![^\n]!;
175 state $FN = qr!(?:"?[^/\n]+/[^\n]+|/dev/null)!;
178 (?: # begin header stuff, don't capture filenames, here,
179 # but instead wait for the --- and +++ lines.
180 (?:^diff\x20--git\x20$FN\x20$FN$LF)
182 # old mode || new mode || copy|rename|deleted|...
184 )? # end of optional stuff, everything below is required
185 ^index\x20($OID_BLOB)\.\.($OID_BLOB)$ANY*$LF
187 ^\+{3}\x20($FN)$LF)/smxo, $$cur);
190 my $linkify = $ctx->{-linkify};
191 my $dctx; # {}, keys: Q, oid_a, oid_b
193 while (defined(my $x = shift @top)) {
194 if (scalar(@top) >= 4 &&
195 $top[1] =~ /\A$OID_BLOB\z/os &&
196 $top[0] =~ /\A$OID_BLOB\z/os) {
197 $dctx = diff_header($dst, \$x, $ctx, \@top);
200 for my $s (split(/((?:(?:^\+[^\n]*\n)+)|
202 (?:^@@ [^\n]+\n))/xsm, $x)) {
203 if (!defined($dctx)) {
205 } elsif ($s =~ s/\A@@ (\S+) (\S+) @@//) {
206 $$dst .= qq(<span\nclass="hunk">);
207 diff_hunk($dst, $dctx, $1, $2);
208 $$dst .= $linkify->to_html($s);
210 } elsif ($s =~ /\A\+/) {
211 $$dst .= qq(<span\nclass="add">);
212 $$dst .= $linkify->to_html($s);
214 } elsif ($s =~ /\A-- $/sm) { # email sig starts
217 } elsif ($s =~ /\A-/) {
218 $$dst .= qq(<span\nclass="del">);
219 $$dst .= $linkify->to_html($s);
222 $$dst .= $linkify->to_html($s);
225 diff_before_or_after($dst, $ctx, \$after) unless $dctx;
227 diff_before_or_after($dst, $ctx, \$x);