1 # Copyright (C) 2019 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;
12 use base qw(Exporter);
13 our @EXPORT_OK = qw(flush_diff);
14 use URI::Escape qw(uri_escape_utf8);
15 use PublicInbox::Hval qw(ascii_html to_attr);
16 use PublicInbox::Git qw(git_unquote);
18 # keep track of state so we can avoid redundant HTML tags for
19 # identically-classed lines
20 sub DSTATE_INIT () { 0 }
21 sub DSTATE_STAT () { 1 }
22 sub DSTATE_HEAD () { 2 } # /^diff --git /, /^index /, /^--- /, /^\+\+\+ /
23 sub DSTATE_CTX () { 3 } # /^ /
24 sub DSTATE_ADD () { 4 } # /^\+/
25 sub DSTATE_DEL () { 5 } # /^\-/
27 # maps the DSTATE_* to CSS class names compatible with what cgit uses:
37 sub UNSAFE () { "^A-Za-z0-9\-\._~/" }
39 my $OID_NULL = '0{7,40}';
40 my $OID_BLOB = '[a-f0-9]{7,40}';
41 my $PATH_A = '"?a/.+|/dev/null';
42 my $PATH_B = '"?b/.+|/dev/null';
44 # cf. git diff.c :: get_compact_summary
45 my $DIFFSTAT_COMMENT = qr/\((?:new|gone|(?:(?:new|mode) [\+\-][lx]))\)/;
47 # link to line numbers in blobs
48 sub diff_hunk ($$$$) {
49 my ($dctx, $spfx, $ca, $cb) = @_;
50 my $oid_a = $dctx->{oid_a};
51 my $oid_b = $dctx->{oid_b};
53 (defined($spfx) && defined($oid_a) && defined($oid_b)) or
54 return "@@ $ca $cb @@";
56 my ($n) = ($ca =~ /^-([0-9]+)/);
57 $n = defined($n) ? do { ++$n; "#n$n" } : '';
59 my $rv = qq(@@ <a\nhref="$spfx$oid_a/s/$dctx->{Q}$n">$ca</a>);
61 ($n) = ($cb =~ /^\+([0-9]+)/);
62 $n = defined($n) ? do { ++$n; "#n$n" } : '';
64 $rv .= qq( <a\nhref="$spfx$oid_b/s/$dctx->{Q}$n">$cb</a> @@);
68 my ($dctx, $spfx, $oid) = @_;
69 defined($spfx) ? qq(<a\nhref="$spfx$oid/s/$dctx->{Q}">$oid</a>) : $oid;
73 my ($dst, $state, $new_state) = @_;
74 $$dst .= '</span>' if $state2class[$state];
76 my $class = $state2class[$new_state] or return;
77 $$dst .= qq(<span\nclass="$class">);
81 my ($dst, $ctx, $linkify, $fn, $rest) = @_;
85 # normal git diffstat output is impossible to parse reliably
86 # without --numstat, and that isn't the default for format-patch.
87 # So only do best-effort handling of renames for common cases;
88 # which works well in practice. If projects put "=>", or trailing
89 # spaces in filenames, oh well :P
90 $fn =~ s/(?: *$DIFFSTAT_COMMENT)? *\z//so;
91 $fn =~ s/{(?:.+) => (.+)}/$1/ or $fn =~ s/.* => (.+)/$1/;
92 $fn = git_unquote($fn);
94 # long filenames will require us to walk backwards in anchor1
95 if ($fn =~ s!\A\.\.\./?!!) {
96 my $lp = $ctx->{-long_path} ||= {};
97 $lp->{$fn} = qr/\Q$fn\E\z/s;
100 if (my $attr = to_attr($ctx->{-apfx}.$fn)) {
101 $ctx->{-anchors}->{$attr} = 1;
102 my $spaces = ($orig =~ s/( +)\z//) ? $1 : '';
103 $$dst .= " <a\nid=i$attr\nhref=#$attr>" .
104 ascii_html($orig) . '</a>' . $spaces .
105 $linkify->to_html($rest);
111 sub anchor1 ($$$$$) {
112 my ($dst, $ctx, $linkify, $pb, $s) = @_;
113 my $attr = to_attr($ctx->{-apfx}.$pb) or return;
114 my $line = $linkify->to_html($s);
116 my $ok = delete $ctx->{-anchors}->{$attr};
118 # unlikely, check the end of all long path names we captured:
120 my $lp = $ctx->{-long_path} or return;
121 foreach my $fn (keys %$lp) {
122 $pb =~ $lp->{$fn} or next;
125 $attr = to_attr($ctx->{-apfx}.$fn) or return;
126 $ok = delete $ctx->{-anchors}->{$attr} or return;
130 if ($ok && $line =~ s/^diff //) {
131 $$dst .= "<a\nhref=#i$attr\nid=$attr>diff</a> ".$line;
137 sub flush_diff ($$$) {
138 my ($dst, $ctx, $linkify) = @_;
139 my $diff = $ctx->{-diff};
140 my $spfx = $ctx->{-spfx};
141 my $state = DSTATE_INIT;
142 my $dctx = { Q => '' }; # {}, keys: oid_a, oid_b, path_a, path_b
144 foreach my $s (@$diff) {
146 to_state($dst, $state, DSTATE_STAT);
148 } elsif ($s =~ /^ / || ($s =~ /^$/ && $state >= DSTATE_CTX)) {
149 # works for common cases, but not weird/long filenames
150 if ($state == DSTATE_STAT &&
151 $s =~ /^ (.+)( +\| .*\z)/s) {
152 anchor0($dst, $ctx, $linkify, $1, $2) and next;
153 } elsif ($state2class[$state]) {
154 to_state($dst, $state, DSTATE_CTX);
156 $$dst .= $linkify->to_html($s);
157 } elsif ($s =~ /^-- $/) { # email signature begins
158 $state == DSTATE_INIT or
159 to_state($dst, $state, DSTATE_INIT);
161 } elsif ($s =~ m!^diff --git ($PATH_A) ($PATH_B)$!o) {
162 my ($pa, $pb) = ($1, $2);
163 if ($state != DSTATE_HEAD) {
164 to_state($dst, $state, DSTATE_HEAD);
166 $pa = (split('/', git_unquote($pa), 2))[1];
167 $pb = (split('/', git_unquote($pb), 2))[1];
169 Q => "?b=".uri_escape_utf8($pb, UNSAFE),
172 $dctx->{Q} .= '&a='.
173 uri_escape_utf8($pa, UNSAFE);
175 anchor1($dst, $ctx, $linkify, $pb, $s) and next;
176 $$dst .= $linkify->to_html($s);
177 } elsif ($s =~ s/^(index $OID_NULL\.\.)($OID_BLOB)\b//o) {
178 $$dst .= $1 . oid($dctx, $spfx, $2);
180 $$dst .= $linkify->to_html($s) ;
181 } elsif ($s =~ s/^index ($OID_BLOB)(\.\.$OID_NULL)\b//o) {
182 $$dst .= 'index ' . oid($dctx, $spfx, $1) . $2;
184 $$dst .= $linkify->to_html($s);
185 } elsif ($s =~ /^index ($OID_BLOB)\.\.($OID_BLOB)/o) {
188 $$dst .= $linkify->to_html($s);
189 } elsif ($s =~ s/^@@ (\S+) (\S+) @@//) {
190 $$dst .= '</span>' if $state2class[$state];
191 $$dst .= qq(<span\nclass="hunk">);
192 $$dst .= diff_hunk($dctx, $spfx, $1, $2);
195 $$dst .= $linkify->to_html($s);
196 } elsif ($s =~ m!^--- (?:$PATH_A)!o ||
197 $s =~ m!^\+{3} (?:$PATH_B)!o) {
198 # color only (no oid link) if missing dctx->{oid_*}
199 $state <= DSTATE_STAT and
200 to_state($dst, $state, DSTATE_HEAD);
201 $$dst .= $linkify->to_html($s);
202 } elsif ($s =~ /^\+/) {
203 if ($state != DSTATE_ADD && $state > DSTATE_STAT) {
204 to_state($dst, $state, DSTATE_ADD);
206 $$dst .= $linkify->to_html($s);
207 } elsif ($s =~ /^-/) {
208 if ($state != DSTATE_DEL && $state > DSTATE_STAT) {
209 to_state($dst, $state, DSTATE_DEL);
211 $$dst .= $linkify->to_html($s);
212 # ignore the following lines in headers:
213 } elsif ($s =~ /^(?:dis)similarity index/ ||
214 $s =~ /^(?:old|new) mode/ ||
215 $s =~ /^(?:deleted|new) file mode/ ||
216 $s =~ /^(?:copy|rename) (?:from|to) / ||
217 $s =~ /^(?:dis)?similarity index /) {
218 $$dst .= $linkify->to_html($s);
220 $state <= DSTATE_STAT or
221 to_state($dst, $state, DSTATE_INIT);
222 $$dst .= $linkify->to_html($s);
226 $$dst .= '</span>' if $state2class[$state];