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 from_attr);
16 use PublicInbox::Git qw(git_unquote);
18 sub DSTATE_INIT () { 0 }
19 sub DSTATE_STAT () { 1 }
20 sub DSTATE_HEAD () { 2 } # /^diff --git /, /^index /, /^--- /, /^\+\+\+ /
21 sub DSTATE_CTX () { 3 } # /^ /
22 sub DSTATE_ADD () { 4 } # /^\+/
23 sub DSTATE_DEL () { 5 } # /^\-/
33 sub UNSAFE () { "^A-Za-z0-9\-\._~/" }
35 my $OID_NULL = '0{7,40}';
36 my $OID_BLOB = '[a-f0-9]{7,40}';
37 my $PATH_A = '"?a/.+|/dev/null';
38 my $PATH_B = '"?b/.+|/dev/null';
41 $_[0]->linkify_1($_[1]);
42 $_[0]->linkify_2(ascii_html($_[1]));
45 # link to line numbers in blobs
46 sub diff_hunk ($$$$) {
47 my ($dctx, $spfx, $ca, $cb) = @_;
48 my $oid_a = $dctx->{oid_a};
49 my $oid_b = $dctx->{oid_b};
51 (defined($spfx) && defined($oid_a) && defined($oid_b)) or
52 return "@@ $ca $cb @@";
54 my ($n) = ($ca =~ /^-(\d+)/);
55 $n = defined($n) ? do { ++$n; "#n$n" } : '';
57 my $rv = qq(@@ <a\nhref="$spfx$oid_a/s/$dctx->{Q}$n">$ca</a>);
59 ($n) = ($cb =~ /^\+(\d+)/);
60 $n = defined($n) ? do { ++$n; "#n$n" } : '';
62 $rv .= qq( <a\nhref="$spfx$oid_b/s/$dctx->{Q}$n">$cb</a> @@);
66 my ($dctx, $spfx, $oid) = @_;
67 defined($spfx) ? qq(<a\nhref="$spfx$oid/s/$dctx->{Q}">$oid</a>) : $oid;
71 my ($dst, $state, $new_state) = @_;
72 $$dst .= '</span>' if $state2class[$state];
74 my $class = $state2class[$new_state] or return;
75 $$dst .= qq(<span\nclass="$class">);
79 my ($dst, $ctx, $linkify, $fn, $rest) = @_;
83 # normal git diffstat output is impossible to parse reliably
84 # without --numstat, and that isn't the default for format-patch.
85 # So only do best-effort handling of renames for common cases;
86 # which works well in practice. If projects put "=>", or trailing
87 # spaces in filenames, oh well :P
89 $fn =~ s/{(?:.+) => (.+)}/$1/ or $fn =~ s/.* => (.+)/$1/;
90 $fn = git_unquote($fn);
92 # long filenames will require us to walk backwards in anchor1
93 if ($fn =~ s!\A\.\.\./?!!) {
94 my $lp = $ctx->{-long_path} ||= {};
95 $lp->{$fn} = qr/\Q$fn\E\z/s;
98 if (my $attr = to_attr($ctx->{-apfx}.$fn)) {
99 $ctx->{-anchors}->{$attr} = 1;
100 $$dst .= " <a\nid=i$attr\nhref=#$attr>" .
101 ascii_html($orig) . '</a>'.
102 to_html($linkify, $rest);
108 sub anchor1 ($$$$$) {
109 my ($dst, $ctx, $linkify, $pb, $s) = @_;
110 my $attr = to_attr($ctx->{-apfx}.$pb) or return;
111 my $line = to_html($linkify, $s);
113 my $ok = delete $ctx->{-anchors}->{$attr};
115 # unlikely, check the end of all long path names we captured:
117 my $lp = $ctx->{-long_path} or return;
118 foreach my $fn (keys %$lp) {
119 $pb =~ $lp->{$fn} or next;
122 $attr = to_attr($ctx->{-apfx}.$fn) or return;
123 $ok = delete $ctx->{-anchors}->{$attr} or return;
127 if ($ok && $line =~ s/^diff //) {
128 $$dst .= "<a\nhref=#i$attr\nid=$attr>diff</a> ".$line;
134 sub flush_diff ($$$) {
135 my ($dst, $ctx, $linkify) = @_;
136 my $diff = $ctx->{-diff};
137 my $spfx = $ctx->{-spfx};
138 my $state = DSTATE_INIT;
139 my $dctx = { Q => '' }; # {}, keys: oid_a, oid_b, path_a, path_b
141 foreach my $s (@$diff) {
143 to_state($dst, $state, DSTATE_STAT);
145 } elsif ($s =~ /^ /) {
146 # works for common cases, but not weird/long filenames
147 if ($state == DSTATE_STAT &&
148 $s =~ /^ (.+)( +\| .*\z)/s) {
149 anchor0($dst, $ctx, $linkify, $1, $2) and next;
150 } elsif ($state2class[$state]) {
151 to_state($dst, $state, DSTATE_CTX);
153 $$dst .= to_html($linkify, $s);
154 } elsif ($s =~ /^-- $/) { # email signature begins
155 $state == DSTATE_INIT or
156 to_state($dst, $state, DSTATE_INIT);
158 } elsif ($s =~ m!^diff --git ($PATH_A) ($PATH_B)$!) {
159 my ($pa, $pb) = ($1, $2);
160 if ($state != DSTATE_HEAD) {
161 to_state($dst, $state, DSTATE_HEAD);
163 $pa = (split('/', git_unquote($pa), 2))[1];
164 $pb = (split('/', git_unquote($pb), 2))[1];
166 Q => "?b=".uri_escape_utf8($pb, UNSAFE),
169 $dctx->{Q} .= '&a='.
170 uri_escape_utf8($pa, UNSAFE);
172 anchor1($dst, $ctx, $linkify, $pb, $s) and next;
173 $$dst .= to_html($linkify, $s);
174 } elsif ($s =~ s/^(index $OID_NULL\.\.)($OID_BLOB)\b//o) {
175 $$dst .= $1 . oid($dctx, $spfx, $2);
177 $$dst .= to_html($linkify, $s) ;
178 } elsif ($s =~ s/^index ($OID_BLOB)(\.\.$OID_NULL)\b//o) {
179 $$dst .= 'index ' . oid($dctx, $spfx, $1) . $2;
181 $$dst .= to_html($linkify, $s);
182 } elsif ($s =~ /^index ($OID_BLOB)\.\.($OID_BLOB)/o) {
185 $$dst .= to_html($linkify, $s);
186 } elsif ($s =~ s/^@@ (\S+) (\S+) @@//) {
187 $$dst .= '</span>' if $state2class[$state];
188 $$dst .= qq(<span\nclass="hunk">);
189 $$dst .= diff_hunk($dctx, $spfx, $1, $2);
192 $$dst .= to_html($linkify, $s);
193 } elsif ($s =~ m!^--- $PATH_A! ||
194 $s =~ m!^\+{3} $PATH_B!) {
195 # color only (no oid link) if missing dctx->{oid_*}
196 $state <= DSTATE_STAT and
197 to_state($dst, $state, DSTATE_HEAD);
198 $$dst .= to_html($linkify, $s);
199 } elsif ($s =~ /^\+/) {
200 if ($state != DSTATE_ADD && $state > DSTATE_STAT) {
201 to_state($dst, $state, DSTATE_ADD);
203 $$dst .= to_html($linkify, $s);
204 } elsif ($s =~ /^-/) {
205 if ($state != DSTATE_DEL && $state > DSTATE_STAT) {
206 to_state($dst, $state, DSTATE_DEL);
208 $$dst .= to_html($linkify, $s);
209 # ignore the following lines in headers:
210 } elsif ($s =~ /^(?:dis)similarity index/ ||
211 $s =~ /^(?:old|new) mode/ ||
212 $s =~ /^(?:deleted|new) file mode/ ||
213 $s =~ /^(?:copy|rename) (?:from|to) / ||
214 $s =~ /^(?:dis)?similarity index /) {
215 $$dst .= to_html($linkify, $s);
217 $state <= DSTATE_STAT or
218 to_state($dst, $state, DSTATE_INIT);
219 $$dst .= to_html($linkify, $s);
223 $$dst .= '</span>' if $state2class[$state];