search: allow searching within mail diffs

This can be tied into a repository browser to browse
in-flight topics on a mailing list.

2 files changed, 148 insertions, 9 deletions

diff --git a/lib/PublicInbox/Search.pm b/lib/PublicInbox/Search.pm
index 67837f4f..c7c5455d 100644
--- a/lib/PublicInbox/Search.pm
+++ b/lib/PublicInbox/Search.pm
@@ -75,6 +75,14 @@ my %prob_prefix = (
 
         q => 'XQUOT',
         nq => 'XNQ',
+        dfn => 'XDFN',
+        dfa => 'XDFA',
+        dfb => 'XDFB',
+        dfhh => 'XDFHH',
+        dfctx => 'XDFCTX',
+        dfpre => 'XDFPRE',
+        dfpost => 'XDFPOST',
+        dfblob => 'XDFPRE XDFPOST',
 
         # default:
         '' => 'XMID S A XNQ XQUOT XFN',
@@ -98,12 +106,16 @@ EOF
         'a:' => 'match within the To, Cc, and From headers',
         'tc:' => 'match within the To and Cc headers',
         'bs:' => 'match within the Subject and body',
+        'dfn:' => 'match filename from diff',
+        'dfa:' => 'match diff removed (-) lines',
+        'dfb:' => 'match diff added (+) lines',
+        'dfhh:' => 'match diff hunk header context (usually a function name)',
+        'dfctx:' => 'match diff context lines',
+        'dfpre:' => 'match pre-image git blob ID',
+        'dfpost:' => 'match post-image git blob ID',
+        'dfblob:' => 'match either pre or post-image git blob ID',
 );
 chomp @HELP;
-# TODO:
-# df (filenames from diff)
-# da (diff a/ removed lines)
-# db (diff b/ added lines)
 
 my $mail_query = Search::Xapian::Query->new('T' . 'mail');
 
diff --git a/lib/PublicInbox/SearchIdx.pm b/lib/PublicInbox/SearchIdx.pm
index 30d3fe92..9ba9437d 100644
--- a/lib/PublicInbox/SearchIdx.pm
+++ b/lib/PublicInbox/SearchIdx.pm
@@ -28,8 +28,28 @@ use constant {
         PERM_GROUP => 0660,
         PERM_EVERYBODY => 0664,
         BATCH_BYTES => 1_000_000,
+        DEBUG => !!$ENV{DEBUG},
 };
 
+my %GIT_ESC = (
+        a => "\a",
+        b => "\b",
+        f => "\f",
+        n => "\n",
+        r => "\r",
+        t => "\t",
+        v => "\013",
+);
+
+sub git_unquote ($) {
+        my ($s) = @_;
+        return $s unless ($s =~ /\A"(.*)"\z/);
+        $s = $1;
+        $s =~ s/\\([abfnrtv])/$GIT_ESC{$1}/g;
+        $s =~ s/\\([0-7]{1,3})/chr(oct($1))/ge;
+        $s;
+}
+
 sub new {
         my ($class, $inbox, $creat) = @_;
         my $git_dir = $inbox;
@@ -134,11 +154,108 @@ sub index_users ($$) {
         $tg->increase_termpos;
 }
 
+sub index_text_inc ($$$) {
+        my ($tg, $text, $pfx) = @_;
+        $tg->index_text($text, 1, $pfx);
+        $tg->increase_termpos;
+}
+
+sub index_old_diff_fn {
+        my ($tg, $seen, $fa, $fb) = @_;
+
+        # no renames or space support for traditional diffs,
+        # find the number of leading common paths to strip:
+        my @fa = split('/', $fa);
+        my @fb = split('/', $fb);
+        while (scalar(@fa) && scalar(@fb)) {
+                $fa = join('/', @fa);
+                $fb = join('/', @fb);
+                if ($fa eq $fb) {
+                        index_text_inc($tg, $fa,'XDFN') unless $seen->{$fa}++;
+                        return 1;
+                }
+                shift @fa;
+                shift @fb;
+        }
+        0;
+}
+
+sub index_diff ($$$) {
+        my ($tg, $lines, $doc) = @_;
+        my %seen;
+        my $in_diff;
+        foreach (@$lines) {
+                if ($in_diff && s/^ //) { # diff context
+                        index_text_inc($tg, $_, 'XDFCTX');
+                } elsif (/^-- $/) { # email signature begins
+                        $in_diff = undef;
+                } elsif (m!^diff --git ("?a/.+) ("?b/.+)\z!) {
+                        my ($fa, $fb) = ($1, $2);
+                        my $fn = (split('/', git_unquote($fa), 2))[1];
+                        index_text_inc($tg, $fn, 'XDFN') unless $seen{$fn}++;
+                        $fn = (split('/', git_unquote($fb), 2))[1];
+                        index_text_inc($tg, $fn, 'XDFN') unless $seen{$fn}++;
+                        $in_diff = 1;
+                # traditional diff:
+                } elsif (m/^diff -(.+) (\S+) (\S+)$/) {
+                        my ($opt, $fa, $fb) = ($1, $2, $3);
+                        # only support unified:
+                        next unless $opt =~ /[uU]/;
+                        $in_diff = index_old_diff_fn($tg, \%seen, $fa, $fb);
+                } elsif (m!^--- ("?a/.+)!) {
+                        my $fn = (split('/', git_unquote($1), 2))[1];
+                        index_text_inc($tg, $fn, 'XDFN') unless $seen{$fn}++;
+                        $in_diff = 1;
+                } elsif (m!^\+\+\+ ("?b/.+)!)  {
+                        my $fn = (split('/', git_unquote($1), 2))[1];
+                        index_text_inc($tg, $fn, 'XDFN') unless $seen{$fn}++;
+                        $in_diff = 1;
+                } elsif (/^--- (\S+)/) {
+                        $in_diff = $1;
+                } elsif (defined $in_diff && /^\+\+\+ (\S+)/) {
+                        $in_diff = index_old_diff_fn($tg, \%seen, $in_diff, $1);
+                } elsif ($in_diff && s/^\+//) { # diff added
+                        index_text_inc($tg, $_, 'XDFB');
+                } elsif ($in_diff && s/^-//) { # diff removed
+                        index_text_inc($tg, $_, 'XDFA');
+                } elsif (m!^index ([a-f0-9]+)\.\.([a-f0-9]+)!) {
+                        my ($ba, $bb) = ($1, $2);
+                        index_git_blob_id($doc, 'XDFPRE', $ba);
+                        index_git_blob_id($doc, 'XDFPOST', $bb);
+                        $in_diff = 1;
+                } elsif (/^@@ (?:\S+) (?:\S+) @@\s*$/) {
+                        # traditional diff w/o -p
+                } elsif (/^@@ (?:\S+) (?:\S+) @@\s*(\S+.*)$/) {
+                        # hunk header context
+                        index_text_inc($tg, $1, 'XDFHH');
+                # ignore the following lines:
+                } elsif (/^(?:dis)similarity index/) {
+                } elsif (/^(?:old|new) mode/) {
+                } elsif (/^(?:deleted|new) file mode/) {
+                } elsif (/^(?:copy|rename) (?:from|to) /) {
+                } elsif (/^(?:dis)?similarity index /) {
+                } elsif (/^\\ No newline at end of file/) {
+                } elsif (/^Binary files .* differ/) {
+                } elsif ($_ eq '') {
+                        $in_diff = undef;
+                } else {
+                        warn "non-diff line: $_\n" if DEBUG && $_ ne '';
+                        $in_diff = undef;
+                }
+        }
+}
+
 sub index_body ($$$) {
-        my ($tg, $lines, $inc) = @_;
-        $tg->index_text(join("\n", @$lines), $inc, $inc ? 'XNQ' : 'XQUOT');
-        @$lines = ();
+        my ($tg, $lines, $doc) = @_;
+        my $txt = join("\n", @$lines);
+        $tg->index_text($txt, !!$doc, $doc ? 'XNQ' : 'XQUOT');
         $tg->increase_termpos;
+        # does it look like a diff?
+        if ($doc && $txt =~ /^(?:diff|---|\+\+\+) /ms) {
+                $txt = undef;
+                index_diff($tg, $lines, $doc);
+        }
+        @$lines = ();
 }
 
 sub add_message {
@@ -205,7 +322,7 @@ sub add_message {
                         my @lines = split(/\n/, $body);
                         while (defined(my $l = shift @lines)) {
                                 if ($l =~ /^>/) {
-                                        index_body($tg, \@orig, 1) if @orig;
+                                        index_body($tg, \@orig, $doc) if @orig;
                                         push @quot, $l;
                                 } else {
                                         index_body($tg, \@quot, 0) if @quot;
@@ -213,7 +330,7 @@ sub add_message {
                                 }
                         }
                         index_body($tg, \@quot, 0) if @quot;
-                        index_body($tg, \@orig, 1) if @orig;
+                        index_body($tg, \@orig, $doc) if @orig;
                 });
 
                 link_message($self, $smsg, $old_tid);
@@ -339,6 +456,16 @@ sub index_blob {
         $self->add_message($mime, $bytes, $num, $blob);
 }
 
+sub index_git_blob_id {
+        my ($doc, $pfx, $objid) = @_;
+
+        my $len = length($objid);
+        for (my $len = length($objid); $len >= 7; ) {
+                $doc->add_term($pfx.$objid);
+                $objid = substr($objid, 0, --$len);
+        }
+}
+
 sub unindex_blob {
         my ($self, $mime) = @_;
         my $mid = eval { mid_clean(mid_mime($mime)) };