Rework usage stats.

6 files changed, 45 insertions, 34 deletions

diff --git a/cgi-bin/lxr b/cgi-bin/lxr
index 28d6635..375b640 100755
--- a/cgi-bin/lxr
+++ b/cgi-bin/lxr
@@ -331,8 +331,9 @@ sub search {
 					  'idents' => \@cooked};
 	}
 	if ($type eq 'ident') {
+	    my $usage  = $context->config->{'usage'};
 	    my ($symname, $symid, $ident, $refs) =
-		$index->get_identifier_info($find, $rel_id);
+		$index->get_identifier_info($usage, $find, $rel_id);
 
 	    $$ident[1] = $LXRng::Lang::deftypes{$$ident[1]};
 	    $$ident[5] &&= $LXRng::Lang::deftypes{$$ident[5]};
diff --git a/lib/LXRng/Context.pm b/lib/LXRng/Context.pm
index dab03e3..3bc7a92 100644
--- a/lib/LXRng/Context.pm
+++ b/lib/LXRng/Context.pm
@@ -41,6 +41,7 @@ sub new {
 	    unless ref($config[0]) eq 'HASH' and exists($config[0]{$tree});
 
 	$$self{'config'} = $config[0]{$tree};
+	$$self{'config'}{'usage'} ||= $$self{'config'}{'index'};
     }
 
     if (exists $$self{'params'}{'v'} and $$self{'params'}{'v'}) {
diff --git a/lib/LXRng/Index/DBI.pm b/lib/LXRng/Index/DBI.pm
index 932202c..2b796bd 100644
--- a/lib/LXRng/Index/DBI.pm
+++ b/lib/LXRng/Index/DBI.pm
@@ -334,19 +334,6 @@ sub _symbols_by_file {
     return \%res;
 }
 
-sub _add_usage {
-    my ($self, $file_id, $line, $symbol_id) = @_;
-    
-    my $dbh = $self->dbh;
-    my $pre = $self->prefix;
-    my $sth = $$self{'sth'}{'_add_usage'} ||=
-	$dbh->prepare(qq{insert into ${pre}usage(id_rfile, line, id_symbol)
-			     values (?, ?, ?)});
-    $sth->execute($file_id, $line, $symbol_id);
-
-    return 1;
-}
-
 sub _usage_by_file {
     my ($self, $rfile_id) = @_;
 
@@ -455,21 +442,28 @@ sub get_symbol_usage {
     return undef unless $symid =~ /^\d+$/s;
     my $sth =
 	$dbh->prepare(qq{
-	    select u.id_rfile, u.line
-		from ${pre}usage u, ${pre}filereleases fr
+	    select f.path, u.lines
+		from ${pre}usage u, ${pre}filereleases fr,
+			${pre}files f, ${pre} revisions r
 		where u.id_symbol = $symid
 		and u.id_rfile = fr.id_rfile and fr.id_release = ?
+		and u.id_rfile = r.id and r.id_file = f.id
 		limit 1000});
 
     $sth->execute($rel_id);
     my $res = $sth->fetchall_arrayref();
     $sth->finish();
 
-    return $res;
+    my %rlines;
+    foreach my $r (@$res) {
+	$rlines{$$r[0]} = [$$r[1] =~ /(\d+),?/g];
+    }
+
+    return \%rlines;
 }
 
 sub get_identifier_info {
-    my ($self, $ident, $rel_id) = @_;
+    my ($self, $usage, $ident, $rel_id) = @_;
 
     my $dbh = $self->dbh;
     my $pre = $self->prefix;
@@ -499,15 +493,16 @@ sub get_identifier_info {
 	    $sth->fetchrow_array();
     $sth->finish();
 
-    my $refs = {$rfile_id => $path};
-    $self->get_referring_files($rel_id, $rfile_id, $refs);
-    my $usage = $self->get_symbol_usage($rel_id, $symid);
+    my $incs = {$rfile_id => $path};
+    $self->get_referring_files($rel_id, $rfile_id, $incs);
+    my %paths; @paths{values %$incs} = ();
 
+    my $refs = $usage->get_symbol_usage($rel_id, $symid);
     my %reflines;
-    foreach my $u (@$usage) {
-	next unless $$refs{$$u[0]};
-	$reflines{$$refs{$$u[0]}} ||= [];
-	push(@{$reflines{$$refs{$$u[0]}}}, $$u[1]);
+    my ($p, $l);
+    while (($p, $l) = each %$refs) {
+	next unless exists $paths{$p};
+	$reflines{$p} = $l;
     }
 
     return ($symname, $symid, 
diff --git a/lib/LXRng/Index/Generic.pm b/lib/LXRng/Index/Generic.pm
index cab2513..1cdaab5 100644
--- a/lib/LXRng/Index/Generic.pm
+++ b/lib/LXRng/Index/Generic.pm
@@ -144,13 +144,12 @@ sub symbols_by_file {
 }
 
 sub add_usage {
-    my ($self, $file_id, $line, $symbol) = @_;
+    my ($self, $doc, $file_id, $sym_id, $lines) = @_;
 
-    my $sym_id = $self->symbol_id($symbol, 1);
-    
-    return $self->_add_usage($file_id, $line, $sym_id);
+    return $self->_add_usage($file_id, $sym_id, $lines);
 }
 
+# TODO: What functionality actually uses this?  Can it be removed?
 sub usage_by_file {
     my ($self, $tree, $release, $path) = @_;
 
diff --git a/lib/LXRng/Index/Pg.pm b/lib/LXRng/Index/Pg.pm
index 1b905c0..a385767 100644
--- a/lib/LXRng/Index/Pg.pm
+++ b/lib/LXRng/Index/Pg.pm
@@ -154,9 +154,9 @@ sub init_db {
     $dbh->do(qq{
 	create table ${pre}usage
 	    (
-	     id_rfile		int references ${pre}revisions(id) deferrable,
-	     id_symbol		int references ${pre}symbols(id) deferrable,
-	     line		int
+	     id_rfile		int,
+	     id_symbol		int,
+	     lines		int[]
 	     )
 	}) or die($dbh->errstr);
 
@@ -407,6 +407,20 @@ sub _add_ident {
     return $id;
 }
 
+sub _add_usage {
+    my ($self, $file_id, $symbol_id, $lines) = @_;
+    
+    my $dbh = $self->dbh;
+    my $pre = $self->prefix;
+    my $sth = $$self{'sth'}{'_add_usage'} ||=
+	$dbh->prepare(qq{insert into ${pre}usage(id_rfile, id_symbol, lines)
+			     values (?, ?, ?)});
+    $sth->execute($file_id, $symbol_id, '{'.join(',', @$lines).'}');
+
+    return 1;
+}
+
+
 sub DESTROY {
     my ($self) = @_;
 
diff --git a/lib/LXRng/Index/PgBatch.pm b/lib/LXRng/Index/PgBatch.pm
index 19c9fa9..40654d2 100644
--- a/lib/LXRng/Index/PgBatch.pm
+++ b/lib/LXRng/Index/PgBatch.pm
@@ -146,9 +146,10 @@ sub _prime_symbol_cache {
 }
 
 sub _add_usage {
-    my ($self, $file_id, $line, $symbol_id) = @_;
+    my ($self, $file_id, $symbol_id, $lines) = @_;
     
-    push(@{$self->_cache('usage')}, "$file_id\t$symbol_id\t$line\n");
+    push(@{$self->_cache('usage')},
+	 "$file_id\t$symbol_id\t\{".join(",", @$lines)."}\n");
 
     return 1;
 }