[dw-free] Site-wide search

[mark]

[commit: http://hg.dwscoalition.org/dw-free/rev/eb5de5207217]

http://bugs.dwscoalition.org/show_bug.cgi?id=1902

Okay, journal search now lets you see all content you can see, not just
public posts. This required changing the search db structure again.

Patch by mark.

Files modified:

• bin/worker/sphinx-copier
• bin/worker/sphinx-search-gm
• cgi-bin/LJ/User.pm
• htdocs/search.bml

--------------------------------------------------------------------------------
diff -r da0e247843a5 -r eb5de5207217 bin/worker/sphinx-copier
--- a/bin/worker/sphinx-copier	Tue Oct 13 16:10:32 2009 +0000
+++ b/bin/worker/sphinx-copier	Wed Oct 14 04:14:57 2009 +0000
@@ -110,10 +110,16 @@ sub work {
         );
         die $dbfrom->errstr if $dbfrom->err;
 
+        # we need extra security bits for some metadata.  we have to do this this way because
+        # it makes it easier to later do searches on various combinations of things at the same
+        # time...  also, even though these are bits, we're not going to ever use them as actual bits.
+        my @extrabits;
+        push @extrabits, 101 if $row->{security} eq 'private';
+        push @extrabits, 102 if $row->{security} eq 'public';
+
         # have to do some more munging
-        $row->{is_public} = $row->{security} eq 'public' ? 1 : 0;
         $row->{edittime} = $db_times->{$jitemid};
-        $row->{allowmask} = join ',', LJ::bit_breakdown( $row->{allowmask} );
+        $row->{allowmask} = join ',', LJ::bit_breakdown( $row->{allowmask} ), @extrabits;
         $row->{allowpublic} = $u->include_in_global_search ? 1 : 0;
 
         # very important, the search engine can't index compressed crap...
@@ -127,10 +133,10 @@ sub work {
 
         # insert
         $dbto->do(
-            q{REPLACE INTO posts_raw (id, journal_id, jitemid, poster_id, is_public, security_bits, allow_global_search, date_posted,
+            q{REPLACE INTO posts_raw (id, journal_id, jitemid, poster_id, security_bits, allow_global_search, date_posted,
                                       title, data, revtime)
-              VALUES (NULL, ?, ?, ?, ?, ?, ?, UNIX_TIMESTAMP(?), ?, ?, ?)},
-            undef, map { $row->{$_} } qw/ journalid jitemid posterid is_public allowmask allowpublic eventtime subject event edittime /
+              VALUES (NULL, ?, ?, ?, ?, ?, UNIX_TIMESTAMP(?), ?, ?, ?)},
+            undef, map { $row->{$_} } qw/ journalid jitemid posterid allowmask allowpublic eventtime subject event edittime /
         );
         die $dbto->errstr if $dbto->err;
 
diff -r da0e247843a5 -r eb5de5207217 bin/worker/sphinx-search-gm
--- a/bin/worker/sphinx-search-gm	Tue Oct 13 16:10:32 2009 +0000
+++ b/bin/worker/sphinx-search-gm	Wed Oct 14 04:14:57 2009 +0000
@@ -34,9 +34,6 @@ sub sphinx_search {
     my $args = Storable::thaw( $job->arg ) || {};
     return undef unless $args->{query};
 
-    # a little sanity check, we don't currently allow a global search on non-public content
-    return undef if $args->{userid} == 0 && $args->{public} != 1;
-
     my $sx = Sphinx::Search->new();
     $sx->SetServer( @LJ::SPHINX_SEARCHD );
 
@@ -51,9 +48,14 @@ sub sphinx_search {
     $sx->SetFilter( 'allow_global_search', [ 1 ] )
         unless $args->{userid};
 
-    # further filter on public items or not
-    $sx->SetFilter( 'is_public', [ 1 ] )
-        if $args->{public};
+    # security filtering is a dangerous game.  basically, the caller tells us whether to
+    # ignore security or gives us a mask of bits to work with.  from that we intuit what
+    # security options to enable on our filters to Sphinx.
+    unless ( $args->{ignore_security} ) {
+        # allow public posts and anything the mask allows
+        my @bits = ( 102, LJ::bit_breakdown( $args->{allowmask} ) );
+        $sx->SetFilter( 'security_bits', \@bits );
+    }
 
     my $res = $sx->Query( $args->{query} );
     return undef unless $res;
diff -r da0e247843a5 -r eb5de5207217 cgi-bin/LJ/User.pm
--- a/cgi-bin/LJ/User.pm	Tue Oct 13 16:10:32 2009 +0000
+++ b/cgi-bin/LJ/User.pm	Wed Oct 14 04:14:57 2009 +0000
@@ -1751,7 +1751,7 @@ sub allow_search_by {
     return 1 if $u->is_community || $u->equals( $by );
 
     # check the userprop for security access
-    my $whocan = $u->prop( 'opt_allowsearchby' );
+    my $whocan = $u->prop( 'opt_allowsearchby' ) || 'F';
     return 1 if $whocan eq 'A';
     return 1 if $whocan eq 'F' && $u->trusts( $by );
     return 1 if $whocan eq 'N' && $u->equals( $by );
diff -r da0e247843a5 -r eb5de5207217 htdocs/search.bml
--- a/htdocs/search.bml	Tue Oct 13 16:10:32 2009 +0000
+++ b/htdocs/search.bml	Wed Oct 14 04:14:57 2009 +0000
@@ -96,12 +96,26 @@ body<=
     return $error->( "Hey, that offset is nonsensical... :(" )
         if $offset < 0 || $offset > 1000;
 
-    # set a public only flag if this is not your journal or a community you are in
-    my $public = 1;
-    $public = 0 if $su && ( $remote->equals( $su ) || ( $su->is_community && $remote->member_of( $su ) ) );
+    # we have to set a few flags on what to search.  default to public and no bits.
+    my ( $ignore_security, $allowmask ) = ( 0, 0 );
+    if ( $su ) {
+        # if it's you, all posts, all bits
+        if ( $remote->equals( $su ) ) {
+            $ignore_security = 1;
+
+        # or a comm you're in ...
+        } elsif ( $su->is_community && $remote->member_of( $su ) ) {
+            $ignore_security = 1;
+
+        # otherwise, if they trust you, get the mask ...
+        } elsif ( $su->trusts( $remote ) ) {
+            $allowmask = $su->trustmask( $remote );
+        }
+    }
 
     # the arguments to the search (userid=0 implies global search)
-    my $args = { userid => $su ? $su->id : 0, query => $q, offset => $offset, public => $public };
+    my $args = { userid => $su ? $su->id : 0, query => $q, offset => $offset,
+                 ignore_security => $ignore_security, allowmask => $allowmask };
     my $arg = Storable::nfreeze( $args );
 
     # so we know that they're searching something valid, send to gearman
--------------------------------------------------------------------------------