[Rt-commit] rt branch, 4.2/mysql-native-fts, repushed

[Alex Vandiver]

The branch 4.2/mysql-native-fts was deleted and repushed:
       was b490888fd9527edb9c5340ae492251e3ba08a410
       now 17850b60d781282fa57fca0b70781db54b5e96a6

 1:  ff018f3 =  1:  ff018f3 Add additional clarification points about Sphinx on MySQL
 2:  7d52ea3 =  2:  7d52ea3 Drop sphinx xmlpipe2 output, which was unusable and undocumented
 3:  3152f64 =  3:  3152f64 Drop finalize and clean functions, which are now unused
 4:  fd038d9 =  4:  fd038d9 Rename Sphinx FTS search
 5:  2374496 =  5:  2374496 Support native FTS on MySQL 5.6 and above
 6:  bfd8dce =  6:  bfd8dce Using a separate MyISAM table, we can also support FTS on MySQL < 5.6
 7:  a4e5362 =  7:  a4e5362 extract_text and extract_html are identical; inline them
 8:  2474f53 =  8:  2474f53 Inline the differences between text/plain and text/html attachment lists
 9:  b276cbe =  9:  b276cbe Stop skipping indexing of text/html within multipart/alternative
10:  b3bf43a = 10:  b3bf43a Use the new, shorterm initialization form
11:  06e22fe = 11:  06e22fe Simplify and condense option parsing
12:  5b8ed2a = 12:  5b8ed2a Documentation has moved out; update --help accordingly
13:  b06fef2 = 13:  b06fef2 Remove AUTHOR section; it is unnecessary in core sbin files
14:  929539a = 14:  929539a Skipping ACL checks yields a sizable performance increase
15:  434556b = 15:  434556b Index attachments in one pass through the database, not two
16:  1f70512 = 16:  1f70512 Index attachments even on deleted tickets
17:  f9c8eae = 17:  f9c8eae mysql and pg share the same last_indexed; unify the method
18:  b01e737 ! 18:  31fe979 Replace the last use of goto_specific with explicit function calls
    @@ -12,7 +12,7 @@
     -        process( $a, \$text );
     +        if ($db_type eq 'mysql') {
     +            process_mysql( $a, \$text );
    -+        } elsif ($db_type eq 'pg') {
    ++        } elsif ($db_type eq 'Pg') {
     +            process_pg( $a, \$text );
     +        }
              debug("Processed attachment #". $a->id );
19:  c876f7c = 19:  0692e70 Simplify last_indexed
20:  5c56ed6 = 20:  2152feb Only call last_indexed once, as it may be heavy
21:  6768ac5 = 21:  6f02246 Index even empty attachments
22:  acb1782 ! 22:  35037b6 As last_indexed is based on the highest insert, there will never be an UPDATE needed
    @@ -20,14 +20,16 @@
          $dbh->do( $query, undef, $$text, $attachment->id );
      }
     @@
    +     my $column = $fts_config->{'Column'};
      
          my $query;
    -     if ( $table ) {
    +-    if ( $table ) {
     -        if ( my ($id) = $dbh->selectrow_array("SELECT id FROM $table WHERE id = ?", undef, $attachment->id) ) {
     -            $query = "UPDATE $table SET $column = to_tsvector(?) WHERE id = ?";
     -        } else {
     -            $query = "INSERT INTO $table($column, id) VALUES(to_tsvector(?), ?)";
     -        }
    ++    if ( $table ne 'Attachments' ) {
     +        $query = "INSERT INTO $table($column, id) VALUES(to_tsvector(?), ?)";
          } else {
              $query = "UPDATE Attachments SET $column = to_tsvector(?) WHERE id = ?";
23:  d2ddcbe ! 23:  6842b22 Inversion of control of main indexing loops
    @@ -1,6 +1,16 @@
     Author: Alex Vandiver <alexmv at bestpractical.com>
     
    -    Inversion of control
    +    Inversion of control of main indexing loops
    +    
    +    Rather than having database-dependent if-statements, followed by a
    +    standard loop which contains further database-dependent if-statements,
    +    instead turn the loop into a function which cane be called from one of
    +    two database-specific functions.  This is important because MySQL's
    +    iteration will further diverge from PostgreSQL's in the following
    +    commits.
    +    
    +    This commit is best viewed with:
    +      git diff --patience --ignore-all-space
     
     diff --git a/sbin/rt-fulltext-indexer.in b/sbin/rt-fulltext-indexer.in
     --- a/sbin/rt-fulltext-indexer.in
    @@ -40,7 +50,7 @@
     -        $found++;
     -        if ($db_type eq 'mysql') {
     -            process_mysql( $a, \$text );
    --        } elsif ($db_type eq 'pg') {
    +-        } elsif ($db_type eq 'Pg') {
     -            process_pg( $a, \$text );
     -        }
     -        debug("Processed attachment #". $a->id );
    @@ -50,7 +60,7 @@
     +my $LAST;
     +if ($db_type eq 'mysql') {
     +    process_mysql();
    -+} elsif ($db_type eq 'pg') {
    ++} elsif ($db_type eq 'Pg') {
     +    process_pg();
      }
      
    @@ -117,14 +127,14 @@
          my $table = $fts_config->{'Table'};
          my $column = $fts_config->{'Column'};
      
    -+    if ($table) {
    ++    if ( $table ne 'Attachments' ) {
     +        ($LAST) = $dbh->selectrow_array("SELECT MAX(id) FROM $table");
     +    } else {
     +        ($LAST) = $dbh->selectrow_array("SELECT MAX(id) FROM Attachments WHERE $column IS NOT NULL");
     +    }
     +
          my $query;
    -     if ( $table ) {
    +     if ( $table ne 'Attachments' ) {
              $query = "INSERT INTO $table($column, id) VALUES(to_tsvector(?), ?)";
     @@
              $query = "UPDATE Attachments SET $column = to_tsvector(?) WHERE id = ?";
24:  7a29799 ! 24:  00037f9 Switch to preparing statements, rather than just setting strings
    @@ -30,7 +30,7 @@
      
     -    my $query;
     +    my $sth;
    -     if ( $table ) {
    +     if ( $table ne 'Attachments' ) {
     -        $query = "INSERT INTO $table($column, id) VALUES(to_tsvector(?), ?)";
     +        $sth = $dbh->prepare("INSERT INTO $table($column, id) VALUES(to_tsvector(?), ?)");
          } else {
25:  7d78413 = 25:  d2b9064 INSERT DELAYED provides notable speed benefits on MyISAM
26:  b490888 = 26:  17850b6 Improve MySQL insert speed by batching inserts into one statement