implement per-feed stemming language setting

[tt-rss.git] / update.php

diff --git a/update.php b/update.php
index 521b956ad33b7d60311928690ff0da1716864042..8fc28973f6dd3ce2dd60a741e47aa4b68f554480 100755 (executable)
--- a/update.php
+++ b/update.php
@@ -33,6 +33,7 @@
                        "update-schema",
                        "convert-filters",
                        "force-update",
+                       "gen-search-idx",
                        "list-plugins",
                        "help");
 
@@ -80,6 +81,7 @@
                print "  --log FILE           - log messages to FILE\n";
                print "  --indexes            - recreate missing schema indexes\n";
                print "  --update-schema      - update database schema\n";
+               print "  --gen-search-idx     - generate basic PostgreSQL fulltext search index\n";
                print "  --convert-filters    - convert type1 filters to type2\n";
                print "  --force-update       - force update of all feeds\n";
                print "  --list-plugins       - list all available plugins\n";
@@ -330,6 +332,40 @@
 
        }
 
+       if (isset($options["gen-search-idx"])) {
+               echo "Generating search index (stemming set to English)...\n";
+
+               $result = db_query("SELECT COUNT(id) AS count FROM ttrss_entries");
+               $count = db_fetch_result($result, 0, "count");
+
+               print "Total entries: $count.\n";
+
+               $offset = 0;
+               $limit = 1000;
+
+               while (true) {
+                       $result = db_query("SELECT id, title, content FROM ttrss_entries WHERE tsvector_combined IS NULL ORDER BY id LIMIT $limit OFFSET $offset");
+
+                       if (db_num_rows($result) != 0) {
+                               echo "Offset $offset...\n";
+
+                               while ($line = db_fetch_assoc($result)) {
+                                       $tsvector_combined = db_escape_string(mb_substr($line['title'] . ' ' . strip_tags($line['content']),
+                                               0, 1000000));
+
+                                       db_query("UPDATE ttrss_entries SET tsvector_combined = to_tsvector('english', '$tsvector_combined') WHERE id = " . $line["id"]);
+                               }
+
+                               $offset += $limit;
+                       } else {
+                               echo "All done.\n";
+                               break;
+                       }
+
+               }
+
+       }
+
        if (isset($options["list-plugins"])) {
                $tmppluginhost = new PluginHost();
                $tmppluginhost->load_all($tmppluginhost::KIND_ALL);