Changeset 442

Timestamp:

10/05/07 18:16:12 (1 year ago)

Author:

ogawa

Message:

apply r440 changes to batch_est_crawler.

Files:

• EstCrawler/trunk/EstCrawler/tools/batch_est_cralwer (modified) (3 diffs)

EstCrawler/trunk/EstCrawler/tools/batch_est_cralwer

@@ -40 +40 @@
-    my $cdate = ts2iso($entry->blog, $entry->authored_on);
-    my $mdate = ts2iso($entry->blog, $entry->modified_on);
+    my $categories = join('', map { '[' . $_->label . ']' } @{$entry->categories})
+        if $entry->categories;
+    my $tags = join('', map { '[' . $_ . ']' } $entry->tags)
+        if $entry->tags;
-
-    # metainfo (attribute, not searchable)
@@ -47 +51 @@
-    $doc->add_attr('@cdate', $cdate);
-    $doc->add_attr('@mdate', $mdate);
+    $doc->add_attr('entry_id', $entry->id);
+    $doc->add_attr('blog_id', $entry->blog_id);
+    $doc->add_attr('categories', $categories) if $categories;
+    $doc->add_attr('tags', $tags) if $tags;
-
-    # document body (searchable)
@@ -55 +63 @@
-    $doc->add_hidden_text($title);
-    $doc->add_hidden_text($author);
-join(', ', map { $_->label } @{$entry->categories})
-join(', ', $entry->tags)
+$categories
+$tags
-    $doc->add_hidden_text(remove_html($entry->keywords) || '');
-    $doc;