diff options
| author | wei <> | 2006-05-07 03:34:25 +0000 |
|---|---|---|
| committer | wei <> | 2006-05-07 03:34:25 +0000 |
| commit | 30eddf57c8de433e8ea02b9e552c8e1744a505a7 (patch) | |
| tree | 9e81f3a15f9a695cb96c5cc4dd80de5a3a0bb7b2 /buildscripts/texbuilder/create_index.php | |
| parent | 0bb2822f68dfe3cf568affd4acf0d8120d9d53c7 (diff) | |
Add search to quickstart demo.
Diffstat (limited to 'buildscripts/texbuilder/create_index.php')
| -rw-r--r-- | buildscripts/texbuilder/create_index.php | 87 |
1 files changed, 87 insertions, 0 deletions
diff --git a/buildscripts/texbuilder/create_index.php b/buildscripts/texbuilder/create_index.php new file mode 100644 index 00000000..db72c453 --- /dev/null +++ b/buildscripts/texbuilder/create_index.php @@ -0,0 +1,87 @@ +<?php
+
+// Create quickstart search index
+require_once (dirname(__FILE__).'/Zend/Search/Lucene.php');
+
+class quickstart_index
+{
+ private $_index;
+ private $_dir;
+
+ public function __construct($index_file)
+ {
+ $this->_index = new Zend_Search_Lucene($index_file, true);
+ $this->_dir = $index_file;
+ echo "Building search index...\n";
+ }
+
+ public function add($content, $section, $mtime)
+ {
+ foreach($this->split_headings($content) as $headers)
+ {
+ $doc = new Zend_Search_Lucene_Document();
+ $link = "index.php?page=".preg_replace('/\/|\\\/', '.', $section);
+ $link = str_replace('.page', '', $link).'#'.$headers['section'];
+
+ //unsearchable text
+ $doc->addField(Zend_Search_Lucene_Field::UnIndexed('link', $link));
+ $doc->addField(Zend_Search_Lucene_Field::UnIndexed('mtime', $mtime));
+ $doc->addField(Zend_Search_Lucene_Field::UnIndexed('title', $headers['title']));
+ $doc->addField(Zend_Search_Lucene_Field::UnIndexed('text', $headers['content']));
+
+ //searchable text
+ $doc->addField(Zend_Search_Lucene_Field::Keyword('page', strtolower($headers['title'])));
+ $body = strtolower($this->sanitize($headers['content']));
+ $doc->addField(Zend_Search_Lucene_Field::Unstored('contents',$body));
+ $this->_index->addDocument($doc);
+ }
+ }
+
+ function sanitize($input)
+ {
+ return htmlentities(strip_tags( $input ));
+ }
+
+ public function index()
+ {
+ return $this->_index;
+ }
+
+ protected function split_headings($html)
+ {
+ $html = preg_replace('/<\/?com:TContent[^<]*>/', '', $html);
+
+ $html = preg_replace('/<b>([^<]*)<\/b>/', '$1', $html);
+ $html = preg_replace('/<i>([^<]*)<\/i>/', '$1', $html);
+ $html = preg_replace('/<tt>([^<]*)<\/tt>/', '$1', $html);
+
+ $html = preg_replace('/<h1([^>]*)>([^<]*)<\/h1>/', '<hh$1>$2</hh>', $html);
+ $html = preg_replace('/<h2([^>]*)>([^<]*)<\/h2>/', '<hh$1>$2</hh>', $html);
+ $html = preg_replace('/<h3([^>]*)>([^<]*)<\/h3>/', '<hh$1>$2</hh>', $html);
+
+
+ $sections = preg_split('/<hh[^>]*>([^<]+)<\/hh>/', $html,-1);
+ $headers = array();
+ preg_match_all('/<hh([^>]*)>([^<]+)<\/hh>/', $html, $headers);
+ $contents = array();
+ for($i = 1, $t = count($sections); $i < $t; $i++)
+ {
+ $content['title'] = trim($this->sanitize($headers[2][$i-1]));
+ $sec = array();
+ preg_match('/"([^"]*)"/', $headers[1][$i-1], $sec);
+ $content['section'] = str_replace('"', '',$sec[0]);
+ $content['content'] = trim($this->sanitize($sections[$i]));
+ $contents[] = $content;
+ }
+
+ return $contents;
+ }
+
+ public function commit()
+ {
+ $this->_index->commit();
+ $count = $this->_index->count();
+ echo "\nSaving search index ({$count}) to {$this->_dir}\n\n";
+ }
+}
+?>
\ No newline at end of file |