4 * GForge Doc Search engine
\r
7 * Fabio Bertagnin November 2005
\r
9 * @version $Id: 04_IMPROVDOC_75_document_specific_search_engine.dpatch,v 1.1 2006/01/11 17:02:45 fabio Exp $
\r
12 require_once('pre.php');
\r
13 require_once('www/docman/include/doc_utils.php');
\r
14 require_once('common/docman/Parsedata.class');
\r
15 require_once('common/docman/Document.class');
\r
16 require_once('common/docman/DocumentFactory.class');
\r
17 require_once('common/docman/DocumentGroupFactory.class');
\r
19 $p = new Parsedata ("$sys_engine_path");
\r
20 // print_debug(print_r($p->get_parser_list (),true));
\r
22 $timestarttrait = microtime_float();
\r
24 $sql = "SELECT docid, group_id, filename, title, createdate, filename, description, filetype, data FROM doc_data \n";
\r
26 $result=db_query($sql);
\r
29 while ($arr = db_fetch_array($result))
\r
37 foreach ($resarr as $item)
\r
40 $timestart = microtime_float();
\r
41 $data1 = base64_decode($item["data"]);
\r
42 $lenin = strlen($data1);
\r
43 $res = $p->get_parse_data ($data1, $item["title"], $item["description"], $item["filetype"]);
\r
44 $len = strlen($res);
\r
45 $sql = "UPDATE doc_data SET data_words = '$res' WHERE docid = $item[docid] ";
\r
47 $timeend = microtime_float();
\r
48 $timetrait = $timeend - $timestart;
\r
49 print_debug ("analyze $item[filename] type=$item[filetype] octets in=$lenin octets out=$len time=$timetrait sec");
\r
51 $timeendtrait = microtime_float();
\r
52 $timetot = $timeendtrait - $timestarttrait;
\r
53 print_debug ("End analyze : $compt files, $timetot secs.");
\r
56 function print_debug ($text)
\r
61 function microtime_float()
\r
63 list($usec, $sec) = explode(" ", microtime());
\r
64 return ((float)$usec + (float)$sec);
\r