xref: /plugin/sphinxsearch-was/xmlall.php (revision 99:5adfbda2edb9) 
1<?php
2/**
3 * XML feed export
4 *
5 * @license    GPL 2 (http://www.gnu.org/licenses/gpl.html)
6 * @author     Andreas Gohr <andi@splitbrain.org>
7 */
8
9
10/* Initialization */
11
12if(!defined('DOKU_INC')) define('DOKU_INC',dirname(__FILE__).'/../../../');
13if(!defined('DOKU_PLUGIN')) define('DOKU_PLUGIN',DOKU_INC.'lib/plugins/');
14
15require_once(DOKU_INC.'inc/init.php');
16require_once(DOKU_INC.'inc/common.php');
17require_once(DOKU_INC.'inc/events.php');
18require_once(DOKU_INC.'inc/parserutils.php');
19require_once(DOKU_INC.'inc/feedcreator.class.php');
20require_once(DOKU_INC.'inc/auth.php');
21require_once(DOKU_INC.'inc/pageutils.php');
22require_once(DOKU_INC.'inc/search.php');
23require_once(DOKU_INC.'inc/parser/parser.php');
24
25
26require_once(DOKU_PLUGIN.'sphinxsearch/PageMapper.php');
27require_once(DOKU_PLUGIN.'sphinxsearch/functions.php');
28
29if (!file_exists(DOKU_INC.$conf['savedir']."/sphinxsearch/")){
30	mkdir(DOKU_INC.$conf['savedir']."/sphinxsearch/");
31}
32
33$pagesList = getPagesList();
34
35echo '<?xml version="1.0" encoding="utf-8"?>
36<sphinx:docset>
37
38<sphinx:schema>
39<sphinx:field name="title"/>
40<sphinx:field name="body"/>
41<sphinx:field name="namespace"/>
42<sphinx:field name="pagename"/>
43<sphinx:field name="level"/>
44<sphinx:field name="modified"/>
45<sphinx:attr name="level" type="int" bits="8" default="1"/>
46</sphinx:schema>
47';
48
49$pageMapper = new PageMapper();
50foreach($pagesList as $row){
51    $dokuPageId = $row['id'];
52    resolve_pageid('',$page,$exists);
53    if (empty($dokuPageId) || !$exists){ //do not include not exists page
54        continue;
55    }
56    //get meta data
57    $metadata = p_get_metadata($dokuPageId);
58
59    $sections = getDocumentsByHeadings($dokuPageId, $metadata);
60
61    if (!empty($sections)){
62        foreach($sections as $hid => $section){
63            //parse meta data for headers, abstract, date, authors
64            $data = array();
65            $data['id'] = crc32($dokuPageId.$hid);
66            $data['namespace'] = getCategories($dokuPageId);
67            $data['pagename'] = getPagename($dokuPageId);
68            $data['level'] = $section['level'];
69            $data['modified'] = $metadata['date']['modified'];
70            $data['title'] = strip_tags($section['title_text']);
71            $data['title_to_index'] = $section['title_to_index'];
72            $data['body'] = $section['section']; //strip_tags(p_render('xhtml',p_get_instructions($section['section']),$info));
73
74            //convert to utf-8 encoding
75            $data['title_to_index'] = mb_convert_encoding($data['title_to_index'], "UTF-8", mb_detect_encoding($data['title_to_index'], "auto"));
76            $data['body'] = mb_convert_encoding($data['body'], "UTF-8", mb_detect_encoding($data['body'], "auto"));
77
78            echo formatXml($data)."\n";
79            $pageMapper->add($dokuPageId, $data['title'], $section['title'], $hid);
80        }
81    } else {
82        $data = array();
83        $data['id'] = crc32($dokuPageId);
84        $data['namespace'] = getCategories($dokuPageId);
85        $data['pagename'] = getPagename($dokuPageId);
86        $data['level'] = 1;
87        $data['modified'] = $metadata['date']['modified'];
88        $data['title'] = strip_tags($metadata['title']);
89        $data['title_to_index'] = $metadata['title'];
90        $data['body'] = io_readFile(wikiFN($dokuPageId)); //strip_tags(p_wiki_xhtml($dokuPageId,$metadata['date']['modified'],false));
91
92        //convert to utf-8 encoding
93        $data['title_to_index'] = mb_convert_encoding($data['title_to_index'], "UTF-8", mb_detect_encoding($data['title_to_index'], "auto"));
94        $data['body'] = mb_convert_encoding($data['body'], "UTF-8", mb_detect_encoding($data['body'], "auto"));
95
96        echo formatXml($data)."\n";
97        $pageMapper->add($dokuPageId, $metadata['title'], $metadata['title']);
98    }
99
100}
101echo '</sphinx:docset>';
102