xref: /plugin/sphinxsearch-was/xmlall.php (revision 39:c93d111c37ff) 
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$pagesList = getPagesList();
34
35echo '<?xml version="1.0" encoding="utf-8"?>
36<sphinx:docset>
37
38<sphinx:schema>
39<sphinx:field name="title"/>
40<sphinx:field name="body"/>
41<sphinx:field name="categories"/>
42<sphinx:field name="level"/>
43<sphinx:field name="modified"/>
44<sphinx:attr name="level" type="int" bits="8" default="1"/>
45</sphinx:schema>
46';
47
48$pageMapper = new PageMapper();
49foreach($pagesList as $row){
50    $dokuPageId = $row['id'];
51    resolve_pageid('',$page,$exists);
52    if (empty($dokuPageId) || !$exists){ //do not include not exists page
53        continue;
54    }
55    //get meta data
56    $metadata = p_get_metadata($dokuPageId);
57
58    $data = array();
59    $data['id'] = crc32($dokuPageId);
60    $data['categories'] = getCategories($dokuPageId);
61    $data['level'] = 1;
62    $data['modified'] = $metadata['date']['modified'];
63    $data['title'] = strip_tags($metadata['title']);
64    $data['body'] = io_readFile(wikiFN($dokuPageId));//strip_tags(p_wiki_xhtml($dokuPageId,$metadata['date']['modified'],false));
65
66    echo formatXml($data)."\n";
67    $pageMapper->add($dokuPageId, $metadata['title'], $metadata['title']);
68
69    $sections = getDocumentsByHeadings($dokuPageId, $metadata);
70
71    if (!empty($sections)){
72        foreach($sections as $hid => $section){
73            //parse meta data for headers, abstract, date, authors
74            if(strlen($section['section']) < 1000){
75                continue;
76            }
77            $data = array();
78            $data['id'] = crc32($dokuPageId.$hid);
79            $data['categories'] = getCategories($dokuPageId) . '#' . $hid;
80            $data['level'] = $section['level'];
81            $data['modified'] = $metadata['date']['modified'];
82            $data['title'] = strip_tags($section['title_text']);
83            $data['body'] = $section['section']; //strip_tags(p_render('xhtml',p_get_instructions($section['section']),$info));
84
85            echo formatXml($data)."\n";
86            $pageMapper->add($dokuPageId, $section['title_text'], $section['title'], $hid);
87        }
88    }
89
90}
91echo '</sphinx:docset>';
92