function forumng_delete_instance($id, $ociskip = true)
{
require_once dirname(__FILE__) . '/forum.php';
try {
$forum = forum::get_from_id($id, forum::CLONE_DIRECT);
// avoid deleting OCI specific forum if running in upload block
if ($ociskip) {
global $restore;
if (isset($restore) && $restore->restoreto == 0 && strpos($_SERVER['HTTP_REFERER'], 'blocks/versions/upload.php') !== false) {
if ($forum->get_name() == get_string('newunitforumname', 'createcourse')) {
//Unit forum
echo ' found forumng ' . $forum->get_id() . ' ' . $forum->get_name();
return true;
}
}
}
$forum->delete_all_data();
if (forum::search_installed()) {
$cm = $forum->get_course_module();
ousearch_document::delete_module_instance_data($cm);
}
} catch (Exception $e) {
return false;
}
return delete_records('forumng', 'id', $id);
}
function test_split_words()
{
// Standard usage and caps
$this->assertEqual(ousearch_document::split_words('Hello I AM a basic test'), array('hello', 'i', 'am', 'a', 'basic', 'test'));
// Numbers
$this->assertEqual(ousearch_document::split_words('13 2by2'), array('13', '2by2'));
// Ignored and accepted punctuation and whitespace
$this->assertEqual(ousearch_document::split_words(' hello,testing!what\'s&up there-by '), array('hello', 'testing', 'what\'s', 'up', 'there', 'by'));
// Unicode letters and nonletter
$this->assertEqual(ousearch_document::split_words('café ßåřĉĕļÅ?ņä※tonight'), array('café', 'ßåřĉĕļÅ?ņä', 'tonight'));
// Unicode caps
$this->assertEqual(ousearch_document::split_words('ĀĒĪŌŪ'), array('Ä?Ä“Ä«Å?Å«'));
// Query mode (keeps " + -)
$this->assertEqual(ousearch_document::split_words('"hello there" +frog -doughnut extra-special', true), array('"hello', 'there"', '+frog', '-doughnut', 'extra-special'));
// Position mode: normal
$this->assertEqual(ousearch_document::split_words('hello test', false, true), array(array('hello', 'test'), array(0, 6, 10)));
// Position mode: whitespace
$this->assertEqual(ousearch_document::split_words(' hello test ', false, true), array(array('hello', 'test'), array(4, 13, 21)));
// Position mode: unicode
$this->assertEqual(ousearch_document::split_words('hĕllo tĕst', false, true), array(array('hĕllo', 'tĕst'), array(0, 7, 12)));
// Positions are in bytes
}
/**
* Obtains a search document relating to a particular blog post.
*
* @param object $post Post object. Required fields: id (optionally also
* groupid, userid save a db query)
* @param object $cm Course-module object. Required fields: id, course
* @return ousearch_doument
*/
function oublog_get_search_document($post, $cm)
{
// Set up 'search document' to refer to this post
$doc = new ousearch_document();
$doc->init_module_instance('oublog', $cm);
if (!isset($post->userid) || !isset($post->groupid)) {
global $CFG;
$results = get_record_sql("\nSELECT\n p.groupid,i.userid\nFROM\n{$CFG->prefix}oublog_posts p\n INNER JOIN {$CFG->prefix}oublog_instances i ON p.oubloginstancesid=i.id\nWHERE\n p.id={$post->id}");
if (!$results) {
error("Can't find details for blog post {$post->id}");
}
$post->userid = $results->userid;
$post->groupid = $results->groupid;
}
if ($post->groupid) {
$doc->set_group_id($post->groupid);
}
$doc->set_user_id($post->userid);
$doc->set_int_refs($post->id);
return $doc;
}
/**
* Update all wiki documents for ousearch.
* @param bool $feedback If true, prints feedback as HTML list items
* @param int $courseid If specified, restricts to particular courseid
*/
function ouwiki_ousearch_update_all($feedback = false, $courseid = 0)
{
global $CFG;
// Get list of all wikis. We need the coursemodule data plus
// the type of subwikis
$coursecriteria = $courseid === 0 ? '' : 'cm.course=' . $courseid . ' AND';
$coursemodules = get_records_sql("\nSELECT\n cm.id,cm.course,cm.instance,w.subwikis\nFROM\n {$CFG->prefix}modules m\n INNER JOIN {$CFG->prefix}course_modules cm ON cm.module=m.id\n INNER JOIN {$CFG->prefix}ouwiki w ON cm.instance=w.id\nWHERE\n {$coursecriteria}\n m.name='ouwiki'");
if (!$coursemodules) {
return;
}
if ($feedback) {
print '<li><strong>' . count($coursemodules) . '</strong> wikis to process.</li>';
$dotcount = 0;
}
$count = 0;
foreach ($coursemodules as $coursemodule) {
// This condition is needed because if somebody creates some stuff
// then changes the wiki type, it actually keeps the old bits
// in the database. Maybe it shouldn't, not sure.
switch ($coursemodule->subwikis) {
case OUWIKI_SUBWIKIS_SINGLE:
$where = "sw.userid IS NULL AND sw.groupid IS NULL";
break;
case OUWIKI_SUBWIKIS_GROUPS:
$where = "sw.userid IS NULL AND sw.groupid IS NOT NULL";
break;
case OUWIKI_SUBWIKIS_INDIVIDUAL:
$where = "sw.userid IS NOT NULL AND sw.groupid IS NULL";
break;
}
// Get all pages in that wiki
$rs = get_recordset_sql("\nSELECT\n p.id,p.title,v.xhtml,v.timecreated,sw.groupid,sw.userid\nFROM\n {$CFG->prefix}ouwiki_subwikis sw\n INNER JOIN {$CFG->prefix}ouwiki_pages p ON p.subwikiid=sw.id\n INNER JOIN {$CFG->prefix}ouwiki_versions v ON v.id=p.currentversionid\nWHERE\n sw.wikiid={$coursemodule->instance} AND {$where}");
while ($page = rs_fetch_next_record($rs)) {
// Update the page for search
$doc = new ousearch_document();
$doc->init_module_instance('ouwiki', $coursemodule);
if ($page->groupid) {
$doc->set_group_id($page->groupid);
}
if ($page->title) {
$doc->set_string_ref($page->title);
}
if ($page->userid) {
$doc->set_user_id($page->userid);
}
$title = $page->title ? $page->title : '';
if (!$doc->update($title, $page->xhtml, $page->timecreated)) {
ouwiki_error('Failed to update database record for page ID ' . $page->id);
}
}
rs_close($rs);
$count++;
if ($feedback) {
if ($dotcount == 0) {
print '<li>';
}
print '.';
$dotcount++;
if ($dotcount == 20 || $count == count($coursemodules)) {
print 'done ' . $count . '</li>';
$dotcount = 0;
}
flush();
}
}
}
/**
* Given an ID of an instance of this module, this function will
* permanently delete the instance and any data that depends on it.
*
* @param int $id The ID of the module instance
* @return boolena true on success, false on failure.
*/
function oublog_delete_instance($oublogid)
{
if (!($oublog = get_record('oublog', 'id', $oublogid))) {
return false;
}
if ($oublog->global) {
error('You can\'t delete the global blog');
}
if ($instances = get_records('oublog_instances', 'oublogid', $oublog->id)) {
foreach ($instances as $oubloginstancesid => $bloginstance) {
// tags
delete_records('oublog_taginstances', 'oubloginstancesid', $oubloginstancesid);
if ($posts = get_records('oublog_posts', 'oubloginstancesid', $oubloginstancesid)) {
foreach ($posts as $postid => $post) {
// comments
delete_records('oublog_comments', 'postid', $postid);
// edits
delete_records('oublog_edits', 'postid', $postid);
}
// posts
delete_records('oublog_posts', 'oubloginstancesid', $oubloginstancesid);
}
}
}
// links
delete_records('oublog_links', 'oublogid', $oublog->id);
// instances
delete_records('oublog_instances', 'oublogid', $oublog->id);
// Fulltext search data
require_once dirname(__FILE__) . '/locallib.php';
if (oublog_search_installed()) {
$moduleid = get_field('modules', 'id', 'name', 'oublog');
$cm = get_record('course_modules', 'module', $moduleid, 'instance', $oublog->id);
if (!$cm) {
error('Can\'t find coursemodule');
}
ousearch_document::delete_module_instance_data($cm);
}
// oublog
return delete_records('oublog', 'id', $oublog->id);
}
$lastreport = 0;
$pos = 0;
$type_none = 0;
$type_user = 0;
$type_group = 0;
$totaltime = 0;
$documents = 0;
while ($pos < $numlines) {
// Pick how many lines to use for one 'document'
$doclines = rand(5, 200);
if ($pos + $doclines > $numlines) {
$doclines = $numlines - $pos;
}
$start = $pos;
// Set up document
$document = new ousearch_document();
$document->init_test('test2');
// Refs define document and place in it
$document->set_string_ref($filename);
$document->set_int_refs($start, $doclines);
// Type (group.user)
$type = rand(0, 99);
if ($type < 50) {
// 50% None
$type_none++;
} else {
if ($type < 90) {
// 40% Group
$document->set_group_id($groups[rand(0, count($groups) - 1)]);
$type_group++;
} else {
/**
* Saves a new version of the given named page within a subwiki. Can create
* a new page or just add a new version to an existing one. In case of
* failure, ends up calling error() rather than returning something.
* @param object $course Course object
* @param object $cm Course-module object
* @param object $ouwiki OU wiki object
* @param object $subwiki Subwiki object
* @param string $pagename Name of page (NO SLASHES)
* @param string $content XHTML Content (NO SLASHES)
* @param int $changestart For section changes. Start position of change. (-1 if not section change)
* @param int $changesize Size of changed section.
* @param int $changeprevsize Previous size of changed section
* @param bool $nouser If true, creates as system
*/
function ouwiki_save_new_version($course, $cm, $ouwiki, $subwiki, $pagename, $content, $changestart = -1, $changesize = -1, $changeprevsize = -1, $nouser = false)
{
$tw = new transaction_wrapper();
// Find page if it exists
$pageversion = ouwiki_get_current_page($subwiki, $pagename, OUWIKI_GETPAGE_CREATE);
// Analyse content for HTML headings that don't already have an ID.
// These are all assigned unique, fairly short IDs.
// Get number of version [guarantees in-page uniqueness of generated IDs]
$versionnumber = count_records('ouwiki_versions', 'pageid', $pageversion->pageid);
// Remove any spaces from annotation tags that were added for editing or by users
// and remove any duplicate annotation tags
$pattern = '~<span\\b.id=\\"annotation(.+?)\\">.*?</span>~';
$replace = '<span id="annotation$1"></span>';
$content = preg_replace($pattern, $replace, $content);
unset($pattern, $replace, $used);
// Get rid of any heading tags that only contain whitespace
$emptypatterns = array();
for ($i = 1; $i <= 6; $i++) {
$emptypatterns[] = '~<h' . $i . '[^>]*>\\s*(<br[^>]*>\\s*)*</h' . $i . '>~';
}
$content = preg_replace($emptypatterns, '', $content);
// List all headings that already have IDs, to check for duplicates
$matches = array();
preg_match_all('|<h[1-9] id="ouw_s(.*?)">(.*?)</h[1-9]>|', $content, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
// Organise list by ID
$byid = array();
foreach ($matches as $index => $data) {
$id = $data[1][0];
if (!array_key_exists($id, $byid)) {
$byid[$id] = array();
}
$byid[$id][] = $index;
}
// Handle any duplicates
$deletebits = array();
foreach ($byid as $id => $duplicates) {
if (count($duplicates) > 1) {
// We have a duplicate. By default, keep the first one
$keep = $duplicates[0];
// See if there is a title entry in the database for it
$knowntitle = get_field('ouwiki_sections', 'title', 'xhtmlid', addslashes($id), 'pageid', $pageversion->pageid);
if ($knowntitle) {
foreach ($duplicates as $duplicate) {
$title = ouwiki_get_section_title(null, null, $matches[$duplicate][2][0]);
if ($title === $knowntitle) {
$keep = $duplicate;
break;
}
}
}
foreach ($duplicates as $duplicate) {
if ($duplicate !== $keep) {
$deletebits[] = (object) array('startbyte' => $matches[$duplicate][1][1] - 10, 'bytes' => strlen($matches[$duplicate][1][0]) + 11);
}
}
}
}
// Were there any?
if (count($deletebits) > 0) {
// Sort in reverse order of starting position
usort($deletebits, 'ouwiki_internal_sort_deletions');
// Delete each bit
foreach ($deletebits as $deletebit) {
$content = substr($content, 0, $deletebit->startbyte) . substr($content, $deletebit->startbyte + $deletebit->bytes);
}
}
// Replace existing empty headings with an ID including version count plus another index
global $ouwiki_count;
// Nasty but I can't think of a better way!
$ouwiki_count = 0;
global $ouwiki_internal_re;
$ouwiki_internal_re->version = $versionnumber;
$ouwiki_internal_re->count = 0;
$sizebefore = strlen($content);
$content = preg_replace_callback('/<h([1-9])>/', 'ouwiki_internal_re_headings', $content);
$sizeafter = strlen($content);
// Replace wiki links to [[Start page]] with the correct (non
// language-specific) format [[]]
$regex = str_replace('.*?', preg_quote(get_string('startpage', 'ouwiki')), OUWIKI_LINKS_SQUAREBRACKETS) . 'ui';
$content = preg_replace($regex, '[[]]', $content);
// Create version
$version = new StdClass();
$version->pageid = $pageversion->pageid;
$version->xhtml = addslashes($content);
$version->timecreated = time();
if (!$nouser) {
global $USER;
$version->userid = $USER->id;
}
if ($changestart != -1) {
$version->changestart = $changestart;
// In tracking the new size, account for any added headings etc
$version->changesize = $changesize + ($sizeafter - $sizebefore);
$version->changeprevsize = $changeprevsize;
}
if (!($versionid = insert_record('ouwiki_versions', $version))) {
$tw->rollback();
ouwiki_dberror();
}
// Update latest version
if (!set_field('ouwiki_pages', 'currentversionid', $versionid, 'id', $pageversion->pageid)) {
$tw->rollback();
ouwiki_dberror();
}
// Analyse for links
$wikilinks = array();
$externallinks = array();
// Wiki links: ordinary [[links]]
$matches = array();
preg_match_all(OUWIKI_LINKS_SQUAREBRACKETS, $content, $matches, PREG_PATTERN_ORDER);
foreach ($matches[1] as $match) {
// Convert to page name (this also removes HTML tags etc)
$wikilinks[] = ouwiki_get_wiki_link_details($match)->page;
}
// Note that we used to support CamelCase links but have removed support because:
// 1. Confusing: students type JavaScript or MySpace and don't expect it to become a link
// 2. Not accessible: screenreaders cannot cope with run-together words, and
// dyslexic students can have difficulty reading them
// External links
preg_match_all('/<a [^>]*href=(?:(?:\'(.*?)\')|(?:"(.*?))")/', $content, $matches, PREG_PATTERN_ORDER);
foreach ($matches[1] as $match) {
if ($match) {
$externallinks[] = html_entity_decode($match);
}
}
foreach ($matches[2] as $match) {
if ($match) {
$externallinks[] = html_entity_decode($match);
}
}
// Add link records
$link = new StdClass();
$link->fromversionid = $versionid;
foreach ($wikilinks as $targetpage) {
if (!empty($targetpage)) {
$pagerecord = get_record_select('ouwiki_pages', "subwikiid='{$subwiki->id}' AND UPPER(title)=UPPER('" . addslashes($targetpage) . "')");
if ($pagerecord) {
$pageid = $pagerecord->id;
} else {
$pageid = false;
}
} else {
$pageid = get_field_select('ouwiki_pages', 'id', "subwikiid={$subwiki->id} AND title IS NULL");
}
if ($pageid) {
$link->topageid = $pageid;
$link->tomissingpage = null;
} else {
$link->topageid = null;
$link->tomissingpage = addslashes(strtoupper($targetpage));
}
if (!($link->id = insert_record('ouwiki_links', $link))) {
$tw->rollback();
ouwiki_dberror();
}
}
$link->topageid = null;
$link->tomissingpage = null;
$tl = textlib_get_instance();
foreach ($externallinks as $url) {
// Restrict length of URL
if ($tl->strlen($url) > 255) {
$url = $tl->substr($url, 0, 255);
}
$link->tourl = addslashes($url);
if (!($link->id = insert_record('ouwiki_links', $link))) {
$tw->rollback();
ouwiki_dberror();
}
}
// Inform search, if installed
if (ouwiki_search_installed()) {
$doc = new ousearch_document();
$doc->init_module_instance('ouwiki', $cm);
if ($subwiki->groupid) {
$doc->set_group_id($subwiki->groupid);
}
$doc->set_string_ref($pageversion->title === '' ? null : $pageversion->title);
if ($subwiki->userid) {
$doc->set_user_id($subwiki->userid);
}
$title = is_null($pageversion->title) ? '' : $pageversion->title;
if (!$doc->update($title, $content)) {
$tw->rollback();
ouwiki_dberror();
}
}
// Inform completion system, if available
if (class_exists('ouflags')) {
if (completion_is_enabled($course, $cm) && ($ouwiki->completionedits || $ouwiki->completionpages)) {
completion_update_state($course, $cm, COMPLETION_COMPLETE);
}
}
$tw->commit();
}
/**
* Update all documents for ousearch.
* @param bool $feedback If true, prints feedback as HTML list items
* @param int $courseid If specified, restricts to particular courseid
* @param int $cmid If specified, restricts to particular cmid
*/
static function search_update_all($feedback = false, $courseid = 0, $cmid = 0)
{
global $CFG;
// If cmid is specified, only retrieve that one
if ($cmid) {
$cmrestrict = "cm.id = {$cmid} AND";
} else {
$cmrestrict = '';
}
// Get module-instances that need updating
$cms = get_records_sql("\nSELECT\n cm.id, cm.course, cm.instance, f.name\nFROM\n {$CFG->prefix}forumng f\n INNER JOIN {$CFG->prefix}course_modules cm ON cm.instance=f.id\nWHERE\n {$cmrestrict}\n cm.module = (SELECT id FROM {$CFG->prefix}modules m WHERE name='forumng')" . ($courseid ? " AND f.course={$courseid}" : ''));
$cms = $cms ? $cms : array();
// Print count
if ($feedback && !$cmid) {
print '<li>' . get_string('search_update_count', 'forumng', '<strong>' . count($cms) . '</strong>') . '</li>';
}
// This can take a while, so let's be sure to reset the time limit.
// Store the existing limit; we will set this existing value again
// each time around the loop. Note: Despite the name, ini_get returns
// the most recently set time limit, not the one from php.ini.
$timelimitbefore = ini_get('max_execution_time');
// Loop around updating
foreach ($cms as $cm) {
forum_utils::start_transaction();
// Wipe existing search data, if any
ousearch_document::delete_module_instance_data($cm);
// Get all discussions for this forum
$discussions = get_records('forumng_discussions', 'forumid', $cm->instance, '', 'id, postid');
$discussions = $discussions ? $discussions : array();
if ($feedback) {
print '<li><strong>' . $cm->name . '</strong> (' . count($discussions) . '):';
}
// Process each discussion
foreach ($discussions as $discussionrec) {
// Ignore discussion with no postid
// (This should not happen, where ther is a $discussionrec->id
// it also shopuld have a $discussionrec->postid. This if-statement
// fixes bug 10497 and would not have any side-effect.)
if (!$discussionrec->postid) {
continue;
}
set_time_limit($timelimitbefore);
$discussion = forum_discussion::get_from_id($discussionrec->id, forum::CLONE_DIRECT, -1);
$root = $discussion->get_root_post();
$root->search_update();
$root->search_update_children();
print '. ';
flush();
}
forum_utils::finish_transaction();
if ($feedback) {
print '</li>';
}
}
}
/**
* Obtains search document representing this post.
* @return ousearch_document Document object
*/
function search_get_document()
{
$doc = new ousearch_document();
$doc->init_module_instance('forumng', $this->get_forum()->get_course_module());
if ($groupid = $this->discussion->get_group_id()) {
$doc->set_group_id($groupid);
}
$doc->set_int_refs($this->get_id());
return $doc;
}
/**
* Filters search results to pick out only the ones that match the query.
* @param array $results Array of results from internal_query
* @param int $desired Number of desired results
* @return object ->results containing actual results and ->dbnext containing
* database position of next set of results.
*/
function internal_filter($results, $desired)
{
global $CFG;
$required = array();
$accepted = array();
$count = 0;
$return = new StdClass();
$return->dbnext = 0;
$tl = textlib_get_instance();
foreach ($results as $result) {
$return->dbnext++;
if (substr($result->plugin, 0, 4) === 'mod/') {
// Module plugins
$module = substr($result->plugin, 4);
$function = $module . '_ousearch_get_document';
if (!array_key_exists($module, $required)) {
require_once $CFG->dirroot . '/mod/' . $module . '/lib.php';
$required[$module] = true;
if (!function_exists($function)) {
error('Missing module search support ' . $function);
}
}
} else {
if (substr($result->plugin, 0, 5) === 'test/') {
// Testing code, assumed to already be included
$function = substr($result->plugin, 5) . '_ousearch_get_document';
} else {
// Nothing else supported yet
error('Unsupported search plugin type ' . $result->plugin);
}
}
// Let's request the document. Note that the 'document' fields of
// $result are those used by this function to find the right one.
$page = $function($result);
// Ignore if we can't find the document
if (!$page) {
debugging('Module ' . $result->plugin . ' can\'t find search document');
ousearch_document::wipe_document($result->id);
continue;
}
// Page option can request that this result is not included
if (!empty($page->hide)) {
continue;
}
// Strip XHTML from content (need this before phrase scan)
$textcontent = ousearch_document::strip_xhtml($page->content);
// Add extra strings to the content after a special don't-show-this
// marker and with another special marker between each (to prevent
// phrases)
if (isset($page->extrastrings) && count($page->extrastrings) > 0) {
$evilmarker = rand();
// This means people can't do it on purpose
$textcontent .= ' xxrealcontentends' . $evilmarker;
foreach ($page->extrastrings as $string) {
$textcontent .= ' ' . $string . ' xxsplit' . $evilmarker;
}
}
// Do quick phrase scan that doesn't deal with Unicode,
// or word-splitting but just discards results that
// don't have the phrase words next to each other without
// ASCII letters in between. This is intended to discard
// results that (fairly) definitely don't have the phrase.
// The further check below will make sure they really do
// have it according to our standard (slow) word-splitting.
$quickcheckcontent = $page->title . ' ' . $textcontent;
$ok = true;
foreach ($this->terms as $term) {
if (count($term->words) < 2) {
continue;
}
$gap = '[^A-Za-z0-9]+';
$pattern = '/(^|' . $gap . ')';
$first = true;
foreach ($term->words as $word) {
if ($first) {
$first = false;
} else {
$pattern .= $gap;
}
$pattern .= $word;
}
$pattern .= '($|' . $gap . ')/i';
if (!preg_match($pattern, $quickcheckcontent)) {
$ok = false;
break;
}
}
if (!$ok) {
continue;
}
// OK, obtain document as linear text
list($contentwords, $contentpositions) = ousearch_document::split_words($textcontent, false, true);
list($titlewords, $titlepositions) = ousearch_document::split_words($page->title, false, true);
$allwords = array_merge($titlewords, $contentwords);
// Check it for phrases
$positivewords = array();
$ok = true;
$DNIfound = -1;
foreach ($this->terms as $term) {
foreach ($term->words as $word) {
$positivewords[$word] = true;
}
if (count($term->words) < 2) {
continue;
}
$pos = 0;
$found = false;
foreach ($allwords as $word) {
if ($word === $term->words[$pos]) {
$pos++;
if ($pos === count($term->words)) {
$found = true;
break;
}
} else {
$pos = 0;
}
}
if (!$found) {
$ok = false;
break;
}
}
foreach ($this->negativeterms as $term) {
if (count($term->words) < 2) {
continue;
}
$pos = 0;
$found = false;
foreach ($allwords as $word) {
if ($word === $term->words[$pos]) {
$pos++;
if ($pos === count($term->words)) {
$found = true;
break;
}
} else {
$pos = 0;
}
}
if ($found) {
$ok = false;
break;
}
}
if (!$ok) {
continue;
}
// Result passes! Make structure holding it...
// We now have list of all positive words, let's mark these
// in title and summary
$result->title = self::internal_highlight_words($page->title, $titlewords, $titlepositions, $positivewords);
// Strip searchable-but-not-displayable content for summary
if (isset($evilmarker)) {
$strippedwords = array();
foreach ($contentwords as $word) {
// Do not include extra strings in summary
if ($word == 'xxrealcontentends' . $evilmarker) {
break;
}
$strippedwords[] = $word;
}
$contentwords = $strippedwords;
}
// Pick a section to include in the summary. This algorithm works as follows:
// * Compute the 'score' (number of highlight words in the previous 20 words
// up to and including this one) at each position in the text
// * Observe where the maximum score is reached and where it is lost.
// * A nice range that contains the most highlight words in the middle of the
// range will end at ($maxstart + $maxlength/2).
$highlights = array();
$pos = 0;
$currentscore = 0;
$maxscore = -1;
$maxstart = 0;
$maxlength = 0;
$run = true;
foreach ($contentwords as $word) {
if (array_key_exists($pos - OUSEARCH_SUMMARYLENGTH, $highlights)) {
unset($highlights[$pos - OUSEARCH_SUMMARYLENGTH]);
$currentscore--;
}
if (array_key_exists($word, $positivewords)) {
$highlights[$pos] = true;
$currentscore++;
}
if ($currentscore > $maxscore) {
$maxscore = $currentscore;
$maxstart = $pos;
$maxlength = 1;
$run = true;
} else {
if ($currentscore === $maxscore && $run) {
$maxlength++;
} else {
$run = false;
}
}
$pos++;
}
$start = $maxstart + $maxlength / 2 - OUSEARCH_SUMMARYLENGTH;
if ($start < 0) {
$start = 0;
}
$end = $start + OUSEARCH_SUMMARYLENGTH;
if ($end > count($contentwords)) {
$end = count($contentwords);
}
// $contentpositions is in characters
$result->summary = $tl->substr($textcontent, $contentpositions[$start], $contentpositions[$end] - $contentpositions[$start]) . ($end < count($contentwords) ? '...' : '');
$offset = -$contentpositions[$start];
$result->summary = self::internal_highlight_words($result->summary, $contentwords, $contentpositions, $positivewords, $offset, $start, $end);
if ($start !== 0) {
$result->summary = '...' . $result->summary;
}
$result->summary = trim($result->summary);
$result->activityname = $page->activityname;
$result->activityurl = $page->activityurl;
$result->url = $page->url;
if (isset($page->data)) {
$result->data = $page->data;
}
// Do user-specified filter if set
if ($this->filter) {
$filter = $this->filter;
if (!$filter($result)) {
continue;
}
}
$accepted[] = $result;
$count++;
if ($count == $desired) {
break;
}
}
$return->results = $accepted;
return $return;
}
