<?php
/* Search snippet.
 * TS, March 2009. v1.4 beta		http://timspencerweb.co.uk/modx-search
 * Last modified: TS, 17/6/2010
 *
 * Released under the GNU GPL as is MODx http://modxcms.com/license.html
 *
 * Can search pagetitle, longtitle, description, alias, introtext, content, menutitle or any custom template variable
 * 
 * Ranks results firstly according to the number of different search terms matched
 * 	(but not according to the number of individual matches of the same term)
 *  	and secondly according to edit date - most recent first
 * 
 * By default ignores search terms with less than three letters
 *
 * Detects database chracter set using the MODx config
 * Expects documents to be standards compliant. Ampersands should be encoded.
 *
 * -------------------------------------------------------------------------------------------------------------------
 *
 * Revisions for 1.4
 *	- Modified to use double quoted search strings
 *	- Re-search form included in default searchNoResultTpl (not previously possible within the templating)
 *	- More to come...TODO...
 *			- Modify re-search form to work with non-friendly URLs
 *			- Prioritise headings (see below)
 *			- NOT search
 *
 * Revisions for 1.3.*:
 *	- Changed so that the plugin works with the current AjaxSearch highlight plugin (implode highlighted terms with ' ' not '+')
 *	- Fixes for PHP 4 (!). Added but not supported. Multi-byte character support required.
 *
 * Revisions for version 1.3:
 *	- searchParents/searchExcludeParents function added 
 *	- fixed bug regarding documents that are not in a document group the current user can view
 *	- Database character set now set manually - defaults to UTF-8.
 *	- Exact phrase matches (i.e. entire search term input) now ranked first
 *	- Facility to include a re-search form in the templates (default searchStartTpl now includes this)
 *
 * Revisions for version 1.2:
 *	- Pagination added
 * 	- Templating extended
 *	- non-UTF8 character sets catered for [afternote - no they were not - oops]
 *	- strip tags from search terms
 *
 * Revisions for version 1.1:
 * 	- No search terms results in message to that effect as opposed to the empty string returned in 1.0
 *	- If search extract code fails to find terms within the maintext, title attributes of all tags (bar <a...> tags) are then searched
 *			(<a...> tag's titles are deemed here to 'belong' to the target/linked page instead)
 *	- Code to construct extracts updated to use mb_* functions. Copes now with multibyte charcters
 * 
 * -------------------------------------------------------------------------------------------------------------------
 *
 * USEAGE:
 *		[!Search!]	- the default behaviour; searches pagetitle, longtitle, introtext, content
 *
 *		[!Search?
 * 			&searchFields=`pagetitle, longtitle, content, Year, RepairNotes`	- search these TVs
 * 			&tpl=`SearchResultsTpl`							- Template for each document found
 * 			&searchStartTpl=`SearchInfoTpl`!] 					- Template to put at the start 
 * 
 * INPUTS:
 *		$searchParents		- optional space and/or comma delimited list of document roots to search. If set but empty, don't search anything.
 *		$searchExcludeParents	- optional space and/or comma delimited list of document roots to not search (overides $searchParents). If set but empty, has no effect.
 *	 	$searchTerms  		- space and/or comma delimited list of search terms (default: searchTerms i.e. looks in $_REQUEST['searchTerms'])
 * 		$searchFields 		- space and/or comma delimited list of TVs to search (default: pagetitle, longtitle, introtext, content)
 *		$searchExtractChars 	- Maximum number of chars in extract (default: 200). If set to zero, the default value is used.
 *		$searchMaxDocs		- Maximum number of docs to find (if unset or zero all are listed). If pagination used this refers to the total found, not just the number on this page.
 * 		$searchMinLetters	- Shortest length of search terms to use (default: 3). Warning issued if shorter lengths used.
 * 		$searchOutputList 	- Output document list as unordered list (default: 1) i.e. surrounds results listing with <ul></ul>
 *		$searchStartTpl 	- Template to output at the start (default: <p class="search-info">[+search.finds+] document[+search.plural+] found containing [+search.terms+].</p>[+search.form+][+search.pagination+]).
 *		$searchEndTpl 		- Template to output at the end (default: [+search.pagination+]).
 * 		$searchNoResultTpl	- Output if no results found (default: <p id="search-no-results">No Results Found</p>[+search.form+])
 *		$searchNoInputTpl	- Output if no input given (default: <p id="search-no-results">No search terms given.</p>
 * 		$tpl 			- Template for individual documents. (default: <h3>[+search.result_link+][+pagetitle+][+/link+]</h3>[+search.terms_found+][+search.result_extract+])
 *		$searchPlural 		- Plural character(s) for templating; applies to [+search.plural+] placeholder. (Default: 's')
 *		$mb_charset		- Character set received from MySQL; sets mb_internal_encoding and used for entity de/encoding. Defaults to 'UTF-8'.
 *
 * INPUTS FOR PAGINATION:
 *		$searchPaginate			- Set to 1 to enable pagination
 *		$searchPaginationTpl		- Template to insert for [+search.pagination+] if needed.
 *							(Default:	<div class="search-pagination">
 *										<p class="search-pagination-message">Displaying [+search.page_start+] to [+search.page_end+] of [+search.finds+] result[+search.plural+].</p>
 *										<div class="search-pagination-itemsperpage">
 *											[+search.itemsperpage+]
 *										</div>
 *										<div class="search-pagination-links">
 *											[+search.goto_first+] | [+search.goto_prev+] | [+search.goto_pages+] | [+search.goto_next+] | [+search.goto_last+]
 *										</div>
 *									</div>)
 *		$searchPaginateLinkText 		- comma delimited link text (default: First,Previous,Next,Last)
 *		$searchPaginateLinksSeparator 		- Separator between page links (default: ' | ')
 *		$searchPaginateItemsPerPage 		- Items per page (default: 10)
 *		$searchPaginateItemsPerPageLabel	- Label for select box (for above options) (default: 'Items per page')
 *		$searchPaginateItemsPerPageOptions 	- Options for 'items per page' select box as a space and/or comma delimited list.
 *									$searchPaginateItemsPerPage will be included by the snippet if missing from this list.
 * 									(default: "5, 10, 25, 50")
 *
 * TEMPLATING:
 *		[+search.finds+]	- total number of documents found (NOTE - if $searchMaxDocs used this will show the number displayed)
 *		[+search.terms+]	- search terms used
 *		[+search.plural+]	- inserts $searchPlural if >1 document found
 *		[+search.pagination+]	- inserts pagination template ($searchPaginationTpl) if used and required
 *
 * TEMPLATING FOR PAGINATION ($searchPaginationTpl chunk only)
 *		[+search.page_start+]	- index of first item displayed
 *		[+search.page_end+]	- index of last item displayed
 *		[+search.page_items+]	- items displayed per page
 *		[+search.goto_first+]   - link to first 'page' or if current page, output a dummy <span> instead of <a>
 *		[+search.goto_last+]	- link to last 'page' or if current page, output a dummy <span> instead of <a>
 *		[+search.goto_prev+]	- link to previous 'page' or if current page, output a dummy <span> instead of <a>
 *		[+search.goto_next+]	- link to next 'page' or if current page, output a dummy <span> instead of <a>
 *		[+search.goto_pages+]	- links to all 'pages' or if current page, output a dummy <span> instead of <a>
 *		[+search.itemsperpage+]	- Select box to change items per page (JS only, not displayed if JS not present)
 * 
 * TEMPLATING WITHIN INDIVIDUAL SEARCH RESULTS ($tpl chunk only):
 *		[+search.result_posn+]			- position in search results
 *		[+search.result_terms+]			- Shows terms found in individual documents if (and only if) more than one search term entered.
 *							  (Wrapped by div.search-terms-found)
 *		[+search.result_link+]....[+/link]	- Link to document
 *		TVs can be inserted using [+TV+] e.g. [+pagetitle+], [+customTV+]
 *
 * NON-TEMPLATED OUTPUT:
 *		EXTRACTS:
 *			p.search-result-extract - extract for each result
 *			div.search-terms-found	- wraps [+search.result_terms+]
 *			span.highlight 		- used in extracts for highlighting of search terms within text
 *		LISTS:
 *			ul#search-result-list and li.search-result-extract
 *	 					- used if $searchOutputList is 1 (the default)
 *		JAVASCRIPT: (items per page functionality)
 *			.search-JS-only 	- elements to be displayed only if JS enabled (this being handled by the snippet)
 *
 * -------------------------------------------------------------------------------------------------------------------
 *
 * ISSUES: 	Will not find words that contain HTML tags - minor
 *		Will not find words that contain &shy; - very minor
 * 		If documents are found correctly that also have characteristics of false positives,
 * 			the false positive search terms are still displayed - very minor
 *
 * TODO		Prioritise on headings (1.4)
 *			- cycle through search finds, checking content fields for <h\d>([^<]|\<(?!\/h))*<term> Cannot use MySQL REGEXP due to Lookahead.
 * 			  Do this prior to uasort and modify uasort's anonymous function. Just need to check terms from $search_terms_found.
 *		- implement NOT/- search
 *		- miss weblink content link out???
 *
 * FUTURE PORTABILITY ISSUES:
 *		Use of mysql_* functions
 *
 * -------------------------------------------------------------------------------------------------------------------
 */
 
// PHP 4 fix.
if (!function_exists("mb_stripos")) {
  function mb_stripos($str,$needle) {
   return mb_strpos(strtolower($str),strtolower($needle));
  }
}

/********************************/
/* Check and process parameters */
/********************************/

// Set character set for mb_* functions
$mb_charset = isset($mb_charset) ? $mb_charset : 'UTF-8';
mb_internal_encoding($mb_charset);

$searchTerms = isset($searchTerms) ? $searchTerms : 'searchTerms';

// Convert searchTerms to actual string
$search_string = trim($_REQUEST[$searchTerms]);
if (get_magic_quotes_gpc()) $search_string = stripslashes($search_string);
$search_string = strip_tags($search_string);
	
// Check search terms are present
if (empty($search_string)) return isset($searchNoInputTpl)  ? $modx->getChunk($searchNoInputTpl)  : '<p id="search-no-results">No search terms given.</p>';  /* !! Early Return !! */

// Convert $search_string to arrays
$search_terms_array_quoted = array();
// Note that the search is case insensitive (uses SQL 'LIKE'). The next line ensures the calls to array_unique() below filter out repeated values that differ in case only.
$search_string_copy = strtolower($search_string);
// First look for double quoted terms.
while (($pos = mb_strpos($search_string_copy, '"')) !== false)
	{
	if ($pos2 = mb_strpos($search_string_copy, '"', $pos+1))
		{
		$search_terms_array_quoted[] = mb_substr($search_string_copy, $pos+1, $pos2 - $pos - 1);
		$search_string_copy = ($pos ? mb_substr($search_string_copy, 0, $pos-1) : '').' '.(($pos2 < mb_strlen($search_string_copy) - 1) ? mb_substr($search_string_copy, $pos2+1) : '');
		}
	else
		{
		// incomplete? Finish looping
		break;
		}
	}
// Filter out repeated values.
$search_terms_array_quoted = array_unique($search_terms_array_quoted);

// If anything is left in $search_string_copy, then secondly split string based on spaces and/or commas.
if ($search_string_copy = trim($search_string_copy))
	{
	$search_terms_array_splitonspaces = array_unique(preg_split('/[\s,]+/', $search_string_copy)); // Note that repeated values are filtered out.
	$search_terms_array = array_unique(array_merge($search_terms_array_splitonspaces, $search_terms_array_quoted)); // ditto
	}
else
	{
	$search_terms_array_splitonspaces = array();
	$search_terms_array = $search_terms_array_quoted;
	}

// Add an extra 'term' - that of the entire search string, unprocessed. This will ensure that exact phrase matches have a $search_finds[n]['count'] higher than all other documents.
// Note that we do not bother if the phrase is already there. This would mean that one word or one quoted phrase has been entered, but we don't check
// if (sizeof($search_terms_array) > 1) as this wouldn't cover situations such as 'around 10' where the 10 would not by default be added to $search_terms_array.
if (!in_array($search_string, $search_terms_array)) $search_terms_array[] = $search_string;

// Terms to highlight
$search_highlight_plugin_parameters = '&searched='.urlencode(implode(' ', $search_terms_array)).'&highlight=highlight';

// Convert $searchFields to array
$search_fields_array = isset($searchFields) ? preg_split('/[\s,]+/', $searchFields) : array('pagetitle', 'longtitle', 'introtext', 'content');

// Split $search_fields_array into default site content fields, and custom template variables. Store strings escaped for MySQL.
$search_std_tplvars = array('pagetitle', 'longtitle', 'description', 'alias', 'introtext', 'content', 'menutitle');
foreach($search_fields_array as $search_field)
	{
	if ($search_field)
		{
		if (in_array($search_field, $search_std_tplvars)) $search_std_fields[] = mysql_real_escape_string($search_field);
		else $search_custom_tv_fields[] = mysql_real_escape_string($search_field);
		}
	}	
	
// Set $searchExtractChars to default if not set
$searchExtractChars = $searchExtractChars ? $searchExtractChars : 200;
$searchExtractCharsHalved = $searchExtractChars >> 1;

// Convert template chunk names to text (or use defaults)
$tpl               = isset($tpl)               ? $modx->getChunk($tpl)               : '<h3>[+search.result_link+][+pagetitle+][+/link+]</h3>[+search.terms_found+][+search.result_extract+]';
$searchStartTpl    = isset($searchStartTpl)    ? $modx->getChunk($searchStartTpl)    : '<p class="search-info">[+search.finds+] document[+search.plural+] found containing [+search.terms+].</p>[+search.form+][+search.pagination+]';
$searchEndTpl      = isset($searchEndTpl)      ? $modx->getChunk($searchEndTpl)      : '[+search.pagination+]';
$searchNoResultTpl = isset($searchNoResultTpl) ? $modx->getChunk($searchNoResultTpl) : '<p id="search-no-results">No Results Found</p>[+search.form+]';

$searchPlural = isset($searchPlural) ? $searchPlural : 's';

// Output as (unordered) list?
$searchOutputList = isset($searchOutputList) ? $searchOutputList : 1;

// Minimum length of search terms
$searchMinLetters = isset($searchMinLetters) ? $searchMinLetters : 3;

// Pagination
if ($searchPaginate)
	{
	// Ensure $_GET params can override snippet parameters
	$searchPaginateItemsPerPage = ($_GET['search_itemsperpage'] && ctype_digit($_GET['search_itemsperpage'])) ?
						$_GET['search_itemsperpage'] : (isset($searchPaginateItemsPerPage) ? $searchPaginateItemsPerPage : 10);
	$search_paginate_start = ($_GET['search_start'] && ctype_digit($_GET['search_start'])) ? $_GET['search_start'] : 1;
	$searchPaginateLinksSeparator = isset($searchPaginateLinksSeparator) ? $searchPaginateLinksSeparator : ' | ';
	
	// Array of link text: 0=>First 1=>Prev 2=>Next 3=>Last
	$search_paginate_linktext_array = explode(',', $searchPaginateLinkText ? $searchPaginateLinkText : 'First,Previous,Next,Last');

	// User selectable items per page
	$search_paginate_itemsPerPage_options = isset($searchPaginateItemsPerPageOptions) ? preg_split('/[\s,]+/', $searchPaginateItemsPerPageOptions) : array (5, 10, 25, 50);
	if (!in_array($searchPaginateItemsPerPage, $search_paginate_itemsPerPage_options))
		{
		// Ensure current itemsperpage value is in option list
		$search_paginate_itemsPerPage_options[] = $searchPaginateItemsPerPage;
		sort($search_paginate_itemsPerPage_options);
		}
	$searchPaginateItemsPerPageLabel = isset($searchPaginateItemsPerPageLabel) ? $searchPaginateItemsPerPageLabel : 'Items per page';

	// Pagination template
	$searchPaginationTpl = isset($searchPaginationTpl) ? $modx->getChunk($searchPaginationTpl) :
				'<div class="search-pagination">
					<p class="search-pagination-message">Displaying [+search.page_start+] to [+search.page_end+] of [+search.finds+] result[+search.plural+].</p>
					<div class="search-pagination-itemsperpage">
						[+search.itemsperpage+]
					</div>
					<div class="search-pagination-links">
						[+search.goto_first+] | [+search.goto_prev+] | [+search.goto_pages+] | [+search.goto_next+] | [+search.goto_last+]
					</div>
				</div>';
				
	// Start of query string to pass on with pagination. Note quotes are encoded now to avoid issues of quotes within (X)HTML attributes.
	$search_pagination_query = $searchTerms.'='.str_replace('"', '%22', $search_string);
	}

// Check for $searchParents/$searchExcludeParents - if set we need to restrict the domain of the search by filtering the MySQL results

function getChildrenInDomain($id)
	{
	global $modx;
	
	$result = array($id);
	$child_docs = $modx->getActiveChildren($id, 'id', 'ASC', 'id');
	if ($child_docs && sizeof($child_docs))
		{
		foreach($child_docs as $child_doc)
			$result = array_merge($result, getChildrenInDomain($child_doc['id']));
		}

	return $result;
	}
	
if (strlen($searchParents))
	{
	$search_parents_array = preg_split('/[\s,]+/', $searchParents);
	$search_domain = array();
	foreach($search_parents_array as $search_parent) $search_domain = array_merge($search_domain, getChildrenInDomain($search_parent));
	}
elseif ($searchParents === '')
	$search_domain = array();

if (strlen($searchExcludeParents))
	{
	if (!isset($search_domain)) $search_domain = getChildrenInDomain(0);
	$search_exclude_parents_array = preg_split('/[\s,]+/', $searchExcludeParents);
	foreach($search_exclude_parents_array as $search_exclude_parent) $search_domain = array_diff($search_domain, getChildrenInDomain($search_exclude_parent));
	}

/**********/
/* Search */
/**********/

$modx_site_content = $modx->getFullTableName('site_content');
$modx_site_tmplvars = $modx->getFullTableName('site_tmplvars');
$modx_site_tmplvar_contentvalues = $modx->getFullTableName('site_tmplvar_contentvalues');

$search_output = '';
$search_output_warnings = array();

// $search_finds[$docid] will be incremented when a search term is found.
// $search_terms_found[$docid] is either unset or will contain an array of search terms.

$search_finds = array();
$search_terms_found = array();

// Loop through search terms
foreach($search_terms_array as $search_term_key=>$search_term)
	{
	if (mb_strlen($search_term, $mb_charset) < $searchMinLetters)
		{
		// Eliminate search terms that are too short 
		// Also gets rid of empty items in $search_terms_array that may have occured due to extraneous spaces and explode()
		$search_output_warnings[] = '<p id="search-warning">Ignoring search terms with less than '.$searchMinLetters.' letters.</p>';
		// Ensure that [+search.terms+] placeholder does not show null or irrelevant strings e.g. "term,,,,term2"
		unset($search_terms_array[$search_term_key]);
		}
	else
		{
		// Note that magic-quotes-derived slashes are stripped above
		// Consider all combinations of entities without duplicating search values
		$search_terms_escaped = array();
		$search_terms_escaped[] = mysql_real_escape_string($search_term);
		$search_terms_escaped[] = mysql_real_escape_string(htmlentities($search_term, ENT_NOQUOTES, $mb_charset));
		$search_terms_escaped[] = mysql_real_escape_string(htmlentities($search_term, ENT_COMPAT, $mb_charset));
		$search_terms_escaped[] = mysql_real_escape_string(htmlentities($search_term, ENT_QUOTES, $mb_charset));
		$search_terms_escaped = array_unique($search_terms_escaped);
		
		// Reset query parts from previous iterations
		$search_query1 = $search_query2 = '';
		
		// Standard TVs as stored in modx_site_content (note that $search_std_fields has already been escaped)
		if (is_array($search_std_fields))
			{
			foreach($search_std_fields as $search_std_field)
				$search_query1 .= " OR {$search_std_field} LIKE '%".implode("%' OR {$search_std_field} LIKE '%", $search_terms_escaped).'%\'';
			}

		// Custom TV fields (note that $search_custom_tv_fields has already been escaped)
		if (is_array($search_custom_tv_fields)) 
			{
			foreach($search_custom_tv_fields as $search_custom_tv_field) 
				$search_query2 .= " OR ({$modx_site_tmplvars}.name = '{$search_custom_tv_field}' AND
								({$modx_site_tmplvar_contentvalues}.value LIKE '%".implode("%' OR {$modx_site_tmplvar_contentvalues}.value LIKE '%", $search_terms_escaped).'%\'))';
			}

		if ($search_query1 && $search_query2) 	// Both standard and custom TV's present
			{			
			$search_result = mysql_query("
					SELECT id, editedon
						FROM {$modx_site_content}
						WHERE published = 1 AND searchable = 1 AND deleted=0 AND (".substr($search_query1, 4).")
					UNION DISTINCT
					SELECT {$modx_site_content}.id, {$modx_site_content}.editedon
						FROM {$modx_site_content}, {$modx_site_tmplvar_contentvalues}, {$modx_site_tmplvars}
						WHERE {$modx_site_content}.id = {$modx_site_tmplvar_contentvalues}.contentid AND {$modx_site_tmplvars}.id = {$modx_site_tmplvar_contentvalues}.tmplvarid
							AND {$modx_site_content}.published = 1 AND {$modx_site_content}.searchable = 1 AND {$modx_site_content}.deleted=0 AND (".substr($search_query2, 4).')');
			}
		elseif ($search_query1)			// Standard TVs only
			{
			$search_result = mysql_query("
					SELECT id, editedon
						FROM {$modx_site_content}
						WHERE published = 1 AND searchable = 1 AND deleted=0 AND (".substr($search_query1, 4).')');
			}
		elseif ($search_query2)			// Custom TVs only
			{
			$search_result = mysql_query("
					SELECT {$modx_site_content}.id, {$modx_site_content}.editedon
						FROM {$modx_site_content}, {$modx_site_tmplvar_contentvalues}, {$modx_site_tmplvars}
						WHERE {$modx_site_content}.id = {$modx_site_tmplvar_contentvalues}.contentid AND {$modx_site_tmplvars}.id = {$modx_site_tmplvar_contentvalues}.tmplvarid
							AND {$modx_site_content}.published = 1 AND {$modx_site_content}.searchable = 1 AND {$modx_site_content}.deleted=0 AND (".substr($search_query2, 4).')');
			}

		if ($search_result)
			{
			while($search_row = mysql_fetch_row($search_result))
				{
				// increase counts of docids found
				$search_finds[$search_row[0]]['count']++;
				// store last edit date for ordering
				$search_finds[$search_row[0]]['date'] = $search_row[1];
				// record search terms found and in which documents
				$search_terms_found[$search_row[0]][] = $search_term;
				}
			}
		}
	}
	
if (sizeof($search_finds))
	{
	
	// Put documents with the most finds (by distinct keywords) first, then by date edited (most recent first)
	uasort($search_finds,
		create_function('$a, $b',
			'if ($a[\'count\'] > $b[\'count\']) return -1;
			elseif ($a[\'count\'] < $b[\'count\']) return 1;
			elseif ($a[\'date\'] > $b[\'date\']) return -1;
			elseif ($a[\'date\'] < $b[\'date\']) return 1;
			else return 0;'));

	/**********/
	/* Output */
	/**********/

	$search_posn = 0;		// Current position in ALL results during construction of output.
					// At the end is the total number of items found
					
	$search_page_posn = 0;		// Current position within page during construction of output.
					// At the end is the number of items on this page.

	foreach(array_keys($search_finds) as $search_find_docid)
		{
		if ($searchMaxDocs && $search_posn >= $searchMaxDocs) break; // !! Early exit from loop !!
		
		if (isset($search_domain) && !in_array($search_find_docid, $search_domain)) continue; // !! Early jump to next iteration !!
		
		$search_template = $tpl; // Reset from previous iteration

		// Get template vars for the document.
		$search_tv_outputs = $modx->getTemplateVarOutput('*', $search_find_docid, 1);

		if (is_array($search_tv_outputs)) // Filters out documents that current user cannot access
			{

			// Parse template vars, and get their lengths (used when constructing extract)
			foreach($search_tv_outputs as $search_tv_name=>$search_tv_output)
				{
				// Parse
				$search_template = str_replace('[+'.$search_tv_name.'+]', $search_tv_output, $search_template);
				// Store lengths of outputs for use by extract code
				$search_tv_output_lengths[$search_tv_name] = strlen($search_tv_output);
				}

			// Parse and construct extract
			// This also acts as a final filter to eliminate documents that have been erroneously found 
			// due to matches within the code of HTML entities e.g. eta with &theta; See notes below re: false positive

			// Go through TVs in descending order of output length so as to try and form an extract using the longest TV
			arsort($search_tv_output_lengths);
		
			// First try to construct extract based on main body of template variables.
			// 	This excludes everything within tags e.g. title attributes
			foreach(array_keys($search_tv_output_lengths) as $search_tv_name)
				{
				// Only try to construct extracts from fields that were searched
				if (in_array($search_tv_name, $search_fields_array))
					{
					// Decode entities to avoid erroneous matchings within the entity codes
					// Strip tags for uniform extract styling
					$search_extract = html_entity_decode(strip_tags($search_tv_outputs[$search_tv_name]), ENT_QUOTES, $mb_charset);

					if ($search_extract) // Avoid mb_stripos error
						{
						foreach ($search_terms_array as $search_term)
							{
							// At some point in these two nested loops this will succeed unless we have a
							// false positive OR the search terms have been found within title attributes.
							// This means that on success we leave these loops with $search_extract_posn as an integer, or false on fail.
							if (($search_extract_posn = mb_stripos($search_extract, $search_term)) !== false)
								{
								// Found - make extract
								if (($search_extract_length = mb_strlen($search_extract)) > $searchExtractChars)
									{
									// Extract is smaller than its source TV
									if ($search_extract_length - $search_extract_posn > $searchExtractCharsHalved)
										{
										if ($search_extract_posn > $searchExtractCharsHalved)
											$search_extract = '...'.mb_substr($search_extract, $search_extract_posn - $searchExtractCharsHalved, $searchExtractChars).'...';
										else
											$search_extract = mb_substr($search_extract, 0, $searchExtractChars).'...';
										}
									else
										$search_extract = '...'.mb_substr($search_extract, -$searchExtractChars);
									}
					
								break 2;						
								}
							}
						}
					}
				}

			// If the above fails, then attempt to construct an extract using the title attributes.
			// Do not search <a...> tags - their titles are related to the linked page, not the source document.
			if ($search_extract_posn === false)
				{
				foreach(array_keys($search_tv_output_lengths) as $search_tv_name)
					{
					// Only try to construct extracts from fields that were searched
					if (in_array($search_tv_name, $search_fields_array))
						{
						// Decode entities to avoid erroneous matchings within the entity codes.
						// Find title attributes - attribute strings go into $title_attributes[2]
						$title_attributes = array();
						preg_match_all('/<\s*(a[a-z]|[^a])[^>]*title="([^"]*)"[^>]*>/', html_entity_decode($search_tv_outputs[$search_tv_name], ENT_QUOTES, $mb_charset), $title_attributes);
					
						if (is_array($title_attributes[2]))
							{
							foreach ($search_terms_array as $search_term)
								{
								foreach($title_attributes[2] as $title_attribute)
									{
									if ($title_attribute && ($search_extract_posn = mb_stripos($title_attribute, $search_term)) !== false)
										{
										$search_extract = $title_attribute;
										break 3;						
										}
									}
								}
							}
						}
					}
				}

			if ($search_extract_posn !== false) // i.e. if not a false positive
				{
				++$search_posn; // This records the first item as 1, not 0, and hence can be later used as a 
						// test for finds vs no finds, as well as showing the total number of finds.
			
				if (!$searchPaginate || ($search_posn >= $search_paginate_start && $search_posn < $search_paginate_start + $searchPaginateItemsPerPage))
					{
					++$search_page_posn;
				
					// Highlight extract
					// 	Ensure all terms highlighted, not just the one that identified the extract
					// 	Need to search for htmlentities in the extract as we have just converted it.
					foreach ($search_terms_array as $search_term) 
						{
						$search_highlight_posn = 0;
						$search_term_len = mb_strlen($search_term);
		
						while (($search_highlight_posn = mb_stripos($search_extract, $search_term, $search_highlight_posn)) !== false)
							{
							// Insert temporary placeholders until HTML entities re-encoded
							$search_extract = mb_substr($search_extract, 0, $search_highlight_posn).
								'[+search_highlight+]'.mb_substr($search_extract, $search_highlight_posn, $search_term_len).'[+/search_highlight+]'.
								mb_substr($search_extract, $search_highlight_posn + $search_term_len);
				
							// Advance offset to after this highlight (including temporary placeholders) for next iteration
							$search_highlight_posn += $search_term_len + 41; // strlen('[+search_highl.....') = 41
							}
						}

					// Re-encode (note this always encodes quotes regardless of the initial state in the TV)
					$search_extract = htmlentities($search_extract, ENT_QUOTES, $mb_charset);
	
					// Replace temporary script-inserted placeholders after running htmlentities
					$search_extract = str_replace('[+search_highlight+]', '<span class="highlight">', $search_extract);
					$search_extract = str_replace('[+/search_highlight+]', '</span>', $search_extract);
				
					// When finished with the individual extract, replace the placeholder in the template with this extract
					$search_template = str_replace('[+search.result_extract+]', '<p class="search-result-extract">'.$search_extract.'</p>', $search_template);				

					// Parse position
					$search_template = str_replace('[+search.result_posn+]', $search_posn, $search_template);

					// Parse search terms found in this document
					$search_template = str_replace('[+search.terms_found+]', sizeof($search_terms_array) > 1 ? '<div class="search-terms-found">Contains '.strtolower(implode(', ', $search_terms_found[$search_find_docid])).'.</div>' : '', $search_template);
	
					// Parse links
					$search_template = str_replace('[+search.result_link+]', '<a href="'.$modx->makeUrl($search_find_docid, '', $search_highlight_plugin_parameters).'" title="'.$search_tv_outputs['pagetitle'].'">', $search_template);
					$search_template = str_replace('[+/link+]', '</a>', $search_template);
	
					// Add this item to the output
					if ($searchOutputList) $search_output .= '<li class="search-result-item">'.$search_template.'</li>';
					else $search_output .= $search_template;
					}
				}
			}
		}

	/*******************************************/
	/* Check for success and make final output */
	/*******************************************/
	if ($search_posn) // Valid finds and no false positives
		{
		// Wrap in <ul>
		if ($searchOutputList) $search_output = '<ul id="search-result-list">'.$search_output.'</ul>';

		// Info at beginning or end e.g. "Found 3 documents containing burble"
		$search_output = $searchStartTpl.implode('', array_unique($search_output_warnings)).$search_output.$searchEndTpl;

		// Pagination parsing
		if ($searchPaginate && $search_posn > $searchPaginateItemsPerPage)
			{
			// Construct array of all possible start indices
			// At the same time build an array of links to these pages
			$search_new_start = 1;
			$search_new_starts = array();
			$search_pages_links = array();
			$search_pages_linktext = 1;
			while ($search_new_start <= $search_posn)
				{
				$search_new_starts[] = $search_new_start;
				$search_pages_links[] = $search_new_start != $search_paginate_start ? 
								'<a href="'.$modx->makeUrl($modx->documentIdentifier, '', "{$search_pagination_query}&search_itemsperpage={$searchPaginateItemsPerPage}&search_start={$search_new_start}")."\">{$search_pages_linktext}</a>" :
								"<span>{$search_pages_linktext}</span>"
								;
				$search_new_start += $searchPaginateItemsPerPage;
				$search_pages_linktext++;
				}
				
			// Numerical info on this page (start, end, number of items)
			$searchPaginationTpl = str_replace('[+search.page_start+]', $search_paginate_start, $searchPaginationTpl);		
			$searchPaginationTpl = str_replace('[+search.page_end+]', $search_paginate_start + $search_page_posn - 1, $searchPaginationTpl);
			$searchPaginationTpl = str_replace('[+search.page_items+]', $search_page_posn, $searchPaginationTpl);
			
			// Page links
			$searchPaginationTpl = str_replace('[+search.goto_pages+]', implode($searchPaginateLinksSeparator, $search_pages_links), $searchPaginationTpl);

			// First does not need to pass a start index
			$searchPaginationTpl = str_replace('[+search.goto_first+]',
							$search_paginate_start > 1 ?
							'<a href="'.$modx->makeUrl($modx->documentIdentifier, '', "{$search_pagination_query}&search_itemsperpage={$searchPaginateItemsPerPage}")."\">{$search_paginate_linktext_array[0]}</a>" :
							"<span>{$search_paginate_linktext_array[0]}</span>"
							, $searchPaginationTpl);
							
			// Last takes the last start index from $search_new__starts[]
			$searchPaginationTpl = str_replace('[+search.goto_last+]',
							$search_paginate_start != end($search_new_starts) ?
								'<a href="'.$modx->makeUrl($modx->documentIdentifier, '', "{$search_pagination_query}&search_itemsperpage={$searchPaginateItemsPerPage}&search_start=".end($search_new_starts))."\">{$search_paginate_linktext_array[3]}</a>" :
								"<span>{$search_paginate_linktext_array[3]}</span>"
								, $searchPaginationTpl);
							
			// Previous and Next go up and down in multiples of $searchPaginateItemsPerPage,
			// regardles of whether we are looking at a page with an expected start index, 
			// unless this would send the new start index out of range.
			$searchPaginationTpl = str_replace('[+search.goto_prev+]',
							$search_paginate_start > 1 ?
								'<a href="'.$modx->makeUrl($modx->documentIdentifier, '', "{$search_pagination_query}&search_itemsperpage={$searchPaginateItemsPerPage}&search_start=".max(1, $search_paginate_start - $searchPaginateItemsPerPage))."\">{$search_paginate_linktext_array[1]}</a>" :
								"<span>{$search_paginate_linktext_array[1]}</span>"
								, $searchPaginationTpl);
			$searchPaginationTpl = str_replace('[+search.goto_next+]',
							$search_paginate_start + $searchPaginateItemsPerPage <= $search_posn ? 
								'<a href="'.$modx->makeUrl($modx->documentIdentifier, '', "{$search_pagination_query}&search_itemsperpage={$searchPaginateItemsPerPage}&search_start=".min($searchPaginateItemsPerPage * floor($search_posn/$searchPaginateItemsPerPage) + 1, $search_paginate_start + $searchPaginateItemsPerPage))."\">{$search_paginate_linktext_array[2]}</a>" :
								"<span>{$search_paginate_linktext_array[2]}</span>"
								, $searchPaginationTpl);

			// User selectable items per page
			$search_paginate_itemsPerPage_select = ($searchPaginateItemsPerPageLabel ? "<label style=\"visibility: hidden; display: inline\" class=\"search_JS_only\" for=\"search_itemsperpage\">{$searchPaginateItemsPerPageLabel}</label>" : '').
								'<select style="visibility: hidden; display: inline" class="search_JS_only" name="search_itemsperpage"
									onchange="window.location.href=\''.$modx->makeUrl($modx->documentIdentifier, '', $search_pagination_query)."&amp;search_start='+(1+Math.floor(({$search_paginate_start}-1)/this.value)*this.value)+'&amp;search_itemsperpage='+this.value;\">";
			foreach($search_paginate_itemsPerPage_options as $search_paginate_itemsPerPage_option)
				$search_paginate_itemsPerPage_select .= '<option'.(($search_paginate_itemsPerPage_option == $searchPaginateItemsPerPage) ? ' selected="selected" ' : '')." value=\"{$search_paginate_itemsPerPage_option}\">{$search_paginate_itemsPerPage_option}</option>";
			$search_paginate_itemsPerPage_select .= '</select>';
			
			$searchPaginationTpl = str_replace('[+search.itemsperpage+]', $search_paginate_itemsPerPage_select, $searchPaginationTpl);
			}
		else
			{
			// Pagination off or not required
			$searchPaginationTpl = '';
			}

		// Insert Pagination - note this will be empty if poagination not used
		$search_output = str_replace('[+search.pagination+]', $searchPaginationTpl, $search_output);
		}
	else
		{
		// False positives elimated and no finds left
		$search_output = implode('', array_unique($search_output_warnings)).$searchNoResultTpl;
		}
	}
else
	{
	// No MySQL finds
	$search_output = implode('', array_unique($search_output_warnings)).$searchNoResultTpl;
	}

// Prepare search string as input by the user, for output in the templates.
$search_string_html = htmlentities($search_string, ENT_QUOTES, $mb_charset);

// The simplest equivalent of the search string actually used accounting for quotes (which are output as &quot;)
// Note that words less than the minimum length are included! <<<<
$search_string_used_html = htmlentities(strtolower(implode(', ', $search_terms_array_splitonspaces).(sizeof($search_terms_array_quoted) ? (sizeof($search_terms_array_splitonspaces) ? ', ' : '').'"'.implode('", "', $search_terms_array_quoted).'"': '')), ENT_QUOTES, $mb_charset);

// Final parsing
$search_output = str_replace('[+search.finds+]', $search_posn, $search_output);
$search_output = str_replace('[+search.terms+]', $search_string_used_html, $search_output);
$search_output = str_replace('[+search.plural+]', $search_posn > 1 ? $searchPlural : '', $search_output);

$search_output = str_replace('[+search.form+]', '<form id="search-re-search" method="get" action="'.$modx->makeUrl($modx->documentIdentifier).'">
							<fieldset>
								<input type="text" id="search-re-search-terms" name="'.$searchTerms.'" value="'.$search_string_html.'" />
								<input type="submit" id="search-re-search-submit" class="button" value="Search Again" />
							</fieldset>
						</form>', $search_output);

// Return and ensure select boxes and labels are displayed if javascript enabled
return $search_output.'<script type="text/javascript">
				(function()
					{
					var i, JS_elements;
					JS_elements = document.getElementsByTagName("select");
					for (i=0; i != JS_elements.length; ++i)
						if (JS_elements[i].className == "search_JS_only") JS_elements[i].style.visibility="visible";
					JS_elements = document.getElementsByTagName("label");
					for (i=0; i != JS_elements.length; ++i)
						if (JS_elements[i].className == "search_JS_only") JS_elements[i].style.visibility="visible";
					})();
			</script>';

?>