Add new find methods to FieldtypePage, FieldtypeTextarea and MarkupQA for finding page references and links specific to a given page. The public interfaces to these will be in the Page class (coming in next commit).

wire/core/MarkupQA.php

@@ -415,6 +415,7 @@ class MarkupQA extends Wire {
 				$languageID = 0;
 			}
 
+			$pageID = (int) $pageID;
 			$full = $matches[0][$key];
 			$start = $matches[1][$key];
 			$href = $matches[3][$key];
@@ -427,7 +428,7 @@ class MarkupQA extends Wire {
 				$language = null;
 			}
 			
-			$livePath = $this->wire('pages')->getPath((int) $pageID, array(
+			$livePath = $this->wire('pages')->getPath($pageID, array(
 				'language' => $language
 			));
 			
@@ -469,6 +470,108 @@ class MarkupQA extends Wire {
 		}
 	}
 
+	/**
+	 * Find pages linking to another
+	 * 
+	 * @param Page $page Page to find links to, or omit to use page specified in constructor
+	 * @param array $fieldNames Field names to look in or omit to use field specified in constructor
+	 * @param string $selector Optional selector to use as a filter
+	 * @param array $options Additional options
+	 *  - `getIDs` (bool): Return array of page IDs rather than Page instances. (default=false)
+	 *  - `getCount` (bool): Return a total count (int) of found pages rather than Page instances. (default=false)
+	 *  - `confirm` (bool): Confirm that the links are present by looking at the actual page field data. (default=true)
+	 *     You can specify false for this option to make it perform faster, but with a potentially less accurate result.
+	 * @return PageArray|array|int
+	 * 
+	 */
+	public function findLinks(Page $page = null, $fieldNames = array(), $selector = '', array $options = array()) {
+		
+		$defaults = array(
+			'getIDs' => false,
+			'getCount' => false, 
+			'confirm' => true 
+		);
+
+		$options = array_merge($defaults, $options);
+		
+		if($options['getIDs']) {
+			$result = array();
+		} else if($options['getCount']) {
+			$result = 0;
+		} else {
+			$result = $this->wire('pages')->newPageArray();
+		}
+		
+		if(!$page) $page = $this->page;
+		if(!$page) return $result;
+		
+		if(empty($fieldNames)) {
+			if($this->field) $fieldNames[] = $this->field->name;
+			if(empty($fieldNames)) return $result;
+		}
+		
+		if($selector === true) $selector = "include=all";
+		$op = strlen("$page->id") > 3 ? "~=" : "%=";
+		$selector = implode('|', $fieldNames) . "$op'$page->id', id!=$page->id, $selector";
+		$selector = trim($selector, ', ');
+		
+		
+		// find pages
+		if($options['getCount'] && !$options['confirm']) {
+			// just return a count
+			return $this->wire('pages')->count($selector);
+		} else {
+			// find the IDs
+			$checkIDs = array();
+			$foundIDs = $this->wire('pages')->findIDs($selector);
+			if(!count($foundIDs)) return $result;
+			if($options['confirm']) {
+				$checkIDs = array_flip($foundIDs);
+				$foundIDs = array();
+			}
+		}
+		
+		// confirm results
+		foreach($fieldNames as $fieldName) {
+			if(!count($checkIDs)) break;
+			$field = $this->wire('fields')->get($fieldName);
+			if(!$field) continue;
+			$table = $field->getTable();
+			$ids = implode(',', array_keys($checkIDs));
+			$sql = "SELECT * FROM `$table` WHERE `pages_id` IN($ids)";
+			$query = $this->wire('database')->prepare($sql);
+			$query->execute();
+
+			while($row = $query->fetch(\PDO::FETCH_ASSOC)) {
+				$pageID = (int) $row['pages_id'];
+				if(isset($foundIDs[$pageID])) continue;
+				$row = implode(' ', $row);
+				$find = "data-pwid=$page->id";
+				// first check if it might be there
+				if(!strpos($row, $find)) continue;
+				// then confirm with a more accurate check
+				if(!strpos($row, "$find ") && !strpos($row, "$find\t") && !strpos($row, "$find-")) continue;
+				// at this point we have confirmed that this item links to $page
+				unset($checkIDs[$pageID]);
+				$foundIDs[$pageID] = $pageID;
+			}
+			
+			$query->closeCursor();
+		}
+	
+		if(count($foundIDs)) {
+			if($options['getIDs']) {
+				$result = $foundIDs;
+			} else if($options['getCount']) {
+				$result = count($foundIDs); 
+			} else {
+				$result = $this->wire('pages')->getById($foundIDs);
+			}
+		}
+	
+		return $result;
+	}
+
 	/**
 	 * Display and log a warning about a path that didn't resolve
 	 * 
@@ -554,7 +657,7 @@ class MarkupQA extends Wire {
 			list($name, $val) = explode('=', $attr);
 
 			$name = strtolower($name);
-			$val = trim($val, "\"' ");
+			$val = trim($val, "\"'> ");
 
 			if($name == 'alt' && !strlen($val)) {
 				$replaceAlt = $attr;

wire/modules/Fieldtype/FieldtypePage.module

@@ -9,7 +9,7 @@
  * /wire/core/Fieldtype.php
  * /wire/core/FieldtypeMulti.php
  * 
- * ProcessWire 3.x, Copyright 2016 by Ryan Cramer
+ * ProcessWire 3.x, Copyright 2018 by Ryan Cramer
  * https://processwire.com
  *
  */
@@ -19,7 +19,7 @@ class FieldtypePage extends FieldtypeMulti implements Module, ConfigurableModule
 	public static function getModuleInfo() {
 		return array(
 			'title' => 'Page Reference',
-			'version' => 104,
+			'version' => 105,
 			'summary' => 'Field that stores one or more references to ProcessWire pages',
 			'permanent' => true, 
 			);
@@ -1367,8 +1367,6 @@ class FieldtypePage extends FieldtypeMulti implements Module, ConfigurableModule
 	/**
 	 * Return pages referencing the given $page, optionally indexed by field name
 	 * 
-	 * Not currently applicable, for future use. 
-	 *
 	 * The default behavior when no arguments are provided (except $page) is to simply return a PageArray of all pages
 	 * referencing the given one (excluding hidden, unpublished, trash, no-access, etc.). Specify "include=all"
 	 * as your $selector if you want to include all pages without filtering. If the quantity may be large,
@@ -1377,26 +1375,37 @@ class FieldtypePage extends FieldtypeMulti implements Module, ConfigurableModule
 	 *
 	 * @param Page $page Page to get references for
 	 * @param string|bool $selector Optionally filter/modify returned result, i.e. "limit=10, include=all", etc.
-	 * 	Or specify boolean TRUE to get a count. 
+	 *  Or boolean TRUE as shortcut for "include=all". 
 	 * @param bool|Field|null $field Optionally specify Field to limit results to (default includes all fields of this type),
 	 * 	Or boolean TRUE to return array indexed by field name.
+	 * @param bool $getCount Specify true to get a count (int) rather than a PageArray (default=false)
 	 * @return PageArray|array|int Returns one of the following, according to the provided arguments:
 	 * 	- returns PageArray as default behavior, including when given a $selector string and/or Field object.
 	 * 	- returns array of PageArray objects if $field argument is TRUE ($selector may be populated string or blank string).
 	 *  - returns int if the count option (boolean true) specified for $selector.
 	 *
 	 */
-	private function findReferences(Page $page, $selector = '', $field = false) {
+	public function findReferences(Page $page, $selector = '', $field = false, $getCount = false) {
+
+		/** @var Pages $pages */
+		$pages = $this->wire('pages');
 		
 		// modifier option defaults
-		$getCount = false;
 		$byField = false;
+		$includeAll = $selector === true || $selector === "include=all";
+		$findLimit = 200;
 		$fieldName = '';
 
-		// determine which modifier options are used
+		// determine whether to use include=all
 		if($selector === true) {
-			$getCount = true;
-			$selector = '';
+			$selector = "include=all";
+		} else if(strlen($selector) && !$includeAll && strpos($selector, "include=all") !== false) {
+			foreach(new Selectors($selector) as $s) {
+				if($s->field() === 'include' && $s->value() === 'all') {
+					$includeAll = true;
+					break;
+				}
+			}
 		}
 
 		if(is_bool($field) || is_null($field)) {
@@ -1411,48 +1420,72 @@ class FieldtypePage extends FieldtypeMulti implements Module, ConfigurableModule
 		}
 
 		// results
-		$fieldNames = array();
-		$fieldCounts = array();
+		$fieldNames = array(); // field names that point to $page, array of [ field_id => field_name ]
+		$fieldCounts = array(); // counts indexed by field name (if count mode)
 		$total = 0;
 	
+		// first determine which fields have references to $page
 		foreach($this->wire('fields') as $field) {
+
 			if($fieldName && $field->name != $fieldName) continue;
 			if(!$field->type instanceof FieldtypePage) continue;
+
 			$table = $field->getTable();
 			$sql = "SELECT COUNT(*) FROM `$table` WHERE data=:id";
 			$query = $this->wire('database')->prepare($sql);
 			$query->bindValue(':id', $page->id, \PDO::PARAM_INT);
 			$query->execute();
+
 			$cnt = (int) $query->fetchColumn();
-			if($cnt) {
+			if($cnt > 0) {
 				$fieldNames[$field->id] = $field->name;
 				$fieldCounts[$field->name] = $cnt;
 				$total += $cnt;
 			}
+
 			$query->closeCursor();
 		}
 
-		// return count or array of counts
-		if($getCount) return $byField ? $fieldCounts : $total;
+		// if they just asked for the count, then we have all that we need to finish now
+		if($getCount && $includeAll) {
+			// return count or array of counts
+			return $byField ? $fieldCounts : $total;
+		}
 
-		// no references found and PageArray requested, return blank PageArray
-		if(!$total) return $this->wire('pages')->newPageArray();
+		// if there was nothing found, finish early
+		if(!$total) {
+			// no references found and PageArray requested, return blank PageArray
+			return $byField ? array() : $pages->newPageArray();
+		}
 	
+		// perform another find() to filter results, and return requested result type
 		if($byField) {
 			// return array of PageArrays indexed by fieldName
 			$result = array();
 			foreach($fieldNames as $fieldName) {
-				$selector = "$fieldName=$page->id, $selector";
-				$items = $this->wire('pages')->find(rtrim($selector, ", "));
-				if($items->count()) {
-					$result[$fieldName] = $items;
+				$s = rtrim("$fieldName=$page->id, $selector", ', ');
+				if($getCount) {
+					$cnt = $pages->count($s);
+					if($cnt) $result[$fieldName] = $cnt;
+				} else {
+					if($total > $findLimit) {
+						$items = $pages->findMany($s);
+					} else {
+						$items = $pages->find($s);
+					}
+					if($items->count()) $result[$fieldName] = $items;
 				}
 			}
-
 		} else {
 			// return PageArray of all references
-			$selector = implode('|', $fieldNames) . "=$page->id, $selector";
-			$result = $this->wire('pages')->find(rtrim($selector, ", "));
+			$selector = rtrim(implode('|', $fieldNames) . "=$page->id, $selector", ', ');
+			if($getCount) {
+				$result = $pages->count($selector);
+			} else if($total > $findLimit) {
+				$result = $pages->findMany($selector);
+			} else {
+				$result = $pages->find($selector);
+			}
 		}
 
 		return $result;

wire/modules/Fieldtype/FieldtypeTextarea.module

@@ -8,7 +8,7 @@
  * For documentation about the fields used in this class, please see:  
  * /wire/core/Fieldtype.php
  * 
- * ProcessWire 3.x, Copyright 2016 by Ryan Cramer
+ * ProcessWire 3.x, Copyright 2018 by Ryan Cramer
  * https://processwire.com
  * 
  * Properties set to $field that is using this type, acceessed by $field->get('property'):
@@ -24,7 +24,7 @@ class FieldtypeTextarea extends FieldtypeText {
 	public static function getModuleInfo() {
 		return array(
 			'title' => 'Textarea',
-			'version' => 106,
+			'version' => 107,
 			'summary' => 'Field that stores multiple lines of text',
 			'permanent' => true,
 		);
@@ -373,5 +373,43 @@ class FieldtypeTextarea extends FieldtypeText {
 		return $value;
 	}
 
+	/**
+	 * Find abstracted HTML/href attribute Textarea links to given $page
+	 * 
+	 * @param Page $page Find links to this page
+	 * @param string|bool $selector Optionally filter by selector or specify boolean true to assume "include=all". 
+	 * @param string|Field $field Optionally limit to searching given field name/instance.
+	 * @param array $options Options to modify return value: 
+	 *  - `getIDs` (bool): Return array of page IDs rather than Page instances. (default=false)
+	 *  - `getCount` (bool): Return a total count (int) of found pages rather than Page instances. (default=false)
+	 *  - `confirm` (bool): Confirm that the links are present by looking at the actual page field data. (default=true)
+	 *     You can specify false for this option to make it perform faster, but with a potentially less accurate result.
+	 * @return PageArray|array|int
+	 * 
+	 */
+	public function findLinks(Page $page, $selector = '', $field = '', array $options = array()) {
+
+		$searchFields = array();
+		if($selector === true) $selector = "include=all";
+		
+		foreach($this->wire('fields') as $f) {
+			if($field) {
+				if("$f" != "$field") continue;
+			} else {
+				// limit to fields with contentTypeHTML and htmlLinkAbstract
+				$contentType = $f->get('contentType');
+				if(empty($contentType)) continue;
+				if($contentType != self::contentTypeHTML && $contentType != self::contentTypeImageHTML) continue;
+				$htmlOptions = $f->get('htmlOptions');
+				if(!is_array($htmlOptions) || !in_array(self::htmlLinkAbstract, $htmlOptions)) continue;
+			}
+			$searchFields[$f->name] = $f->name;
+		}
+		
+		if(!count($searchFields)) return $this->wire('pages')->newPageArray();
+		
+		return $this->markupQA()->findLinks($page, $searchFields, $selector, $options); 
+	}
+
 }