AbstractUriElement.php

@@ -115,6 +115,8 @@ protected function canonicalizePath(string $path): string
      *
      * @param \DOMElement $node A \DOMElement instance
      *
+     * @return void
+     *
      * @throws \LogicException If given node is not an anchor
      */
     abstract protected function setNode(\DOMElement $node);

CHANGELOG.md

@@ -1,6 +1,14 @@
 CHANGELOG
 =========
 
+6.3
+---
+
+ * Add `$useHtml5Parser` argument to `Crawler`
+ * Add `CrawlerSelectorCount` test constraint
+ * Add argument `$normalizeWhitespace` to `Crawler::innerText()`
+ * Make `Crawler::innerText()` return the first non-empty text
+
 6.0
 ---
 

Crawler.php

@@ -58,16 +58,17 @@ class Crawler implements \Countable, \IteratorAggregate
      */
     private bool $isHtml = true;
 
-    private HTML5 $html5Parser;
+
+    private ?HTML5 $html5Parser = null;
 
     /**
      * @param \DOMNodeList|\DOMNode|\DOMNode[]|string|null $node A Node to use as the base for the crawling
      */
-    public function __construct(\DOMNodeList|\DOMNode|array|string $node = null, string $uri = null, string $baseHref = null)
+    public function __construct(\DOMNodeList|\DOMNode|array|string $node = null, string $uri = null, string $baseHref = null, bool $useHtml5Parser = true)
     {
         $this->uri = $uri;
         $this->baseHref = $baseHref ?: $uri;
-        $this->html5Parser = new HTML5(['disable_html_ns' => true]);
+        $this->html5Parser = $useHtml5Parser ? new HTML5(['disable_html_ns' => true]) : null;
         $this->cachedNamespaces = new \ArrayObject();
 
         $this->add($node);
@@ -91,6 +92,8 @@ public function getBaseHref(): ?string
 
     /**
      * Removes all the nodes.
+     *
+     * @return void
      */
     public function clear()
     {
@@ -107,6 +110,8 @@ public function clear()
      *
      * @param \DOMNodeList|\DOMNode|\DOMNode[]|string|null $node A node
      *
+     * @return void
+     *
      * @throws \InvalidArgumentException when node is not the expected type
      */
     public function add(\DOMNodeList|\DOMNode|array|string|null $node)
@@ -130,6 +135,8 @@ public function add(\DOMNodeList|\DOMNode|array|string|null $node)
      * If the charset is not set via the content type, it is assumed to be UTF-8,
      * or ISO-8859-1 as a fallback, which is the default charset defined by the
      * HTTP 1.1 specification.
+     *
+     * @return void
      */
     public function addContent(string $content, string $type = null)
     {
@@ -170,6 +177,8 @@ public function addContent(string $content, string $type = null)
      * internal errors via libxml_use_internal_errors(true)
      * and then, get the errors via libxml_get_errors(). Be
      * sure to clear errors with libxml_clear_errors() afterward.
+     *
+     * @return void
      */
     public function addHtmlContent(string $content, string $charset = 'UTF-8')
     {
@@ -204,6 +213,8 @@ public function addHtmlContent(string $content, string $charset = 'UTF-8')
      * @param int $options Bitwise OR of the libxml option constants
      *                     LIBXML_PARSEHUGE is dangerous, see
      *                     http://symfony.com/blog/security-release-symfony-2-0-17-released
+     *
+     * @return void
      */
     public function addXmlContent(string $content, string $charset = 'UTF-8', int $options = \LIBXML_NONET)
     {
@@ -232,6 +243,8 @@ public function addXmlContent(string $content, string $charset = 'UTF-8', int $o
      * Adds a \DOMDocument to the list of nodes.
      *
      * @param \DOMDocument $dom A \DOMDocument instance
+     *
+     * @return void
      */
     public function addDocument(\DOMDocument $dom)
     {
@@ -244,6 +257,8 @@ public function addDocument(\DOMDocument $dom)
      * Adds a \DOMNodeList to the list of nodes.
      *
      * @param \DOMNodeList $nodes A \DOMNodeList instance
+     *
+     * @return void
      */
     public function addNodeList(\DOMNodeList $nodes)
     {
@@ -258,6 +273,8 @@ public function addNodeList(\DOMNodeList $nodes)
      * Adds an array of \DOMNode instances to the list of nodes.
      *
      * @param \DOMNode[] $nodes An array of \DOMNode instances
+     *
+     * @return void
      */
     public function addNodes(array $nodes)
     {
@@ -270,6 +287,8 @@ public function addNodes(array $nodes)
      * Adds a \DOMNode instance to the list of nodes.
      *
      * @param \DOMNode $node A \DOMNode instance
+     *
+     * @return void
      */
     public function addNode(\DOMNode $node)
     {
@@ -553,18 +572,34 @@ public function text(string $default = null, bool $normalizeWhitespace = true):
         $text = $this->getNode(0)->nodeValue;
 
         if ($normalizeWhitespace) {
-            return trim(preg_replace("/(?:[ \n\r\t\x0C]{2,}+|[\n\r\t\x0C])/", ' ', $text), " \n\r\t\x0C");
+            return $this->normalizeWhitespace($text);
         }
 
         return $text;
     }
 
     /**
      * Returns only the inner text that is the direct descendent of the current node, excluding any child nodes.
+     *
+     * @param bool $normalizeWhitespace Whether whitespaces should be trimmed and normalized to single spaces
      */
-    public function innerText(): string
+    public function innerText(/* bool $normalizeWhitespace = true */): string
     {
-        return $this->filterXPath('.//text()')->text();
+        $normalizeWhitespace = 1 <= \func_num_args() ? func_get_arg(0) : true;
+
+        foreach ($this->getNode(0)->childNodes as $childNode) {
+            if (\XML_TEXT_NODE !== $childNode->nodeType) {
+                continue;
+            }
+            if (!$normalizeWhitespace) {
+                return $childNode->nodeValue;
+            }
+            if ('' !== trim($childNode->nodeValue)) {
+                return $this->normalizeWhitespace($childNode->nodeValue);
+            }
+        }
+
+        return '';
     }
 
     /**
@@ -587,7 +622,7 @@ public function html(string $default = null): string
         $node = $this->getNode(0);
         $owner = $node->ownerDocument;
 
-        if ('<!DOCTYPE html>' === $owner->saveXML($owner->childNodes[0])) {
+        if ($this->html5Parser && '<!DOCTYPE html>' === $owner->saveXML($owner->childNodes[0])) {
             $owner = $this->html5Parser;
         }
 
@@ -608,7 +643,7 @@ public function outerHtml(): string
         $node = $this->getNode(0);
         $owner = $node->ownerDocument;
 
-        if ('<!DOCTYPE html>' === $owner->saveXML($owner->childNodes[0])) {
+        if ($this->html5Parser && '<!DOCTYPE html>' === $owner->saveXML($owner->childNodes[0])) {
             $owner = $this->html5Parser;
         }
 
@@ -846,12 +881,17 @@ public function form(array $values = null, string $method = null): Form
 
     /**
      * Overloads a default namespace prefix to be used with XPath and CSS expressions.
+     *
+     * @return void
      */
     public function setDefaultNamespacePrefix(string $prefix)
     {
         $this->defaultNamespacePrefix = $prefix;
     }
 
+    /**
+     * @return void
+     */
     public function registerNamespace(string $prefix, string $namespace)
     {
         $this->namespaces[$prefix] = $namespace;
@@ -1176,6 +1216,10 @@ private function parseHtmlString(string $content, string $charset): \DOMDocument
 
     private function canParseHtml5String(string $content): bool
     {
+        if (!$this->html5Parser) {
+            return false;
+        }
+
         if (false === ($pos = stripos($content, '<!doctype html>'))) {
             return false;
         }
@@ -1189,4 +1233,9 @@ private function isValidHtml5Heading(string $heading): bool
     {
         return 1 === preg_match('/^\x{FEFF}?\s*(<!--[^>]*?-->\s*)*$/u', $heading);
     }
+
+    private function normalizeWhitespace(string $string): string
+    {
+        return trim(preg_replace("/(?:[ \n\r\t\x0C]{2,}+|[\n\r\t\x0C])/", ' ', $string), " \n\r\t\x0C");
+    }
 }

Field/ChoiceFormField.php

@@ -60,6 +60,8 @@ public function isDisabled(): bool
 
     /**
      * Sets the value of the field.
+     *
+     * @return void
      */
     public function select(string|array|bool $value)
     {
@@ -69,6 +71,8 @@ public function select(string|array|bool $value)
     /**
      * Ticks a checkbox.
      *
+     * @return void
+     *
      * @throws \LogicException When the type provided is not correct
      */
     public function tick()
@@ -83,6 +87,8 @@ public function tick()
     /**
      * Unticks a checkbox.
      *
+     * @return void
+     *
      * @throws \LogicException When the type provided is not correct
      */
     public function untick()
@@ -97,6 +103,8 @@ public function untick()
     /**
      * Sets the value of the field.
      *
+     * @return void
+     *
      * @throws \InvalidArgumentException When value type provided is not correct
      */
     public function setValue(string|array|bool|null $value)
@@ -141,7 +149,7 @@ public function setValue(string|array|bool|null $value)
      *
      * @internal
      */
-    public function addChoice(\DOMElement $node)
+    public function addChoice(\DOMElement $node): void
     {
         if (!$this->multiple && 'radio' !== $this->type) {
             throw new \LogicException(sprintf('Unable to add a choice for "%s" as it is not multiple or is not a radio button.', $this->name));
@@ -174,6 +182,8 @@ public function isMultiple(): bool
     /**
      * Initializes the form field.
      *
+     * @return void
+     *
      * @throws \LogicException When node type is incorrect
      */
     protected function initialize()

Field/FileFormField.php

@@ -23,6 +23,8 @@ class FileFormField extends FormField
      *
      * @param int $error The error code (one of UPLOAD_ERR_INI_SIZE, UPLOAD_ERR_FORM_SIZE, UPLOAD_ERR_PARTIAL, UPLOAD_ERR_NO_FILE, UPLOAD_ERR_NO_TMP_DIR, UPLOAD_ERR_CANT_WRITE, or UPLOAD_ERR_EXTENSION)
      *
+     * @return void
+     *
      * @throws \InvalidArgumentException When error code doesn't exist
      */
     public function setErrorCode(int $error)
@@ -37,6 +39,8 @@ public function setErrorCode(int $error)
 
     /**
      * Sets the value of the field.
+     *
+     * @return void
      */
     public function upload(?string $value)
     {
@@ -45,6 +49,8 @@ public function upload(?string $value)
 
     /**
      * Sets the value of the field.
+     *
+     * @return void
      */
     public function setValue(?string $value)
     {
@@ -76,6 +82,8 @@ public function setValue(?string $value)
 
     /**
      * Sets path to the file as string for simulating HTTP request.
+     *
+     * @return void
      */
     public function setFilePath(string $path)
     {
@@ -85,6 +93,8 @@ public function setFilePath(string $path)
     /**
      * Initializes the form field.
      *
+     * @return void
+     *
      * @throws \LogicException When node type is incorrect
      */
     protected function initialize()

Field/FormField.php

@@ -92,6 +92,8 @@ public function getValue(): string|array|null
 
     /**
      * Sets the value of the field.
+     *
+     * @return void
      */
     public function setValue(?string $value)
     {
@@ -116,6 +118,8 @@ public function isDisabled(): bool
 
     /**
      * Initializes the form field.
+     *
+     * @return void
      */
     abstract protected function initialize();
 }

Field/InputFormField.php

@@ -24,6 +24,8 @@ class InputFormField extends FormField
     /**
      * Initializes the form field.
      *
+     * @return void
+     *
      * @throws \LogicException When node type is incorrect
      */
     protected function initialize()

Field/TextareaFormField.php

@@ -21,6 +21,8 @@ class TextareaFormField extends FormField
     /**
      * Initializes the form field.
      *
+     * @return void
+     *
      * @throws \LogicException When node type is incorrect
      */
     protected function initialize()