lochmueller / sourceopt

Classes/Service/CleanHtmlService.php

Spacing +1 added lines, -1 removed lines

@@ -191,7 +191,7 @@
             '/(<(?:[^<>]+(?:"[^"]*"|\'[^\']*\')?)+>)/',
             $html,
             -1,
-            \PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY
+            \PREG_SPLIT_DELIM_CAPTURE|\PREG_SPLIT_NO_EMPTY
         );
 
         if (false === $htmlArrayTemp) {

Indentation +411 added lines, -411 removed lines

@@ -17,415 +17,415 @@
  */
 class CleanHtmlService implements SingletonInterface
 {
-    /**
-     * Enable Debug comment in footer.
-     *
-     * @var bool
-     */
-    protected $debugComment = false;
-
-    /**
-     * Format Type.
-     *
-     * @var int
-     */
-    protected $formatType = 0;
-
-    /**
-     * Tab character.
-     *
-     * @var string
-     */
-    protected $tab = "\t";
-
-    /**
-     * Newline character.
-     *
-     * @var string
-     */
-    protected $newline = "\n";
-
-    /**
-     * Configured extra header comment.
-     *
-     * @var string
-     */
-    protected $headerComment = '';
-
-    /**
-     * Empty space char.
-     *
-     * @var string
-     */
-    protected $emptySpaceChar = ' ';
-
-    /**
-     * Set variables based on given config.
-     */
-    public function setVariables(array $config): void
-    {
-        if (!empty($config)) {
-            if ($config['formatHtml'] && is_numeric($config['formatHtml'])) {
-                $this->formatType = (int) $config['formatHtml'];
-            }
-
-            if ($config['formatHtml.']['tabSize'] && is_numeric($config['formatHtml.']['tabSize'])) {
-                $this->tab = str_pad('', (int) $config['formatHtml.']['tabSize'], ' ');
-            }
-
-            if (isset($config['formatHtml.']['debugComment'])) {
-                $this->debugComment = (bool) $config['formatHtml.']['debugComment'];
-            }
-
-            if (isset($config['headerComment'])) {
-                $this->headerComment = $config['headerComment'];
-            }
-
-            if (isset($config['dropEmptySpaceChar']) && (bool) $config['dropEmptySpaceChar']) {
-                $this->emptySpaceChar = '';
-            }
-        }
-    }
-
-    /**
-     * Clean given HTML with formatter.
-     *
-     * @param string $html
-     * @param array  $config
-     *
-     * @return string
-     */
-    public function clean($html, $config = [])
-    {
-        if (!empty($config)) {
-            $this->setVariables($config);
-        }
-        // convert line-breaks to UNIX
-        $this->convNlOs($html);
-
-        $manipulations = [];
-
-        if (isset($config['removeGenerator']) && (bool) $config['removeGenerator']) {
-            $manipulations['removeGenerator'] = GeneralUtility::makeInstance(RemoveGenerator::class);
-        }
-
-        if (isset($config['removeComments']) && (bool) $config['removeComments']) {
-            $manipulations['removeComments'] = GeneralUtility::makeInstance(RemoveComments::class);
-        }
-
-        if (!empty($this->headerComment)) {
-            $this->includeHeaderComment($html);
-        }
-
-        foreach ($manipulations as $key => $manipulation) {
-            /** @var ManipulationInterface $manipulation */
-            $configuration = isset($config[$key.'.']) && \is_array($config[$key.'.']) ? $config[$key.'.'] : [];
-            $html = $manipulation->manipulate($html, $configuration);
-        }
-
-        // cleanup HTML5 self-closing elements
-        if (!isset($GLOBALS['TSFE']->config['config']['doctype'])
-            || 'x' !== substr($GLOBALS['TSFE']->config['config']['doctype'], 0, 1)) {
-            $html = preg_replace(
-                '/<((?:area|base|br|col|embed|hr|img|input|link|meta|param|source|track|wbr)\s[^>]+?)\s?\/>/',
-                '<$1>',
-                $html
-            );
-        }
-
-        if ($this->formatType > 0) {
-            $html = $this->formatHtml($html);
-        }
-        // remove white space after line ending
-        $this->rTrimLines($html);
-
-        // recover line-breaks
-        if (Environment::isWindows()) {
-            $html = str_replace($this->newline, "\r\n", $html);
-        }
-
-        return $html;
-    }
-
-    /**
-     * Formats the (X)HTML code:
-     *  - taps according to the hirarchy of the tags
-     *  - removes empty spaces between tags
-     *  - removes linebreaks within tags (spares where necessary: pre, textarea, comments, ..)
-     *  choose from five options:
-     *    0 => off
-     *    1 => no line break at all  (code in one line)
-     *    2 => minimalistic line breaks (structure defining box-elements)
-     *    3 => aesthetic line breaks (important box-elements)
-     *    4 => logic line breaks (all box-elements)
-     *    5 => max line breaks (all elements).
-     *
-     * @param string $html
-     *
-     * @return string
-     */
-    protected function formatHtml($html)
-    {
-        // Save original formated comments, pre, textarea, styles and java-scripts & replace them with markers
-        preg_match_all(
-            '/(?s)((<!--.*?-->)|(<[ \n\r]*pre[^>]*>.*?<[ \n\r]*\/pre[^>]*>)|(<[ \n\r]*textarea[^>]*>.*?<[ \n\r]*\/textarea[^>]*>)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im',
-            $html,
-            $matches
-        );
-        $noFormat = $matches[0]; // do not format these block elements
-        for ($i = 0; $i < \count($noFormat); ++$i) {
-            $html = str_replace($noFormat[$i], "\n<!-- ELEMENT {$i} -->", $html);
-        }
-
-        // define box elements for formatting
-        $trueBoxElements = 'address|blockquote|center|dir|div|dl|fieldset|form|h1|h2|h3|h4|h5|h6|hr|isindex|menu|noframes|noscript|ol|p|pre|table|ul|article|aside|details|figcaption|figure|footer|header|hgroup|menu|nav|section';
-        $functionalBoxElements = 'dd|dt|frameset|li|tbody|td|tfoot|th|thead|tr|colgroup';
-        $usableBoxElements = 'applet|button|del|iframe|ins|map|object|script';
-        $imagineBoxElements = 'html|body|head|meta|title|link|script|base|!--';
-        $allBoxLikeElements = '(?>'.$trueBoxElements.'|'.$functionalBoxElements.'|'.$usableBoxElements.'|'.$imagineBoxElements.')';
-        $esteticBoxLikeElements = '(?>html|head|body|meta name|title|div|table|h1|h2|h3|h4|h5|h6|p|form|pre|center|!--)';
-        $structureBoxLikeElements = '(?>html|head|body|div|!--)';
-
-        // split html into it's elements
-        $htmlArrayTemp = preg_split(
-            '/(<(?:[^<>]+(?:"[^"]*"|\'[^\']*\')?)+>)/',
-            $html,
-            -1,
-            \PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY
-        );
-
-        if (false === $htmlArrayTemp) {
-            // Restore saved comments, styles and java-scripts
-            for ($i = 0; $i < \count($noFormat); ++$i) {
-                $html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html);
-            }
-
-            return $html;
-        }
-        // remove empty lines
-        $htmlArray = [''];
-        $index = 1;
-        for ($x = 0; $x < \count($htmlArrayTemp); ++$x) {
-            $text = trim($htmlArrayTemp[$x]);
-            $htmlArray[$index] = '' !== $text ? $htmlArrayTemp[$x] : $this->emptySpaceChar;
-            ++$index;
-        }
-
-        // rebuild html
-        $html = '';
-        $tabs = 0;
-        for ($x = 0; $x < \count($htmlArray); ++$x) {
-            $htmlArrayBefore = $htmlArray[$x - 1] ?? '';
-            $htmlArrayCurrent = $htmlArray[$x] ?? '';
-
-            // check if the element should stand in a new line
-            $newline = false;
-            if ('<?xml' == substr($htmlArrayBefore, 0, 5)) {
-                $newline = true;
-            } elseif (2 == $this->formatType && ( // minimalistic line break
-                    // this element has a line break before itself
-                    preg_match(
-                        '/<'.$structureBoxLikeElements.'(.*)>/Usi',
-                        $htmlArrayCurrent
-                    ) || preg_match(
-                        '/<'.$structureBoxLikeElements.'(.*) \/>/Usi',
-                        $htmlArrayCurrent
-                    ) // one element before is a element that has a line break after
-                    || preg_match(
-                        '/<\/'.$structureBoxLikeElements.'(.*)>/Usi',
-                        $htmlArrayBefore
-                    ) || '<!--' == substr(
-                        $htmlArrayBefore,
-                        0,
-                        4
-                    ) || preg_match('/<'.$structureBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore))
-            ) {
-                $newline = true;
-            } elseif (3 == $this->formatType && ( // aestetic line break
-                    // this element has a line break before itself
-                    preg_match(
-                        '/<'.$esteticBoxLikeElements.'(.*)>/Usi',
-                        $htmlArrayCurrent
-                    ) || preg_match(
-                        '/<'.$esteticBoxLikeElements.'(.*) \/>/Usi',
-                        $htmlArrayCurrent
-                    ) // one element before is a element that has a line break after
-                    || preg_match('/<\/'.$esteticBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr(
-                        $htmlArrayBefore,
-                        0,
-                        4
-                    ) || preg_match('/<'.$esteticBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore))
-            ) {
-                $newline = true;
-            } elseif ($this->formatType >= 4 && ( // logical line break
-                    // this element has a line break before itself
-                    preg_match(
-                        '/<'.$allBoxLikeElements.'(.*)>/Usi',
-                        $htmlArrayCurrent
-                    ) || preg_match(
-                        '/<'.$allBoxLikeElements.'(.*) \/>/Usi',
-                        $htmlArrayCurrent
-                    ) // one element before is a element that has a line break after
-                    || preg_match('/<\/'.$allBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr(
-                        $htmlArrayBefore,
-                        0,
-                        4
-                    ) || preg_match('/<'.$allBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore))
-            ) {
-                $newline = true;
-            }
-
-            // count down a tab
-            if ('</' == substr($htmlArrayCurrent, 0, 2)) {
-                --$tabs;
-            }
-
-            // add tabs and line breaks in front of the current tag
-            if ($newline) {
-                $html .= $this->newline;
-                for ($y = 0; $y < $tabs; ++$y) {
-                    $html .= $this->tab;
-                }
-            }
-
-            // remove white spaces and line breaks and add current tag to the html-string
-            if ('<![CDATA[' == substr($htmlArrayCurrent, 0, 9) // remove multiple white space in CDATA / XML
-                || '<?xml' == substr($htmlArrayCurrent, 0, 5)
-            ) {
-                $html .= $this->killWhiteSpace($htmlArrayCurrent);
-            } else { // remove all line breaks
-                $html .= $this->killLineBreaks($htmlArrayCurrent);
-            }
-
-            // count up a tab
-            if ('<' == substr($htmlArrayCurrent, 0, 1) && '/' != substr($htmlArrayCurrent, 1, 1)) {
-                if (' ' !== substr($htmlArrayCurrent, 1, 1)
-                    && 'img' !== substr($htmlArrayCurrent, 1, 3)
-                    && 'source' !== substr($htmlArrayCurrent, 1, 6)
-                    && 'br' !== substr($htmlArrayCurrent, 1, 2)
-                    && 'hr' !== substr($htmlArrayCurrent, 1, 2)
-                    && 'input' !== substr($htmlArrayCurrent, 1, 5)
-                    && 'link' !== substr($htmlArrayCurrent, 1, 4)
-                    && 'meta' !== substr($htmlArrayCurrent, 1, 4)
-                    && 'col ' !== substr($htmlArrayCurrent, 1, 4)
-                    && 'frame' !== substr($htmlArrayCurrent, 1, 5)
-                    && 'isindex' !== substr($htmlArrayCurrent, 1, 7)
-                    && 'param' !== substr($htmlArrayCurrent, 1, 5)
-                    && 'area' !== substr($htmlArrayCurrent, 1, 4)
-                    && 'base' !== substr($htmlArrayCurrent, 1, 4)
-                    && '<!' !== substr($htmlArrayCurrent, 0, 2)
-                    && '<?xml' !== substr($htmlArrayCurrent, 0, 5)
-                ) {
-                    ++$tabs;
-                }
-            }
-        }
-
-        // Remove empty lines
-        if ($this->formatType > 1) {
-            $this->removeEmptyLines($html);
-        }
-
-        // Restore saved comments, styles and java-scripts
-        for ($i = 0; $i < \count($noFormat); ++$i) {
-            $html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html);
-        }
-
-        // include debug comment at the end
-        if (0 != $tabs && true === $this->debugComment) {
-            $html .= "<!-- {$tabs} open elements found -->";
-        }
-
-        return $html;
-    }
-
-    /**
-     * Remove ALL line breaks and multiple white space.
-     *
-     * @param string $html
-     *
-     * @return string
-     */
-    protected function killLineBreaks($html)
-    {
-        $html = str_replace($this->newline, '', $html);
-
-        return preg_replace('/\s\s+/u', ' ', $html);
-        //? return preg_replace('/\n|\s+(\s)/u', '$1', $html);
-    }
-
-    /**
-     * Remove multiple white space, keeps line breaks.
-     *
-     * @param string $html
-     *
-     * @return string
-     */
-    protected function killWhiteSpace($html)
-    {
-        $temp = explode($this->newline, $html);
-        for ($i = 0; $i < \count($temp); ++$i) {
-            if (!trim($temp[$i])) {
-                unset($temp[$i]);
-                continue;
-            }
-
-            $temp[$i] = trim($temp[$i]);
-            $temp[$i] = preg_replace('/\s\s+/', ' ', $temp[$i]);
-        }
-
-        return implode($this->newline, $temp);
-    }
-
-    /**
-     * Remove white space at the end of lines, keeps other white space and line breaks.
-     *
-     * @param string $html
-     *
-     * @return string
-     */
-    protected function rTrimLines(& $html)
-    {
-        $html = preg_replace('/\s+$/m', '', $html);
-    }
-
-    /**
-     * Convert newlines according to the current OS.
-     *
-     * @param string $html
-     *
-     * @return string
-     */
-    protected function convNlOs(& $html)
-    {
-        $html = preg_replace("(\r\n|\r)", $this->newline, $html);
-    }
-
-    /**
-     * Remove empty lines.
-     *
-     * @param string $html
-     */
-    protected function removeEmptyLines(& $html): void
-    {
-        $temp = explode($this->newline, $html);
-        $result = [];
-        for ($i = 0; $i < \count($temp); ++$i) {
-            if ('' == trim($temp[$i])) {
-                continue;
-            }
-            $result[] = $temp[$i];
-        }
-        $html = implode($this->newline, $result);
-    }
-
-    /**
-     * Include configured header comment in HTML content block.
-     *
-     * @param $html
-     */
-    public function includeHeaderComment(& $html): void
-    {
-        $html = preg_replace('/^(-->)$/m', "\n\t".$this->headerComment."\n$1", $html);
-    }
+	/**
+	 * Enable Debug comment in footer.
+	 *
+	 * @var bool
+	 */
+	protected $debugComment = false;
+
+	/**
+	 * Format Type.
+	 *
+	 * @var int
+	 */
+	protected $formatType = 0;
+
+	/**
+	 * Tab character.
+	 *
+	 * @var string
+	 */
+	protected $tab = "\t";
+
+	/**
+	 * Newline character.
+	 *
+	 * @var string
+	 */
+	protected $newline = "\n";
+
+	/**
+	 * Configured extra header comment.
+	 *
+	 * @var string
+	 */
+	protected $headerComment = '';
+
+	/**
+	 * Empty space char.
+	 *
+	 * @var string
+	 */
+	protected $emptySpaceChar = ' ';
+
+	/**
+	 * Set variables based on given config.
+	 */
+	public function setVariables(array $config): void
+	{
+		if (!empty($config)) {
+			if ($config['formatHtml'] && is_numeric($config['formatHtml'])) {
+				$this->formatType = (int) $config['formatHtml'];
+			}
+
+			if ($config['formatHtml.']['tabSize'] && is_numeric($config['formatHtml.']['tabSize'])) {
+				$this->tab = str_pad('', (int) $config['formatHtml.']['tabSize'], ' ');
+			}
+
+			if (isset($config['formatHtml.']['debugComment'])) {
+				$this->debugComment = (bool) $config['formatHtml.']['debugComment'];
+			}
+
+			if (isset($config['headerComment'])) {
+				$this->headerComment = $config['headerComment'];
+			}
+
+			if (isset($config['dropEmptySpaceChar']) && (bool) $config['dropEmptySpaceChar']) {
+				$this->emptySpaceChar = '';
+			}
+		}
+	}
+
+	/**
+	 * Clean given HTML with formatter.
+	 *
+	 * @param string $html
+	 * @param array  $config
+	 *
+	 * @return string
+	 */
+	public function clean($html, $config = [])
+	{
+		if (!empty($config)) {
+			$this->setVariables($config);
+		}
+		// convert line-breaks to UNIX
+		$this->convNlOs($html);
+
+		$manipulations = [];
+
+		if (isset($config['removeGenerator']) && (bool) $config['removeGenerator']) {
+			$manipulations['removeGenerator'] = GeneralUtility::makeInstance(RemoveGenerator::class);
+		}
+
+		if (isset($config['removeComments']) && (bool) $config['removeComments']) {
+			$manipulations['removeComments'] = GeneralUtility::makeInstance(RemoveComments::class);
+		}
+
+		if (!empty($this->headerComment)) {
+			$this->includeHeaderComment($html);
+		}
+
+		foreach ($manipulations as $key => $manipulation) {
+			/** @var ManipulationInterface $manipulation */
+			$configuration = isset($config[$key.'.']) && \is_array($config[$key.'.']) ? $config[$key.'.'] : [];
+			$html = $manipulation->manipulate($html, $configuration);
+		}
+
+		// cleanup HTML5 self-closing elements
+		if (!isset($GLOBALS['TSFE']->config['config']['doctype'])
+			|| 'x' !== substr($GLOBALS['TSFE']->config['config']['doctype'], 0, 1)) {
+			$html = preg_replace(
+				'/<((?:area|base|br|col|embed|hr|img|input|link|meta|param|source|track|wbr)\s[^>]+?)\s?\/>/',
+				'<$1>',
+				$html
+			);
+		}
+
+		if ($this->formatType > 0) {
+			$html = $this->formatHtml($html);
+		}
+		// remove white space after line ending
+		$this->rTrimLines($html);
+
+		// recover line-breaks
+		if (Environment::isWindows()) {
+			$html = str_replace($this->newline, "\r\n", $html);
+		}
+
+		return $html;
+	}
+
+	/**
+	 * Formats the (X)HTML code:
+	 *  - taps according to the hirarchy of the tags
+	 *  - removes empty spaces between tags
+	 *  - removes linebreaks within tags (spares where necessary: pre, textarea, comments, ..)
+	 *  choose from five options:
+	 *    0 => off
+	 *    1 => no line break at all  (code in one line)
+	 *    2 => minimalistic line breaks (structure defining box-elements)
+	 *    3 => aesthetic line breaks (important box-elements)
+	 *    4 => logic line breaks (all box-elements)
+	 *    5 => max line breaks (all elements).
+	 *
+	 * @param string $html
+	 *
+	 * @return string
+	 */
+	protected function formatHtml($html)
+	{
+		// Save original formated comments, pre, textarea, styles and java-scripts & replace them with markers
+		preg_match_all(
+			'/(?s)((<!--.*?-->)|(<[ \n\r]*pre[^>]*>.*?<[ \n\r]*\/pre[^>]*>)|(<[ \n\r]*textarea[^>]*>.*?<[ \n\r]*\/textarea[^>]*>)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im',
+			$html,
+			$matches
+		);
+		$noFormat = $matches[0]; // do not format these block elements
+		for ($i = 0; $i < \count($noFormat); ++$i) {
+			$html = str_replace($noFormat[$i], "\n<!-- ELEMENT {$i} -->", $html);
+		}
+
+		// define box elements for formatting
+		$trueBoxElements = 'address|blockquote|center|dir|div|dl|fieldset|form|h1|h2|h3|h4|h5|h6|hr|isindex|menu|noframes|noscript|ol|p|pre|table|ul|article|aside|details|figcaption|figure|footer|header|hgroup|menu|nav|section';
+		$functionalBoxElements = 'dd|dt|frameset|li|tbody|td|tfoot|th|thead|tr|colgroup';
+		$usableBoxElements = 'applet|button|del|iframe|ins|map|object|script';
+		$imagineBoxElements = 'html|body|head|meta|title|link|script|base|!--';
+		$allBoxLikeElements = '(?>'.$trueBoxElements.'|'.$functionalBoxElements.'|'.$usableBoxElements.'|'.$imagineBoxElements.')';
+		$esteticBoxLikeElements = '(?>html|head|body|meta name|title|div|table|h1|h2|h3|h4|h5|h6|p|form|pre|center|!--)';
+		$structureBoxLikeElements = '(?>html|head|body|div|!--)';
+
+		// split html into it's elements
+		$htmlArrayTemp = preg_split(
+			'/(<(?:[^<>]+(?:"[^"]*"|\'[^\']*\')?)+>)/',
+			$html,
+			-1,
+			\PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY
+		);
+
+		if (false === $htmlArrayTemp) {
+			// Restore saved comments, styles and java-scripts
+			for ($i = 0; $i < \count($noFormat); ++$i) {
+				$html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html);
+			}
+
+			return $html;
+		}
+		// remove empty lines
+		$htmlArray = [''];
+		$index = 1;
+		for ($x = 0; $x < \count($htmlArrayTemp); ++$x) {
+			$text = trim($htmlArrayTemp[$x]);
+			$htmlArray[$index] = '' !== $text ? $htmlArrayTemp[$x] : $this->emptySpaceChar;
+			++$index;
+		}
+
+		// rebuild html
+		$html = '';
+		$tabs = 0;
+		for ($x = 0; $x < \count($htmlArray); ++$x) {
+			$htmlArrayBefore = $htmlArray[$x - 1] ?? '';
+			$htmlArrayCurrent = $htmlArray[$x] ?? '';
+
+			// check if the element should stand in a new line
+			$newline = false;
+			if ('<?xml' == substr($htmlArrayBefore, 0, 5)) {
+				$newline = true;
+			} elseif (2 == $this->formatType && ( // minimalistic line break
+					// this element has a line break before itself
+					preg_match(
+						'/<'.$structureBoxLikeElements.'(.*)>/Usi',
+						$htmlArrayCurrent
+					) || preg_match(
+						'/<'.$structureBoxLikeElements.'(.*) \/>/Usi',
+						$htmlArrayCurrent
+					) // one element before is a element that has a line break after
+					|| preg_match(
+						'/<\/'.$structureBoxLikeElements.'(.*)>/Usi',
+						$htmlArrayBefore
+					) || '<!--' == substr(
+						$htmlArrayBefore,
+						0,
+						4
+					) || preg_match('/<'.$structureBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore))
+			) {
+				$newline = true;
+			} elseif (3 == $this->formatType && ( // aestetic line break
+					// this element has a line break before itself
+					preg_match(
+						'/<'.$esteticBoxLikeElements.'(.*)>/Usi',
+						$htmlArrayCurrent
+					) || preg_match(
+						'/<'.$esteticBoxLikeElements.'(.*) \/>/Usi',
+						$htmlArrayCurrent
+					) // one element before is a element that has a line break after
+					|| preg_match('/<\/'.$esteticBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr(
+						$htmlArrayBefore,
+						0,
+						4
+					) || preg_match('/<'.$esteticBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore))
+			) {
+				$newline = true;
+			} elseif ($this->formatType >= 4 && ( // logical line break
+					// this element has a line break before itself
+					preg_match(
+						'/<'.$allBoxLikeElements.'(.*)>/Usi',
+						$htmlArrayCurrent
+					) || preg_match(
+						'/<'.$allBoxLikeElements.'(.*) \/>/Usi',
+						$htmlArrayCurrent
+					) // one element before is a element that has a line break after
+					|| preg_match('/<\/'.$allBoxLikeElements.'(.*)>/Usi', $htmlArrayBefore) || '<!--' == substr(
+						$htmlArrayBefore,
+						0,
+						4
+					) || preg_match('/<'.$allBoxLikeElements.'(.*) \/>/Usi', $htmlArrayBefore))
+			) {
+				$newline = true;
+			}
+
+			// count down a tab
+			if ('</' == substr($htmlArrayCurrent, 0, 2)) {
+				--$tabs;
+			}
+
+			// add tabs and line breaks in front of the current tag
+			if ($newline) {
+				$html .= $this->newline;
+				for ($y = 0; $y < $tabs; ++$y) {
+					$html .= $this->tab;
+				}
+			}
+
+			// remove white spaces and line breaks and add current tag to the html-string
+			if ('<![CDATA[' == substr($htmlArrayCurrent, 0, 9) // remove multiple white space in CDATA / XML
+				|| '<?xml' == substr($htmlArrayCurrent, 0, 5)
+			) {
+				$html .= $this->killWhiteSpace($htmlArrayCurrent);
+			} else { // remove all line breaks
+				$html .= $this->killLineBreaks($htmlArrayCurrent);
+			}
+
+			// count up a tab
+			if ('<' == substr($htmlArrayCurrent, 0, 1) && '/' != substr($htmlArrayCurrent, 1, 1)) {
+				if (' ' !== substr($htmlArrayCurrent, 1, 1)
+					&& 'img' !== substr($htmlArrayCurrent, 1, 3)
+					&& 'source' !== substr($htmlArrayCurrent, 1, 6)
+					&& 'br' !== substr($htmlArrayCurrent, 1, 2)
+					&& 'hr' !== substr($htmlArrayCurrent, 1, 2)
+					&& 'input' !== substr($htmlArrayCurrent, 1, 5)
+					&& 'link' !== substr($htmlArrayCurrent, 1, 4)
+					&& 'meta' !== substr($htmlArrayCurrent, 1, 4)
+					&& 'col ' !== substr($htmlArrayCurrent, 1, 4)
+					&& 'frame' !== substr($htmlArrayCurrent, 1, 5)
+					&& 'isindex' !== substr($htmlArrayCurrent, 1, 7)
+					&& 'param' !== substr($htmlArrayCurrent, 1, 5)
+					&& 'area' !== substr($htmlArrayCurrent, 1, 4)
+					&& 'base' !== substr($htmlArrayCurrent, 1, 4)
+					&& '<!' !== substr($htmlArrayCurrent, 0, 2)
+					&& '<?xml' !== substr($htmlArrayCurrent, 0, 5)
+				) {
+					++$tabs;
+				}
+			}
+		}
+
+		// Remove empty lines
+		if ($this->formatType > 1) {
+			$this->removeEmptyLines($html);
+		}
+
+		// Restore saved comments, styles and java-scripts
+		for ($i = 0; $i < \count($noFormat); ++$i) {
+			$html = str_replace("<!-- ELEMENT {$i} -->", $noFormat[$i], $html);
+		}
+
+		// include debug comment at the end
+		if (0 != $tabs && true === $this->debugComment) {
+			$html .= "<!-- {$tabs} open elements found -->";
+		}
+
+		return $html;
+	}
+
+	/**
+	 * Remove ALL line breaks and multiple white space.
+	 *
+	 * @param string $html
+	 *
+	 * @return string
+	 */
+	protected function killLineBreaks($html)
+	{
+		$html = str_replace($this->newline, '', $html);
+
+		return preg_replace('/\s\s+/u', ' ', $html);
+		//? return preg_replace('/\n|\s+(\s)/u', '$1', $html);
+	}
+
+	/**
+	 * Remove multiple white space, keeps line breaks.
+	 *
+	 * @param string $html
+	 *
+	 * @return string
+	 */
+	protected function killWhiteSpace($html)
+	{
+		$temp = explode($this->newline, $html);
+		for ($i = 0; $i < \count($temp); ++$i) {
+			if (!trim($temp[$i])) {
+				unset($temp[$i]);
+				continue;
+			}
+
+			$temp[$i] = trim($temp[$i]);
+			$temp[$i] = preg_replace('/\s\s+/', ' ', $temp[$i]);
+		}
+
+		return implode($this->newline, $temp);
+	}
+
+	/**
+	 * Remove white space at the end of lines, keeps other white space and line breaks.
+	 *
+	 * @param string $html
+	 *
+	 * @return string
+	 */
+	protected function rTrimLines(& $html)
+	{
+		$html = preg_replace('/\s+$/m', '', $html);
+	}
+
+	/**
+	 * Convert newlines according to the current OS.
+	 *
+	 * @param string $html
+	 *
+	 * @return string
+	 */
+	protected function convNlOs(& $html)
+	{
+		$html = preg_replace("(\r\n|\r)", $this->newline, $html);
+	}
+
+	/**
+	 * Remove empty lines.
+	 *
+	 * @param string $html
+	 */
+	protected function removeEmptyLines(& $html): void
+	{
+		$temp = explode($this->newline, $html);
+		$result = [];
+		for ($i = 0; $i < \count($temp); ++$i) {
+			if ('' == trim($temp[$i])) {
+				continue;
+			}
+			$result[] = $temp[$i];
+		}
+		$html = implode($this->newline, $result);
+	}
+
+	/**
+	 * Include configured header comment in HTML content block.
+	 *
+	 * @param $html
+	 */
+	public function includeHeaderComment(& $html): void
+	{
+		$html = preg_replace('/^(-->)$/m', "\n\t".$this->headerComment."\n$1", $html);
+	}
 }

Classes/Manipulation/RemoveComments.php

Indentation +57 added lines, -57 removed lines

@@ -14,67 +14,67 @@
  */
 class RemoveComments implements ManipulationInterface
 {
-    /**
-     * Patterns for white-listing comments inside content.
-     *
-     * @var array
-     */
-    protected $whiteListCommentsPatterns = [];
+	/**
+	 * Patterns for white-listing comments inside content.
+	 *
+	 * @var array
+	 */
+	protected $whiteListCommentsPatterns = [];
 
-    /**
-     * @param string $html          The original HTML
-     * @param array  $configuration Configuration
-     *
-     * @return string the manipulated HTML
-     */
-    public function manipulate($html, array $configuration = [])
-    {
-        if (isset($configuration['keep.'])) {
-            $this->whiteListCommentsPatterns = $configuration['keep.'];
-        }
+	/**
+	 * @param string $html          The original HTML
+	 * @param array  $configuration Configuration
+	 *
+	 * @return string the manipulated HTML
+	 */
+	public function manipulate($html, array $configuration = [])
+	{
+		if (isset($configuration['keep.'])) {
+			$this->whiteListCommentsPatterns = $configuration['keep.'];
+		}
 
-        // match all styles, scripts and comments
-        $matches = [];
-        preg_match_all(
-            '/(?s)((<!--.*?-->)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im',
-            $html,
-            $matches
-        );
-        foreach ($matches[0] as $tag) {
-            if (false === $this->keepComment($tag)) {
-                $html = str_replace($tag, '', $html);
-            }
-        }
+		// match all styles, scripts and comments
+		$matches = [];
+		preg_match_all(
+			'/(?s)((<!--.*?-->)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im',
+			$html,
+			$matches
+		);
+		foreach ($matches[0] as $tag) {
+			if (false === $this->keepComment($tag)) {
+				$html = str_replace($tag, '', $html);
+			}
+		}
 
-        return $html;
-    }
+		return $html;
+	}
 
-    /**
-     * Check if a comment is defined to be kept in a pattern whiteListOfComments.
-     *
-     * @param string $commentHtml
-     *
-     * @return bool
-     */
-    protected function keepComment($commentHtml)
-    {
-        // if not even a comment, skip this
-        if (!preg_match('/^\<\!\-\-(.*?)\-\-\>$/usi', $commentHtml)) {
-            return true;
-        }
+	/**
+	 * Check if a comment is defined to be kept in a pattern whiteListOfComments.
+	 *
+	 * @param string $commentHtml
+	 *
+	 * @return bool
+	 */
+	protected function keepComment($commentHtml)
+	{
+		// if not even a comment, skip this
+		if (!preg_match('/^\<\!\-\-(.*?)\-\-\>$/usi', $commentHtml)) {
+			return true;
+		}
 
-        // if not defined in white list
-        if (!empty($this->whiteListCommentsPatterns)) {
-            $commentHtml = str_replace('<!--', '', $commentHtml);
-            $commentHtml = str_replace('-->', '', $commentHtml);
-            $commentHtml = trim($commentHtml);
-            foreach ($this->whiteListCommentsPatterns as $pattern) {
-                if (!empty($pattern) && preg_match($pattern, $commentHtml)) {
-                    return true;
-                }
-            }
-        }
+		// if not defined in white list
+		if (!empty($this->whiteListCommentsPatterns)) {
+			$commentHtml = str_replace('<!--', '', $commentHtml);
+			$commentHtml = str_replace('-->', '', $commentHtml);
+			$commentHtml = trim($commentHtml);
+			foreach ($this->whiteListCommentsPatterns as $pattern) {
+				if (!empty($pattern) && preg_match($pattern, $commentHtml)) {
+					return true;
+				}
+			}
+		}
 
-        return false;
-    }
+		return false;
+	}
 }

Classes/Manipulation/ManipulationInterface.php

Indentation +7 added lines, -7 removed lines

@@ -14,11 +14,11 @@
  */
 interface ManipulationInterface
 {
-    /**
-     * @param string $html          The original HTML
-     * @param array  $configuration Configuration
-     *
-     * @return string the manipulated HTML
-     */
-    public function manipulate($html, array $configuration = []);
+	/**
+	 * @param string $html          The original HTML
+	 * @param array  $configuration Configuration
+	 *
+	 * @return string the manipulated HTML
+	 */
+	public function manipulate($html, array $configuration = []);
 }

Classes/Manipulation/RemoveGenerator.php

Indentation +11 added lines, -11 removed lines

@@ -14,16 +14,16 @@
  */
 class RemoveGenerator implements ManipulationInterface
 {
-    /**
-     * @param string $html          The original HTML
-     * @param array  $configuration Configuration
-     *
-     * @return string the manipulated HTML
-     */
-    public function manipulate($html, array $configuration = [])
-    {
-        $regex = '<meta name=["\']?generator["\']? [^>]+>';
+	/**
+	 * @param string $html          The original HTML
+	 * @param array  $configuration Configuration
+	 *
+	 * @return string the manipulated HTML
+	 */
+	public function manipulate($html, array $configuration = [])
+	{
+		$regex = '<meta name=["\']?generator["\']? [^>]+>';
 
-        return preg_replace('/'.$regex.'/is', '', $html);
-    }
+		return preg_replace('/'.$regex.'/is', '', $html);
+	}
 }

Classes/Middleware/CleanHtmlMiddleware.php

Indentation +35 added lines, -35 removed lines

@@ -19,39 +19,39 @@
  */
 class CleanHtmlMiddleware implements MiddlewareInterface
 {
-    /**
-     * @var CleanHtmlService
-     */
-    protected $cleanHtmlService;
-
-    public function __construct()
-    {
-        $this->cleanHtmlService = GeneralUtility::makeInstance(CleanHtmlService::class);
-    }
-
-    /**
-     * Clean the HTML output.
-     */
-    public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
-    {
-        $response = $handler->handle($request);
-
-        if (!($response instanceof NullResponse)
-        && $GLOBALS['TSFE'] instanceof TypoScriptFrontendController
-        && false !== (bool) $GLOBALS['TSFE']->config['config']['sourceopt.']['enabled']
-        && 'text/html' == substr($response->getHeaderLine('Content-Type'), 0, 9)
-        ) {
-            $processedHtml = $this->cleanHtmlService->clean(
-                $response->getBody()->__toString(),
-                $GLOBALS['TSFE']->config['config']['sourceopt.']
-            );
-
-            // Replace old body with $processedHtml
-            $responseBody = new Stream('php://temp', 'rw');
-            $responseBody->write($processedHtml);
-            $response = $response->withBody($responseBody);
-        }
-
-        return $response;
-    }
+	/**
+	 * @var CleanHtmlService
+	 */
+	protected $cleanHtmlService;
+
+	public function __construct()
+	{
+		$this->cleanHtmlService = GeneralUtility::makeInstance(CleanHtmlService::class);
+	}
+
+	/**
+	 * Clean the HTML output.
+	 */
+	public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
+	{
+		$response = $handler->handle($request);
+
+		if (!($response instanceof NullResponse)
+		&& $GLOBALS['TSFE'] instanceof TypoScriptFrontendController
+		&& false !== (bool) $GLOBALS['TSFE']->config['config']['sourceopt.']['enabled']
+		&& 'text/html' == substr($response->getHeaderLine('Content-Type'), 0, 9)
+		) {
+			$processedHtml = $this->cleanHtmlService->clean(
+				$response->getBody()->__toString(),
+				$GLOBALS['TSFE']->config['config']['sourceopt.']
+			);
+
+			// Replace old body with $processedHtml
+			$responseBody = new Stream('php://temp', 'rw');
+			$responseBody->write($processedHtml);
+			$response = $response->withBody($responseBody);
+		}
+
+		return $response;
+	}
 }

Classes/Middleware/SvgStoreMiddleware.php

Indentation +20 added lines, -20 removed lines

@@ -18,27 +18,27 @@
  */
 class SvgStoreMiddleware implements MiddlewareInterface
 {
-    /**
-     * Search/Extract/Merge SVGs @ HTML output.
-     */
-    public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
-    {
-        $response = $handler->handle($request);
+	/**
+	 * Search/Extract/Merge SVGs @ HTML output.
+	 */
+	public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
+	{
+		$response = $handler->handle($request);
 
-        if (!($response instanceof NullResponse)
-        && $GLOBALS['TSFE'] instanceof TypoScriptFrontendController
-        && false !== (bool) $GLOBALS['TSFE']->config['config']['svgstore.']['enabled']
-        && 'text/html' == substr($response->getHeaderLine('Content-Type'), 0, 9)
-        ) {
-            $processedHtml = GeneralUtility::makeInstance(\HTML\Sourceopt\Service\SvgStoreService::class)
-                ->process($response->getBody()->__toString())
-            ;
+		if (!($response instanceof NullResponse)
+		&& $GLOBALS['TSFE'] instanceof TypoScriptFrontendController
+		&& false !== (bool) $GLOBALS['TSFE']->config['config']['svgstore.']['enabled']
+		&& 'text/html' == substr($response->getHeaderLine('Content-Type'), 0, 9)
+		) {
+			$processedHtml = GeneralUtility::makeInstance(\HTML\Sourceopt\Service\SvgStoreService::class)
+				->process($response->getBody()->__toString())
+			;
 
-            $responseBody = new Stream('php://temp', 'rw');
-            $responseBody->write($processedHtml);
-            $response = $response->withBody($responseBody);
-        }
+			$responseBody = new Stream('php://temp', 'rw');
+			$responseBody->write($processedHtml);
+			$response = $response->withBody($responseBody);
+		}
 
-        return $response;
-    }
+		return $response;
+	}
 }

Classes/Service/SvgStoreService.php

Spacing +4 added lines, -4 removed lines

@@ -45,7 +45,7 @@ 
         }
 
         // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/img#attributes
-        $html['body'] = preg_replace_callback('/<img(?<pre>[^>]*)src="(?<src>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?!\s*<\/picture>)/s', function (array $match): string { // ^[/]
+        $html['body'] = preg_replace_callback('/<img(?<pre>[^>]*)src="(?<src>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?!\s*<\/picture>)/s', function(array $match): string { // ^[/]
             if (!isset($this->svgFileArr[$match['src']])) { // check usage
                 return $match[0];
             }
@@ -55,7 +55,7 @@ 
         }, $html['body']);
 
         // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/object#attributes
-        $html['body'] = preg_replace_callback('/<object(?<pre>[^>]*)data="(?<data>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?:<\/object>)/s', function (array $match): string { // ^[/]
+        $html['body'] = preg_replace_callback('/<object(?<pre>[^>]*)data="(?<data>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?:<\/object>)/s', function(array $match): string { // ^[/]
             if (!isset($this->svgFileArr[$match['data']])) { // check usage
                 return $match[0];
             }
@@ -102,7 +102,7 @@ 
         $svg = preg_replace('/.*<svg|<\/svg>.*|xlink:|\s(?:(?:version|xmlns)|(?:[a-z\-]+\:[a-z\-]+))="[^"]*"/s', '', $svg); // cleanup
 
         // https://developer.mozilla.org/en-US/docs/Web/SVG/Element/svg#attributes
-        $svg = preg_replace_callback('/([^>]+)\s*(?=>)/s', function (array $match) use (&$attr): string {
+        $svg = preg_replace_callback('/([^>]+)\s*(?=>)/s', function(array $match) use (&$attr): string {
             if (false === preg_match_all('/\s(?<attr>[\w\-]+)="\s*(?<value>[^"]+)\s*"/', $match[1], $matches)) {
                 return $match[0];
             }
@@ -146,7 +146,7 @@ 
 
         $svg = preg_replace_callback(
             '/<use(?<pre>.*?)(?:xlink:)?href="(?<href>\/.+?\.svg)#[^"]+"(?<post>.*?)[\s\/]*>(?:<\/use>)?/s',
-            function (array $match): string {
+            function(array $match): string {
                 if (!isset($this->svgFileArr[$match['href']])) { // check usage
                     return $match[0];
                 }

Indentation +182 added lines, -182 removed lines

@@ -13,186 +13,186 @@
  */
 class SvgStoreService implements \TYPO3\CMS\Core\SingletonInterface
 {
-    /**
-     * SVG-Sprite storage directory.
-     *
-     * @var string
-     */
-    protected $outputDir = '/typo3temp/assets/svg/';
-
-    public function __construct()
-    {
-        //$this->styl = []; # https://stackoverflow.com/questions/39583880/external-svg-fails-to-apply-internal-css
-        //$this->defs = []; # https://bugs.chromium.org/p/chromium/issues/detail?id=751733#c14
-        $this->svgs = [];
-
-        $this->sitePath = \TYPO3\CMS\Core\Core\Environment::getPublicPath(); // [^/]$
-        $this->svgCache = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Cache\CacheManager::class)->getCache('svgstore');
-    }
-
-    public function process(string $html): string
-    {
-        $this->spritePath = $this->svgCache->get('spritePath');
-        $this->svgFileArr = $this->svgCache->get('svgFileArr');
-
-        if (empty($this->spritePath) && !$this->populateCache()) {
-            throw new \Exception('could not write file: '.$this->sitePath.$this->spritePath);
-        }
-
-        if (!file_exists($this->sitePath.$this->spritePath)) {
-            throw new \Exception('file does not exists: '.$this->sitePath.$this->spritePath);
-        }
-
-        if (!preg_match('/(?<head>.+?<\/head>)(?<body>.+)/s', $html, $html) && 5 == \count($html)) {
-            throw new \Exception('fix HTML!');
-        }
-
-        // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/img#attributes
-        $html['body'] = preg_replace_callback('/<img(?<pre>[^>]*)src="(?<src>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?!\s*<\/picture>)/s', function (array $match): string { // ^[/]
-            if (!isset($this->svgFileArr[$match['src']])) { // check usage
-                return $match[0];
-            }
-            $attr = preg_replace('/\s(?:alt|ismap|loading|title|sizes|srcset|usemap|crossorigin|decoding|referrerpolicy)="[^"]*"/', '', $match['pre'].$match['post']); // cleanup
-
-            return sprintf('<svg %s %s><use href="%s#%s"/></svg>', $this->svgFileArr[$match['src']]['attr'], trim($attr), $this->spritePath, $this->convertFilePath($match['src']));
-        }, $html['body']);
-
-        // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/object#attributes
-        $html['body'] = preg_replace_callback('/<object(?<pre>[^>]*)data="(?<data>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?:<\/object>)/s', function (array $match): string { // ^[/]
-            if (!isset($this->svgFileArr[$match['data']])) { // check usage
-                return $match[0];
-            }
-            $attr = preg_replace('/\s(?:form|name|type|usemap)="[^"]*"/', '', $match['pre'].$match['post']); // cleanup
-
-            return sprintf('<svg %s %s><use href="%s#%s"/></svg>', $this->svgFileArr[$match['data']]['attr'], trim($attr), $this->spritePath, $this->convertFilePath($match['data']));
-        }, $html['body']);
-
-        return $html['head'].$html['body'];
-    }
-
-    private function convertFilePath(string $path): string
-    {
-        return preg_replace('/.svg$|[^\w\-]/', '', str_replace('/', '-', ltrim($path, '/'))); // ^[^/]
-    }
-
-    private function addFileToSpriteArr(string $hash, string $path): ?array
-    {
-        if (1 === preg_match('/(?:;base64|i:a?i?pgf)/', $svg = file_get_contents($this->sitePath.$path))) { // noop!
-            return null;
-        }
-
-        if (1 === preg_match('/<(?:style|defs)|url\(/', $svg)) {
-            return null; // check links @ __construct
-        }
-
-        //$svg = preg_replace('/((?:id|class)=")/', '$1'.$hash.'__', $svg); // extend  IDs
-        //$svg = preg_replace('/(href="|url\()#/', '$1#'.$hash.'__', $svg); // recover IDs
-
-        //$svg = preg_replace_callback('/<style[^>]*>(?<styl>.+?)<\/style>|<defs[^>]*>(?<defs>.+?)<\/defs>/s', function(array $match) use($hash): string {
-        //
-        //    if(isset($match['styl']))
-        //    {
-        //        $this->styl[] = preg_replace('/\s*(\.|#){1}(.+?)\s*\{/', '$1'.$hash.'__$2{', $match['styl']); // patch CSS # https://mathiasbynens.be/notes/css-escapes
-        //    }
-        //    if(isset($match['defs']))
-        //    {
-        //        $this->defs[] = trim($match['defs']);
-        //    }
-        //    return '';
-        //}, $svg);
-
-        // https://developer.mozilla.org/en-US/docs/Web/SVG/Attribute/xlink:href
-        $svg = preg_replace('/.*<svg|<\/svg>.*|xlink:|\s(?:(?:version|xmlns)|(?:[a-z\-]+\:[a-z\-]+))="[^"]*"/s', '', $svg); // cleanup
-
-        // https://developer.mozilla.org/en-US/docs/Web/SVG/Element/svg#attributes
-        $svg = preg_replace_callback('/([^>]+)\s*(?=>)/s', function (array $match) use (&$attr): string {
-            if (false === preg_match_all('/\s(?<attr>[\w\-]+)="\s*(?<value>[^"]+)\s*"/', $match[1], $matches)) {
-                return $match[0];
-            }
-            foreach ($matches['attr'] as $index => $attribute) {
-                switch ($attribute) {
-                  case 'id':
-                  case 'width':
-                  case 'height':
-                      unset($matches[0][$index]);
-                      break;
-
-                  case 'viewBox':
-                      $attr[] = sprintf('%s="%s"', $attribute, $matches['value'][$index]); // save!
-                      // no break
-                  default:
-                      $matches[0][$index] = sprintf('%s="%s"', $attribute, $matches['value'][$index]); // cleanup
-                }
-            }
-
-            return implode(' ', $matches[0]);
-        }, $svg, 1);
-
-        if ($attr) { // TODO; beautify
-            $this->svgs[] = sprintf('id="%s" %s', $this->convertFilePath($path), $svg); // append ID
-        }
-
-        return !$attr ?: ['attr' => implode(' ', $attr), 'hash' => $hash];
-    }
-
-    private function populateCache(): bool
-    {
-        $storageArr = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Resource\StorageRepository::class)->findAll();
-        foreach ($storageArr as $storage) {
-            if ('relative' == $storage->getConfiguration()['pathType']) {
-                $storageArr[$storage->getUid()] = rtrim($storage->getConfiguration()['basePath'], '/'); // [^/]$
-            }
-        }
-        unset($storageArr[0]); // keep!
-
-        $svgFileArr = GeneralUtility::makeInstance(\HTML\Sourceopt\Resource\SvgFileRepository::class)->findAllByStorageUids(array_keys($storageArr));
-        foreach ($svgFileArr as $index => $row) {
-            if (!$this->svgFileArr[($row['path'] = '/'.$storageArr[$row['storage']].$row['identifier'])] = $this->addFileToSpriteArr($row['sha1'], $row['path'])) { // ^[/]
-                unset($this->svgFileArr[$row['path']]);
-            }
-        }
-
-        unset($storageArr); // save MEM
-        unset($svgFileArr); // save MEM
-
-        $svg = preg_replace_callback(
-            '/<use(?<pre>.*?)(?:xlink:)?href="(?<href>\/.+?\.svg)#[^"]+"(?<post>.*?)[\s\/]*>(?:<\/use>)?/s',
-            function (array $match): string {
-                if (!isset($this->svgFileArr[$match['href']])) { // check usage
-                    return $match[0];
-                }
-                return sprintf('<use%s href="#%s"/>', $match['pre'].$match['post'], $this->convertFilePath($match['href']));
-            },
-            '<svg xmlns="http://www.w3.org/2000/svg">'
-            //."\n<style>\n".implode("\n", $this->styl)."\n</style>"
-            //."\n<defs>\n".implode("\n", $this->defs)."\n</defs>"
-            ."\n<symbol ".implode("</symbol>\n<symbol ", $this->svgs)."</symbol>\n"
-            .'</svg>'
-        );
-
-        //unset($this->styl); // save MEM
-        //unset($this->defs); // save MEM
-        unset($this->svgs); // save MEM
-
-        if (\is_int($var = $GLOBALS['TSFE']->config['config']['sourceopt.']['formatHtml']) && 1 == $var) {
-            $svg = preg_replace('/[\n\r\t\v\0]|\s{2,}/', '', $svg);
-        }
-
-        $svg = preg_replace('/<([a-z]+)\s*(\/|>\s*<\/\1)>\s*/i', '', $svg); // remove emtpy
-        $svg = preg_replace('/<((circle|ellipse|line|path|polygon|polyline|rect|stop|use)\s[^>]+?)\s*>\s*<\/\2>/', '<$1/>', $svg); // shorten/minify
-
-        if (!is_dir($this->sitePath.$this->outputDir)) {
-            GeneralUtility::mkdir_deep($this->sitePath.$this->outputDir);
-        }
-
-        $this->spritePath = $this->outputDir.hash('sha1', serialize($this->svgFileArr)).'.svg';
-        if (false === file_put_contents($this->sitePath.$this->spritePath, $svg)) {
-            return false;
-        }
-
-        $this->svgCache->set('svgFileArr', $this->svgFileArr);
-        $this->svgCache->set('spritePath', $this->spritePath);
-
-        return true;
-    }
+	/**
+	 * SVG-Sprite storage directory.
+	 *
+	 * @var string
+	 */
+	protected $outputDir = '/typo3temp/assets/svg/';
+
+	public function __construct()
+	{
+		//$this->styl = []; # https://stackoverflow.com/questions/39583880/external-svg-fails-to-apply-internal-css
+		//$this->defs = []; # https://bugs.chromium.org/p/chromium/issues/detail?id=751733#c14
+		$this->svgs = [];
+
+		$this->sitePath = \TYPO3\CMS\Core\Core\Environment::getPublicPath(); // [^/]$
+		$this->svgCache = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Cache\CacheManager::class)->getCache('svgstore');
+	}
+
+	public function process(string $html): string
+	{
+		$this->spritePath = $this->svgCache->get('spritePath');
+		$this->svgFileArr = $this->svgCache->get('svgFileArr');
+
+		if (empty($this->spritePath) && !$this->populateCache()) {
+			throw new \Exception('could not write file: '.$this->sitePath.$this->spritePath);
+		}
+
+		if (!file_exists($this->sitePath.$this->spritePath)) {
+			throw new \Exception('file does not exists: '.$this->sitePath.$this->spritePath);
+		}
+
+		if (!preg_match('/(?<head>.+?<\/head>)(?<body>.+)/s', $html, $html) && 5 == \count($html)) {
+			throw new \Exception('fix HTML!');
+		}
+
+		// https://developer.mozilla.org/en-US/docs/Web/HTML/Element/img#attributes
+		$html['body'] = preg_replace_callback('/<img(?<pre>[^>]*)src="(?<src>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?!\s*<\/picture>)/s', function (array $match): string { // ^[/]
+			if (!isset($this->svgFileArr[$match['src']])) { // check usage
+				return $match[0];
+			}
+			$attr = preg_replace('/\s(?:alt|ismap|loading|title|sizes|srcset|usemap|crossorigin|decoding|referrerpolicy)="[^"]*"/', '', $match['pre'].$match['post']); // cleanup
+
+			return sprintf('<svg %s %s><use href="%s#%s"/></svg>', $this->svgFileArr[$match['src']]['attr'], trim($attr), $this->spritePath, $this->convertFilePath($match['src']));
+		}, $html['body']);
+
+		// https://developer.mozilla.org/en-US/docs/Web/HTML/Element/object#attributes
+		$html['body'] = preg_replace_callback('/<object(?<pre>[^>]*)data="(?<data>\/[^"]+\.svg)"(?<post>[^>]*?)[\s\/]*>(?:<\/object>)/s', function (array $match): string { // ^[/]
+			if (!isset($this->svgFileArr[$match['data']])) { // check usage
+				return $match[0];
+			}
+			$attr = preg_replace('/\s(?:form|name|type|usemap)="[^"]*"/', '', $match['pre'].$match['post']); // cleanup
+
+			return sprintf('<svg %s %s><use href="%s#%s"/></svg>', $this->svgFileArr[$match['data']]['attr'], trim($attr), $this->spritePath, $this->convertFilePath($match['data']));
+		}, $html['body']);
+
+		return $html['head'].$html['body'];
+	}
+
+	private function convertFilePath(string $path): string
+	{
+		return preg_replace('/.svg$|[^\w\-]/', '', str_replace('/', '-', ltrim($path, '/'))); // ^[^/]
+	}
+
+	private function addFileToSpriteArr(string $hash, string $path): ?array
+	{
+		if (1 === preg_match('/(?:;base64|i:a?i?pgf)/', $svg = file_get_contents($this->sitePath.$path))) { // noop!
+			return null;
+		}
+
+		if (1 === preg_match('/<(?:style|defs)|url\(/', $svg)) {
+			return null; // check links @ __construct
+		}
+
+		//$svg = preg_replace('/((?:id|class)=")/', '$1'.$hash.'__', $svg); // extend  IDs
+		//$svg = preg_replace('/(href="|url\()#/', '$1#'.$hash.'__', $svg); // recover IDs
+
+		//$svg = preg_replace_callback('/<style[^>]*>(?<styl>.+?)<\/style>|<defs[^>]*>(?<defs>.+?)<\/defs>/s', function(array $match) use($hash): string {
+		//
+		//    if(isset($match['styl']))
+		//    {
+		//        $this->styl[] = preg_replace('/\s*(\.|#){1}(.+?)\s*\{/', '$1'.$hash.'__$2{', $match['styl']); // patch CSS # https://mathiasbynens.be/notes/css-escapes
+		//    }
+		//    if(isset($match['defs']))
+		//    {
+		//        $this->defs[] = trim($match['defs']);
+		//    }
+		//    return '';
+		//}, $svg);
+
+		// https://developer.mozilla.org/en-US/docs/Web/SVG/Attribute/xlink:href
+		$svg = preg_replace('/.*<svg|<\/svg>.*|xlink:|\s(?:(?:version|xmlns)|(?:[a-z\-]+\:[a-z\-]+))="[^"]*"/s', '', $svg); // cleanup
+
+		// https://developer.mozilla.org/en-US/docs/Web/SVG/Element/svg#attributes
+		$svg = preg_replace_callback('/([^>]+)\s*(?=>)/s', function (array $match) use (&$attr): string {
+			if (false === preg_match_all('/\s(?<attr>[\w\-]+)="\s*(?<value>[^"]+)\s*"/', $match[1], $matches)) {
+				return $match[0];
+			}
+			foreach ($matches['attr'] as $index => $attribute) {
+				switch ($attribute) {
+				  case 'id':
+				  case 'width':
+				  case 'height':
+					  unset($matches[0][$index]);
+					  break;
+
+				  case 'viewBox':
+					  $attr[] = sprintf('%s="%s"', $attribute, $matches['value'][$index]); // save!
+					  // no break
+				  default:
+					  $matches[0][$index] = sprintf('%s="%s"', $attribute, $matches['value'][$index]); // cleanup
+				}
+			}
+
+			return implode(' ', $matches[0]);
+		}, $svg, 1);
+
+		if ($attr) { // TODO; beautify
+			$this->svgs[] = sprintf('id="%s" %s', $this->convertFilePath($path), $svg); // append ID
+		}
+
+		return !$attr ?: ['attr' => implode(' ', $attr), 'hash' => $hash];
+	}
+
+	private function populateCache(): bool
+	{
+		$storageArr = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Resource\StorageRepository::class)->findAll();
+		foreach ($storageArr as $storage) {
+			if ('relative' == $storage->getConfiguration()['pathType']) {
+				$storageArr[$storage->getUid()] = rtrim($storage->getConfiguration()['basePath'], '/'); // [^/]$
+			}
+		}
+		unset($storageArr[0]); // keep!
+
+		$svgFileArr = GeneralUtility::makeInstance(\HTML\Sourceopt\Resource\SvgFileRepository::class)->findAllByStorageUids(array_keys($storageArr));
+		foreach ($svgFileArr as $index => $row) {
+			if (!$this->svgFileArr[($row['path'] = '/'.$storageArr[$row['storage']].$row['identifier'])] = $this->addFileToSpriteArr($row['sha1'], $row['path'])) { // ^[/]
+				unset($this->svgFileArr[$row['path']]);
+			}
+		}
+
+		unset($storageArr); // save MEM
+		unset($svgFileArr); // save MEM
+
+		$svg = preg_replace_callback(
+			'/<use(?<pre>.*?)(?:xlink:)?href="(?<href>\/.+?\.svg)#[^"]+"(?<post>.*?)[\s\/]*>(?:<\/use>)?/s',
+			function (array $match): string {
+				if (!isset($this->svgFileArr[$match['href']])) { // check usage
+					return $match[0];
+				}
+				return sprintf('<use%s href="#%s"/>', $match['pre'].$match['post'], $this->convertFilePath($match['href']));
+			},
+			'<svg xmlns="http://www.w3.org/2000/svg">'
+			//."\n<style>\n".implode("\n", $this->styl)."\n</style>"
+			//."\n<defs>\n".implode("\n", $this->defs)."\n</defs>"
+			."\n<symbol ".implode("</symbol>\n<symbol ", $this->svgs)."</symbol>\n"
+			.'</svg>'
+		);
+
+		//unset($this->styl); // save MEM
+		//unset($this->defs); // save MEM
+		unset($this->svgs); // save MEM
+
+		if (\is_int($var = $GLOBALS['TSFE']->config['config']['sourceopt.']['formatHtml']) && 1 == $var) {
+			$svg = preg_replace('/[\n\r\t\v\0]|\s{2,}/', '', $svg);
+		}
+
+		$svg = preg_replace('/<([a-z]+)\s*(\/|>\s*<\/\1)>\s*/i', '', $svg); // remove emtpy
+		$svg = preg_replace('/<((circle|ellipse|line|path|polygon|polyline|rect|stop|use)\s[^>]+?)\s*>\s*<\/\2>/', '<$1/>', $svg); // shorten/minify
+
+		if (!is_dir($this->sitePath.$this->outputDir)) {
+			GeneralUtility::mkdir_deep($this->sitePath.$this->outputDir);
+		}
+
+		$this->spritePath = $this->outputDir.hash('sha1', serialize($this->svgFileArr)).'.svg';
+		if (false === file_put_contents($this->sitePath.$this->spritePath, $svg)) {
+			return false;
+		}
+
+		$this->svgCache->set('svgFileArr', $this->svgFileArr);
+		$this->svgCache->set('spritePath', $this->spritePath);
+
+		return true;
+	}
 }

Classes/Resource/SvgFileRepository.php

Indentation +42 added lines, -42 removed lines

@@ -13,46 +13,46 @@
  */
 class SvgFileRepository extends \TYPO3\CMS\Core\Resource\FileRepository
 {
-    /**
-     * Retrieves all used SVGs within given storage-array
-     *
-     * @param array $storageUids
-     * @return array
-     */
-    public function findAllByStorageUids(array $storageUids): array
-    {
-        return
-            ($queryBuilder = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Database\ConnectionPool::class)->getQueryBuilderForTable($this->table))
-                ->select('sys_file.storage', 'sys_file.identifier', 'sys_file.sha1')
-                ->from($this->table)
-                ->innerJoin(
-                    'sys_file',
-                    'sys_file_reference',
-                    'sys_file_reference',
-                    $queryBuilder->expr()->eq(
-                        'sys_file_reference.uid_local',
-                        $queryBuilder->quoteIdentifier('sys_file.uid')
-                    )
-                )
-                ->where(
-                    $queryBuilder->expr()->in(
-                        'sys_file.storage',
-                        $queryBuilder->createNamedParameter($storageUids, \Doctrine\DBAL\Connection::PARAM_INT_ARRAY)
-                    ),
-                    $queryBuilder->expr()->lt(
-                        'sys_file.size',
-                        $queryBuilder->createNamedParameter((int) $GLOBALS['TSFE']->config['config']['svgstore.']['fileSize'], \PDO::PARAM_INT)
-                    ),
-                    $queryBuilder->expr()->eq(
-                        'sys_file.mime_type',
-                        $queryBuilder->createNamedParameter('image/svg+xml', \PDO::PARAM_STR)
-                    )
-                )
-                ->groupBy('sys_file.uid')
-                ->orderBy('sys_file.storage')
-                ->addOrderBy('sys_file.identifier')
-                ->execute()
-                ->fetchAll() // TODO; use stdClass
-            ;
-    }
+	/**
+	 * Retrieves all used SVGs within given storage-array
+	 *
+	 * @param array $storageUids
+	 * @return array
+	 */
+	public function findAllByStorageUids(array $storageUids): array
+	{
+		return
+			($queryBuilder = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Database\ConnectionPool::class)->getQueryBuilderForTable($this->table))
+				->select('sys_file.storage', 'sys_file.identifier', 'sys_file.sha1')
+				->from($this->table)
+				->innerJoin(
+					'sys_file',
+					'sys_file_reference',
+					'sys_file_reference',
+					$queryBuilder->expr()->eq(
+						'sys_file_reference.uid_local',
+						$queryBuilder->quoteIdentifier('sys_file.uid')
+					)
+				)
+				->where(
+					$queryBuilder->expr()->in(
+						'sys_file.storage',
+						$queryBuilder->createNamedParameter($storageUids, \Doctrine\DBAL\Connection::PARAM_INT_ARRAY)
+					),
+					$queryBuilder->expr()->lt(
+						'sys_file.size',
+						$queryBuilder->createNamedParameter((int) $GLOBALS['TSFE']->config['config']['svgstore.']['fileSize'], \PDO::PARAM_INT)
+					),
+					$queryBuilder->expr()->eq(
+						'sys_file.mime_type',
+						$queryBuilder->createNamedParameter('image/svg+xml', \PDO::PARAM_STR)
+					)
+				)
+				->groupBy('sys_file.uid')
+				->orderBy('sys_file.storage')
+				->addOrderBy('sys_file.identifier')
+				->execute()
+				->fetchAll() // TODO; use stdClass
+			;
+	}
 }