update

lib/common/HTMLIndent.php

@@ -0,0 +1,141 @@
+<?php
+
+namespace Common;
+
+/**
+ * @link https://github.com/gajus/dintent for the canonical source repository
+ * @license https://github.com/gajus/dintent/blob/master/LICENSE BSD 3-Clause
+ */
+class HTMLIndent {
+    private
+        $log = array(),
+        $indent = '    ',
+        $temporary_replacements_script = array(),
+        $temporary_replacements_inline = array();
+
+    const MATCH_INDENT_NO = 0;
+    const MATCH_INDENT_DECREASE = 1;
+    const MATCH_INDENT_INCREASE = 2;
+    const MATCH_DISCARD = 3;
+
+    public function indent ($input) {
+        $this->log = array();
+
+        // Dindent does not indent <script> body. Instead, it temporary removes it from the code, indents the input, and restores the script body.
+        if (preg_match_all('/<script\b[^>]*>([\s\S]*?)<\/script>/mi', $input, $matches)) {
+            $this->temporary_replacements_script = $matches[0];
+            foreach ($matches[0] as $i => $match) {
+                $input = str_replace($match, '<script>' . ($i + 1) . '</script>', $input);
+            }
+        }
+
+        // Removing double whitespaces to make the source code easier to read.
+        // With exception of <pre>/ CSS white-space changing the default behaviour, double whitespace is meaningless in HTML output.
+        // This reason alone is sufficient not to use Dindent in production.
+        $input = str_replace("\t", '', $input);
+        $input = preg_replace('/\s{2,}/', ' ', $input);
+
+        // Remove inline tags and replace them with text entities.
+        if (preg_match_all('/<(b|i|abbr|em|strong|a|span)[^>]*>(?:[^<]*)<\/\1>/', $input, $matches)) {
+            $this->temporary_replacements_inline = $matches[0];
+            foreach ($matches[0] as $i => $match) {
+                $input = str_replace($match, 'ᐃ' . ($i + 1) . 'ᐃ', $input);
+            }
+        }
+
+        $subject = $input;
+
+        $output = '';
+
+        $next_line_indentation_level = 0;
+
+        do {
+            $indentation_level = $next_line_indentation_level;
+
+            $patterns = array(
+                // block tag
+                '/^(<([a-z]+)(?:[^>]*)>(?:[^<]*)<\/(?:\2)>)/' => static::MATCH_INDENT_NO,
+                // DOCTYPE
+                '/^<!([^>]*)>/' => static::MATCH_INDENT_NO,
+                // tag with implied closing
+                '/^<(input|link|meta|base|br|img|hr)([^>]*)>/' => static::MATCH_INDENT_NO,
+                // opening tag
+                '/^<[^\/]([^>]*)>/' => static::MATCH_INDENT_INCREASE,
+                // closing tag
+                '/^<\/([^>]*)>/' => static::MATCH_INDENT_DECREASE,
+                // self-closing tag
+                '/^<(.+)\/>/' => static::MATCH_INDENT_DECREASE,
+                // whitespace
+                '/^(\s+)/' => static::MATCH_DISCARD,
+                // text node
+                '/([^<]+)/' => static::MATCH_INDENT_NO
+            );
+            $rules = array('NO', 'DECREASE', 'INCREASE', 'DISCARD');
+
+            foreach ($patterns as $pattern => $rule) {
+                if ($match = preg_match($pattern, $subject, $matches)) {
+                    $this->log[] = array(
+                        'rule' => $rules[$rule],
+                        'pattern' => $pattern,
+                        'subject' => $subject,
+                        'match' => $matches[0]
+                    );
+
+                    $subject = mb_substr($subject, mb_strlen($matches[0]));
+
+                    if ($rule === static::MATCH_DISCARD) {
+                        break;
+                    }
+
+                    if ($rule === static::MATCH_INDENT_NO) {
+
+                    } else if ($rule === static::MATCH_INDENT_DECREASE) {
+                        $next_line_indentation_level--;
+                        $indentation_level--;
+                    } else {
+                        $next_line_indentation_level++;
+                    }
+
+                    if ($indentation_level < 0) {
+                        $indentation_level = 0;
+                    }
+
+                    #$output .= str_repeat($this->indent, $indentation_level) . 'A:' . $indentation_level . "\n";
+                    $output .= str_repeat($this->indent, $indentation_level) . $matches[0] . "\n";
+
+                    break;
+                }
+            }
+        } while ($match);
+
+        $interpreted_input = '';
+        foreach ($this->log as $e) {
+            $interpreted_input .= $e['match'];
+        }
+
+        if ($interpreted_input !== $input) {
+            throw new \RuntimeException('Did not reproduce the exact input.');
+        }
+
+        $output = preg_replace('/(<(\w+)[^>]*>)\s*(<\/\2>)/', '\\1\\3', $output);
+
+        foreach ($this->temporary_replacements_script as $i => $original) {
+            $output = str_replace('<script>' . ($i + 1) . '</script>', $original, $output);
+        }
+
+        foreach ($this->temporary_replacements_inline as $i => $original) {
+            $output = str_replace('ᐃ' . ($i + 1) . 'ᐃ', $original, $output);
+        }
+
+        return trim($output);
+    }
+
+    /**
+     * Debugging utility. Get log for the last indent operation.
+     *
+     * @return array
+     */
+    public function getLog () {
+        return $this->log;
+    }
+}