tags. } } } } else { // State 2.2:

• // ---- // State 2.4:

  // --- } } /** * Splits up a text in paragraph tokens and appends them * to the result stream that will replace the original * @param $data String text data that will be processed * into paragraphs * @param $result Reference to array of tokens that the * tags will be appended onto * @param $config Instance of HTMLPurifier_Config * @param $context Instance of HTMLPurifier_Context */ private function _splitText($data, &$result) { $raw_paragraphs = explode("\n\n", $data); $paragraphs = array(); // without empty paragraphs $needs_start = false; $needs_end = false; $c = count($raw_paragraphs); if ($c == 1) { // There were no double-newlines, abort quickly. In theory this // should never happen. $result[] = new HTMLPurifier_Token_Text($data); return; } for ($i = 0; $i < $c; $i++) { $par = $raw_paragraphs[$i]; if (trim($par) !== '') { $paragraphs[] = $par; } else { if ($i == 0) { // Double newline at the front if (empty($result)) { // The empty result indicates that the AutoParagraph // injector did not add any start paragraph tokens. // This means that we have been in a paragraph for // a while, and the newline means we should start a new one. $result[] = new HTMLPurifier_Token_End('p'); $result[] = new HTMLPurifier_Token_Text("\n\n"); // However, the start token should only be added if // there is more processing to be done (i.e. there are // real paragraphs in here). If there are none, the // next start paragraph tag will be handled by the // next call to the injector $needs_start = true; } else { // We just started a new paragraph! // Reinstate a double-newline for presentation's sake, since // it was in the source code. array_unshift($result, new HTMLPurifier_Token_Text("\n\n")); } } elseif ($i + 1 == $c) { // Double newline at the end // There should be a trailing

  when we're finally done. $needs_end = true; } } } // Check if this was just a giant blob of whitespace. Move this earlier, // perhaps? if (empty($paragraphs)) { return; } // Add the start tag indicated by \n\n at the beginning of $data if ($needs_start) { $result[] = $this->_pStart(); } // Append the paragraphs onto the result foreach ($paragraphs as $par) { $result[] = new HTMLPurifier_Token_Text($par); $result[] = new HTMLPurifier_Token_End('p'); $result[] = new HTMLPurifier_Token_Text("\n\n"); $result[] = $this->_pStart(); } // Remove trailing start token; Injector will handle this later if // it was indeed needed. This prevents from needing to do a lookahead, // at the cost of a lookbehind later. array_pop($result); // If there is no need for an end tag, remove all of it and let // MakeWellFormed close it later. if (!$needs_end) { array_pop($result); // removes \n\n array_pop($result); // removes
  } } /** * Returns true if passed token is inline (and, ergo, allowed in * paragraph tags) */ private function _isInline($token) { return isset($this->htmlDefinition->info['p']->child->elements[$token->name]); } /** * Looks ahead in the token list and determines whether or not we need * to insert a

  tag. */ private function _pLookAhead() { $this->_current($i, $current); if ($current instanceof HTMLPurifier_Token_Start) $nesting = 1; else $nesting = 0; $ok = false; while ($this->_forwardUntilEndToken($i, $current, $nesting)) { $result = $this->_checkNeedsP($current); if ($result !== null) { $ok = $result; break; } } return $ok; } /** * Iterator function, which starts with the next token and continues until * you reach the end of the input tokens. * @warning Please prevent previous references from interfering with this * functions by setting $i = null beforehand! * @param &$i Current integer index variable for inputTokens * @param &$current Current token variable. Do NOT use $token, as that variable is also a reference */ private function _forward(&$i, &$current) { if ($i === null) $i = $this->inputIndex + 1; else $i++; if (!isset($this->inputTokens[$i])) return false; $current = $this->inputTokens[$i]; return true; } /** * Similar to _forward, but accepts a third parameter $nesting (which * should be initialized at 0) and stops when we hit the end tag * for the node $this->inputIndex starts in. */ private function _forwardUntilEndToken(&$i, &$current, &$nesting) { $result = $this->_forward($i, $current); if (!$result) return false; if ($nesting === null) $nesting = 0; if ($current instanceof HTMLPurifier_Token_Start) $nesting++; elseif ($current instanceof HTMLPurifier_Token_End) { if ($nesting <= 0) return false; $nesting--; } return true; } /** * Iterator function, starts with the previous token and continues until * you reach the beginning of input tokens. * @warning Please prevent previous references from interfering with this * functions by setting $i = null beforehand! * @param &$i Current integer index variable for inputTokens * @param &$current Current token variable. Do NOT use $token, as that variable is also a reference */ private function _backward(&$i, &$current) { if ($i === null) $i = $this->inputIndex - 1; else $i--; if ($i < 0) return false; $current = $this->inputTokens[$i]; return true; } /** * Initializes the iterator at the current position. Use in a do {} while; * loop to force the _forward and _backward functions to start at the * current location. * @warning Please prevent previous references from interfering with this * functions by setting $i = null beforehand! * @param &$i Current integer index variable for inputTokens * @param &$current Current token variable. Do NOT use $token, as that variable is also a reference */ private function _current(&$i, &$current) { if ($i === null) $i = $this->inputIndex; $current = $this->inputTokens[$i]; } /** * Determines if a particular token requires an earlier inline token * to get a paragraph. This should be used with _forwardUntilEndToken */ private function _checkNeedsP($current) { if ($current instanceof HTMLPurifier_Token_Start){ if (!$this->_isInline($current)) { //

  PAR1

  // ---- // Terminate early, since we hit a block element return false; } } elseif ($current instanceof HTMLPurifier_Token_Text) { if (strpos($current->data, "\n\n") !== false) { //

  PAR1PAR1\n\nPAR2 // ---- return true; } else { //

  PAR1PAR1... // ---- } } return null; } }