From 7a40f4354b32809af3d0cfd6e3af0eda02ab0e0a Mon Sep 17 00:00:00 2001 From: friendica Date: Sat, 12 May 2012 17:57:41 -0700 Subject: some important stuff we'll need --- .../HTMLPurifier/Injector/AutoParagraph.php | 345 +++++++++++++++++++++ .../HTMLPurifier/Injector/DisplayLinkURI.php | 26 ++ .../library/HTMLPurifier/Injector/Linkify.php | 46 +++ .../HTMLPurifier/Injector/PurifierLinkify.php | 45 +++ .../library/HTMLPurifier/Injector/RemoveEmpty.php | 51 +++ .../Injector/RemoveSpansWithoutAttributes.php | 60 ++++ .../library/HTMLPurifier/Injector/SafeObject.php | 91 ++++++ 7 files changed, 664 insertions(+) create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/DisplayLinkURI.php create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/Linkify.php create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/PurifierLinkify.php create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveEmpty.php create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php create mode 100644 lib/htmlpurifier/library/HTMLPurifier/Injector/SafeObject.php (limited to 'lib/htmlpurifier/library/HTMLPurifier/Injector') diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php new file mode 100644 index 000000000..afa760892 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php @@ -0,0 +1,345 @@ +armor['MakeWellFormed_TagClosedError'] = true; + return $par; + } + + public function handleText(&$token) { + $text = $token->data; + // Does the current parent allow

tags? + if ($this->allowsElement('p')) { + if (empty($this->currentNesting) || strpos($text, "\n\n") !== false) { + // Note that we have differing behavior when dealing with text + // in the anonymous root node, or a node inside the document. + // If the text as a double-newline, the treatment is the same; + // if it doesn't, see the next if-block if you're in the document. + + $i = $nesting = null; + if (!$this->forwardUntilEndToken($i, $current, $nesting) && $token->is_whitespace) { + // State 1.1: ... ^ (whitespace, then document end) + // ---- + // This is a degenerate case + } else { + if (!$token->is_whitespace || $this->_isInline($current)) { + // State 1.2: PAR1 + // ---- + + // State 1.3: PAR1\n\nPAR2 + // ------------ + + // State 1.4:

PAR1\n\nPAR2 (see State 2) + // ------------ + $token = array($this->_pStart()); + $this->_splitText($text, $token); + } else { + // State 1.5: \n
+ // -- + } + } + } else { + // State 2:
PAR1... (similar to 1.4) + // ---- + + // We're in an element that allows paragraph tags, but we're not + // sure if we're going to need them. + if ($this->_pLookAhead()) { + // State 2.1:
PAR1PAR1\n\nPAR2 + // ---- + // Note: This will always be the first child, since any + // previous inline element would have triggered this very + // same routine, and found the double newline. One possible + // exception would be a comment. + $token = array($this->_pStart(), $token); + } else { + // State 2.2.1:
PAR1
+ // ---- + + // State 2.2.2:
PAR1PAR1
+ // ---- + } + } + // Is the current parent a

tag? + } elseif ( + !empty($this->currentNesting) && + $this->currentNesting[count($this->currentNesting)-1]->name == 'p' + ) { + // State 3.1: ...

PAR1 + // ---- + + // State 3.2: ...

PAR1\n\nPAR2 + // ------------ + $token = array(); + $this->_splitText($text, $token); + // Abort! + } else { + // State 4.1: ...PAR1 + // ---- + + // State 4.2: ...PAR1\n\nPAR2 + // ------------ + } + } + + public function handleElement(&$token) { + // We don't have to check if we're already in a

tag for block + // tokens, because the tag would have been autoclosed by MakeWellFormed. + if ($this->allowsElement('p')) { + if (!empty($this->currentNesting)) { + if ($this->_isInline($token)) { + // State 1:

... + // --- + + // Check if this token is adjacent to the parent token + // (seek backwards until token isn't whitespace) + $i = null; + $this->backward($i, $prev); + + if (!$prev instanceof HTMLPurifier_Token_Start) { + // Token wasn't adjacent + + if ( + $prev instanceof HTMLPurifier_Token_Text && + substr($prev->data, -2) === "\n\n" + ) { + // State 1.1.4:

PAR1

\n\n + // --- + + // Quite frankly, this should be handled by splitText + $token = array($this->_pStart(), $token); + } else { + // State 1.1.1:

PAR1

+ // --- + + // State 1.1.2:

+ // --- + + // State 1.1.3:
PAR + // --- + } + + } else { + // State 1.2.1:
+ // --- + + // Lookahead to see if

is needed. + if ($this->_pLookAhead()) { + // State 1.3.1:

PAR1\n\nPAR2 + // --- + $token = array($this->_pStart(), $token); + } else { + // State 1.3.2:
PAR1
+ // --- + + // State 1.3.3:
PAR1
\n\n
+ // --- + } + } + } else { + // State 2.3: ...
+ // ----- + } + } else { + if ($this->_isInline($token)) { + // State 3.1: + // --- + // This is where the {p} tag is inserted, not reflected in + // inputTokens yet, however. + $token = array($this->_pStart(), $token); + } else { + // State 3.2:
+ // ----- + } + + $i = null; + if ($this->backward($i, $prev)) { + if ( + !$prev instanceof HTMLPurifier_Token_Text + ) { + // State 3.1.1: ...

{p} + // --- + + // State 3.2.1: ...

+ // ----- + + if (!is_array($token)) $token = array($token); + array_unshift($token, new HTMLPurifier_Token_Text("\n\n")); + } else { + // State 3.1.2: ...

\n\n{p} + // --- + + // State 3.2.2: ...

\n\n
+ // ----- + + // Note: PAR cannot occur because PAR would have been + // wrapped in

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 $ok; + } + + /** + * Determines if a particular token requires an earlier inline token + * to get a paragraph. This should be used with _forwardUntilEndToken + */ + private function _checkNeedsP($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; + } + +} + +// vim: et sw=4 sts=4 diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/DisplayLinkURI.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/DisplayLinkURI.php new file mode 100644 index 000000000..9dce9bd08 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/DisplayLinkURI.php @@ -0,0 +1,26 @@ +start->attr['href'])){ + $url = $token->start->attr['href']; + unset($token->start->attr['href']); + $token = array($token, new HTMLPurifier_Token_Text(" ($url)")); + } else { + // nothing to display + } + } +} + +// vim: et sw=4 sts=4 diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/Linkify.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/Linkify.php new file mode 100644 index 000000000..296dac282 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/Linkify.php @@ -0,0 +1,46 @@ + array('href')); + + public function handleText(&$token) { + if (!$this->allowsElement('a')) return; + + if (strpos($token->data, '://') === false) { + // our really quick heuristic failed, abort + // this may not work so well if we want to match things like + // "google.com", but then again, most people don't + return; + } + + // there is/are URL(s). Let's split the string: + // Note: this regex is extremely permissive + $bits = preg_split('#((?:https?|ftp)://[^\s\'"<>()]+)#S', $token->data, -1, PREG_SPLIT_DELIM_CAPTURE); + + $token = array(); + + // $i = index + // $c = count + // $l = is link + for ($i = 0, $c = count($bits), $l = false; $i < $c; $i++, $l = !$l) { + if (!$l) { + if ($bits[$i] === '') continue; + $token[] = new HTMLPurifier_Token_Text($bits[$i]); + } else { + $token[] = new HTMLPurifier_Token_Start('a', array('href' => $bits[$i])); + $token[] = new HTMLPurifier_Token_Text($bits[$i]); + $token[] = new HTMLPurifier_Token_End('a'); + } + } + + } + +} + +// vim: et sw=4 sts=4 diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/PurifierLinkify.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/PurifierLinkify.php new file mode 100644 index 000000000..ad2455a91 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/PurifierLinkify.php @@ -0,0 +1,45 @@ + array('href')); + + public function prepare($config, $context) { + $this->docURL = $config->get('AutoFormat.PurifierLinkify.DocURL'); + return parent::prepare($config, $context); + } + + public function handleText(&$token) { + if (!$this->allowsElement('a')) return; + if (strpos($token->data, '%') === false) return; + + $bits = preg_split('#%([a-z0-9]+\.[a-z0-9]+)#Si', $token->data, -1, PREG_SPLIT_DELIM_CAPTURE); + $token = array(); + + // $i = index + // $c = count + // $l = is link + for ($i = 0, $c = count($bits), $l = false; $i < $c; $i++, $l = !$l) { + if (!$l) { + if ($bits[$i] === '') continue; + $token[] = new HTMLPurifier_Token_Text($bits[$i]); + } else { + $token[] = new HTMLPurifier_Token_Start('a', + array('href' => str_replace('%s', $bits[$i], $this->docURL))); + $token[] = new HTMLPurifier_Token_Text('%' . $bits[$i]); + $token[] = new HTMLPurifier_Token_End('a'); + } + } + + } + +} + +// vim: et sw=4 sts=4 diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveEmpty.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveEmpty.php new file mode 100644 index 000000000..638bfca03 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveEmpty.php @@ -0,0 +1,51 @@ +config = $config; + $this->context = $context; + $this->removeNbsp = $config->get('AutoFormat.RemoveEmpty.RemoveNbsp'); + $this->removeNbspExceptions = $config->get('AutoFormat.RemoveEmpty.RemoveNbsp.Exceptions'); + $this->attrValidator = new HTMLPurifier_AttrValidator(); + } + + public function handleElement(&$token) { + if (!$token instanceof HTMLPurifier_Token_Start) return; + $next = false; + for ($i = $this->inputIndex + 1, $c = count($this->inputTokens); $i < $c; $i++) { + $next = $this->inputTokens[$i]; + if ($next instanceof HTMLPurifier_Token_Text) { + if ($next->is_whitespace) continue; + if ($this->removeNbsp && !isset($this->removeNbspExceptions[$token->name])) { + $plain = str_replace("\xC2\xA0", "", $next->data); + $isWsOrNbsp = $plain === '' || ctype_space($plain); + if ($isWsOrNbsp) continue; + } + } + break; + } + if (!$next || ($next instanceof HTMLPurifier_Token_End && $next->name == $token->name)) { + if ($token->name == 'colgroup') return; + $this->attrValidator->validateToken($token, $this->config, $this->context); + $token->armor['ValidateAttributes'] = true; + if (isset($token->attr['id']) || isset($token->attr['name'])) return; + $token = $i - $this->inputIndex + 1; + for ($b = $this->inputIndex - 1; $b > 0; $b--) { + $prev = $this->inputTokens[$b]; + if ($prev instanceof HTMLPurifier_Token_Text && $prev->is_whitespace) continue; + break; + } + // This is safe because we removed the token that triggered this. + $this->rewind($b - 1); + return; + } + } + +} + +// vim: et sw=4 sts=4 diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php new file mode 100644 index 000000000..b21313470 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php @@ -0,0 +1,60 @@ +attrValidator = new HTMLPurifier_AttrValidator(); + $this->config = $config; + $this->context = $context; + return parent::prepare($config, $context); + } + + public function handleElement(&$token) { + if ($token->name !== 'span' || !$token instanceof HTMLPurifier_Token_Start) { + return; + } + + // We need to validate the attributes now since this doesn't normally + // happen until after MakeWellFormed. If all the attributes are removed + // the span needs to be removed too. + $this->attrValidator->validateToken($token, $this->config, $this->context); + $token->armor['ValidateAttributes'] = true; + + if (!empty($token->attr)) { + return; + } + + $nesting = 0; + $spanContentTokens = array(); + while ($this->forwardUntilEndToken($i, $current, $nesting)) {} + + if ($current instanceof HTMLPurifier_Token_End && $current->name === 'span') { + // Mark closing span tag for deletion + $current->markForDeletion = true; + // Delete open span tag + $token = false; + } + } + + public function handleEnd(&$token) { + if ($token->markForDeletion) { + $token = false; + } + } +} + +// vim: et sw=4 sts=4 diff --git a/lib/htmlpurifier/library/HTMLPurifier/Injector/SafeObject.php b/lib/htmlpurifier/library/HTMLPurifier/Injector/SafeObject.php new file mode 100644 index 000000000..c1d8b0412 --- /dev/null +++ b/lib/htmlpurifier/library/HTMLPurifier/Injector/SafeObject.php @@ -0,0 +1,91 @@ + 'never', + 'allowNetworking' => 'internal', + ); + protected $allowedParam = array( + 'wmode' => true, + 'movie' => true, + 'flashvars' => true, + 'src' => true, + 'allowFullScreen' => true, // if omitted, assume to be 'false' + ); + + public function prepare($config, $context) { + parent::prepare($config, $context); + } + + public function handleElement(&$token) { + if ($token->name == 'object') { + $this->objectStack[] = $token; + $this->paramStack[] = array(); + $new = array($token); + foreach ($this->addParam as $name => $value) { + $new[] = new HTMLPurifier_Token_Empty('param', array('name' => $name, 'value' => $value)); + } + $token = $new; + } elseif ($token->name == 'param') { + $nest = count($this->currentNesting) - 1; + if ($nest >= 0 && $this->currentNesting[$nest]->name === 'object') { + $i = count($this->objectStack) - 1; + if (!isset($token->attr['name'])) { + $token = false; + return; + } + $n = $token->attr['name']; + // We need this fix because YouTube doesn't supply a data + // attribute, which we need if a type is specified. This is + // *very* Flash specific. + if (!isset($this->objectStack[$i]->attr['data']) && + ($token->attr['name'] == 'movie' || $token->attr['name'] == 'src')) { + $this->objectStack[$i]->attr['data'] = $token->attr['value']; + } + // Check if the parameter is the correct value but has not + // already been added + if ( + !isset($this->paramStack[$i][$n]) && + isset($this->addParam[$n]) && + $token->attr['name'] === $this->addParam[$n] + ) { + // keep token, and add to param stack + $this->paramStack[$i][$n] = true; + } elseif (isset($this->allowedParam[$n])) { + // keep token, don't do anything to it + // (could possibly check for duplicates here) + } else { + $token = false; + } + } else { + // not directly inside an object, DENY! + $token = false; + } + } + } + + public function handleEnd(&$token) { + // This is the WRONG way of handling the object and param stacks; + // we should be inserting them directly on the relevant object tokens + // so that the global stack handling handles it. + if ($token->name == 'object') { + array_pop($this->objectStack); + array_pop($this->paramStack); + } + } + +} + +// vim: et sw=4 sts=4 -- cgit v1.2.3