From 6854cb3f4d8219cf1829e32122eb2502a916eae9 Mon Sep 17 00:00:00 2001 From: Andreas Baumann Date: Sat, 1 Feb 2020 09:05:48 +0100 Subject: initial checkin --- .../HTMLPurifier/Injector/AutoParagraph.php | 356 +++++++++++++++++++++ 1 file changed, 356 insertions(+) create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php (limited to 'vendor/ezyang/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php') diff --git a/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php new file mode 100644 index 0000000..4afdd12 --- /dev/null +++ b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Injector/AutoParagraph.php @@ -0,0 +1,356 @@ +armor['MakeWellFormed_TagClosedError'] = true; + return $par; + } + + /** + * @param HTMLPurifier_Token_Text $token + */ + public function handleText(&$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 + // ------------ + } + } + + /** + * @param HTMLPurifier_Token $token + */ + public function handleElement(&$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 string $data String text data that will be processed + * into paragraphs + * @param HTMLPurifier_Token[] $result Reference to array of tokens that the + * tags will be appended onto + */ + private function _splitText($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) + * @param HTMLPurifier_Token $token + * @return bool + */ + private function _isInline($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. + * @return bool + */ + private function _pLookAhead() + { + if ($this->currentToken instanceof HTMLPurifier_Token_Start) { + $nesting = 1; + } else { + $nesting = 0; + } + $ok = false; + $i = null; + while ($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 + * @param HTMLPurifier_Token $current + * @return bool + */ + private function _checkNeedsP($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 -- cgit v1.2.3-70-g09d2