/**
* @return RegexHelper
*/
public static function getInstance()
{
if (self::$instance === null) {
self::$instance = new CommonMark_Util_RegexHelper();
}
return self::$instance;
}
/**
* @param string $ln
* @param int $lineNumber
*/
protected function incorporateLine($ln, $lineNumber)
{
$allMatched = true;
$offset = 0;
$blank = false;
$container = $this->doc;
$oldTip = $this->tip;
// Convert tabs to spaces:
$ln = self::detabLine($ln);
// For each containing block, try to parse the associated line start.
// Bail out on failure: container will point to the last matching block.
// Set all_matched to false if not all containers match.
while ($container->hasChildren()) {
/** @var BlockElement $lastChild */
$lastChild = $container->getChildren()->last();
if (!$lastChild->getIsOpen()) {
break;
}
$container = $lastChild;
$match = CommonMark_Util_RegexHelper::matchAt('/[^ ]/', $ln, $offset);
if ($match === null) {
$firstNonSpace = strlen($ln);
$blank = true;
} else {
$firstNonSpace = $match;
$blank = false;
}
$indent = $firstNonSpace - $offset;
switch ($container->getType()) {
case CommonMark_Element_BlockElement::TYPE_BLOCK_QUOTE:
$matched = $indent <= 3 && isset($ln[$firstNonSpace]) && $ln[$firstNonSpace] === '>';
if ($matched) {
$offset = $firstNonSpace + 1;
if (isset($ln[$offset]) && $ln[$offset] === ' ') {
$offset++;
}
} else {
$allMatched = false;
}
break;
case CommonMark_Element_BlockElement::TYPE_LIST_ITEM:
$listData = $container->getExtra('list_data');
$increment = $listData['marker_offset'] + $listData['padding'];
if ($indent >= $increment) {
$offset += $increment;
} elseif ($blank) {
$offset = $firstNonSpace;
} else {
$allMatched = false;
}
break;
case CommonMark_Element_BlockElement::TYPE_INDENTED_CODE:
if ($indent >= self::CODE_INDENT) {
$offset += self::CODE_INDENT;
} elseif ($blank) {
$offset = $firstNonSpace;
} else {
$allMatched = false;
}
break;
case CommonMark_Element_BlockElement::TYPE_ATX_HEADER:
case CommonMark_Element_BlockElement::TYPE_SETEXT_HEADER:
case CommonMark_Element_BlockElement::TYPE_HORIZONTAL_RULE:
// a header can never contain > 1 line, so fail to match:
$allMatched = false;
break;
case CommonMark_Element_BlockElement::TYPE_FENCED_CODE:
// skip optional spaces of fence offset
$i = $container->getExtra('fence_offset');
while ($i > 0 && $ln[$offset] === ' ') {
$offset++;
$i--;
}
break;
case CommonMark_Element_BlockElement::TYPE_HTML_BLOCK:
if ($blank) {
$allMatched = false;
}
break;
case CommonMark_Element_BlockElement::TYPE_PARAGRAPH:
if ($blank) {
$container->setIsLastLineBlank(true);
$allMatched = false;
}
break;
default:
// Nothing
}
if (!$allMatched) {
$container = $container->getParent();
// back up to the last matching block
break;
}
}
$lastMatchedContainer = $container;
// This function is used to finalize and close any unmatched
// blocks. We aren't ready to do this now, because we might
// have a lazy paragraph continuation, in which case we don't
// want to close unmatched blocks. So we store this closure for
// use later, when we have more information.
$closeUnmatchedBlocksAlreadyDone = false;
// Check to see if we've hit 2nd blank line; if so break out of list:
if ($blank && $container->getIsLastLineBlank()) {
$this->breakOutOfLists($container, $lineNumber);
}
// Unless last matched container is a code block, try new container starts,
// adding children to the last matched container:
while ($container->getType() != CommonMark_Element_BlockElement::TYPE_FENCED_CODE && $container->getType() != CommonMark_Element_BlockElement::TYPE_INDENTED_CODE && $container->getType() != CommonMark_Element_BlockElement::TYPE_HTML_BLOCK && CommonMark_Util_RegexHelper::matchAt('/^[ #`~*+_=<>0-9-]/', $ln, $offset) !== null) {
$match = CommonMark_Util_RegexHelper::matchAt('/[^ ]/', $ln, $offset);
if ($match === null) {
$firstNonSpace = strlen($ln);
$blank = true;
} else {
$firstNonSpace = $match;
$blank = false;
}
$indent = $firstNonSpace - $offset;
if ($indent >= self::CODE_INDENT) {
// indented code
if ($this->tip->getType() != CommonMark_Element_BlockElement::TYPE_PARAGRAPH && !$blank) {
$offset += self::CODE_INDENT;
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_INDENTED_CODE, $lineNumber, $offset);
} else {
// ident > 4 in a lazy paragraph continuation
break;
}
} elseif (!$blank && $ln[$firstNonSpace] === '>') {
// blockquote
$offset = $firstNonSpace + 1;
// optional following space
if (isset($ln[$offset]) && $ln[$offset] === ' ') {
$offset++;
}
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_BLOCK_QUOTE, $lineNumber, $offset);
} elseif ($match = CommonMark_Util_RegexHelper::matchAll('/^#{1,6}(?: +|$)/', $ln, $firstNonSpace)) {
// ATX header
$offset = $firstNonSpace + strlen($match[0]);
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_ATX_HEADER, $lineNumber, $firstNonSpace);
$container->setExtra('level', strlen(trim($match[0])));
// number of #s
// remove trailing ###s
$container->getStrings()->add(preg_replace('/(?:(\\\\#) *#*| *#+) *$/', '$1', substr($ln, $offset)));
break;
} elseif ($match = CommonMark_Util_RegexHelper::matchAll('/^`{3,}(?!.*`)|^~{3,}(?!.*~)/', $ln, $firstNonSpace)) {
// fenced code block
$fenceLength = strlen($match[0]);
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_FENCED_CODE, $lineNumber, $firstNonSpace);
$container->setExtra('fence_length', $fenceLength);
$container->setExtra('fence_char', $match[0][0]);
$container->setExtra('fence_offset', $firstNonSpace - $offset);
$offset = $firstNonSpace + $fenceLength;
break;
} elseif (CommonMark_Util_RegexHelper::matchAt(CommonMark_Util_RegexHelper::getInstance()->getHtmlBlockOpenRegex(), $ln, $firstNonSpace) !== null) {
// html block
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_HTML_BLOCK, $lineNumber, $firstNonSpace);
// note, we don't adjust offset because the tag is part of the text
break;
} elseif ($container->getType() === CommonMark_Element_BlockElement::TYPE_PARAGRAPH && $container->getStrings()->count() === 1 && ($match = CommonMark_Util_RegexHelper::matchAll('/^(?:=+|-+) *$/', $ln, $firstNonSpace))) {
// setext header line
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container->setType(CommonMark_Element_BlockElement::TYPE_SETEXT_HEADER);
$container->setExtra('level', $match[0][0] === '=' ? 1 : 2);
$offset = strlen($ln);
} elseif (CommonMark_Util_RegexHelper::matchAt(CommonMark_Util_RegexHelper::getInstance()->getHRuleRegex(), $ln, $firstNonSpace) !== null) {
// hrule
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_HORIZONTAL_RULE, $lineNumber, $firstNonSpace);
$offset = strlen($ln) - 1;
break;
} elseif ($data = $this->parseListMarker($ln, $firstNonSpace)) {
// list item
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
$data['marker_offset'] = $indent;
$offset = $firstNonSpace + $data['padding'];
// add the list if needed
if ($container->getType() !== CommonMark_Element_BlockElement::TYPE_LIST || !$this->listsMatch($container->getExtra('list_data'), $data)) {
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_LIST, $lineNumber, $firstNonSpace);
$container->setExtra('list_data', $data);
}
// add the list item
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_LIST_ITEM, $lineNumber, $firstNonSpace);
$container->setExtra('list_data', $data);
} else {
break;
}
if ($container->acceptsLines()) {
// if it's a line container, it can't contain other containers
break;
}
}
// What remains at the offset is a text line. Add the text to the appropriate container.
$match = CommonMark_Util_RegexHelper::matchAt('/[^ ]/', $ln, $offset);
if ($match === null) {
$firstNonSpace = strlen($ln);
$blank = true;
} else {
$firstNonSpace = $match;
$blank = false;
}
$indent = $firstNonSpace - $offset;
// First check for a lazy paragraph continuation:
if ($this->tip !== $lastMatchedContainer && !$blank && $this->tip->getType() == CommonMark_Element_BlockElement::TYPE_PARAGRAPH && $this->tip->getStrings()->count() > 0) {
// lazy paragraph continuation
$this->lastLineBlank = false;
// TODO: really? (see line 1152)
$this->addLine($ln, $offset);
} else {
// not a lazy continuation
//finalize any blocks not matched
$this->closeUnmatchedBlocks($this, $oldTip, $lastMatchedContainer, $lineNumber, $closeUnmatchedBlocksAlreadyDone);
// Block quote lines are never blank as they start with >
// and we don't count blanks in fenced code for purposes of tight/loose
// lists or breaking out of lists. We also don't set last_line_blank
// on an empty list item.
$container->setIsLastLineBlank($blank && !($container->getType() == CommonMark_Element_BlockElement::TYPE_BLOCK_QUOTE || $container->getType() == CommonMark_Element_BlockElement::TYPE_FENCED_CODE || $container->getType() == CommonMark_Element_BlockElement::TYPE_LIST_ITEM && $container->getChildren()->count() === 0 && $container->getStartLine() == $lineNumber));
$cont = $container;
while ($cont->getParent()) {
$cont->getParent()->setIsLastLineBlank(false);
$cont = $cont->getParent();
}
switch ($container->getType()) {
case CommonMark_Element_BlockElement::TYPE_INDENTED_CODE:
case CommonMark_Element_BlockElement::TYPE_HTML_BLOCK:
$this->addLine($ln, $offset);
break;
case CommonMark_Element_BlockElement::TYPE_FENCED_CODE:
// check for closing code fence
$test = $indent <= 3 && isset($ln[$firstNonSpace]) && $ln[$firstNonSpace] == $container->getExtra('fence_char') && ($match = CommonMark_Util_RegexHelper::matchAll('/^(?:`{3,}|~{3,})(?= *$)/', $ln, $firstNonSpace));
if ($test && strlen($match[0]) >= $container->getExtra('fence_length')) {
// don't add closing fence to container; instead, close it:
$this->finalize($container, $lineNumber);
} else {
$this->addLine($ln, $offset);
}
break;
case CommonMark_Element_BlockElement::TYPE_ATX_HEADER:
case CommonMark_Element_BlockElement::TYPE_SETEXT_HEADER:
case CommonMark_Element_BlockElement::TYPE_HORIZONTAL_RULE:
// nothing to do; we already added the contents.
break;
default:
if ($container->acceptsLines()) {
$this->addLine($ln, $firstNonSpace);
} elseif ($blank) {
// do nothing
} elseif ($container->getType() != CommonMark_Element_BlockElement::TYPE_HORIZONTAL_RULE && $container->getType() != CommonMark_Element_BlockElement::TYPE_SETEXT_HEADER) {
// create paragraph container for line
$container = $this->addChild(CommonMark_Element_BlockElement::TYPE_PARAGRAPH, $lineNumber, $firstNonSpace);
$this->addLine($ln, $firstNonSpace);
} else {
// TODO: throw exception?
}
}
}
}
/**
* Finalize the block; mark it closed for modification
*
* @param int $lineNumber
* @param InlineParser $inlineParser
* @param ReferenceMap $refMap
*/
public function finalize($lineNumber, CommonMark_InlineParser $inlineParser, CommonMark_Reference_ReferenceMap $refMap)
{
if (!$this->open) {
return;
}
$this->open = false;
if ($lineNumber > $this->startLine) {
$this->endLine = $lineNumber - 1;
} else {
$this->endLine = $lineNumber;
}
switch ($this->getType()) {
case self::TYPE_PARAGRAPH:
$this->stringContent = preg_replace('/^ */m', '', implode("\n", $this->strings->toArray()));
// Try parsing the beginning as link reference definitions:
while ($this->stringContent[0] === '[' && ($pos = $inlineParser->parseReference($this->stringContent, $refMap))) {
$this->stringContent = substr($this->stringContent, $pos);
if ($this->isStringContentBlank()) {
//RegexHelper::getInstance()->isBlank($this->stringContent)) {
$this->type = self::TYPE_REFERENCE_DEF;
break;
}
}
break;
case self::TYPE_ATX_HEADER:
case self::TYPE_SETEXT_HEADER:
case self::TYPE_HTML_BLOCK:
$this->stringContent = implode("\n", $this->strings->toArray());
break;
case self::TYPE_INDENTED_CODE:
$reversed = array_reverse($this->strings->toArray(), true);
foreach ($reversed as $index => $line) {
if ($line == '' || $line === "\n" || preg_match('/^(\\n *)$/', $line)) {
unset($reversed[$index]);
} else {
break;
}
}
$fixed = array_reverse($reversed);
$tmp = implode("\n", $fixed);
if (substr($tmp, -1) !== "\n") {
$tmp .= "\n";
}
$this->stringContent = $tmp;
break;
case self::TYPE_FENCED_CODE:
// first line becomes info string
$this->setExtra('info', CommonMark_Util_RegexHelper::unescape(trim($this->strings->first())));
if ($this->strings->count() == 1) {
$this->stringContent = '';
} else {
$this->stringContent = implode("\n", $this->strings->slice(1)) . "\n";
}
break;
case self::TYPE_LIST:
$this->setExtra('tight', true);
// tight by default
$numItems = $this->children->count();
$i = 0;
while ($i < $numItems) {
/** @var BlockElement $item */
$item = $this->children->get($i);
// check for non-final list item ending with blank line:
$lastItem = $i == $numItems - 1;
if ($item->endsWithBlankLine() && !$lastItem) {
$this->setExtra('tight', false);
break;
}
// Recurse into children of list item, to see if there are
// spaces between any of them:
$numSubItems = $item->getChildren()->count();
$j = 0;
while ($j < $numSubItems) {
$subItem = $item->getChildren()->get($j);
$lastSubItem = $j == $numSubItems - 1;
if ($subItem->endsWithBlankLine() && !($lastItem && $lastSubItem)) {
$this->setExtra('tight', false);
break;
}
$j++;
}
$i++;
}
break;
default:
break;
}
}
/**
* Parse a run of ordinary characters, or a single character with
* a special meaning in markdown, as a plain string, adding to inlines.
*
* @param \ColinODell\CommonMark\Util\ArrayCollection $inlines
*
* @return int
*/
protected function parseString(CommonMark_Util_ArrayCollection $inlines)
{
if ($m = $this->match(CommonMark_Util_RegexHelper::getInstance()->getMainRegex())) {
$inlines->add(CommonMark_Element_InlineCreator::createString($m));
return strlen($m);
}
return 0;
}
