/**
* Parse the given content.
*
* Any newly created nodes should be appended to the given target. Any remaining content should be passed to the
* next parser in the chain.
*
* @param Text $content
* @param InlineNodeAcceptorInterface $target
* @return void
*/
public function parseInline(Text $content, InlineNodeAcceptorInterface $target)
{
$content->handle('{
(?<!\\\\)
(
\\<[A-Z][A-Z0-9]* # an opening HTML tag with
(?: # multiple attributes...
\\s+
[A-Z_:][A-Z0-9_:.\\-]*
(?:
\\s*=\\s*
(?:
[^ "\'=<>`]+| # unquoted attribute values
\'[^\']*\'| # single-quoted attribute values
"[^"]*" # double-quoted attribute values
)
)?
)*
\\s*/?\\>| # ...or
\\</[A-Z][A-Z0-9]*\\s*\\>| # a closing HTML tag, or
\\<!--(-(?!-)|[^\\-])*?--\\>| # a HTML comment, or
\\<\\?.*?\\?\\>| # a processing instruction, or
\\<![A-Z]+\\s+[^>]+\\>| # an element type declaration, or
\\<!\\[CDATA\\[.*?\\]\\]\\> # a CDATA section
)
}isx', function (Text $content) use($target) {
$target->addInline(new RawHTML($content));
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('/
(
(?: # We are either at the beginning of the match
^|\\n # or at a newline (for multi-line quotes)
)
(?:
[ ]{0,3}\\>[ ]* # either a blank line
|
[ ]{0,3}\\>[ ]?[`~]{3,}.* # or a code fence
|
[ ]{0,3}\\> # or non-blank lines...
[ ]*[^\\n\\ ][^\\n]*[ ]*
( # ...with lazy continuation
\\n
[ ]{0,3}
[^>\\-+*=\\ \\n][^\\n]*
)*
)
)+
$
/mx', function (Text $content) use($target) {
// Remove block quote marker plus surrounding whitespace on each line
$content->replace('/^[ ]{0,3}\\>[ ]?/m', '');
$blockquote = new Blockquote();
$target->addChild($blockquote);
$this->first->parseBlock($content, $blockquote);
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
protected function parseEmail(Text $content, InlineNodeAcceptorInterface $target)
{
if ($content->contains('@')) {
$content->handle('{
<
(?:mailto:)?
(
[a-zA-Z0-9.!#$%&\'*+/=?^_`{|}~-]+
\\@
[a-zA-Z0-9]
(?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?
(?:\\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*
)
>
}ix', function (Text $w, Text $email) use($target) {
$text = $email->copy();
$link = new Link($email->prepend('mailto:'));
$link->addInline(new String($text));
$target->addInline($link);
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
} else {
$this->next->parseInline($content, $target);
}
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('{
^(?:
([ ]{0,3}) #1 Initial indentation
( #2 Fence
([`~]) #3 The fence character (` or ~)
\\3{2,} # At least two remaining fence characters
)
([^`\\n]*?)? #4 Code language [optional]
(?:
\\n(.*?) #5 Code block
)?
(?:(?<=\\n)([ ]{0,3}\\2\\3*[ ]*)|\\z) # Closing fence or end of document
)$
}msx', function (Text $whole, Text $whitespace, Text $fence, Text $fenceChar, Text $lang, Text $code = null) use($target) {
$leading = $whitespace->getLength();
$code = $code ?: new Text();
$language = new Text(explode(' ', $lang->trim())[0]);
$language->decodeEntities();
// Remove all leading whitespace from content lines
if ($leading > 0) {
$code->replace("/^[ ]{0,{$leading}}/m", '');
}
$target->addChild(new CodeBlock($code, $language));
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('{
(?: # Ensure blank line before (or beginning of subject)
\\A\\s*\\n?|
\\n\\s*\\n
)
[ ]{4} # four leading spaces
.+
(?: # optionally more spaces
(?: # blank lines in between are okay
\\n[ ]*
)*
\\n
[ ]{4}
.+
)*
$
}mx', function (Text $code) use($target) {
// Remove leading blank lines
$code->replace('/^(\\s*\\n)*/', '');
// Remove indent
$code->replace('/^[ ]{1,4}/m', '');
$code->append("\n");
$target->addChild(new CodeBlock($code));
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be appended to the given target. Any remaining content should be passed to the
* next parser in the chain.
*
* @param Text $content
* @param InlineNodeAcceptorInterface $target
* @return void
*/
public function parseInline(Text $content, InlineNodeAcceptorInterface $target)
{
$content->handle('/(\\\\\\n)|( {2,}\\n)/', function () use($target) {
$target->addInline(new HardBreak());
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
}
protected function parseUnderscores(Text $content, InlineNodeAcceptorInterface $target)
{
$content->handle('{ (?<![A-Za-z0-9]) (__) (?![\\s_]) (.+) (?<![\\s_]) \\1 (?![A-Za-z0-9]) }sx', function (Text $w, Text $a, Text $inner) use($target) {
$emphasis = new StrongEmphasis($inner);
$this->context->queue($emphasis->getContent(), $emphasis);
$target->addInline($emphasis);
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('/^\\s*$/m', function (Text $line) use($target) {
$target->addChild(new BlankLine($line));
}, function (Text $part) use($target) {
$paragraph = new Paragraph($part);
$target->addChild($paragraph);
$this->inlineParser->queue($paragraph->getText(), $paragraph);
});
}
/**
* Preprocess the text.
*
* This unifies line endings and replaces tabs by spaces.
*
* @param Text $text
* @return void
*/
protected function prepare(Text $text)
{
// Unify line endings
$text->replaceString("\r\n", "\n");
$text->replaceString("\r", "\n");
// Replace tabs by spaces
$text->replace('/(.*?)\\t/', function (Text $whole, Text $string) {
$tabWidth = 4;
return $string . str_repeat(' ', $tabWidth - $string->getLength() % $tabWidth);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('{
^
[ ]{0,3}\\[(.+)\\]: # id = $1
[ \\t]*
\\n? # maybe *one* newline
[ \\t]*
(?| # url = $2
([^\\s<>]+) # either without spaces
|
<([^<>\\n]*)> # or enclosed in angle brackets
)
(?:
[ \\t]*
\\n? # maybe one newline
[ \\t]*
(?<=\\s) # lookbehind for whitespace
["\'(]
(.+?) # title = $3
["\')]
[ \\t]*
)? # title is optional
$
}xm', function (Text $whole, Text $id, Text $url, Text $title = null) use($target) {
if (!$this->queue->isEmpty()) {
$lastPart = $this->queue->dequeue();
// If the previous line was not empty, we should not parse this as a link reference
if (!$lastPart->match('/(^|\\n *)\\n$/')) {
$lastPart->append($whole);
$this->queue->enqueue($lastPart);
return;
}
$this->parsePart($lastPart, $target);
}
$id = $id->lower()->getString();
// Throw away duplicate reference definitions
if (!$this->links->exists($id)) {
$url->decodeEntities();
// Replace special characters in the URL
$url->encodeUrl();
$this->links->set($id, $url);
if ($title) {
$title->decodeEntities();
$this->titles->set($id, $title);
}
}
}, function (Text $part) {
if (!$this->queue->isEmpty()) {
$lastPart = $this->queue->dequeue();
$part->prepend($lastPart);
}
$this->queue->enqueue($part);
});
if (!$this->queue->isEmpty()) {
$lastPart = $this->queue->dequeue();
$this->parsePart($lastPart, $target);
}
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('{
^[ ]{0,3} # Optional leading spaces
(\\#{1,6}) # $1 = string of #\'s
(([ ].+?)??) # $2 = Header text
([ ]\\#*[ ]*)? # optional closing #\'s (not counted)
$
}mx', function (Text $whole, Text $marks, Text $content) use($target) {
$level = $marks->getLength();
$heading = new Heading($content->trim(), $level);
$target->addChild($heading);
$this->inlineParser->queue($heading->getText(), $heading);
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
protected function parseCodeSpan(Text $content, InlineNodeAcceptorInterface $target)
{
$content->handle('{
(?<![`\\\\])
(`+) # $1 = Opening run of `
(?!`)
(.+?) # $2 = The code block
(?<!`)
\\1 # Matching closer
(?!`)
}sx', function (Text $whole, Text $b, Text $code) use($target) {
// Replace multiple whitespace characters in a row with a single space
$code->trim()->replaceString("\n", ' ')->replace('/[\\s]{2,}/', ' ');
$target->addInline(new Code($code->escapeHtml()));
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$content->handle('{
^
[ ]{0,3}
([^>\\-*=\\ \\n].*)
[ ]*
\\n
[ ]{0,3}
(=+|-+)
[ ]*
\\n*
$
}mx', function (Text $whole, Text $content, Text $mark) use($target) {
$level = substr($mark, 0, 1) == '=' ? 1 : 2;
$heading = new Heading($content->trim(), $level);
$target->addChild($heading);
$this->inlineParser->queue($heading->getText(), $heading);
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
/**
* Parse the given content.
*
* Any newly created nodes should be pushed to the stack. Any remaining content should be passed to the next parser
* in the chain.
*
* @param Text $content
* @param Container $target
* @return void
*/
public function parseBlock(Text $content, Container $target)
{
$tags = implode('|', $this->getValidTags());
$content->handle('{
^ # starts at the beginning or with a newline
[ ]{0,3} # up to 3 leading spaces allowed
(?: # start with one of the following...
\\<(?:' . $tags . ')\\s*/?\\>?| # an opening HTML tag, or
\\</(?:' . $tags . ')\\s*\\>| # a closing HTML tag, or
\\<!--(-(?!-)|[^\\-])*?--\\>| # a HTML comment, or
\\<\\?.*?\\?\\>| # a processing instruction, or
\\<![A-Z]+\\s+[^>]+\\>| # an element type declaration, or
\\<!\\[CDATA\\[.*?\\]\\]\\> # a CDATA section
)
.*? # match everything until...
((?=\\n[ ]*\\n)|\\Z) # we encounter an empty line or the end
}imsx', function (Text $content) use($target) {
$block = new HTMLBlock($content);
$target->addChild($block);
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
protected function parseReferenceLink(Text $content, InlineNodeAcceptorInterface $target)
{
$references = implode('|', array_map(function ($reference) {
return str_replace(' ', '[ ]', preg_quote($reference));
}, $this->context->getReferences()));
$content->handle('/
(?<!\\\\)
(?|
()
\\[
(' . $references . ')
\\]
[\\n\\ ]*
\\[\\]
|
(?:
\\[
(' . $this->getNestedBrackets() . ') # link text = $1
\\]
[ \\t\\n]*
)?
\\[
(' . $references . ') # label = $2
\\]
)
/iux', function (Text $whole, Text $linkText, Text $label) use($target) {
$reference = $label->copy()->lower();
$url = $this->context->getReferenceUrl($reference);
$title = $this->context->getReferenceTitle($reference);
if ($linkText->isEmpty()) {
$linkText = $label;
}
$link = new Link($url, $linkText, $title);
$target->addInline($link);
$this->context->queue($link->getContent(), $link);
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
}
protected function parseReferenceImage(Text $content, InlineNodeAcceptorInterface $target)
{
$references = implode('|', array_map(function ($reference) {
return str_replace(' ', '[ ]', preg_quote($reference));
}, $this->context->getReferences()));
$content->handle('{
(?<!\\\\)!
(?|
()
\\[
(' . $references . ')
\\]
[\\n\\ ]*
\\[\\]
|
(?:
\\[
(.*?) # alt text = $1
\\]
[ \\t\\n]*
)?
\\[
(' . $references . ')
\\]
)
}iux', function (Text $whole, Text $alt, Text $label) use($target) {
$reference = $label->copy()->lower();
$url = $this->context->getReferenceUrl($reference);
$title = $this->context->getReferenceTitle($reference);
if ($alt->isEmpty()) {
$alt = $label;
}
$image = new Image($url, $alt, $title);
$target->addInline($image);
}, function (Text $part) use($target) {
$this->next->parseInline($part, $target);
});
}
protected function parseOrderedLists(Text $content, Container $target)
{
$content->handle('{
^
([ ]{0,3}) # $1 - initial indent
([0-9]+)([.)]) # $2 - list marker; $3 - punctuation
(?|
([ ]{1,4}) # $4 - list indent
[^ ].*
|
() # ... which can also be empty
)
(
\\n\\n?
\\1[ ]{2}\\4
.*
)*
(
(?:
\\n
\\1[0-9]+\\3\\4
[^ ].*
| # empty items
\\n
\\1\\2
| # Lazy continuation lines
\\n
[ ]*
(?:
[^0-9>\\-+*=\\ \\n]
|
[0-9]+[^.)]
)
[^\\n]*
)
(
\\n\\n?
\\1[ ]{2}\\4
.*
)*
)*
$
}mx', function (Text $content, Text $i, Text $start, Text $punctuation, Text $indent) use($target) {
$isTerse = !$content->match('/^[ ]*$/m');
$lines = explode("\n", $content->getString());
$start = $start->getString();
$indentLength = $i->getLength() + $indent->getLength() + 2;
$list = new ListBlock('ol', $isTerse, $start);
// Go through all the lines to assemble the list items
$curItem = substr(array_shift($lines), $indentLength) . "\n";
foreach ($lines as $line) {
if (preg_match('/^[0-9]+' . preg_quote($punctuation) . '/', $line)) {
$this->addItemToList($curItem, $list);
$curItem = substr($line, $indentLength) . "\n";
} else {
$curItem .= $this->unindentLine($line, $indentLength) . "\n";
}
}
$this->addItemToList($curItem, $list);
$target->addChild($list);
}, function (Text $part) use($target) {
$this->next->parseBlock($part, $target);
});
}
public function handleBackslashes(Text $content)
{
$content->replaceString(['\\\\', '\\!', '\\"', '\\#', '\\$', '\\%', '\\&', '\\\'', '\\(', '\\)', '\\*', '\\+', '\\,', '\\-', '\\.', '\\/', '\\:', '\\;', '\\<', '\\=', '\\>', '\\?', '\\@', '\\[', '\\]', '\\^', '\\_', '\\`', '\\{', '\\|', '\\}', '\\~'], ['\\', '!', '"', '#', '$', '%', '&', '\'', '(', ')', '*', '+', ',', '-', '.', '/', ':', ';', '<', '=', '>', '?', '@', '[', ']', '^', '_', '`', '{', '|', '}', '~']);
}
public function visitHardBreak(HardBreak $softBreak)
{
$this->buffer->append(Tag::inline('br'))->append("\n");
}
public function hasLanguage()
{
return !$this->language->isEmpty();
}
/**
* @return Text
*/
public function getContent()
{
return $this->content->copy();
}
protected function handle(Text $content, $mark, Container $target, callable $next)
{
$content->handle($this->getPattern($mark), function () use($target) {
$target->addChild(new HorizontalRule());
}, $next);
}
protected function makeLines(Text $text)
{
return $text->trim()->split('/\\n/')->apply(function (Text $line) {
return $line->copy()->ltrim();
});
}
