MarkdownParser::processInlines() - Code Metrics - Inspection of "Fix ability to register block parsers that identif..." - thephpleague/commonmark - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — 2.0 ( 040f68...c13575 )

by Colin

created 2021-08-14 14:04 UTC

MarkdownParser::processInlines() A

↳ Parent: MarkdownParser

Complexity

Conditions	2
Paths	2

Size

Total Lines	6
Code Lines	3

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	4
CRAP Score	2

Importance

Changes	3
Bugs	0	Features	0

Metric	Value
eloc	3
c	3
b	0
f	0
dl	0
loc	6
ccs	4
cts	4
cp	1
rs	10
cc	2
nc	2
nop	0
crap	2

<?php

declare(strict_types=1);

/*
 * This file is part of the league/commonmark package.
 *
 * (c) Colin O'Dell <[email protected]>
 *
 * Original code based on the CommonMark JS reference parser (https://bitly.com/commonmark-js)
 *  - (c) John MacFarlane
 *
 * Additional code based on commonmark-java (https://github.com/commonmark/commonmark-java)
 *  - (c) Atlassian Pty Ltd
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 */

namespace League\CommonMark\Parser;

use League\CommonMark\Environment\EnvironmentInterface;
use League\CommonMark\Event\DocumentParsedEvent;
use League\CommonMark\Event\DocumentPreParsedEvent;
use League\CommonMark\Input\MarkdownInput;
use League\CommonMark\Node\Block\Document;
use League\CommonMark\Node\Block\Paragraph;
use League\CommonMark\Parser\Block\BlockContinueParserInterface;
use League\CommonMark\Parser\Block\BlockContinueParserWithInlinesInterface;
use League\CommonMark\Parser\Block\BlockStart;
use League\CommonMark\Parser\Block\BlockStartParserInterface;
use League\CommonMark\Parser\Block\DocumentBlockParser;
use League\CommonMark\Parser\Block\ParagraphParser;
use League\CommonMark\Reference\ReferenceInterface;
use League\CommonMark\Reference\ReferenceMap;

final class MarkdownParser implements MarkdownParserInterface
{
    /** @psalm-readonly */
    private EnvironmentInterface $environment;

    /** @psalm-readonly-allow-private-mutation */
    private int $maxNestingLevel;

    /** @psalm-readonly-allow-private-mutation */
    private ReferenceMap $referenceMap;

    /** @psalm-readonly-allow-private-mutation */
    private int $lineNumber = 0;

    /** @psalm-readonly-allow-private-mutation */
    private Cursor $cursor;

    /**
     * @var array<int, BlockContinueParserInterface>
     *
     * @psalm-readonly-allow-private-mutation
     */
    private array $activeBlockParsers = [];

    /**
     * @var array<int, BlockContinueParserWithInlinesInterface>
     *
     * @psalm-readonly-allow-private-mutation
     */
    private array $closedBlockParsers = [];

    public function __construct(EnvironmentInterface $environment)
    {
        $this->environment = $environment;
    }

    private function initialize(): void
    {
        $this->referenceMap       = new ReferenceMap();
        $this->lineNumber         = 0;
        $this->activeBlockParsers = [];
        $this->closedBlockParsers = [];

        $this->maxNestingLevel = $this->environment->getConfiguration()->get('max_nesting_level');
    }

    /**
     * @throws \RuntimeException
     */
    public function parse(string $input): Document
    {
        $this->initialize();

        $documentParser = new DocumentBlockParser($this->referenceMap);
        $this->activateBlockParser($documentParser);

        $preParsedEvent = new DocumentPreParsedEvent($documentParser->getBlock(), new MarkdownInput($input));
        $this->environment->dispatch($preParsedEvent);
        $markdownInput = $preParsedEvent->getMarkdown();

        foreach ($markdownInput->getLines() as $lineNumber => $line) {
            $this->lineNumber = $lineNumber;
            $this->parseLine($line);
        }

        // finalizeAndProcess
        $this->closeBlockParsers(\count($this->activeBlockParsers), $this->lineNumber);
        $this->processInlines();

        $this->environment->dispatch(new DocumentParsedEvent($documentParser->getBlock()));

        return $documentParser->getBlock();
    }

    /**
     * Analyze a line of text and update the document appropriately. We parse markdown text by calling this on each
     * line of input, then finalizing the document.
     */
    private function parseLine(string $line): void
    {
        $this->cursor = new Cursor($line);

        $matches = $this->parseBlockContinuation();
        if ($matches === null) {
            return;
        }

        $unmatchedBlocks = \count($this->activeBlockParsers) - $matches;
        $blockParser     = $this->activeBlockParsers[$matches - 1];
        $startedNewBlock = false;

        // Unless last matched container is a code block, try new container starts,
        // adding children to the last matched container:
        $tryBlockStarts = $blockParser->getBlock() instanceof Paragraph || $blockParser->isContainer();
        while ($tryBlockStarts) {
            // this is a little performance optimization
            if ($this->cursor->isBlank()) {
                $this->cursor->advanceToEnd();
                break;
            }

            if ($blockParser->getBlock()->getDepth() >= $this->maxNestingLevel) {
                break;
            }

            $blockStart = $this->findBlockStart($blockParser);
            if ($blockStart === null || $blockStart->isAborting()) {
                $this->cursor->advanceToNextNonSpaceOrTab();
                break;
            }

            if (($state = $blockStart->getCursorState()) !== null) {
                $this->cursor->restoreState($state);
            }

            $startedNewBlock = true;

            // We're starting a new block. If we have any previous blocks that need to be closed, we need to do it now.
            if ($unmatchedBlocks > 0) {
                $this->closeBlockParsers($unmatchedBlocks, $this->lineNumber - 1);
                $unmatchedBlocks = 0;
            }

            if ($blockStart->isReplaceActiveBlockParser()) {
                $this->prepareActiveBlockParserForReplacement();
            }

            foreach ($blockStart->getBlockParsers() as $newBlockParser) {
                $blockParser    = $this->addChild($newBlockParser);
                $tryBlockStarts = $newBlockParser->isContainer();
            }
        }

        // What remains at the offset is a text line. Add the text to the appropriate block.

        // First check for a lazy paragraph continuation:
        if (! $startedNewBlock && ! $this->cursor->isBlank() && $this->getActiveBlockParser()->canHaveLazyContinuationLines()) {
            $this->getActiveBlockParser()->addLine($this->cursor->getRemainder());
        } else {
            // finalize any blocks not matched
            if ($unmatchedBlocks > 0) {
                $this->closeBlockParsers($unmatchedBlocks, $this->lineNumber);
            }

            if (! $blockParser->isContainer()) {
                $this->getActiveBlockParser()->addLine($this->cursor->getRemainder());
            } elseif (! $this->cursor->isBlank()) {
                $this->addChild(new ParagraphParser());
                $this->getActiveBlockParser()->addLine($this->cursor->getRemainder());
            }
        }
    }

    private function parseBlockContinuation(): ?int
    {
        // For each containing block, try to parse the associated line start.
        // The document will always match, so we can skip the first block parser and start at 1 matches
        $matches = 1;
        for ($i = 1; $i < \count($this->activeBlockParsers); $i++) {
            $blockParser   = $this->activeBlockParsers[$i];
            $blockContinue = $blockParser->tryContinue(clone $this->cursor, $this->getActiveBlockParser());
            if ($blockContinue === null) {
                break;
            }

            if ($blockContinue->isFinalize()) {
                $this->closeBlockParsers(\count($this->activeBlockParsers) - $i, $this->lineNumber);

                return null;
            }

            if (($state = $blockContinue->getCursorState()) !== null) {
                $this->cursor->restoreState($state);
            }

            $matches++;
        }

        return $matches;
    }

    private function findBlockStart(BlockContinueParserInterface $lastMatchedBlockParser): ?BlockStart
    {
        $matchedBlockParser = new MarkdownParserState($this->getActiveBlockParser(), $lastMatchedBlockParser);

        foreach ($this->environment->getBlockStartParsers() as $blockStartParser) {
            \assert($blockStartParser instanceof BlockStartParserInterface);
            if (($result = $blockStartParser->tryStart(clone $this->cursor, $matchedBlockParser)) !== null) {
                return $result;
            }
        }

        return null;
    }

    private function closeBlockParsers(int $count, int $endLineNumber): void
    {
        for ($i = 0; $i < $count; $i++) {
            $blockParser = $this->deactivateBlockParser();
            $this->finalize($blockParser, $endLineNumber);

            // phpcs:disable SlevomatCodingStandard.ControlStructures.EarlyExit.EarlyExitNotUsed
            if ($blockParser instanceof BlockContinueParserWithInlinesInterface) {
                // Remember for inline parsing
                $this->closedBlockParsers[] = $blockParser;
            }
        }
    }

    /**
     * Finalize a block. Close it and do any necessary postprocessing, e.g. creating string_content from strings,
     * setting the 'tight' or 'loose' status of a list, and parsing the beginnings of paragraphs for reference
     * definitions.
     */
    private function finalize(BlockContinueParserInterface $blockParser, int $endLineNumber): void
    {
        if ($blockParser instanceof ParagraphParser) {
            $this->updateReferenceMap($blockParser->getReferences());
        }

        $blockParser->getBlock()->setEndLine($endLineNumber);
        $blockParser->closeBlock();
    }

    /**
     * Walk through a block & children recursively, parsing string content into inline content where appropriate.
     */
    private function processInlines(): void
    {
        $p = new InlineParserEngine($this->environment, $this->referenceMap);

        foreach ($this->closedBlockParsers as $blockParser) {
            $blockParser->parseInlines($p);
        }
    }

    /**
     * Add block of type tag as a child of the tip. If the tip can't accept children, close and finalize it and try
     * its parent, and so on til we find a block that can accept children.
     */
    private function addChild(BlockContinueParserInterface $blockParser): BlockContinueParserInterface
    {
        $blockParser->getBlock()->setStartLine($this->lineNumber);

        while (! $this->getActiveBlockParser()->canContain($blockParser->getBlock())) {
            $this->closeBlockParsers(1, $this->lineNumber - 1);
        }

        $this->getActiveBlockParser()->getBlock()->appendChild($blockParser->getBlock());
        $this->activateBlockParser($blockParser);

        return $blockParser;
    }

    private function activateBlockParser(BlockContinueParserInterface $blockParser): void
    {
        $this->activeBlockParsers[] = $blockParser;
    }

    private function deactivateBlockParser(): BlockContinueParserInterface
    {
        $popped = \array_pop($this->activeBlockParsers);
        if ($popped === null) {
            throw new \RuntimeException('The last block parser should not be deactivated');
        }

        return $popped;
    }

    private function prepareActiveBlockParserForReplacement(): void
    {
        // Note that we don't want to parse inlines or finalize this block, as it's getting replaced.
        $old = $this->deactivateBlockParser();

        if ($old instanceof ParagraphParser) {
            $this->updateReferenceMap($old->getReferences());
        }

        $old->getBlock()->detach();
    }

    /**
     * @param ReferenceInterface[] $references
     */
    private function updateReferenceMap(iterable $references): void
    {
        foreach ($references as $reference) {
            if (! $this->referenceMap->contains($reference->getLabel())) {
                $this->referenceMap->add($reference);
            }
        }
    }

    public function getActiveBlockParser(): BlockContinueParserInterface
    {
        $active = \end($this->activeBlockParsers);
        if ($active === false) {
            throw new \RuntimeException('No active block parsers are available');
        }

        return $active;
    }
}


1		<?php
2
3		declare(strict_types=1);
4
5		/*
6		* This file is part of the league/commonmark package.
7		*
8		* (c) Colin O'Dell <[email protected]>
9		*
10		* Original code based on the CommonMark JS reference parser (https://bitly.com/commonmark-js)
11		* - (c) John MacFarlane
12		*
13		* Additional code based on commonmark-java (https://github.com/commonmark/commonmark-java)
14		* - (c) Atlassian Pty Ltd
15		*
16		* For the full copyright and license information, please view the LICENSE
17		* file that was distributed with this source code.
18		*/
19
20		namespace League\CommonMark\Parser;
21
22		use League\CommonMark\Environment\EnvironmentInterface;
23		use League\CommonMark\Event\DocumentParsedEvent;
24		use League\CommonMark\Event\DocumentPreParsedEvent;
25		use League\CommonMark\Input\MarkdownInput;
26		use League\CommonMark\Node\Block\Document;
27		use League\CommonMark\Node\Block\Paragraph;
28		use League\CommonMark\Parser\Block\BlockContinueParserInterface;
29		use League\CommonMark\Parser\Block\BlockContinueParserWithInlinesInterface;
30		use League\CommonMark\Parser\Block\BlockStart;
31		use League\CommonMark\Parser\Block\BlockStartParserInterface;
32		use League\CommonMark\Parser\Block\DocumentBlockParser;
33		use League\CommonMark\Parser\Block\ParagraphParser;
34		use League\CommonMark\Reference\ReferenceInterface;
35		use League\CommonMark\Reference\ReferenceMap;
36
37		final class MarkdownParser implements MarkdownParserInterface
38		{
39		/** @psalm-readonly */
40		private EnvironmentInterface $environment;
41
42		/** @psalm-readonly-allow-private-mutation */
43		private int $maxNestingLevel;
44
45		/** @psalm-readonly-allow-private-mutation */
46		private ReferenceMap $referenceMap;
47
48		/** @psalm-readonly-allow-private-mutation */
49		private int $lineNumber = 0;
50
51		/** @psalm-readonly-allow-private-mutation */
52		private Cursor $cursor;
53
54		/**
55		* @var array<int, BlockContinueParserInterface>
56		*
57		* @psalm-readonly-allow-private-mutation
58		*/
59		private array $activeBlockParsers = [];
60
61		/**
62		* @var array<int, BlockContinueParserWithInlinesInterface>
63		*
64		* @psalm-readonly-allow-private-mutation
65		*/
66		private array $closedBlockParsers = [];
67
68	2166	public function __construct(EnvironmentInterface $environment)
69		{
70	2166	$this->environment = $environment;
71	2166	}
72
73	2152	private function initialize(): void
74		{
75	2152	$this->referenceMap = new ReferenceMap();
76	2152	$this->lineNumber = 0;
77	2152	$this->activeBlockParsers = [];
78	2152	$this->closedBlockParsers = [];
79
80	2152	$this->maxNestingLevel = $this->environment->getConfiguration()->get('max_nesting_level');
81	2144	}
82
83		/**
84		* @throws \RuntimeException
85		*/
86	2152	public function parse(string $input): Document
87		{
88	2152	$this->initialize();
89
90	2144	$documentParser = new DocumentBlockParser($this->referenceMap);
91	2144	$this->activateBlockParser($documentParser);
92
93	2144	$preParsedEvent = new DocumentPreParsedEvent($documentParser->getBlock(), new MarkdownInput($input));
94	2138	$this->environment->dispatch($preParsedEvent);
95	2136	$markdownInput = $preParsedEvent->getMarkdown();
96
97	2136	foreach ($markdownInput->getLines() as $lineNumber => $line) {
98	2136	$this->lineNumber = $lineNumber;
99	2136	$this->parseLine($line);
100		}
101
102		// finalizeAndProcess
103	2136	$this->closeBlockParsers(\count($this->activeBlockParsers), $this->lineNumber);
104	2136	$this->processInlines();
105
106	2136	$this->environment->dispatch(new DocumentParsedEvent($documentParser->getBlock()));
107
108	2136	return $documentParser->getBlock();
109		}
110
111		/**
112		* Analyze a line of text and update the document appropriately. We parse markdown text by calling this on each
113		* line of input, then finalizing the document.
114		*/
115	2136	private function parseLine(string $line): void
116		{
117	2136	$this->cursor = new Cursor($line);
118
119	2136	$matches = $this->parseBlockContinuation();
120	2136	if ($matches === null) {
121	64	return;
122		}
123
124	2136	$unmatchedBlocks = \count($this->activeBlockParsers) - $matches;
125	2136	$blockParser = $this->activeBlockParsers[$matches - 1];
126	2136	$startedNewBlock = false;
127
128		// Unless last matched container is a code block, try new container starts,
129		// adding children to the last matched container:
130	2136	$tryBlockStarts = $blockParser->getBlock() instanceof Paragraph \|\| $blockParser->isContainer();
131	2136	while ($tryBlockStarts) {
132		// this is a little performance optimization
133	2136	if ($this->cursor->isBlank()) {
134	596	$this->cursor->advanceToEnd();
135	596	break;
136		}
137
138	2136	if ($blockParser->getBlock()->getDepth() >= $this->maxNestingLevel) {
139	4	break;
140		}
141
142	2136	$blockStart = $this->findBlockStart($blockParser);
143	2136	if ($blockStart === null \|\| $blockStart->isAborting()) {
144	1858	$this->cursor->advanceToNextNonSpaceOrTab();
145	1858	break;
146		}
147
148	824	if (($state = $blockStart->getCursorState()) !== null) {
149	824	$this->cursor->restoreState($state);
150		}
151
152	824	$startedNewBlock = true;
153
154		// We're starting a new block. If we have any previous blocks that need to be closed, we need to do it now.
155	824	if ($unmatchedBlocks > 0) {
156	204	$this->closeBlockParsers($unmatchedBlocks, $this->lineNumber - 1);
157	204	$unmatchedBlocks = 0;
158		}
159
160	824	if ($blockStart->isReplaceActiveBlockParser()) {
161	130	$this->prepareActiveBlockParserForReplacement();
162		}
163
164	824	foreach ($blockStart->getBlockParsers() as $newBlockParser) {
165	824	$blockParser = $this->addChild($newBlockParser);
166	824	$tryBlockStarts = $newBlockParser->isContainer();
167		}
168		}
169
170		// What remains at the offset is a text line. Add the text to the appropriate block.
171
172		// First check for a lazy paragraph continuation:
173	2136	if (! $startedNewBlock && ! $this->cursor->isBlank() && $this->getActiveBlockParser()->canHaveLazyContinuationLines()) {
174	346	$this->getActiveBlockParser()->addLine($this->cursor->getRemainder());
175		} else {
176		// finalize any blocks not matched
177	2136	if ($unmatchedBlocks > 0) {
178	620	$this->closeBlockParsers($unmatchedBlocks, $this->lineNumber);
179		}
180
181	2136	if (! $blockParser->isContainer()) {
182	586	$this->getActiveBlockParser()->addLine($this->cursor->getRemainder());
183	1904	} elseif (! $this->cursor->isBlank()) {
184	1862	$this->addChild(new ParagraphParser());
185	1862	$this->getActiveBlockParser()->addLine($this->cursor->getRemainder());
186		}
187		}
188	2136	}
189
190	2136	private function parseBlockContinuation(): ?int
191		{
192		// For each containing block, try to parse the associated line start.
193		// The document will always match, so we can skip the first block parser and start at 1 matches
194	2136	$matches = 1;
195	2136	for ($i = 1; $i < \count($this->activeBlockParsers); $i++) {
196	1086	$blockParser = $this->activeBlockParsers[$i];
197	1086	$blockContinue = $blockParser->tryContinue(clone $this->cursor, $this->getActiveBlockParser());
198	1086	if ($blockContinue === null) {
199	742	break;
200		}
201
202	714	if ($blockContinue->isFinalize()) {
203	64	$this->closeBlockParsers(\count($this->activeBlockParsers) - $i, $this->lineNumber);
204
205	64	return null;
206		}
207
208	710	if (($state = $blockContinue->getCursorState()) !== null) {
209	710	$this->cursor->restoreState($state);
210		}
211
212	710	$matches++;
213		}
214
215	2136	return $matches;
216		}
217
218	2136	private function findBlockStart(BlockContinueParserInterface $lastMatchedBlockParser): ?BlockStart
219		{
220	2136	$matchedBlockParser = new MarkdownParserState($this->getActiveBlockParser(), $lastMatchedBlockParser);
221
222	2136	foreach ($this->environment->getBlockStartParsers() as $blockStartParser) {
223		\assert($blockStartParser instanceof BlockStartParserInterface);
224	2136	if (($result = $blockStartParser->tryStart(clone $this->cursor, $matchedBlockParser)) !== null) {
225	1394	return $result;
226		}
227		}
228
229	980	return null;
230		}
231
232	2136	private function closeBlockParsers(int $count, int $endLineNumber): void
233		{
234	2136	for ($i = 0; $i < $count; $i++) {
235	2136	$blockParser = $this->deactivateBlockParser();
236	2136	$this->finalize($blockParser, $endLineNumber);
237
238		// phpcs:disable SlevomatCodingStandard.ControlStructures.EarlyExit.EarlyExitNotUsed
239	2136	if ($blockParser instanceof BlockContinueParserWithInlinesInterface) {
240		// Remember for inline parsing
241	1932	$this->closedBlockParsers[] = $blockParser;
242		}
243		}
244	2136	}
245
246		/**
247		* Finalize a block. Close it and do any necessary postprocessing, e.g. creating string_content from strings,
248		* setting the 'tight' or 'loose' status of a list, and parsing the beginnings of paragraphs for reference
249		* definitions.
250		*/
251	2136	private function finalize(BlockContinueParserInterface $blockParser, int $endLineNumber): void
252		{
253	2136	if ($blockParser instanceof ParagraphParser) {
254	1780	$this->updateReferenceMap($blockParser->getReferences());
255		}
256
257	2136	$blockParser->getBlock()->setEndLine($endLineNumber);
258	2136	$blockParser->closeBlock();
259	2136	}
260
261		/**
262		* Walk through a block & children recursively, parsing string content into inline content where appropriate.
263		*/
264	2136	private function processInlines(): void
265		{
266	2136	$p = new InlineParserEngine($this->environment, $this->referenceMap);
267
268	2136	foreach ($this->closedBlockParsers as $blockParser) {
269	1932	$blockParser->parseInlines($p);
270		}
271	2136	}
272
273		/**
274		* Add block of type tag as a child of the tip. If the tip can't accept children, close and finalize it and try
275		* its parent, and so on til we find a block that can accept children.
276		*/
277	2136	private function addChild(BlockContinueParserInterface $blockParser): BlockContinueParserInterface
278		{
279	2136	$blockParser->getBlock()->setStartLine($this->lineNumber);
280
281	2136	while (! $this->getActiveBlockParser()->canContain($blockParser->getBlock())) {
282	120	$this->closeBlockParsers(1, $this->lineNumber - 1);
283		}
284
285	2136	$this->getActiveBlockParser()->getBlock()->appendChild($blockParser->getBlock());
286	2136	$this->activateBlockParser($blockParser);
287
288	2136	return $blockParser;
289		}
290
291	2144	private function activateBlockParser(BlockContinueParserInterface $blockParser): void
292		{
293	2144	$this->activeBlockParsers[] = $blockParser;
294	2144	}
295
296	2136	private function deactivateBlockParser(): BlockContinueParserInterface
297		{
298	2136	$popped = \array_pop($this->activeBlockParsers);
299	2136	if ($popped === null) {
300		throw new \RuntimeException('The last block parser should not be deactivated');
301		}
302
303	2136	return $popped;
304		}
305
306	130	private function prepareActiveBlockParserForReplacement(): void
307		{
308		// Note that we don't want to parse inlines or finalize this block, as it's getting replaced.
309	130	$old = $this->deactivateBlockParser();
310
311	130	if ($old instanceof ParagraphParser) {
312	130	$this->updateReferenceMap($old->getReferences());
313		}
314
315	130	$old->getBlock()->detach();
316	130	}
317
318		/**
319		* @param ReferenceInterface[] $references
320		*/
321	1862	private function updateReferenceMap(iterable $references): void
322		{
323	1862	foreach ($references as $reference) {
324	162	if (! $this->referenceMap->contains($reference->getLabel())) {
325	162	$this->referenceMap->add($reference);
326		}
327		}
328	1862	}
329
330	2136	public function getActiveBlockParser(): BlockContinueParserInterface
331		{
332	2136	$active = \end($this->activeBlockParsers);
333	2136	if ($active === false) {
334		throw new \RuntimeException('No active block parsers are available');
335		}
336
337	2136	return $active;
338		}
339		}
340

thephpleague / commonmark

Push — 2.0 ( 040f68...c13575 )

MarkdownParser::processInlines() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like