Lexer::getNextToken() - Code Metrics - Behat/Gherkin - Measure and Improve Code Quality continuously with Scrutinizer

Lexer::getNextToken() F
last analyzed 2021-01-14 15:16 UTC

↳ Parent: Lexer

Complexity

Conditions	16
Paths	> 20000

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	17
CRAP Score	16

Importance

Changes

Metric	Value
dl	0
loc	19
ccs	17
cts	17
cp	1
rs	1.4
c	0
b	0
f	0
cc	16
nc	32768
nop	0
crap	16

How to fix Complexity

<?php

/*
 * This file is part of the Behat Gherkin.
 * (c) Konstantin Kudryashov <[email protected]>
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 */

namespace Behat\Gherkin;

use Behat\Gherkin\Exception\LexerException;
use Behat\Gherkin\Keywords\KeywordsInterface;

/**
 * Gherkin lexer.
 *
 * @author Konstantin Kudryashov <[email protected]>
 */
class Lexer
{
    private $language;
    private $lines;
    private $linesCount;
    private $line;
    private $trimmedLine;
    private $lineNumber;
    private $eos;
    private $keywords;
    private $keywordsCache = array();
    private $stepKeywordTypesCache = array();
    private $deferredObjects = array();
    private $deferredObjectsCount = 0;
    private $stashedToken;
    private $inPyString = false;
    private $pyStringSwallow = 0;
    private $featureStarted = false;
    private $allowMultilineArguments = false;
    private $allowSteps = false;

    /**
     * Initializes lexer.
     *
     * @param KeywordsInterface $keywords Keywords holder
     */
    public function __construct(KeywordsInterface $keywords)
    {
        $this->keywords = $keywords;
    }

    /**
     * Sets lexer input.
     *
     * @param string $input    Input string
     * @param string $language Language name
     *
     * @throws Exception\LexerException
     */
    public function analyse($input, $language = 'en')
    {
        // try to detect unsupported encoding
        if ('UTF-8' !== mb_detect_encoding($input, 'UTF-8', true)) {
            throw new LexerException('Feature file is not in UTF8 encoding');
        }

        $input = strtr($input, array("\r\n" => "\n", "\r" => "\n"));

        $this->lines = explode("\n", $input);
        $this->linesCount = count($this->lines);
        $this->line = $this->lines[0];
        $this->lineNumber = 1;
        $this->trimmedLine = null;
        $this->eos = false;

        $this->deferredObjects = array();
        $this->deferredObjectsCount = 0;
        $this->stashedToken = null;
        $this->inPyString = false;
        $this->pyStringSwallow = 0;

        $this->featureStarted = false;
        $this->allowMultilineArguments = false;
        $this->allowSteps = false;

        $this->keywords->setLanguage($this->language = $language);
        $this->keywordsCache = array();
        $this->stepKeywordTypesCache = array();
    }

    /**
     * Returns current lexer language.
     *
     * @return string
     */
    public function getLanguage()
    {
        return $this->language;
    }

    /**
     * Returns next token or previously stashed one.
     *
     * @return array
     */
    public function getAdvancedToken()
    {
        return $this->getStashedToken() ?: $this->getNextToken();
    }

    /**
     * Defers token.
     *
     * @param array $token Token to defer
     */
    public function deferToken(array $token)
    {
        $token['deferred'] = true;
        $this->deferredObjects[] = $token;
        ++$this->deferredObjectsCount;
    }

    /**
     * Predicts for number of tokens.
     *
     * @return array
     */
    public function predictToken()
    {
        if (null === $this->stashedToken) {
            $this->stashedToken = $this->getNextToken();
        }

        return $this->stashedToken;
    }

    /**
     * Constructs token with specified parameters.
     *
     * @param string $type  Token type
     * @param string $value Token value
     *
     * @return array
     */
    public function takeToken($type, $value = null)
    {
        return array(
            'type'     => $type,
            'line'     => $this->lineNumber,
            'value'    => $value ?: null,
            'deferred' => false
        );
    }

    /**
     * Consumes line from input & increments line counter.
     */
    protected function consumeLine()
    {
        ++$this->lineNumber;

        if (($this->lineNumber - 1) === $this->linesCount) {
            $this->eos = true;

            return;
        }

        $this->line = $this->lines[$this->lineNumber - 1];
        $this->trimmedLine = null;
    }

    /**
     * Returns trimmed version of line.
     *
     * @return string
     */
    protected function getTrimmedLine()
    {
        return null !== $this->trimmedLine ? $this->trimmedLine : $this->trimmedLine = trim($this->line);
    }

    /**
     * Returns stashed token or null if hasn't.
     *
     * @return array|null
     */
    protected function getStashedToken()
    {
        $stashedToken = $this->stashedToken;
        $this->stashedToken = null;

        return $stashedToken;
    }

    /**
     * Returns deferred token or null if hasn't.
     *
     * @return array|null
     */
    protected function getDeferredToken()
    {
        if (!$this->deferredObjectsCount) {
            return null;
        }

        --$this->deferredObjectsCount;

        return array_shift($this->deferredObjects);
    }

    /**
     * Returns next token from input.
     *
     * @return array
     */
    protected function getNextToken()
    {
        return $this->getDeferredToken()
            ?: $this->scanEOS()
            ?: $this->scanLanguage()
            ?: $this->scanComment()
            ?: $this->scanPyStringOp()
            ?: $this->scanPyStringContent()
            ?: $this->scanStep()
            ?: $this->scanScenario()
            ?: $this->scanBackground()
            ?: $this->scanOutline()
            ?: $this->scanExamples()
            ?: $this->scanFeature()
            ?: $this->scanTags()
            ?: $this->scanTableRow()
            ?: $this->scanNewline()
            ?: $this->scanText();
    }

    /**
     * Scans for token with specified regex.
     *
     * @param string $regex Regular expression
     * @param string $type  Expected token type
     *
     * @return null|array
     */
    protected function scanInput($regex, $type)
    {
        if (!preg_match($regex, $this->line, $matches)) {
            return null;
        }

        $token = $this->takeToken($type, $matches[1]);
        $this->consumeLine();

        return $token;
    }

    /**
     * Scans for token with specified keywords.
     *
     * @param string $keywords Keywords (splitted with |)
     * @param string $type     Expected token type
     *
     * @return null|array
     */
    protected function scanInputForKeywords($keywords, $type)
    {
        if (!preg_match('/^(\s*)(' . $keywords . '):\s*(.*)/u', $this->line, $matches)) {
            return null;
        }

        $token = $this->takeToken($type, $matches[3]);
        $token['keyword'] = $matches[2];
        $token['indent'] = mb_strlen($matches[1], 'utf8');

        $this->consumeLine();

        // turn off language searching
        if ('Feature' === $type) {
            $this->featureStarted = true;
        }

        // turn off PyString and Table searching
        if ('Feature' === $type || 'Scenario' === $type || 'Outline' === $type) {
            $this->allowMultilineArguments = false;
        } elseif ('Examples' === $type) {
            $this->allowMultilineArguments = true;
        }

        // turn on steps searching
        if ('Scenario' === $type || 'Background' === $type || 'Outline' === $type) {
            $this->allowSteps = true;
        }

        return $token;
    }

    /**
     * Scans EOS from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanEOS()
    {
        if (!$this->eos) {
            return null;
        }

        return $this->takeToken('EOS');
    }

    /**
     * Returns keywords for provided type.
     *
     * @param string $type Keyword type
     *
     * @return string
     */
    protected function getKeywords($type)
    {
        if (!isset($this->keywordsCache[$type])) {
            $getter = 'get' . $type . 'Keywords';
            $keywords = $this->keywords->$getter();

            if ('Step' === $type) {
                $padded = array();
                foreach (explode('|', $keywords) as $keyword) {
                    $padded[] = false !== mb_strpos($keyword, '<', 0, 'utf8')
                        ? preg_quote(mb_substr($keyword, 0, -1, 'utf8'), '/') . '\s*'
                        : preg_quote($keyword, '/') . '\s+';
                }

                $keywords = implode('|', $padded);
            }

            $this->keywordsCache[$type] = $keywords;
        }

        return $this->keywordsCache[$type];
    }

    /**
     * Scans Feature from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanFeature()
    {
        return $this->scanInputForKeywords($this->getKeywords('Feature'), 'Feature');
    }

    /**
     * Scans Background from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanBackground()
    {
        return $this->scanInputForKeywords($this->getKeywords('Background'), 'Background');
    }

    /**
     * Scans Scenario from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanScenario()
    {
        return $this->scanInputForKeywords($this->getKeywords('Scenario'), 'Scenario');
    }

    /**
     * Scans Scenario Outline from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanOutline()
    {
        return $this->scanInputForKeywords($this->getKeywords('Outline'), 'Outline');
    }

    /**
     * Scans Scenario Outline Examples from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanExamples()
    {
        return $this->scanInputForKeywords($this->getKeywords('Examples'), 'Examples');
    }

    /**
     * Scans Step from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanStep()
    {
        if (!$this->allowSteps) {
            return null;
        }

        $keywords = $this->getKeywords('Step');
        if (!preg_match('/^\s*(' . $keywords . ')([^\s].*)/u', $this->line, $matches)) {
            return null;
        }

        $keyword = trim($matches[1]);
        $token = $this->takeToken('Step', $keyword);
        $token['keyword_type'] = $this->getStepKeywordType($keyword);
        $token['text'] = $matches[2];

        $this->consumeLine();
        $this->allowMultilineArguments = true;

        return $token;
    }

    /**
     * Scans PyString from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanPyStringOp()
    {
        if (!$this->allowMultilineArguments) {
            return null;
        }

        if (false === ($pos = mb_strpos($this->line, '"""', 0, 'utf8'))) {
            return null;
        }

        $this->inPyString = !$this->inPyString;
        $token = $this->takeToken('PyStringOp');
        $this->pyStringSwallow = $pos;

        $this->consumeLine();

        return $token;
    }

    /**
     * Scans PyString content.
     *
     * @return null|array
     */
    protected function scanPyStringContent()
    {
        if (!$this->inPyString) {
            return null;
        }

        $token = $this->scanText();
        // swallow trailing spaces
        $token['value'] = preg_replace('/^\s{0,' . $this->pyStringSwallow . '}/u', '', $token['value']);

        return $token;
    }

    /**
     * Scans Table Row from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanTableRow()
    {
        if (!$this->allowMultilineArguments) {
            return null;
        }

        $line = $this->getTrimmedLine();
        if (!isset($line[0]) || '|' !== $line[0] || '|' !== substr($line, -1)) {
            return null;
        }

        $token = $this->takeToken('TableRow');
        $line = mb_substr($line, 1, mb_strlen($line, 'utf8') - 2, 'utf8');
        $columns = array_map(function ($column) {
            return trim(str_replace('\\|', '|', $column));
        }, preg_split('/(?<!\\\)\|/u', $line));
        $token['columns'] = $columns;

        $this->consumeLine();

        return $token;
    }

    /**
     * Scans Tags from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanTags()
    {
        $line = $this->getTrimmedLine();
        if (!isset($line[0]) || '@' !== $line[0]) {
            return null;
        }

        $token = $this->takeToken('Tag');
        $tags = explode('@', mb_substr($line, 1, mb_strlen($line, 'utf8') - 1, 'utf8'));
        $tags = array_map('trim', $tags);
        $token['tags'] = $tags;

        $this->consumeLine();

        return $token;
    }

    /**
     * Scans Language specifier from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanLanguage()
    {
        if ($this->featureStarted) {
            return null;
        }

        if ($this->inPyString) {
            return null;
        }

        if (0 !== mb_strpos(ltrim($this->line), '#', 0, 'utf8')) {
            return null;
        }

        return $this->scanInput('/^\s*\#\s*language:\s*([\w_\-]+)\s*$/', 'Language');
    }

    /**
     * Scans Comment from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanComment()
    {
        if ($this->inPyString) {
            return null;
        }

        $line = $this->getTrimmedLine();
        if (0 !== mb_strpos($line, '#', 0, 'utf8')) {
            return null;
        }

        $token = $this->takeToken('Comment', $line);
        $this->consumeLine();

        return $token;
    }

    /**
     * Scans Newline from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanNewline()
    {
        if ('' !== $this->getTrimmedLine()) {
            return null;
        }

        $token = $this->takeToken('Newline', mb_strlen($this->line, 'utf8'));
        $this->consumeLine();

        return $token;
    }

    /**
     * Scans text from input & returns it if found.
     *
     * @return null|array
     */
    protected function scanText()
    {
        $token = $this->takeToken('Text', $this->line);
        $this->consumeLine();

        return $token;
    }

    /**
     * Returns step type keyword (Given, When, Then, etc.).
     *
     * @param string $native Step keyword in provided language
     * @return string
     */
    private function getStepKeywordType($native)
    {
        // Consider "*" as a AND keyword so that it is normalized to the previous step type
        if ('*' === $native) {
            return 'And';
        }

        if (empty($this->stepKeywordTypesCache)) {
            $this->stepKeywordTypesCache = array(
                'Given' => explode('|', $this->keywords->getGivenKeywords()),
                'When' => explode('|', $this->keywords->getWhenKeywords()),
                'Then' => explode('|', $this->keywords->getThenKeywords()),
                'And' => explode('|', $this->keywords->getAndKeywords()),
                'But' => explode('|', $this->keywords->getButKeywords())
            );
        }

        foreach ($this->stepKeywordTypesCache as $type => $keywords) {
            if (in_array($native, $keywords) || in_array($native . '<', $keywords)) {
                return $type;
            }
        }

        return 'Given';
    }
}


1		<?php
2
3		/*
4		* This file is part of the Behat Gherkin.
5		* (c) Konstantin Kudryashov <[email protected]>
6		*
7		* For the full copyright and license information, please view the LICENSE
8		* file that was distributed with this source code.
9		*/
10
11		namespace Behat\Gherkin;
12
13		use Behat\Gherkin\Exception\LexerException;
14		use Behat\Gherkin\Keywords\KeywordsInterface;
15
16		/**
17		* Gherkin lexer.
18		*
19		* @author Konstantin Kudryashov <[email protected]>
20		*/
21		class Lexer
22		{
23		private $language;
24		private $lines;
25		private $linesCount;
26		private $line;
27		private $trimmedLine;
28		private $lineNumber;
29		private $eos;
30		private $keywords;
31		private $keywordsCache = array();
32		private $stepKeywordTypesCache = array();
33		private $deferredObjects = array();
34		private $deferredObjectsCount = 0;
35		private $stashedToken;
36		private $inPyString = false;
37		private $pyStringSwallow = 0;
38		private $featureStarted = false;
39		private $allowMultilineArguments = false;
40		private $allowSteps = false;
41
42		/**
43		* Initializes lexer.
44		*
45		* @param KeywordsInterface $keywords Keywords holder
46		*/
47	245	public function __construct(KeywordsInterface $keywords)
48		{
49	245	$this->keywords = $keywords;
50	245	}
51
52		/**
53		* Sets lexer input.
54		*
55		* @param string $input Input string
56		* @param string $language Language name
57		*
58		* @throws Exception\LexerException
59		*/
60	243	public function analyse($input, $language = 'en')
61		{
62		// try to detect unsupported encoding
63	243	if ('UTF-8' !== mb_detect_encoding($input, 'UTF-8', true)) {
64		throw new LexerException('Feature file is not in UTF8 encoding');
65		}
66
67	243	$input = strtr($input, array("\r\n" => "\n", "\r" => "\n"));
68
69	243	$this->lines = explode("\n", $input);
70	243	$this->linesCount = count($this->lines);
71	243	$this->line = $this->lines[0];
72	243	$this->lineNumber = 1;
73	243	$this->trimmedLine = null;
74	243	$this->eos = false;
75
76	243	$this->deferredObjects = array();
77	243	$this->deferredObjectsCount = 0;
78	243	$this->stashedToken = null;
79	243	$this->inPyString = false;
80	243	$this->pyStringSwallow = 0;
81
82	243	$this->featureStarted = false;
83	243	$this->allowMultilineArguments = false;
84	243	$this->allowSteps = false;
85
86	243	$this->keywords->setLanguage($this->language = $language);
87	243	$this->keywordsCache = array();
88	243	$this->stepKeywordTypesCache = array();
89	243	}
90
91		/**
92		* Returns current lexer language.
93		*
94		* @return string
95		*/
96	242	public function getLanguage()
97		{
98	242	return $this->language;
99		}
100
101		/**
102		* Returns next token or previously stashed one.
103		*
104		* @return array
105		*/
106	243	public function getAdvancedToken()
107		{
108	243	return $this->getStashedToken() ?: $this->getNextToken();
109		}
110
111		/**
112		* Defers token.
113		*
114		* @param array $token Token to defer
115		*/
116		public function deferToken(array $token)
117		{
118		$token['deferred'] = true;
119		$this->deferredObjects[] = $token;
120		++$this->deferredObjectsCount;
121		}
122
123		/**
124		* Predicts for number of tokens.
125		*
126		* @return array
127		*/
128	243	public function predictToken()
129		{
130	243	if (null === $this->stashedToken) {
131	243	$this->stashedToken = $this->getNextToken();
132		}
133
134	243	return $this->stashedToken;
135		}
136
137		/**
138		* Constructs token with specified parameters.
139		*
140		* @param string $type Token type
141		* @param string $value Token value
142		*
143		* @return array
144		*/
145	243	public function takeToken($type, $value = null)
146		{
147		return array(
148	243	'type' => $type,
149	243	'line' => $this->lineNumber,
150		'value' => $value ?: null,
151		'deferred' => false
152		);
153		}
154
155		/**
156		* Consumes line from input & increments line counter.
157		*/
158	243	protected function consumeLine()
159		{
160	243	++$this->lineNumber;
161
162	243	if (($this->lineNumber - 1) === $this->linesCount) {
163	240	$this->eos = true;
164
165	240	return;
166		}
167
168	243	$this->line = $this->lines[$this->lineNumber - 1];
169	243	$this->trimmedLine = null;
170	243	}
171
172		/**
173		* Returns trimmed version of line.
174		*
175		* @return string
176		*/
177	243	protected function getTrimmedLine()
178		{
179	243	return null !== $this->trimmedLine ? $this->trimmedLine : $this->trimmedLine = trim($this->line);
180		}
181
182		/**
183		* Returns stashed token or null if hasn't.
184		*
185		* @return array\|null
186		*/
187	243	protected function getStashedToken()
188		{
189	243	$stashedToken = $this->stashedToken;
190	243	$this->stashedToken = null;
191
192	243	return $stashedToken;
193		}
194
195		/**
196		* Returns deferred token or null if hasn't.
197		*
198		* @return array\|null
199		*/
200	243	protected function getDeferredToken()
201		{
202	243	if (!$this->deferredObjectsCount) {
203	243	return null;
204		}
205
206		--$this->deferredObjectsCount;
207
208		return array_shift($this->deferredObjects);
209		}
210
211		/**
212		* Returns next token from input.
213		*
214		* @return array
215		*/
216	243	protected function getNextToken()
217		{
218	243	return $this->getDeferredToken()
219	243	?: $this->scanEOS()
220	243	?: $this->scanLanguage()
221	243	?: $this->scanComment()
222	243	?: $this->scanPyStringOp()
223	243	?: $this->scanPyStringContent()
224	243	?: $this->scanStep()
225	243	?: $this->scanScenario()
226	243	?: $this->scanBackground()
227	243	?: $this->scanOutline()
228	243	?: $this->scanExamples()
229	243	?: $this->scanFeature()
230	241	?: $this->scanTags()
231	240	?: $this->scanTableRow()
232	240	?: $this->scanNewline()
233	243	?: $this->scanText();
234		}
235
236		/**
237		* Scans for token with specified regex.
238		*
239		* @param string $regex Regular expression
240		* @param string $type Expected token type
241		*
242		* @return null\|array
243		*/
244	195	protected function scanInput($regex, $type)
245		{
246	195	if (!preg_match($regex, $this->line, $matches)) {
247	4	return null;
248		}
249
250	193	$token = $this->takeToken($type, $matches[1]);
251	193	$this->consumeLine();
252
253	193	return $token;
254		}
255
256		/**
257		* Scans for token with specified keywords.
258		*
259		* @param string $keywords Keywords (splitted with \|)
260		* @param string $type Expected token type
261		*
262		* @return null\|array
263		*/
264	243	protected function scanInputForKeywords($keywords, $type)
265		{
266	243	if (!preg_match('/^(\s)(' . $keywords . '):\s(.*)/u', $this->line, $matches)) {
267	243	return null;
268		}
269
270	242	$token = $this->takeToken($type, $matches[3]);
271	242	$token['keyword'] = $matches[2];
272	242	$token['indent'] = mb_strlen($matches[1], 'utf8');
273
274	242	$this->consumeLine();
275
276		// turn off language searching
277	242	if ('Feature' === $type) {
278	242	$this->featureStarted = true;
279		}
280
281		// turn off PyString and Table searching
282	242	if ('Feature' === $type \|\| 'Scenario' === $type \|\| 'Outline' === $type) {
283	242	$this->allowMultilineArguments = false;
284	211	} elseif ('Examples' === $type) {
285	205	$this->allowMultilineArguments = true;
286		}
287
288		// turn on steps searching
289	242	if ('Scenario' === $type \|\| 'Background' === $type \|\| 'Outline' === $type) {
290	237	$this->allowSteps = true;
291		}
292
293	242	return $token;
294		}
295
296		/**
297		* Scans EOS from input & returns it if found.
298		*
299		* @return null\|array
300		*/
301	243	protected function scanEOS()
302		{
303	243	if (!$this->eos) {
304	243	return null;
305		}
306
307	239	return $this->takeToken('EOS');
308		}
309
310		/**
311		* Returns keywords for provided type.
312		*
313		* @param string $type Keyword type
314		*
315		* @return string
316		*/
317	243	protected function getKeywords($type)
318		{
319	243	if (!isset($this->keywordsCache[$type])) {
320	243	$getter = 'get' . $type . 'Keywords';
321	243	$keywords = $this->keywords->$getter();
322
323	243	if ('Step' === $type) {
324	234	$padded = array();
325	234	foreach (explode('\|', $keywords) as $keyword) {
326	234	$padded[] = false !== mb_strpos($keyword, '<', 0, 'utf8')
327	19	? preg_quote(mb_substr($keyword, 0, -1, 'utf8'), '/') . '\s*'
328	228	: preg_quote($keyword, '/') . '\s+';
329		}
330
331	234	$keywords = implode('\|', $padded);
332		}
333
334	243	$this->keywordsCache[$type] = $keywords;
335		}
336
337	243	return $this->keywordsCache[$type];
338		}
339
340		/**
341		* Scans Feature from input & returns it if found.
342		*
343		* @return null\|array
344		*/
345	243	protected function scanFeature()
346		{
347	243	return $this->scanInputForKeywords($this->getKeywords('Feature'), 'Feature');
348		}
349
350		/**
351		* Scans Background from input & returns it if found.
352		*
353		* @return null\|array
354		*/
355	243	protected function scanBackground()
356		{
357	243	return $this->scanInputForKeywords($this->getKeywords('Background'), 'Background');
358		}
359
360		/**
361		* Scans Scenario from input & returns it if found.
362		*
363		* @return null\|array
364		*/
365	243	protected function scanScenario()
366		{
367	243	return $this->scanInputForKeywords($this->getKeywords('Scenario'), 'Scenario');
368		}
369
370		/**
371		* Scans Scenario Outline from input & returns it if found.
372		*
373		* @return null\|array
374		*/
375	243	protected function scanOutline()
376		{
377	243	return $this->scanInputForKeywords($this->getKeywords('Outline'), 'Outline');
378		}
379
380		/**
381		* Scans Scenario Outline Examples from input & returns it if found.
382		*
383		* @return null\|array
384		*/
385	243	protected function scanExamples()
386		{
387	243	return $this->scanInputForKeywords($this->getKeywords('Examples'), 'Examples');
388		}
389
390		/**
391		* Scans Step from input & returns it if found.
392		*
393		* @return null\|array
394		*/
395	243	protected function scanStep()
396		{
397	243	if (!$this->allowSteps) {
398	243	return null;
399		}
400
401	234	$keywords = $this->getKeywords('Step');
402	234	if (!preg_match('/^\s(' . $keywords . ')([^\s].)/u', $this->line, $matches)) {
403	230	return null;
404		}
405
406	230	$keyword = trim($matches[1]);
407	230	$token = $this->takeToken('Step', $keyword);
408	230	$token['keyword_type'] = $this->getStepKeywordType($keyword);
409	230	$token['text'] = $matches[2];
410
411	230	$this->consumeLine();
412	230	$this->allowMultilineArguments = true;
413
414	230	return $token;
415		}
416
417		/**
418		* Scans PyString from input & returns it if found.
419		*
420		* @return null\|array
421		*/
422	243	protected function scanPyStringOp()
423		{
424	243	if (!$this->allowMultilineArguments) {
425	243	return null;
426		}
427
428	230	if (false === ($pos = mb_strpos($this->line, '"""', 0, 'utf8'))) {
429	230	return null;
430		}
431
432	10	$this->inPyString = !$this->inPyString;
433	10	$token = $this->takeToken('PyStringOp');
434	10	$this->pyStringSwallow = $pos;
435
436	10	$this->consumeLine();
437
438	10	return $token;
439		}
440
441		/**
442		* Scans PyString content.
443		*
444		* @return null\|array
445		*/
446	243	protected function scanPyStringContent()
447		{
448	243	if (!$this->inPyString) {
449	243	return null;
450		}
451
452	10	$token = $this->scanText();
453		// swallow trailing spaces
454	10	$token['value'] = preg_replace('/^\s{0,' . $this->pyStringSwallow . '}/u', '', $token['value']);
455
456	10	return $token;
457		}
458
459		/**
460		* Scans Table Row from input & returns it if found.
461		*
462		* @return null\|array
463		*/
464	240	protected function scanTableRow()
465		{
466	240	if (!$this->allowMultilineArguments) {
467	234	return null;
468		}
469
470	225	$line = $this->getTrimmedLine();
471	225	if (!isset($line[0]) \|\| '\|' !== $line[0] \|\| '\|' !== substr($line, -1)) {
472	225	return null;
473		}
474
475	206	$token = $this->takeToken('TableRow');
476	206	$line = mb_substr($line, 1, mb_strlen($line, 'utf8') - 2, 'utf8');
477		$columns = array_map(function ($column) {
478	206	return trim(str_replace('\\\|', '\|', $column));
479	206	}, preg_split('/(?<!\\\)\\|/u', $line));
480	206	$token['columns'] = $columns;
481
482	206	$this->consumeLine();
483
484	206	return $token;
485		}
486
487		/**
488		* Scans Tags from input & returns it if found.
489		*
490		* @return null\|array
491		*/
492	241	protected function scanTags()
493		{
494	241	$line = $this->getTrimmedLine();
495	241	if (!isset($line[0]) \|\| '@' !== $line[0]) {
496	240	return null;
497		}
498
499	10	$token = $this->takeToken('Tag');
500	10	$tags = explode('@', mb_substr($line, 1, mb_strlen($line, 'utf8') - 1, 'utf8'));
501	10	$tags = array_map('trim', $tags);
502	10	$token['tags'] = $tags;
503
504	10	$this->consumeLine();
505
506	10	return $token;
507		}
508
509		/**
510		* Scans Language specifier from input & returns it if found.
511		*
512		* @return null\|array
513		*/
514	243	protected function scanLanguage()
515		{
516	243	if ($this->featureStarted) {
517	242	return null;
518		}
519
520	243	if ($this->inPyString) {
521		return null;
522		}
523
524	243	if (0 !== mb_strpos(ltrim($this->line), '#', 0, 'utf8')) {
525	243	return null;
526		}
527
528	195	return $this->scanInput('/^\s\#\slanguage:\s([\w_\-]+)\s$/', 'Language');
529		}
530
531		/**
532		* Scans Comment from input & returns it if found.
533		*
534		* @return null\|array
535		*/
536	243	protected function scanComment()
537		{
538	243	if ($this->inPyString) {
539	10	return null;
540		}
541
542	243	$line = $this->getTrimmedLine();
543	243	if (0 !== mb_strpos($line, '#', 0, 'utf8')) {
544	243	return null;
545		}
546
547	8	$token = $this->takeToken('Comment', $line);
548	8	$this->consumeLine();
549
550	8	return $token;
551		}
552
553		/**
554		* Scans Newline from input & returns it if found.
555		*
556		* @return null\|array
557		*/
558	240	protected function scanNewline()
559		{
560	240	if ('' !== $this->getTrimmedLine()) {
561	217	return null;
562		}
563
564	240	$token = $this->takeToken('Newline', mb_strlen($this->line, 'utf8'));
565	240	$this->consumeLine();
566
567	240	return $token;
568		}
569
570		/**
571		* Scans text from input & returns it if found.
572		*
573		* @return null\|array
574		*/
575	224	protected function scanText()
576		{
577	224	$token = $this->takeToken('Text', $this->line);
578	224	$this->consumeLine();
579
580	224	return $token;
581		}
582
583		/**
584		* Returns step type keyword (Given, When, Then, etc.).
585		*
586		* @param string $native Step keyword in provided language
587		* @return string
588		*/
589	230	private function getStepKeywordType($native)
590		{
591		// Consider "*" as a AND keyword so that it is normalized to the previous step type
592	230	if ('*' === $native) {
593		return 'And';
594		}
595
596	230	if (empty($this->stepKeywordTypesCache)) {
597	230	$this->stepKeywordTypesCache = array(
598	230	'Given' => explode('\|', $this->keywords->getGivenKeywords()),
599	230	'When' => explode('\|', $this->keywords->getWhenKeywords()),
600	230	'Then' => explode('\|', $this->keywords->getThenKeywords()),
601	230	'And' => explode('\|', $this->keywords->getAndKeywords()),
602	230	'But' => explode('\|', $this->keywords->getButKeywords())
603		);
604		}
605
606	230	foreach ($this->stepKeywordTypesCache as $type => $keywords) {
607	230	if (in_array($native, $keywords) \|\| in_array($native . '<', $keywords)) {
608	230	return $type;
609		}
610		}
611
612		return 'Given';
613		}
614		}
615

Behat / Gherkin

Lexer::getNextToken() F last analyzed 2021-01-14 15:16 UTC

Complexity

Size

Duplication

Code Coverage

Importance

How to fix Complexity

Long Method

Duplication Side-by-Side

Filter issues like

Lexer::getNextToken() F
last analyzed 2021-01-14 15:16 UTC