Indentation::unindent() - Code Metrics - colinodell/indentation - Measure and Improve Code Quality continuously with Scrutinizer

Indentation::unindent() B
last analyzed 2023-09-05 08:43 UTC

↳ Parent: Indentation

Complexity

Conditions	9
Paths	11

Size

Total Lines	32
Code Lines	17

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	17
CRAP Score	9.0139

Importance

Changes

Metric	Value
cc	9
eloc	17
c	0
b	0
f	0
nc	11
nop	1
dl	0
loc	32
ccs	17
cts	18
cp	0.9444
crap	9.0139
rs	8.0555

<?php

declare(strict_types=1);

/*
 * This file is part of the colinodell/indentation package.
 *
 * (c) Colin O'Dell <[email protected]>
 *
 * detect() method forked from detect-indent,
 * (c) Sindre Sorhus <[email protected]> (https://sindresorhus.com)
 *
 * For the full copyright and license information, please view the LICENSE
 * file that was distributed with this source code.
 */

namespace ColinODell\Indentation;

final class Indentation
{
    public const TYPE_SPACE   = 'space';
    public const TYPE_TAB     = 'tab';
    public const TYPE_UNKNOWN = 'unknown';

    /** @var int<0, max> */
    public int $amount;

    /** @var self::TYPE_* */
    public string $type;

    /**
     * @param int<0, max>  $amount
     * @param self::TYPE_* $type
     */
    public function __construct(int $amount, string $type)
    {
        $this->amount = $amount;
        $this->type   = $type;
    }

    /**
     * @return int<0, max>
     */
    public function getAmount(): int
    {
        return $this->amount;
    }

    /**
     * @return self::TYPE_*
     */
    public function getType(): string
    {
        return $this->type;

    }

    public function __toString(): string
    {
        if ($this->amount === 0 || $this->type === self::TYPE_UNKNOWN) {
            return '';
        }

        $indentCharacter = $this->type === self::TYPE_SPACE ? ' ' : "\t";

        return \str_repeat($indentCharacter, $this->amount);
    }

    /**
     * Detect the indentation of the given string.
     */
    public static function detect(string $string): Indentation
    {
        // Identify indents while skipping single space indents to avoid common edge cases (e.g. code comments)
        $indents = self::makeIndentsMap($string, true);
        // If no indents are identified, run again and include all indents for comprehensive detection
        if (\count($indents) === 0) {
            $indents = self::makeIndentsMap($string, false);
        }

        $keyOfMostUsedIndent = self::getMostUsedKey($indents);
        if ($keyOfMostUsedIndent === null) {
            return new self(0, self::TYPE_UNKNOWN);
        }

        [$amount, $type] = self::decodeIndentsKey($keyOfMostUsedIndent);

        return new self(\max(0, $amount), $type);
    }

    /**
     * Change the indentation from one style to another
     */
    public static function change(string $string, Indentation $newStyle): string
    {
        $oldStyle = self::detect($string);

        if ($oldStyle->type === self::TYPE_UNKNOWN || $oldStyle->amount === 0) {
            return $string;
        }

        $lines = \preg_split('/(\R)/', $string, flags: \PREG_SPLIT_DELIM_CAPTURE);
        if ($lines === false) {
            throw new \InvalidArgumentException('Bad input string');
        }

        $newContent = '';
        foreach ($lines as $i => $line) {
            // Newline characters are in the odd-numbered positions
            if ($i % 2 === 1) {
                $newContent .= $line;
                continue;
            }

            if (\preg_match('/^(?:' . \preg_quote($oldStyle->__toString(), '/') . ')+/', $line, $matches) !== 1) {
                $newContent .= $line;
                continue;
            }

            $indentLevel = (int) (\strlen($matches[0]) / $oldStyle->amount);
            $newContent .= \str_repeat($newStyle->__toString(), $indentLevel) . \substr($line, $indentLevel * $oldStyle->amount);
        }

        return $newContent;
    }

    /**
     * Adds the given $indentation to the beginning of each line in the given $string
     *
     * @throws \InvalidArgumentException if $indentation type is not spaces or tabs
     */
    public static function indent(string $string, Indentation $indentation): string
    {
        $toAdd = (string) $indentation;

        $result = \preg_replace('/^(?=.)/m', $toAdd, $string);
        if ($result === null) {
            return $string;
        }

        return $result;
    }

    /**
     * De-indent the given string, removing any leading indentation that is common to all lines.
     */
    public static function unindent(string $string): string
    {
        $leadingIndent     = PHP_INT_MAX;
        $leadingIndentType = self::TYPE_UNKNOWN;
        foreach (self::iterateLines($string) as $indentation) {
            // Any lines with no leading indentation means we can't trim the entire string
            if ($indentation === null) {
                return $string;
            }

            $leadingIndent = \min($leadingIndent, $indentation[0]);
            if ($leadingIndentType === self::TYPE_UNKNOWN) {
                $leadingIndentType = $indentation[1];
            } elseif ($leadingIndentType !== $indentation[1]) {
                // Don't trim if the leading indent types are different
                return $string;
            }
        }

        // Don't trim if there's no leading indents or if the types are inconsistent
        if ($leadingIndent === 0 || $leadingIndent === PHP_INT_MAX || $leadingIndentType === self::TYPE_UNKNOWN) {
            return $string;
        }

        $leadingIndent = new Indentation($leadingIndent, $leadingIndentType);

        $trimmed = \preg_replace('/^' . \preg_quote((string) $leadingIndent, '/') . '/m', '', $string);
        if (! \is_string($trimmed)) {
            return $string;
        }

        return $trimmed;
    }

    /**
     * @return array<string, array{0: int, 1: int}>
     */
    private static function makeIndentsMap(string $string, bool $ignoreSingleSpaces): array
    {
        $indents = [];

        // Remember the size of previous line's indentation
        $previousSize       = 0;
        $previousIndentType = null;

        // Indents key (ident type + size of the indents/unindents)
        $key = null;

        foreach (self::iterateLines($string) as $indentation) {
            // Detect either spaces or tabs but not both to properly handle tabs for indentation and spaces for alignment
            if ($indentation === null) {
                $previousSize       = 0;
                $previousIndentType = '';
                continue;
            }

            [$indent, $indentType] = $indentation;
            // Ignore single space unless it's the only indent detected to prevent common false positives
            if ($ignoreSingleSpaces && $indentType === self::TYPE_SPACE && $indent === 1) {
                continue;
            }

            if ($indentType !== $previousIndentType) {
                $previousSize = 0;
            }

            $previousIndentType = $indentType;
            $weight             = 0;
            $indentDifference   = $indent - $previousSize;
            $previousSize       = $indent;

            // Previous line have same indent?
            if ($indentDifference === 0) {
                $weight++;
                // We use the key from previous loop
                \assert(isset($key) && \is_string($key));
            } else {
                $key = self::encodeIndentsKey($indentType, $indentDifference > 0 ? $indentDifference : -$indentDifference);
            }

            // Update the stats
            if (! isset($indents[$key])) {
                $indents[$key] = [1, 0];
            } else {
                $indents[$key][0]++;
                $indents[$key][1] += $weight;
            }
        }

        return $indents;
    }

    /**
     * @return iterable<int, array{0: int<0, max>, 1: self::TYPE_*}|null>
     */
    private static function iterateLines(string $string): iterable
    {
        $lines = \preg_split('/\R/', $string);
        if ($lines === false) {
            throw new \InvalidArgumentException('Invalid string');
        }

        foreach ($lines as $i => $line) {
            if ($line === '') {
                // Ignore empty lines
                continue;
            }

            // Detect either spaces or tabs but not both to properly handle tabs for indentation and spaces for alignment
            if (\preg_match('/^(?:( )+|\t+)/', $line, $matches) !== 1) {
                yield $i => null;

                continue;
            }

            $indent     = \strlen($matches[0]);
            $indentType = isset($matches[1]) ? self::TYPE_SPACE : self::TYPE_TAB;

            yield $i => [$indent, $indentType];
        }
    }

    /**
     * Encode the indent type and amount as a string (e.g. 's4') for use as a compound key in the indents map.
     *
     * @param self::TYPE_* $indentType
     */
    private static function encodeIndentsKey(string $indentType, int $indentAmount): string
    {
        $typeCharacter = $indentType === self::TYPE_SPACE ? 's' : 't';

        return $typeCharacter . $indentAmount;
    }

    /**
     * Extract the indent type and amount from a key of the indents map.
     *
     * @return array{0: int, 1: self::TYPE_*}
     */
    private static function decodeIndentsKey(string $indentsKey): array
    {
        $keyHasTypeSpace = $indentsKey[0] === 's';
        $type            = $keyHasTypeSpace ? self::TYPE_SPACE : self::TYPE_TAB;

        $amount = \intval(\substr($indentsKey, 1));

        return [$amount, $type];
    }

    /**
     * Return the key (e.g. 's4') from the indents map that represents the most common indent,
     * or return undefined if there are no indents.
     *
     * @param array<string, array{int, int}> $indents
     */
    private static function getMostUsedKey(array $indents): string|null
    {
        $result    = null;
        $maxUsed   = 0;
        $maxWeight = 0;

        foreach ($indents as $key => [$usedCount, $weight]) {
            if ($usedCount <= $maxUsed && ($usedCount !== $maxUsed || $weight <= $maxWeight)) {
                continue;
            }

            $maxUsed   = $usedCount;
            $maxWeight = $weight;
            $result    = $key;
        }

        return $result;
    }
}


1		<?php
2
3		declare(strict_types=1);
4
5		/*
6		* This file is part of the colinodell/indentation package.
7		*
8		* (c) Colin O'Dell <[email protected]>
9		*
10		* detect() method forked from detect-indent,
11		* (c) Sindre Sorhus <[email protected]> (https://sindresorhus.com)
12		*
13		* For the full copyright and license information, please view the LICENSE
14		* file that was distributed with this source code.
15		*/
16
17		namespace ColinODell\Indentation;
18
19		final class Indentation
20		{
21		public const TYPE_SPACE = 'space';
22		public const TYPE_TAB = 'tab';
23		public const TYPE_UNKNOWN = 'unknown';
24
25		/** @var int<0, max> */
26		public int $amount;
27
28		/** @var self::TYPE_* */
29		public string $type;
30
31		/**
32		* @param int<0, max> $amount
33		* @param self::TYPE_* $type
34		*/
35	30	public function __construct(int $amount, string $type)
36		{
37	30	$this->amount = $amount;
38	30	$this->type = $type;
39		}
40
41		/**
42		* @return int<0, max>
43		*/
44	1	public function getAmount(): int
45		{
46	1	return $this->amount;
47		}
48
49		/**
50		* @return self::TYPE_*
51		*/
52	1	public function getType(): string
53		{
54	1	return $this->type;
		0 ignored issues – show Bug Best Practice introduced 2022-03-16 21:50 UTC by Report Bug Copy Issue Report The expression `return $this->type` returns the type `string` which is incompatible with the documented return type `ColinODell\Indentation\Indentation`. Loading history...
55		}
56
57	34	public function __toString(): string
58		{
59	34	if ($this->amount === 0 \|\| $this->type === self::TYPE_UNKNOWN) {
60	3	return '';
61		}
62
63	31	$indentCharacter = $this->type === self::TYPE_SPACE ? ' ' : "\t";
64
65	31	return \str_repeat($indentCharacter, $this->amount);
66		}
67
68		/**
69		* Detect the indentation of the given string.
70		*/
71	21	public static function detect(string $string): Indentation
72		{
73		// Identify indents while skipping single space indents to avoid common edge cases (e.g. code comments)
74	21	$indents = self::makeIndentsMap($string, true);
75		// If no indents are identified, run again and include all indents for comprehensive detection
76	21	if (\count($indents) === 0) {
77	4	$indents = self::makeIndentsMap($string, false);
78		}
79
80	21	$keyOfMostUsedIndent = self::getMostUsedKey($indents);
81	21	if ($keyOfMostUsedIndent === null) {
82	3	return new self(0, self::TYPE_UNKNOWN);
83		}
84
85	18	[$amount, $type] = self::decodeIndentsKey($keyOfMostUsedIndent);
86
87	18	return new self(\max(0, $amount), $type);
88		}
89
90		/**
91		* Change the indentation from one style to another
92		*/
93	9	public static function change(string $string, Indentation $newStyle): string
94		{
95	9	$oldStyle = self::detect($string);
96
97	9	if ($oldStyle->type === self::TYPE_UNKNOWN \|\| $oldStyle->amount === 0) {
98	2	return $string;
99		}
100
101	7	$lines = \preg_split('/(\R)/', $string, flags: \PREG_SPLIT_DELIM_CAPTURE);
102	7	if ($lines === false) {
103		throw new \InvalidArgumentException('Bad input string');
104		}
105
106	7	$newContent = '';
107	7	foreach ($lines as $i => $line) {
108		// Newline characters are in the odd-numbered positions
109	7	if ($i % 2 === 1) {
110	7	$newContent .= $line;
111	7	continue;
112		}
113
114	7	if (\preg_match('/^(?:' . \preg_quote($oldStyle->__toString(), '/') . ')+/', $line, $matches) !== 1) {
115	7	$newContent .= $line;
116	7	continue;
117		}
118
119	7	$indentLevel = (int) (\strlen($matches[0]) / $oldStyle->amount);
120	7	$newContent .= \str_repeat($newStyle->__toString(), $indentLevel) . \substr($line, $indentLevel * $oldStyle->amount);
121		}
122
123	7	return $newContent;
124		}
125
126		/**
127		* Adds the given $indentation to the beginning of each line in the given $string
128		*
129		* @throws \InvalidArgumentException if $indentation type is not spaces or tabs
130		*/
131	6	public static function indent(string $string, Indentation $indentation): string
132		{
133	6	$toAdd = (string) $indentation;
134
135	6	$result = \preg_replace('/^(?=.)/m', $toAdd, $string);
136	6	if ($result === null) {
137		return $string;
138		}
139
140	6	return $result;
141		}
142
143		/**
144		* De-indent the given string, removing any leading indentation that is common to all lines.
145		*/
146	12	public static function unindent(string $string): string
147		{
148	12	$leadingIndent = PHP_INT_MAX;
149	12	$leadingIndentType = self::TYPE_UNKNOWN;
150	12	foreach (self::iterateLines($string) as $indentation) {
151		// Any lines with no leading indentation means we can't trim the entire string
152	11	if ($indentation === null) {
153	1	return $string;
154		}
155
156	10	$leadingIndent = \min($leadingIndent, $indentation[0]);
157	10	if ($leadingIndentType === self::TYPE_UNKNOWN) {
158	10	$leadingIndentType = $indentation[1];
159	10	} elseif ($leadingIndentType !== $indentation[1]) {
160		// Don't trim if the leading indent types are different
161	2	return $string;
162		}
163		}
164
165		// Don't trim if there's no leading indents or if the types are inconsistent
166	9	if ($leadingIndent === 0 \|\| $leadingIndent === PHP_INT_MAX \|\| $leadingIndentType === self::TYPE_UNKNOWN) {
167	1	return $string;
168		}
169
170	8	$leadingIndent = new Indentation($leadingIndent, $leadingIndentType);
171
172	8	$trimmed = \preg_replace('/^' . \preg_quote((string) $leadingIndent, '/') . '/m', '', $string);
173	8	if (! \is_string($trimmed)) {
174		return $string;
175		}
176
177	8	return $trimmed;
178		}
179
180		/**
181		* @return array<string, array{0: int, 1: int}>
182		*/
183	21	private static function makeIndentsMap(string $string, bool $ignoreSingleSpaces): array
184		{
185	21	$indents = [];
186
187		// Remember the size of previous line's indentation
188	21	$previousSize = 0;
189	21	$previousIndentType = null;
190
191		// Indents key (ident type + size of the indents/unindents)
192	21	$key = null;
193
194	21	foreach (self::iterateLines($string) as $indentation) {
195		// Detect either spaces or tabs but not both to properly handle tabs for indentation and spaces for alignment
196	20	if ($indentation === null) {
197	15	$previousSize = 0;
198	15	$previousIndentType = '';
199	15	continue;
200		}
201
202	18	[$indent, $indentType] = $indentation;
203		// Ignore single space unless it's the only indent detected to prevent common false positives
204	18	if ($ignoreSingleSpaces && $indentType === self::TYPE_SPACE && $indent === 1) {
205	2	continue;
206		}
207
208	18	if ($indentType !== $previousIndentType) {
209	18	$previousSize = 0;
210		}
211
212	18	$previousIndentType = $indentType;
213	18	$weight = 0;
214	18	$indentDifference = $indent - $previousSize;
215	18	$previousSize = $indent;
216
217		// Previous line have same indent?
218	18	if ($indentDifference === 0) {
219	11	$weight++;
220		// We use the key from previous loop
221	11	\assert(isset($key) && \is_string($key));
222		} else {
223	18	$key = self::encodeIndentsKey($indentType, $indentDifference > 0 ? $indentDifference : -$indentDifference);
224		}
225
226		// Update the stats
227	18	if (! isset($indents[$key])) {
228	18	$indents[$key] = [1, 0];
229		} else {
230	18	$indents[$key][0]++;
231	18	$indents[$key][1] += $weight;
232		}
233		}
234
235	21	return $indents;
236		}
237
238		/**
239		* @return iterable<int, array{0: int<0, max>, 1: self::TYPE_*}\|null>
240		*/
241	33	private static function iterateLines(string $string): iterable
242		{
243	33	$lines = \preg_split('/\R/', $string);
244	33	if ($lines === false) {
245		throw new \InvalidArgumentException('Invalid string');
246		}
247
248	33	foreach ($lines as $i => $line) {
249	33	if ($line === '') {
250		// Ignore empty lines
251	19	continue;
252		}
253
254		// Detect either spaces or tabs but not both to properly handle tabs for indentation and spaces for alignment
255	31	if (\preg_match('/^(?:( )+\|\t+)/', $line, $matches) !== 1) {
256	16	yield $i => null;
257
258	15	continue;
259		}
260
261	28	$indent = \strlen($matches[0]);
262	28	$indentType = isset($matches[1]) ? self::TYPE_SPACE : self::TYPE_TAB;
263
264	28	yield $i => [$indent, $indentType];
265		}
266		}
267
268		/**
269		* Encode the indent type and amount as a string (e.g. 's4') for use as a compound key in the indents map.
270		*
271		* @param self::TYPE_* $indentType
272		*/
273	18	private static function encodeIndentsKey(string $indentType, int $indentAmount): string
274		{
275	18	$typeCharacter = $indentType === self::TYPE_SPACE ? 's' : 't';
276
277	18	return $typeCharacter . $indentAmount;
278		}
279
280		/**
281		* Extract the indent type and amount from a key of the indents map.
282		*
283		* @return array{0: int, 1: self::TYPE_*}
284		*/
285	18	private static function decodeIndentsKey(string $indentsKey): array
286		{
287	18	$keyHasTypeSpace = $indentsKey[0] === 's';
288	18	$type = $keyHasTypeSpace ? self::TYPE_SPACE : self::TYPE_TAB;
289
290	18	$amount = \intval(\substr($indentsKey, 1));
291
292	18	return [$amount, $type];
293		}
294
295		/**
296		* Return the key (e.g. 's4') from the indents map that represents the most common indent,
297		* or return undefined if there are no indents.
298		*
299		* @param array<string, array{int, int}> $indents
300		*/
301	21	private static function getMostUsedKey(array $indents): string\|null
302		{
303	21	$result = null;
304	21	$maxUsed = 0;
305	21	$maxWeight = 0;
306
307	21	foreach ($indents as $key => [$usedCount, $weight]) {
308	18	if ($usedCount <= $maxUsed && ($usedCount !== $maxUsed \|\| $weight <= $maxWeight)) {
309	6	continue;
310		}
311
312	18	$maxUsed = $usedCount;
313	18	$maxWeight = $weight;
314	18	$result = $key;
315		}
316
317	21	return $result;
318		}
319		}
320

colinodell / indentation

Indentation::unindent() B last analyzed 2023-09-05 08:43 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

Indentation::unindent() B
last analyzed 2023-09-05 08:43 UTC