Exception
1 year ago
Output
1 year ago
Chunk.php
1 year ago
Diff.php
1 year ago
Differ.php
1 year ago
Line.php
1 year ago
LongestCommonSubsequenceCalculator.php
1 year ago
MemoryEfficientLongestCommonSubsequenceCalculator.php
1 year ago
Parser.php
1 year ago
TimeEfficientLongestCommonSubsequenceCalculator.php
1 year ago
Differ.php
328 lines
| 1 | <?php declare(strict_types=1); |
| 2 | /* |
| 3 | * This file is part of sebastian/diff. |
| 4 | * |
| 5 | * (c) Sebastian Bergmann <sebastian@phpunit.de> |
| 6 | * |
| 7 | * For the full copyright and license information, please view the LICENSE |
| 8 | * file that was distributed with this source code. |
| 9 | */ |
| 10 | namespace SebastianBergmann\Diff; |
| 11 | |
| 12 | use const PHP_INT_SIZE; |
| 13 | use const PREG_SPLIT_DELIM_CAPTURE; |
| 14 | use const PREG_SPLIT_NO_EMPTY; |
| 15 | use function array_shift; |
| 16 | use function array_unshift; |
| 17 | use function array_values; |
| 18 | use function count; |
| 19 | use function current; |
| 20 | use function end; |
| 21 | use function get_class; |
| 22 | use function gettype; |
| 23 | use function is_array; |
| 24 | use function is_object; |
| 25 | use function is_string; |
| 26 | use function key; |
| 27 | use function min; |
| 28 | use function preg_split; |
| 29 | use function prev; |
| 30 | use function reset; |
| 31 | use function sprintf; |
| 32 | use function substr; |
| 33 | use SebastianBergmann\Diff\Output\DiffOutputBuilderInterface; |
| 34 | use SebastianBergmann\Diff\Output\UnifiedDiffOutputBuilder; |
| 35 | |
| 36 | final class Differ |
| 37 | { |
| 38 | public const OLD = 0; |
| 39 | |
| 40 | public const ADDED = 1; |
| 41 | |
| 42 | public const REMOVED = 2; |
| 43 | |
| 44 | public const DIFF_LINE_END_WARNING = 3; |
| 45 | |
| 46 | public const NO_LINE_END_EOF_WARNING = 4; |
| 47 | |
| 48 | /** |
| 49 | * @var DiffOutputBuilderInterface |
| 50 | */ |
| 51 | private $outputBuilder; |
| 52 | |
| 53 | /** |
| 54 | * @param DiffOutputBuilderInterface $outputBuilder |
| 55 | * |
| 56 | * @throws InvalidArgumentException |
| 57 | */ |
| 58 | public function __construct($outputBuilder = null) |
| 59 | { |
| 60 | if ($outputBuilder instanceof DiffOutputBuilderInterface) { |
| 61 | $this->outputBuilder = $outputBuilder; |
| 62 | } elseif (null === $outputBuilder) { |
| 63 | $this->outputBuilder = new UnifiedDiffOutputBuilder; |
| 64 | } elseif (is_string($outputBuilder)) { |
| 65 | // PHPUnit 6.1.4, 6.2.0, 6.2.1, 6.2.2, and 6.2.3 support |
| 66 | // @see https://github.com/sebastianbergmann/phpunit/issues/2734#issuecomment-314514056 |
| 67 | // @deprecated |
| 68 | $this->outputBuilder = new UnifiedDiffOutputBuilder($outputBuilder); |
| 69 | } else { |
| 70 | throw new InvalidArgumentException( |
| 71 | sprintf( |
| 72 | 'Expected builder to be an instance of DiffOutputBuilderInterface, <null> or a string, got %s.', |
| 73 | is_object($outputBuilder) ? 'instance of "' . get_class($outputBuilder) . '"' : gettype($outputBuilder) . ' "' . $outputBuilder . '"' |
| 74 | ) |
| 75 | ); |
| 76 | } |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * Returns the diff between two arrays or strings as string. |
| 81 | * |
| 82 | * @param array|string $from |
| 83 | * @param array|string $to |
| 84 | */ |
| 85 | public function diff($from, $to, ?LongestCommonSubsequenceCalculator $lcs = null): string |
| 86 | { |
| 87 | $diff = $this->diffToArray( |
| 88 | $this->normalizeDiffInput($from), |
| 89 | $this->normalizeDiffInput($to), |
| 90 | $lcs |
| 91 | ); |
| 92 | |
| 93 | return $this->outputBuilder->getDiff($diff); |
| 94 | } |
| 95 | |
| 96 | /** |
| 97 | * Returns the diff between two arrays or strings as array. |
| 98 | * |
| 99 | * Each array element contains two elements: |
| 100 | * - [0] => mixed $token |
| 101 | * - [1] => 2|1|0 |
| 102 | * |
| 103 | * - 2: REMOVED: $token was removed from $from |
| 104 | * - 1: ADDED: $token was added to $from |
| 105 | * - 0: OLD: $token is not changed in $to |
| 106 | * |
| 107 | * @param array|string $from |
| 108 | * @param array|string $to |
| 109 | * @param LongestCommonSubsequenceCalculator $lcs |
| 110 | */ |
| 111 | public function diffToArray($from, $to, ?LongestCommonSubsequenceCalculator $lcs = null): array |
| 112 | { |
| 113 | if (is_string($from)) { |
| 114 | $from = $this->splitStringByLines($from); |
| 115 | } elseif (!is_array($from)) { |
| 116 | throw new InvalidArgumentException('"from" must be an array or string.'); |
| 117 | } |
| 118 | |
| 119 | if (is_string($to)) { |
| 120 | $to = $this->splitStringByLines($to); |
| 121 | } elseif (!is_array($to)) { |
| 122 | throw new InvalidArgumentException('"to" must be an array or string.'); |
| 123 | } |
| 124 | |
| 125 | [$from, $to, $start, $end] = self::getArrayDiffParted($from, $to); |
| 126 | |
| 127 | if ($lcs === null) { |
| 128 | $lcs = $this->selectLcsImplementation($from, $to); |
| 129 | } |
| 130 | |
| 131 | $common = $lcs->calculate(array_values($from), array_values($to)); |
| 132 | $diff = []; |
| 133 | |
| 134 | foreach ($start as $token) { |
| 135 | $diff[] = [$token, self::OLD]; |
| 136 | } |
| 137 | |
| 138 | reset($from); |
| 139 | reset($to); |
| 140 | |
| 141 | foreach ($common as $token) { |
| 142 | while (($fromToken = reset($from)) !== $token) { |
| 143 | $diff[] = [array_shift($from), self::REMOVED]; |
| 144 | } |
| 145 | |
| 146 | while (($toToken = reset($to)) !== $token) { |
| 147 | $diff[] = [array_shift($to), self::ADDED]; |
| 148 | } |
| 149 | |
| 150 | $diff[] = [$token, self::OLD]; |
| 151 | |
| 152 | array_shift($from); |
| 153 | array_shift($to); |
| 154 | } |
| 155 | |
| 156 | while (($token = array_shift($from)) !== null) { |
| 157 | $diff[] = [$token, self::REMOVED]; |
| 158 | } |
| 159 | |
| 160 | while (($token = array_shift($to)) !== null) { |
| 161 | $diff[] = [$token, self::ADDED]; |
| 162 | } |
| 163 | |
| 164 | foreach ($end as $token) { |
| 165 | $diff[] = [$token, self::OLD]; |
| 166 | } |
| 167 | |
| 168 | if ($this->detectUnmatchedLineEndings($diff)) { |
| 169 | array_unshift($diff, ["#Warning: Strings contain different line endings!\n", self::DIFF_LINE_END_WARNING]); |
| 170 | } |
| 171 | |
| 172 | return $diff; |
| 173 | } |
| 174 | |
| 175 | /** |
| 176 | * Casts variable to string if it is not a string or array. |
| 177 | * |
| 178 | * @return array|string |
| 179 | */ |
| 180 | private function normalizeDiffInput($input) |
| 181 | { |
| 182 | if (!is_array($input) && !is_string($input)) { |
| 183 | return (string) $input; |
| 184 | } |
| 185 | |
| 186 | return $input; |
| 187 | } |
| 188 | |
| 189 | /** |
| 190 | * Checks if input is string, if so it will split it line-by-line. |
| 191 | */ |
| 192 | private function splitStringByLines(string $input): array |
| 193 | { |
| 194 | return preg_split('/(.*\R)/', $input, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); |
| 195 | } |
| 196 | |
| 197 | private function selectLcsImplementation(array $from, array $to): LongestCommonSubsequenceCalculator |
| 198 | { |
| 199 | // We do not want to use the time-efficient implementation if its memory |
| 200 | // footprint will probably exceed this value. Note that the footprint |
| 201 | // calculation is only an estimation for the matrix and the LCS method |
| 202 | // will typically allocate a bit more memory than this. |
| 203 | $memoryLimit = 100 * 1024 * 1024; |
| 204 | |
| 205 | if ($this->calculateEstimatedFootprint($from, $to) > $memoryLimit) { |
| 206 | return new MemoryEfficientLongestCommonSubsequenceCalculator; |
| 207 | } |
| 208 | |
| 209 | return new TimeEfficientLongestCommonSubsequenceCalculator; |
| 210 | } |
| 211 | |
| 212 | /** |
| 213 | * Calculates the estimated memory footprint for the DP-based method. |
| 214 | * |
| 215 | * @return float|int |
| 216 | */ |
| 217 | private function calculateEstimatedFootprint(array $from, array $to) |
| 218 | { |
| 219 | $itemSize = PHP_INT_SIZE === 4 ? 76 : 144; |
| 220 | |
| 221 | return $itemSize * min(count($from), count($to)) ** 2; |
| 222 | } |
| 223 | |
| 224 | /** |
| 225 | * Returns true if line ends don't match in a diff. |
| 226 | */ |
| 227 | private function detectUnmatchedLineEndings(array $diff): bool |
| 228 | { |
| 229 | $newLineBreaks = ['' => true]; |
| 230 | $oldLineBreaks = ['' => true]; |
| 231 | |
| 232 | foreach ($diff as $entry) { |
| 233 | if (self::OLD === $entry[1]) { |
| 234 | $ln = $this->getLinebreak($entry[0]); |
| 235 | $oldLineBreaks[$ln] = true; |
| 236 | $newLineBreaks[$ln] = true; |
| 237 | } elseif (self::ADDED === $entry[1]) { |
| 238 | $newLineBreaks[$this->getLinebreak($entry[0])] = true; |
| 239 | } elseif (self::REMOVED === $entry[1]) { |
| 240 | $oldLineBreaks[$this->getLinebreak($entry[0])] = true; |
| 241 | } |
| 242 | } |
| 243 | |
| 244 | // if either input or output is a single line without breaks than no warning should be raised |
| 245 | if (['' => true] === $newLineBreaks || ['' => true] === $oldLineBreaks) { |
| 246 | return false; |
| 247 | } |
| 248 | |
| 249 | // two way compare |
| 250 | foreach ($newLineBreaks as $break => $set) { |
| 251 | if (!isset($oldLineBreaks[$break])) { |
| 252 | return true; |
| 253 | } |
| 254 | } |
| 255 | |
| 256 | foreach ($oldLineBreaks as $break => $set) { |
| 257 | if (!isset($newLineBreaks[$break])) { |
| 258 | return true; |
| 259 | } |
| 260 | } |
| 261 | |
| 262 | return false; |
| 263 | } |
| 264 | |
| 265 | private function getLinebreak($line): string |
| 266 | { |
| 267 | if (!is_string($line)) { |
| 268 | return ''; |
| 269 | } |
| 270 | |
| 271 | $lc = substr($line, -1); |
| 272 | |
| 273 | if ("\r" === $lc) { |
| 274 | return "\r"; |
| 275 | } |
| 276 | |
| 277 | if ("\n" !== $lc) { |
| 278 | return ''; |
| 279 | } |
| 280 | |
| 281 | if ("\r\n" === substr($line, -2)) { |
| 282 | return "\r\n"; |
| 283 | } |
| 284 | |
| 285 | return "\n"; |
| 286 | } |
| 287 | |
| 288 | private static function getArrayDiffParted(array &$from, array &$to): array |
| 289 | { |
| 290 | $start = []; |
| 291 | $end = []; |
| 292 | |
| 293 | reset($to); |
| 294 | |
| 295 | foreach ($from as $k => $v) { |
| 296 | $toK = key($to); |
| 297 | |
| 298 | if ($toK === $k && $v === $to[$k]) { |
| 299 | $start[$k] = $v; |
| 300 | |
| 301 | unset($from[$k], $to[$k]); |
| 302 | } else { |
| 303 | break; |
| 304 | } |
| 305 | } |
| 306 | |
| 307 | end($from); |
| 308 | end($to); |
| 309 | |
| 310 | do { |
| 311 | $fromK = key($from); |
| 312 | $toK = key($to); |
| 313 | |
| 314 | if (null === $fromK || null === $toK || current($from) !== current($to)) { |
| 315 | break; |
| 316 | } |
| 317 | |
| 318 | prev($from); |
| 319 | prev($to); |
| 320 | |
| 321 | $end = [$fromK => $from[$fromK]] + $end; |
| 322 | unset($from[$fromK], $to[$toK]); |
| 323 | } while (true); |
| 324 | |
| 325 | return [$from, $to, $start, $end]; |
| 326 | } |
| 327 | } |
| 328 |