Chris@0: <?php
Chris@0: /*
Chris@12:  * This file is part of sebastian/diff.
Chris@0:  *
Chris@0:  * (c) Sebastian Bergmann <sebastian@phpunit.de>
Chris@0:  *
Chris@0:  * For the full copyright and license information, please view the LICENSE
Chris@0:  * file that was distributed with this source code.
Chris@0:  */
Chris@0: 
Chris@0: namespace SebastianBergmann\Diff;
Chris@0: 
Chris@0: use SebastianBergmann\Diff\LCS\LongestCommonSubsequence;
Chris@0: use SebastianBergmann\Diff\LCS\TimeEfficientImplementation;
Chris@0: use SebastianBergmann\Diff\LCS\MemoryEfficientImplementation;
Chris@0: 
Chris@0: /**
Chris@0:  * Diff implementation.
Chris@0:  */
Chris@0: class Differ
Chris@0: {
Chris@0:     /**
Chris@0:      * @var string
Chris@0:      */
Chris@0:     private $header;
Chris@0: 
Chris@0:     /**
Chris@0:      * @var bool
Chris@0:      */
Chris@0:     private $showNonDiffLines;
Chris@0: 
Chris@0:     /**
Chris@0:      * @param string $header
Chris@12:      * @param bool   $showNonDiffLines
Chris@0:      */
Chris@0:     public function __construct($header = "--- Original\n+++ New\n", $showNonDiffLines = true)
Chris@0:     {
Chris@0:         $this->header           = $header;
Chris@0:         $this->showNonDiffLines = $showNonDiffLines;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the diff between two arrays or strings as string.
Chris@0:      *
Chris@0:      * @param array|string             $from
Chris@0:      * @param array|string             $to
Chris@0:      * @param LongestCommonSubsequence $lcs
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     public function diff($from, $to, LongestCommonSubsequence $lcs = null)
Chris@0:     {
Chris@12:         $from  = $this->validateDiffInput($from);
Chris@12:         $to    = $this->validateDiffInput($to);
Chris@12:         $diff  = $this->diffToArray($from, $to, $lcs);
Chris@12:         $old   = $this->checkIfDiffInOld($diff);
Chris@12:         $start = isset($old[0]) ? $old[0] : 0;
Chris@12:         $end   = \count($diff);
Chris@12: 
Chris@12:         if ($tmp = \array_search($end, $old)) {
Chris@12:             $end = $tmp;
Chris@0:         }
Chris@0: 
Chris@12:         return $this->getBuffer($diff, $old, $start, $end);
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@12:      * Casts variable to string if it is not a string or array.
Chris@12:      *
Chris@12:      * @param mixed $input
Chris@12:      *
Chris@12:      * @return string
Chris@12:      */
Chris@12:     private function validateDiffInput($input)
Chris@12:     {
Chris@12:         if (!\is_array($input) && !\is_string($input)) {
Chris@12:             return (string) $input;
Chris@0:         }
Chris@0: 
Chris@12:         return $input;
Chris@12:     }
Chris@0: 
Chris@12:     /**
Chris@12:      * Takes input of the diff array and returns the old array.
Chris@12:      * Iterates through diff line by line,
Chris@12:      *
Chris@12:      * @param array $diff
Chris@12:      *
Chris@12:      * @return array
Chris@12:      */
Chris@12:     private function checkIfDiffInOld(array $diff)
Chris@12:     {
Chris@0:         $inOld = false;
Chris@0:         $i     = 0;
Chris@0:         $old   = array();
Chris@0: 
Chris@0:         foreach ($diff as $line) {
Chris@12:             if ($line[1] === 0 /* OLD */) {
Chris@0:                 if ($inOld === false) {
Chris@0:                     $inOld = $i;
Chris@0:                 }
Chris@0:             } elseif ($inOld !== false) {
Chris@0:                 if (($i - $inOld) > 5) {
Chris@0:                     $old[$inOld] = $i - 1;
Chris@0:                 }
Chris@0: 
Chris@0:                 $inOld = false;
Chris@0:             }
Chris@0: 
Chris@0:             ++$i;
Chris@0:         }
Chris@0: 
Chris@12:         return $old;
Chris@12:     }
Chris@0: 
Chris@12:     /**
Chris@12:      * Generates buffer in string format, returning the patch.
Chris@12:      *
Chris@12:      * @param array $diff
Chris@12:      * @param array $old
Chris@12:      * @param int   $start
Chris@12:      * @param int   $end
Chris@12:      *
Chris@12:      * @return string
Chris@12:      */
Chris@12:     private function getBuffer(array $diff, array $old, $start, $end)
Chris@12:     {
Chris@12:         $buffer = $this->header;
Chris@12: 
Chris@12:         if (!isset($old[$start])) {
Chris@12:             $buffer = $this->getDiffBufferElementNew($diff, $buffer, $start);
Chris@12:             ++$start;
Chris@0:         }
Chris@0: 
Chris@0:         for ($i = $start; $i < $end; $i++) {
Chris@0:             if (isset($old[$i])) {
Chris@12:                 $i      = $old[$i];
Chris@12:                 $buffer = $this->getDiffBufferElementNew($diff, $buffer, $i);
Chris@12:             } else {
Chris@12:                 $buffer = $this->getDiffBufferElement($diff, $buffer, $i);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $buffer;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@12:      * Gets individual buffer element.
Chris@12:      *
Chris@12:      * @param array  $diff
Chris@12:      * @param string $buffer
Chris@12:      * @param int    $diffIndex
Chris@12:      *
Chris@12:      * @return string
Chris@12:      */
Chris@12:     private function getDiffBufferElement(array $diff, $buffer, $diffIndex)
Chris@12:     {
Chris@12:         if ($diff[$diffIndex][1] === 1 /* ADDED */) {
Chris@12:             $buffer .= '+' . $diff[$diffIndex][0] . "\n";
Chris@12:         } elseif ($diff[$diffIndex][1] === 2 /* REMOVED */) {
Chris@12:             $buffer .= '-' . $diff[$diffIndex][0] . "\n";
Chris@12:         } elseif ($this->showNonDiffLines === true) {
Chris@12:             $buffer .= ' ' . $diff[$diffIndex][0] . "\n";
Chris@12:         }
Chris@12: 
Chris@12:         return $buffer;
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@12:      * Gets individual buffer element with opening.
Chris@12:      *
Chris@12:      * @param array  $diff
Chris@12:      * @param string $buffer
Chris@12:      * @param int    $diffIndex
Chris@12:      *
Chris@12:      * @return string
Chris@12:      */
Chris@12:     private function getDiffBufferElementNew(array $diff, $buffer, $diffIndex)
Chris@12:     {
Chris@12:         if ($this->showNonDiffLines === true) {
Chris@12:             $buffer .= "@@ @@\n";
Chris@12:         }
Chris@12: 
Chris@12:         return $this->getDiffBufferElement($diff, $buffer, $diffIndex);
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@0:      * Returns the diff between two arrays or strings as array.
Chris@0:      *
Chris@0:      * Each array element contains two elements:
Chris@12:      *   - [0] => mixed $token
Chris@0:      *   - [1] => 2|1|0
Chris@0:      *
Chris@0:      * - 2: REMOVED: $token was removed from $from
Chris@0:      * - 1: ADDED: $token was added to $from
Chris@0:      * - 0: OLD: $token is not changed in $to
Chris@0:      *
Chris@0:      * @param array|string             $from
Chris@0:      * @param array|string             $to
Chris@0:      * @param LongestCommonSubsequence $lcs
Chris@0:      *
Chris@0:      * @return array
Chris@0:      */
Chris@0:     public function diffToArray($from, $to, LongestCommonSubsequence $lcs = null)
Chris@0:     {
Chris@12:         if (\is_string($from)) {
Chris@12:             $fromMatches = $this->getNewLineMatches($from);
Chris@12:             $from        = $this->splitStringByLines($from);
Chris@12:         } elseif (\is_array($from)) {
Chris@12:             $fromMatches = array();
Chris@12:         } else {
Chris@12:             throw new \InvalidArgumentException('"from" must be an array or string.');
Chris@0:         }
Chris@0: 
Chris@12:         if (\is_string($to)) {
Chris@12:             $toMatches = $this->getNewLineMatches($to);
Chris@12:             $to        = $this->splitStringByLines($to);
Chris@12:         } elseif (\is_array($to)) {
Chris@12:             $toMatches = array();
Chris@12:         } else {
Chris@12:             throw new \InvalidArgumentException('"to" must be an array or string.');
Chris@0:         }
Chris@0: 
Chris@12:         list($from, $to, $start, $end) = self::getArrayDiffParted($from, $to);
Chris@0: 
Chris@0:         if ($lcs === null) {
Chris@0:             $lcs = $this->selectLcsImplementation($from, $to);
Chris@0:         }
Chris@0: 
Chris@12:         $common = $lcs->calculate(\array_values($from), \array_values($to));
Chris@0:         $diff   = array();
Chris@0: 
Chris@12:         if ($this->detectUnmatchedLineEndings($fromMatches, $toMatches)) {
Chris@0:             $diff[] = array(
Chris@12:                 '#Warning: Strings contain different line endings!',
Chris@12:                 0
Chris@0:             );
Chris@0:         }
Chris@0: 
Chris@0:         foreach ($start as $token) {
Chris@0:             $diff[] = array($token, 0 /* OLD */);
Chris@0:         }
Chris@0: 
Chris@12:         \reset($from);
Chris@12:         \reset($to);
Chris@0: 
Chris@0:         foreach ($common as $token) {
Chris@12:             while (($fromToken = \reset($from)) !== $token) {
Chris@12:                 $diff[] = array(\array_shift($from), 2 /* REMOVED */);
Chris@0:             }
Chris@0: 
Chris@12:             while (($toToken = \reset($to)) !== $token) {
Chris@12:                 $diff[] = array(\array_shift($to), 1 /* ADDED */);
Chris@0:             }
Chris@0: 
Chris@0:             $diff[] = array($token, 0 /* OLD */);
Chris@0: 
Chris@12:             \array_shift($from);
Chris@12:             \array_shift($to);
Chris@0:         }
Chris@0: 
Chris@12:         while (($token = \array_shift($from)) !== null) {
Chris@0:             $diff[] = array($token, 2 /* REMOVED */);
Chris@0:         }
Chris@0: 
Chris@12:         while (($token = \array_shift($to)) !== null) {
Chris@0:             $diff[] = array($token, 1 /* ADDED */);
Chris@0:         }
Chris@0: 
Chris@0:         foreach ($end as $token) {
Chris@0:             $diff[] = array($token, 0 /* OLD */);
Chris@0:         }
Chris@0: 
Chris@0:         return $diff;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@12:      * Get new strings denoting new lines from a given string.
Chris@12:      *
Chris@12:      * @param string $string
Chris@12:      *
Chris@12:      * @return array
Chris@12:      */
Chris@12:     private function getNewLineMatches($string)
Chris@12:     {
Chris@12:         \preg_match_all('(\r\n|\r|\n)', $string, $stringMatches);
Chris@12: 
Chris@12:         return $stringMatches;
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@12:      * Checks if input is string, if so it will split it line-by-line.
Chris@12:      *
Chris@12:      * @param string $input
Chris@12:      *
Chris@12:      * @return array
Chris@12:      */
Chris@12:     private function splitStringByLines($input)
Chris@12:     {
Chris@12:         return \preg_split('(\r\n|\r|\n)', $input);
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@0:      * @param array $from
Chris@0:      * @param array $to
Chris@0:      *
Chris@0:      * @return LongestCommonSubsequence
Chris@0:      */
Chris@0:     private function selectLcsImplementation(array $from, array $to)
Chris@0:     {
Chris@0:         // We do not want to use the time-efficient implementation if its memory
Chris@0:         // footprint will probably exceed this value. Note that the footprint
Chris@0:         // calculation is only an estimation for the matrix and the LCS method
Chris@0:         // will typically allocate a bit more memory than this.
Chris@0:         $memoryLimit = 100 * 1024 * 1024;
Chris@0: 
Chris@0:         if ($this->calculateEstimatedFootprint($from, $to) > $memoryLimit) {
Chris@0:             return new MemoryEfficientImplementation;
Chris@0:         }
Chris@0: 
Chris@0:         return new TimeEfficientImplementation;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Calculates the estimated memory footprint for the DP-based method.
Chris@0:      *
Chris@0:      * @param array $from
Chris@0:      * @param array $to
Chris@0:      *
Chris@12:      * @return int|float
Chris@0:      */
Chris@0:     private function calculateEstimatedFootprint(array $from, array $to)
Chris@0:     {
Chris@12:         $itemSize = PHP_INT_SIZE === 4 ? 76 : 144;
Chris@0: 
Chris@12:         return $itemSize * \pow(\min(\count($from), \count($to)), 2);
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@12:      * Returns true if line ends don't match on fromMatches and toMatches.
Chris@12:      *
Chris@12:      * @param array $fromMatches
Chris@12:      * @param array $toMatches
Chris@12:      *
Chris@12:      * @return bool
Chris@12:      */
Chris@12:     private function detectUnmatchedLineEndings(array $fromMatches, array $toMatches)
Chris@12:     {
Chris@12:         return isset($fromMatches[0], $toMatches[0]) &&
Chris@12:                \count($fromMatches[0]) === \count($toMatches[0]) &&
Chris@12:                $fromMatches[0] !== $toMatches[0];
Chris@12:     }
Chris@12: 
Chris@12:     /**
Chris@12:      * @param array $from
Chris@12:      * @param array $to
Chris@12:      *
Chris@12:      * @return array
Chris@12:      */
Chris@12:     private static function getArrayDiffParted(array &$from, array &$to)
Chris@12:     {
Chris@12:         $start = array();
Chris@12:         $end   = array();
Chris@12: 
Chris@12:         \reset($to);
Chris@12: 
Chris@12:         foreach ($from as $k => $v) {
Chris@12:             $toK = \key($to);
Chris@12: 
Chris@12:             if ($toK === $k && $v === $to[$k]) {
Chris@12:                 $start[$k] = $v;
Chris@12: 
Chris@12:                 unset($from[$k], $to[$k]);
Chris@12:             } else {
Chris@12:                 break;
Chris@12:             }
Chris@12:         }
Chris@12: 
Chris@12:         \end($from);
Chris@12:         \end($to);
Chris@12: 
Chris@12:         do {
Chris@12:             $fromK = \key($from);
Chris@12:             $toK   = \key($to);
Chris@12: 
Chris@12:             if (null === $fromK || null === $toK || \current($from) !== \current($to)) {
Chris@12:                 break;
Chris@12:             }
Chris@12: 
Chris@12:             \prev($from);
Chris@12:             \prev($to);
Chris@12: 
Chris@12:             $end = array($fromK => $from[$fromK]) + $end;
Chris@12:             unset($from[$fromK], $to[$toK]);
Chris@12:         } while (true);
Chris@12: 
Chris@12:         return array($from, $to, $start, $end);
Chris@0:     }
Chris@0: }