Chris@0: <?php
Chris@0: 
Chris@0: /*
Chris@0:  * This file is part of the Symfony package.
Chris@0:  *
Chris@0:  * (c) Fabien Potencier <fabien@symfony.com>
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 Symfony\Component\Yaml;
Chris@0: 
Chris@0: use Symfony\Component\Yaml\Exception\ParseException;
Chris@0: 
Chris@0: /**
Chris@0:  * Parser parses YAML strings to convert them to PHP arrays.
Chris@0:  *
Chris@0:  * @author Fabien Potencier <fabien@symfony.com>
Chris@0:  */
Chris@0: class Parser
Chris@0: {
Chris@0:     const TAG_PATTERN = '((?P<tag>![\w!.\/:-]+) +)?';
Chris@0:     const BLOCK_SCALAR_HEADER_PATTERN = '(?P<separator>\||>)(?P<modifiers>\+|\-|\d+|\+\d+|\-\d+|\d+\+|\d+\-)?(?P<comments> +#.*)?';
Chris@0: 
Chris@0:     private $offset = 0;
Chris@0:     private $totalNumberOfLines;
Chris@0:     private $lines = array();
Chris@0:     private $currentLineNb = -1;
Chris@0:     private $currentLine = '';
Chris@0:     private $refs = array();
Chris@0:     private $skippedLineNumbers = array();
Chris@0:     private $locallySkippedLineNumbers = array();
Chris@0: 
Chris@0:     /**
Chris@0:      * Constructor.
Chris@0:      *
Chris@0:      * @param int      $offset             The offset of YAML document (used for line numbers in error messages)
Chris@0:      * @param int|null $totalNumberOfLines The overall number of lines being parsed
Chris@0:      * @param int[]    $skippedLineNumbers Number of comment lines that have been skipped by the parser
Chris@0:      */
Chris@0:     public function __construct($offset = 0, $totalNumberOfLines = null, array $skippedLineNumbers = array())
Chris@0:     {
Chris@0:         $this->offset = $offset;
Chris@0:         $this->totalNumberOfLines = $totalNumberOfLines;
Chris@0:         $this->skippedLineNumbers = $skippedLineNumbers;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses a YAML string to a PHP value.
Chris@0:      *
Chris@0:      * @param string $value A YAML string
Chris@0:      * @param int    $flags A bit field of PARSE_* constants to customize the YAML parser behavior
Chris@0:      *
Chris@0:      * @return mixed A PHP value
Chris@0:      *
Chris@0:      * @throws ParseException If the YAML is not valid
Chris@0:      */
Chris@0:     public function parse($value, $flags = 0)
Chris@0:     {
Chris@0:         if (is_bool($flags)) {
Chris@0:             @trigger_error('Passing a boolean flag to toggle exception handling is deprecated since version 3.1 and will be removed in 4.0. Use the Yaml::PARSE_EXCEPTION_ON_INVALID_TYPE flag instead.', E_USER_DEPRECATED);
Chris@0: 
Chris@0:             if ($flags) {
Chris@0:                 $flags = Yaml::PARSE_EXCEPTION_ON_INVALID_TYPE;
Chris@0:             } else {
Chris@0:                 $flags = 0;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if (func_num_args() >= 3) {
Chris@0:             @trigger_error('Passing a boolean flag to toggle object support is deprecated since version 3.1 and will be removed in 4.0. Use the Yaml::PARSE_OBJECT flag instead.', E_USER_DEPRECATED);
Chris@0: 
Chris@0:             if (func_get_arg(2)) {
Chris@0:                 $flags |= Yaml::PARSE_OBJECT;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if (func_num_args() >= 4) {
Chris@0:             @trigger_error('Passing a boolean flag to toggle object for map support is deprecated since version 3.1 and will be removed in 4.0. Use the Yaml::PARSE_OBJECT_FOR_MAP flag instead.', E_USER_DEPRECATED);
Chris@0: 
Chris@0:             if (func_get_arg(3)) {
Chris@0:                 $flags |= Yaml::PARSE_OBJECT_FOR_MAP;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if (false === preg_match('//u', $value)) {
Chris@0:             throw new ParseException('The YAML value does not appear to be valid UTF-8.');
Chris@0:         }
Chris@0: 
Chris@0:         $this->refs = array();
Chris@0: 
Chris@0:         $mbEncoding = null;
Chris@0:         $e = null;
Chris@0:         $data = null;
Chris@0: 
Chris@0:         if (2 /* MB_OVERLOAD_STRING */ & (int) ini_get('mbstring.func_overload')) {
Chris@0:             $mbEncoding = mb_internal_encoding();
Chris@0:             mb_internal_encoding('UTF-8');
Chris@0:         }
Chris@0: 
Chris@0:         try {
Chris@0:             $data = $this->doParse($value, $flags);
Chris@0:         } catch (\Exception $e) {
Chris@0:         } catch (\Throwable $e) {
Chris@0:         }
Chris@0: 
Chris@0:         if (null !== $mbEncoding) {
Chris@0:             mb_internal_encoding($mbEncoding);
Chris@0:         }
Chris@0: 
Chris@0:         $this->lines = array();
Chris@0:         $this->currentLine = '';
Chris@0:         $this->refs = array();
Chris@0:         $this->skippedLineNumbers = array();
Chris@0:         $this->locallySkippedLineNumbers = array();
Chris@0: 
Chris@0:         if (null !== $e) {
Chris@0:             throw $e;
Chris@0:         }
Chris@0: 
Chris@0:         return $data;
Chris@0:     }
Chris@0: 
Chris@0:     private function doParse($value, $flags)
Chris@0:     {
Chris@0:         $this->currentLineNb = -1;
Chris@0:         $this->currentLine = '';
Chris@0:         $value = $this->cleanup($value);
Chris@0:         $this->lines = explode("\n", $value);
Chris@0:         $this->locallySkippedLineNumbers = array();
Chris@0: 
Chris@0:         if (null === $this->totalNumberOfLines) {
Chris@0:             $this->totalNumberOfLines = count($this->lines);
Chris@0:         }
Chris@0: 
Chris@0:         $data = array();
Chris@0:         $context = null;
Chris@0:         $allowOverwrite = false;
Chris@0: 
Chris@0:         while ($this->moveToNextLine()) {
Chris@0:             if ($this->isCurrentLineEmpty()) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             // tab?
Chris@0:             if ("\t" === $this->currentLine[0]) {
Chris@0:                 throw new ParseException('A YAML file cannot contain tabs as indentation.', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:             }
Chris@0: 
Chris@0:             $isRef = $mergeNode = false;
Chris@0:             if (self::preg_match('#^\-((?P<leadspaces>\s+)(?P<value>.+))?$#u', rtrim($this->currentLine), $values)) {
Chris@0:                 if ($context && 'mapping' == $context) {
Chris@0:                     throw new ParseException('You cannot define a sequence item when in a mapping', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:                 }
Chris@0:                 $context = 'sequence';
Chris@0: 
Chris@0:                 if (isset($values['value']) && self::preg_match('#^&(?P<ref>[^ ]+) *(?P<value>.*)#u', $values['value'], $matches)) {
Chris@0:                     $isRef = $matches['ref'];
Chris@0:                     $values['value'] = $matches['value'];
Chris@0:                 }
Chris@0: 
Chris@0:                 // array
Chris@0:                 if (!isset($values['value']) || '' == trim($values['value'], ' ') || 0 === strpos(ltrim($values['value'], ' '), '#')) {
Chris@0:                     $data[] = $this->parseBlock($this->getRealCurrentLineNb() + 1, $this->getNextEmbedBlock(null, true), $flags);
Chris@0:                 } else {
Chris@0:                     if (isset($values['leadspaces'])
Chris@0:                         && self::preg_match('#^(?P<key>'.Inline::REGEX_QUOTED_STRING.'|[^ \'"\{\[].*?) *\:(\s+(?P<value>.+))?$#u', rtrim($values['value']), $matches)
Chris@0:                     ) {
Chris@0:                         // this is a compact notation element, add to next block and parse
Chris@0:                         $block = $values['value'];
Chris@0:                         if ($this->isNextLineIndented()) {
Chris@0:                             $block .= "\n".$this->getNextEmbedBlock($this->getCurrentLineIndentation() + strlen($values['leadspaces']) + 1);
Chris@0:                         }
Chris@0: 
Chris@0:                         $data[] = $this->parseBlock($this->getRealCurrentLineNb(), $block, $flags);
Chris@0:                     } else {
Chris@0:                         $data[] = $this->parseValue($values['value'], $flags, $context);
Chris@0:                     }
Chris@0:                 }
Chris@0:                 if ($isRef) {
Chris@0:                     $this->refs[$isRef] = end($data);
Chris@0:                 }
Chris@0:             } elseif (
Chris@0:                 self::preg_match('#^(?P<key>'.Inline::REGEX_QUOTED_STRING.'|[^ \'"\[\{].*?) *\:(\s+(?P<value>.+))?$#u', rtrim($this->currentLine), $values)
Chris@0:                 && (false === strpos($values['key'], ' #') || in_array($values['key'][0], array('"', "'")))
Chris@0:             ) {
Chris@0:                 if ($context && 'sequence' == $context) {
Chris@0:                     throw new ParseException('You cannot define a mapping item when in a sequence', $this->currentLineNb + 1, $this->currentLine);
Chris@0:                 }
Chris@0:                 $context = 'mapping';
Chris@0: 
Chris@0:                 // force correct settings
Chris@0:                 Inline::parse(null, $flags, $this->refs);
Chris@0:                 try {
Chris@0:                     Inline::$parsedLineNumber = $this->getRealCurrentLineNb();
Chris@0:                     $key = Inline::parseScalar($values['key']);
Chris@0:                 } catch (ParseException $e) {
Chris@0:                     $e->setParsedLine($this->getRealCurrentLineNb() + 1);
Chris@0:                     $e->setSnippet($this->currentLine);
Chris@0: 
Chris@0:                     throw $e;
Chris@0:                 }
Chris@0: 
Chris@0:                 // Convert float keys to strings, to avoid being converted to integers by PHP
Chris@0:                 if (is_float($key)) {
Chris@0:                     $key = (string) $key;
Chris@0:                 }
Chris@0: 
Chris@0:                 if ('<<' === $key) {
Chris@0:                     $mergeNode = true;
Chris@0:                     $allowOverwrite = true;
Chris@0:                     if (isset($values['value']) && 0 === strpos($values['value'], '*')) {
Chris@0:                         $refName = substr($values['value'], 1);
Chris@0:                         if (!array_key_exists($refName, $this->refs)) {
Chris@0:                             throw new ParseException(sprintf('Reference "%s" does not exist.', $refName), $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:                         }
Chris@0: 
Chris@0:                         $refValue = $this->refs[$refName];
Chris@0: 
Chris@0:                         if (!is_array($refValue)) {
Chris@0:                             throw new ParseException('YAML merge keys used with a scalar value instead of an array.', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:                         }
Chris@0: 
Chris@0:                         $data += $refValue; // array union
Chris@0:                     } else {
Chris@0:                         if (isset($values['value']) && $values['value'] !== '') {
Chris@0:                             $value = $values['value'];
Chris@0:                         } else {
Chris@0:                             $value = $this->getNextEmbedBlock();
Chris@0:                         }
Chris@0:                         $parsed = $this->parseBlock($this->getRealCurrentLineNb() + 1, $value, $flags);
Chris@0: 
Chris@0:                         if (!is_array($parsed)) {
Chris@0:                             throw new ParseException('YAML merge keys used with a scalar value instead of an array.', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:                         }
Chris@0: 
Chris@0:                         if (isset($parsed[0])) {
Chris@0:                             // If the value associated with the merge key is a sequence, then this sequence is expected to contain mapping nodes
Chris@0:                             // and each of these nodes is merged in turn according to its order in the sequence. Keys in mapping nodes earlier
Chris@0:                             // in the sequence override keys specified in later mapping nodes.
Chris@0:                             foreach ($parsed as $parsedItem) {
Chris@0:                                 if (!is_array($parsedItem)) {
Chris@0:                                     throw new ParseException('Merge items must be arrays.', $this->getRealCurrentLineNb() + 1, $parsedItem);
Chris@0:                                 }
Chris@0: 
Chris@0:                                 $data += $parsedItem; // array union
Chris@0:                             }
Chris@0:                         } else {
Chris@0:                             // If the value associated with the key is a single mapping node, each of its key/value pairs is inserted into the
Chris@0:                             // current mapping, unless the key already exists in it.
Chris@0:                             $data += $parsed; // array union
Chris@0:                         }
Chris@0:                     }
Chris@0:                 } elseif (isset($values['value']) && self::preg_match('#^&(?P<ref>[^ ]+) *(?P<value>.*)#u', $values['value'], $matches)) {
Chris@0:                     $isRef = $matches['ref'];
Chris@0:                     $values['value'] = $matches['value'];
Chris@0:                 }
Chris@0: 
Chris@0:                 if ($mergeNode) {
Chris@0:                     // Merge keys
Chris@0:                 } elseif (!isset($values['value']) || '' == trim($values['value'], ' ') || 0 === strpos(ltrim($values['value'], ' '), '#')) {
Chris@0:                     // hash
Chris@0:                     // if next line is less indented or equal, then it means that the current value is null
Chris@0:                     if (!$this->isNextLineIndented() && !$this->isNextLineUnIndentedCollection()) {
Chris@0:                         // Spec: Keys MUST be unique; first one wins.
Chris@0:                         // But overwriting is allowed when a merge node is used in current block.
Chris@0:                         if ($allowOverwrite || !isset($data[$key])) {
Chris@0:                             $data[$key] = null;
Chris@0:                         } else {
Chris@0:                             @trigger_error(sprintf('Duplicate key "%s" detected on line %d whilst parsing YAML. Silent handling of duplicate mapping keys in YAML is deprecated since version 3.2 and will throw \Symfony\Component\Yaml\Exception\ParseException in 4.0.', $key, $this->getRealCurrentLineNb() + 1), E_USER_DEPRECATED);
Chris@0:                         }
Chris@0:                     } else {
Chris@0:                         // remember the parsed line number here in case we need it to provide some contexts in error messages below
Chris@0:                         $realCurrentLineNbKey = $this->getRealCurrentLineNb();
Chris@0:                         $value = $this->parseBlock($this->getRealCurrentLineNb() + 1, $this->getNextEmbedBlock(), $flags);
Chris@0:                         // Spec: Keys MUST be unique; first one wins.
Chris@0:                         // But overwriting is allowed when a merge node is used in current block.
Chris@0:                         if ($allowOverwrite || !isset($data[$key])) {
Chris@0:                             $data[$key] = $value;
Chris@0:                         } else {
Chris@0:                             @trigger_error(sprintf('Duplicate key "%s" detected on line %d whilst parsing YAML. Silent handling of duplicate mapping keys in YAML is deprecated since version 3.2 and will throw \Symfony\Component\Yaml\Exception\ParseException in 4.0.', $key, $realCurrentLineNbKey + 1), E_USER_DEPRECATED);
Chris@0:                         }
Chris@0:                     }
Chris@0:                 } else {
Chris@0:                     $value = $this->parseValue($values['value'], $flags, $context);
Chris@0:                     // Spec: Keys MUST be unique; first one wins.
Chris@0:                     // But overwriting is allowed when a merge node is used in current block.
Chris@0:                     if ($allowOverwrite || !isset($data[$key])) {
Chris@0:                         $data[$key] = $value;
Chris@0:                     } else {
Chris@0:                         @trigger_error(sprintf('Duplicate key "%s" detected on line %d whilst parsing YAML. Silent handling of duplicate mapping keys in YAML is deprecated since version 3.2 and will throw \Symfony\Component\Yaml\Exception\ParseException in 4.0.', $key, $this->getRealCurrentLineNb() + 1), E_USER_DEPRECATED);
Chris@0:                     }
Chris@0:                 }
Chris@0:                 if ($isRef) {
Chris@0:                     $this->refs[$isRef] = $data[$key];
Chris@0:                 }
Chris@0:             } else {
Chris@0:                 // multiple documents are not supported
Chris@0:                 if ('---' === $this->currentLine) {
Chris@0:                     throw new ParseException('Multiple documents are not supported.', $this->currentLineNb + 1, $this->currentLine);
Chris@0:                 }
Chris@0: 
Chris@0:                 // 1-liner optionally followed by newline(s)
Chris@0:                 if (is_string($value) && $this->lines[0] === trim($value)) {
Chris@0:                     try {
Chris@0:                         Inline::$parsedLineNumber = $this->getRealCurrentLineNb();
Chris@0:                         $value = Inline::parse($this->lines[0], $flags, $this->refs);
Chris@0:                     } catch (ParseException $e) {
Chris@0:                         $e->setParsedLine($this->getRealCurrentLineNb() + 1);
Chris@0:                         $e->setSnippet($this->currentLine);
Chris@0: 
Chris@0:                         throw $e;
Chris@0:                     }
Chris@0: 
Chris@0:                     return $value;
Chris@0:                 }
Chris@0: 
Chris@0:                 throw new ParseException('Unable to parse.', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if (Yaml::PARSE_OBJECT_FOR_MAP & $flags && !is_object($data) && 'mapping' === $context) {
Chris@0:             $object = new \stdClass();
Chris@0: 
Chris@0:             foreach ($data as $key => $value) {
Chris@0:                 $object->$key = $value;
Chris@0:             }
Chris@0: 
Chris@0:             $data = $object;
Chris@0:         }
Chris@0: 
Chris@0:         return empty($data) ? null : $data;
Chris@0:     }
Chris@0: 
Chris@0:     private function parseBlock($offset, $yaml, $flags)
Chris@0:     {
Chris@0:         $skippedLineNumbers = $this->skippedLineNumbers;
Chris@0: 
Chris@0:         foreach ($this->locallySkippedLineNumbers as $lineNumber) {
Chris@0:             if ($lineNumber < $offset) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             $skippedLineNumbers[] = $lineNumber;
Chris@0:         }
Chris@0: 
Chris@0:         $parser = new self($offset, $this->totalNumberOfLines, $skippedLineNumbers);
Chris@0:         $parser->refs = &$this->refs;
Chris@0: 
Chris@0:         return $parser->doParse($yaml, $flags);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the current line number (takes the offset into account).
Chris@0:      *
Chris@0:      * @return int The current line number
Chris@0:      */
Chris@0:     private function getRealCurrentLineNb()
Chris@0:     {
Chris@0:         $realCurrentLineNumber = $this->currentLineNb + $this->offset;
Chris@0: 
Chris@0:         foreach ($this->skippedLineNumbers as $skippedLineNumber) {
Chris@0:             if ($skippedLineNumber > $realCurrentLineNumber) {
Chris@0:                 break;
Chris@0:             }
Chris@0: 
Chris@0:             ++$realCurrentLineNumber;
Chris@0:         }
Chris@0: 
Chris@0:         return $realCurrentLineNumber;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the current line indentation.
Chris@0:      *
Chris@0:      * @return int The current line indentation
Chris@0:      */
Chris@0:     private function getCurrentLineIndentation()
Chris@0:     {
Chris@0:         return strlen($this->currentLine) - strlen(ltrim($this->currentLine, ' '));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the next embed block of YAML.
Chris@0:      *
Chris@0:      * @param int  $indentation The indent level at which the block is to be read, or null for default
Chris@0:      * @param bool $inSequence  True if the enclosing data structure is a sequence
Chris@0:      *
Chris@0:      * @return string A YAML string
Chris@0:      *
Chris@0:      * @throws ParseException When indentation problem are detected
Chris@0:      */
Chris@0:     private function getNextEmbedBlock($indentation = null, $inSequence = false)
Chris@0:     {
Chris@0:         $oldLineIndentation = $this->getCurrentLineIndentation();
Chris@0:         $blockScalarIndentations = array();
Chris@0: 
Chris@0:         if ($this->isBlockScalarHeader()) {
Chris@0:             $blockScalarIndentations[] = $this->getCurrentLineIndentation();
Chris@0:         }
Chris@0: 
Chris@0:         if (!$this->moveToNextLine()) {
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         if (null === $indentation) {
Chris@0:             $newIndent = $this->getCurrentLineIndentation();
Chris@0: 
Chris@0:             $unindentedEmbedBlock = $this->isStringUnIndentedCollectionItem();
Chris@0: 
Chris@0:             if (!$this->isCurrentLineEmpty() && 0 === $newIndent && !$unindentedEmbedBlock) {
Chris@0:                 throw new ParseException('Indentation problem.', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:             }
Chris@0:         } else {
Chris@0:             $newIndent = $indentation;
Chris@0:         }
Chris@0: 
Chris@0:         $data = array();
Chris@0:         if ($this->getCurrentLineIndentation() >= $newIndent) {
Chris@0:             $data[] = substr($this->currentLine, $newIndent);
Chris@0:         } else {
Chris@0:             $this->moveToPreviousLine();
Chris@0: 
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         if ($inSequence && $oldLineIndentation === $newIndent && isset($data[0][0]) && '-' === $data[0][0]) {
Chris@0:             // the previous line contained a dash but no item content, this line is a sequence item with the same indentation
Chris@0:             // and therefore no nested list or mapping
Chris@0:             $this->moveToPreviousLine();
Chris@0: 
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         $isItUnindentedCollection = $this->isStringUnIndentedCollectionItem();
Chris@0: 
Chris@0:         if (empty($blockScalarIndentations) && $this->isBlockScalarHeader()) {
Chris@0:             $blockScalarIndentations[] = $this->getCurrentLineIndentation();
Chris@0:         }
Chris@0: 
Chris@0:         $previousLineIndentation = $this->getCurrentLineIndentation();
Chris@0: 
Chris@0:         while ($this->moveToNextLine()) {
Chris@0:             $indent = $this->getCurrentLineIndentation();
Chris@0: 
Chris@0:             // terminate all block scalars that are more indented than the current line
Chris@0:             if (!empty($blockScalarIndentations) && $indent < $previousLineIndentation && trim($this->currentLine) !== '') {
Chris@0:                 foreach ($blockScalarIndentations as $key => $blockScalarIndentation) {
Chris@0:                     if ($blockScalarIndentation >= $this->getCurrentLineIndentation()) {
Chris@0:                         unset($blockScalarIndentations[$key]);
Chris@0:                     }
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             if (empty($blockScalarIndentations) && !$this->isCurrentLineComment() && $this->isBlockScalarHeader()) {
Chris@0:                 $blockScalarIndentations[] = $this->getCurrentLineIndentation();
Chris@0:             }
Chris@0: 
Chris@0:             $previousLineIndentation = $indent;
Chris@0: 
Chris@0:             if ($isItUnindentedCollection && !$this->isCurrentLineEmpty() && !$this->isStringUnIndentedCollectionItem() && $newIndent === $indent) {
Chris@0:                 $this->moveToPreviousLine();
Chris@0:                 break;
Chris@0:             }
Chris@0: 
Chris@0:             if ($this->isCurrentLineBlank()) {
Chris@0:                 $data[] = substr($this->currentLine, $newIndent);
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             // we ignore "comment" lines only when we are not inside a scalar block
Chris@0:             if (empty($blockScalarIndentations) && $this->isCurrentLineComment()) {
Chris@0:                 // remember ignored comment lines (they are used later in nested
Chris@0:                 // parser calls to determine real line numbers)
Chris@0:                 //
Chris@0:                 // CAUTION: beware to not populate the global property here as it
Chris@0:                 // will otherwise influence the getRealCurrentLineNb() call here
Chris@0:                 // for consecutive comment lines and subsequent embedded blocks
Chris@0:                 $this->locallySkippedLineNumbers[] = $this->getRealCurrentLineNb();
Chris@0: 
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             if ($indent >= $newIndent) {
Chris@0:                 $data[] = substr($this->currentLine, $newIndent);
Chris@0:             } elseif (0 == $indent) {
Chris@0:                 $this->moveToPreviousLine();
Chris@0: 
Chris@0:                 break;
Chris@0:             } else {
Chris@0:                 throw new ParseException('Indentation problem.', $this->getRealCurrentLineNb() + 1, $this->currentLine);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return implode("\n", $data);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Moves the parser to the next line.
Chris@0:      *
Chris@0:      * @return bool
Chris@0:      */
Chris@0:     private function moveToNextLine()
Chris@0:     {
Chris@0:         if ($this->currentLineNb >= count($this->lines) - 1) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $this->currentLine = $this->lines[++$this->currentLineNb];
Chris@0: 
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Moves the parser to the previous line.
Chris@0:      *
Chris@0:      * @return bool
Chris@0:      */
Chris@0:     private function moveToPreviousLine()
Chris@0:     {
Chris@0:         if ($this->currentLineNb < 1) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $this->currentLine = $this->lines[--$this->currentLineNb];
Chris@0: 
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses a YAML value.
Chris@0:      *
Chris@0:      * @param string $value   A YAML value
Chris@0:      * @param int    $flags   A bit field of PARSE_* constants to customize the YAML parser behavior
Chris@0:      * @param string $context The parser context (either sequence or mapping)
Chris@0:      *
Chris@0:      * @return mixed A PHP value
Chris@0:      *
Chris@0:      * @throws ParseException When reference does not exist
Chris@0:      */
Chris@0:     private function parseValue($value, $flags, $context)
Chris@0:     {
Chris@0:         if (0 === strpos($value, '*')) {
Chris@0:             if (false !== $pos = strpos($value, '#')) {
Chris@0:                 $value = substr($value, 1, $pos - 2);
Chris@0:             } else {
Chris@0:                 $value = substr($value, 1);
Chris@0:             }
Chris@0: 
Chris@0:             if (!array_key_exists($value, $this->refs)) {
Chris@0:                 throw new ParseException(sprintf('Reference "%s" does not exist.', $value), $this->currentLineNb + 1, $this->currentLine);
Chris@0:             }
Chris@0: 
Chris@0:             return $this->refs[$value];
Chris@0:         }
Chris@0: 
Chris@0:         if (self::preg_match('/^'.self::TAG_PATTERN.self::BLOCK_SCALAR_HEADER_PATTERN.'$/', $value, $matches)) {
Chris@0:             $modifiers = isset($matches['modifiers']) ? $matches['modifiers'] : '';
Chris@0: 
Chris@0:             $data = $this->parseBlockScalar($matches['separator'], preg_replace('#\d+#', '', $modifiers), (int) abs($modifiers));
Chris@0: 
Chris@0:             if (isset($matches['tag']) && '!!binary' === $matches['tag']) {
Chris@0:                 return Inline::evaluateBinaryScalar($data);
Chris@0:             }
Chris@0: 
Chris@0:             return $data;
Chris@0:         }
Chris@0: 
Chris@0:         try {
Chris@0:             $quotation = '' !== $value && ('"' === $value[0] || "'" === $value[0]) ? $value[0] : null;
Chris@0: 
Chris@0:             // do not take following lines into account when the current line is a quoted single line value
Chris@0:             if (null !== $quotation && preg_match('/^'.$quotation.'.*'.$quotation.'(\s*#.*)?$/', $value)) {
Chris@0:                 return Inline::parse($value, $flags, $this->refs);
Chris@0:             }
Chris@0: 
Chris@0:             while ($this->moveToNextLine()) {
Chris@0:                 // unquoted strings end before the first unindented line
Chris@0:                 if (null === $quotation && $this->getCurrentLineIndentation() === 0) {
Chris@0:                     $this->moveToPreviousLine();
Chris@0: 
Chris@0:                     break;
Chris@0:                 }
Chris@0: 
Chris@0:                 $value .= ' '.trim($this->currentLine);
Chris@0: 
Chris@0:                 // quoted string values end with a line that is terminated with the quotation character
Chris@0:                 if ('' !== $this->currentLine && substr($this->currentLine, -1) === $quotation) {
Chris@0:                     break;
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             Inline::$parsedLineNumber = $this->getRealCurrentLineNb();
Chris@0:             $parsedValue = Inline::parse($value, $flags, $this->refs);
Chris@0: 
Chris@0:             if ('mapping' === $context && is_string($parsedValue) && '"' !== $value[0] && "'" !== $value[0] && '[' !== $value[0] && '{' !== $value[0] && '!' !== $value[0] && false !== strpos($parsedValue, ': ')) {
Chris@0:                 throw new ParseException('A colon cannot be used in an unquoted mapping value.');
Chris@0:             }
Chris@0: 
Chris@0:             return $parsedValue;
Chris@0:         } catch (ParseException $e) {
Chris@0:             $e->setParsedLine($this->getRealCurrentLineNb() + 1);
Chris@0:             $e->setSnippet($this->currentLine);
Chris@0: 
Chris@0:             throw $e;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses a block scalar.
Chris@0:      *
Chris@0:      * @param string $style       The style indicator that was used to begin this block scalar (| or >)
Chris@0:      * @param string $chomping    The chomping indicator that was used to begin this block scalar (+ or -)
Chris@0:      * @param int    $indentation The indentation indicator that was used to begin this block scalar
Chris@0:      *
Chris@0:      * @return string The text value
Chris@0:      */
Chris@0:     private function parseBlockScalar($style, $chomping = '', $indentation = 0)
Chris@0:     {
Chris@0:         $notEOF = $this->moveToNextLine();
Chris@0:         if (!$notEOF) {
Chris@0:             return '';
Chris@0:         }
Chris@0: 
Chris@0:         $isCurrentLineBlank = $this->isCurrentLineBlank();
Chris@0:         $blockLines = array();
Chris@0: 
Chris@0:         // leading blank lines are consumed before determining indentation
Chris@0:         while ($notEOF && $isCurrentLineBlank) {
Chris@0:             // newline only if not EOF
Chris@0:             if ($notEOF = $this->moveToNextLine()) {
Chris@0:                 $blockLines[] = '';
Chris@0:                 $isCurrentLineBlank = $this->isCurrentLineBlank();
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         // determine indentation if not specified
Chris@0:         if (0 === $indentation) {
Chris@0:             if (self::preg_match('/^ +/', $this->currentLine, $matches)) {
Chris@0:                 $indentation = strlen($matches[0]);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if ($indentation > 0) {
Chris@0:             $pattern = sprintf('/^ {%d}(.*)$/', $indentation);
Chris@0: 
Chris@0:             while (
Chris@0:                 $notEOF && (
Chris@0:                     $isCurrentLineBlank ||
Chris@0:                     self::preg_match($pattern, $this->currentLine, $matches)
Chris@0:                 )
Chris@0:             ) {
Chris@0:                 if ($isCurrentLineBlank && strlen($this->currentLine) > $indentation) {
Chris@0:                     $blockLines[] = substr($this->currentLine, $indentation);
Chris@0:                 } elseif ($isCurrentLineBlank) {
Chris@0:                     $blockLines[] = '';
Chris@0:                 } else {
Chris@0:                     $blockLines[] = $matches[1];
Chris@0:                 }
Chris@0: 
Chris@0:                 // newline only if not EOF
Chris@0:                 if ($notEOF = $this->moveToNextLine()) {
Chris@0:                     $isCurrentLineBlank = $this->isCurrentLineBlank();
Chris@0:                 }
Chris@0:             }
Chris@0:         } elseif ($notEOF) {
Chris@0:             $blockLines[] = '';
Chris@0:         }
Chris@0: 
Chris@0:         if ($notEOF) {
Chris@0:             $blockLines[] = '';
Chris@0:             $this->moveToPreviousLine();
Chris@0:         } elseif (!$notEOF && !$this->isCurrentLineLastLineInDocument()) {
Chris@0:             $blockLines[] = '';
Chris@0:         }
Chris@0: 
Chris@0:         // folded style
Chris@0:         if ('>' === $style) {
Chris@0:             $text = '';
Chris@0:             $previousLineIndented = false;
Chris@0:             $previousLineBlank = false;
Chris@0: 
Chris@0:             for ($i = 0, $blockLinesCount = count($blockLines); $i < $blockLinesCount; ++$i) {
Chris@0:                 if ('' === $blockLines[$i]) {
Chris@0:                     $text .= "\n";
Chris@0:                     $previousLineIndented = false;
Chris@0:                     $previousLineBlank = true;
Chris@0:                 } elseif (' ' === $blockLines[$i][0]) {
Chris@0:                     $text .= "\n".$blockLines[$i];
Chris@0:                     $previousLineIndented = true;
Chris@0:                     $previousLineBlank = false;
Chris@0:                 } elseif ($previousLineIndented) {
Chris@0:                     $text .= "\n".$blockLines[$i];
Chris@0:                     $previousLineIndented = false;
Chris@0:                     $previousLineBlank = false;
Chris@0:                 } elseif ($previousLineBlank || 0 === $i) {
Chris@0:                     $text .= $blockLines[$i];
Chris@0:                     $previousLineIndented = false;
Chris@0:                     $previousLineBlank = false;
Chris@0:                 } else {
Chris@0:                     $text .= ' '.$blockLines[$i];
Chris@0:                     $previousLineIndented = false;
Chris@0:                     $previousLineBlank = false;
Chris@0:                 }
Chris@0:             }
Chris@0:         } else {
Chris@0:             $text = implode("\n", $blockLines);
Chris@0:         }
Chris@0: 
Chris@0:         // deal with trailing newlines
Chris@0:         if ('' === $chomping) {
Chris@0:             $text = preg_replace('/\n+$/', "\n", $text);
Chris@0:         } elseif ('-' === $chomping) {
Chris@0:             $text = preg_replace('/\n+$/', '', $text);
Chris@0:         }
Chris@0: 
Chris@0:         return $text;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the next line is indented.
Chris@0:      *
Chris@0:      * @return bool Returns true if the next line is indented, false otherwise
Chris@0:      */
Chris@0:     private function isNextLineIndented()
Chris@0:     {
Chris@0:         $currentIndentation = $this->getCurrentLineIndentation();
Chris@0:         $EOF = !$this->moveToNextLine();
Chris@0: 
Chris@0:         while (!$EOF && $this->isCurrentLineEmpty()) {
Chris@0:             $EOF = !$this->moveToNextLine();
Chris@0:         }
Chris@0: 
Chris@0:         if ($EOF) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $ret = $this->getCurrentLineIndentation() > $currentIndentation;
Chris@0: 
Chris@0:         $this->moveToPreviousLine();
Chris@0: 
Chris@0:         return $ret;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the current line is blank or if it is a comment line.
Chris@0:      *
Chris@0:      * @return bool Returns true if the current line is empty or if it is a comment line, false otherwise
Chris@0:      */
Chris@0:     private function isCurrentLineEmpty()
Chris@0:     {
Chris@0:         return $this->isCurrentLineBlank() || $this->isCurrentLineComment();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the current line is blank.
Chris@0:      *
Chris@0:      * @return bool Returns true if the current line is blank, false otherwise
Chris@0:      */
Chris@0:     private function isCurrentLineBlank()
Chris@0:     {
Chris@0:         return '' == trim($this->currentLine, ' ');
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the current line is a comment line.
Chris@0:      *
Chris@0:      * @return bool Returns true if the current line is a comment line, false otherwise
Chris@0:      */
Chris@0:     private function isCurrentLineComment()
Chris@0:     {
Chris@0:         //checking explicitly the first char of the trim is faster than loops or strpos
Chris@0:         $ltrimmedLine = ltrim($this->currentLine, ' ');
Chris@0: 
Chris@0:         return '' !== $ltrimmedLine && $ltrimmedLine[0] === '#';
Chris@0:     }
Chris@0: 
Chris@0:     private function isCurrentLineLastLineInDocument()
Chris@0:     {
Chris@0:         return ($this->offset + $this->currentLineNb) >= ($this->totalNumberOfLines - 1);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Cleanups a YAML string to be parsed.
Chris@0:      *
Chris@0:      * @param string $value The input YAML string
Chris@0:      *
Chris@0:      * @return string A cleaned up YAML string
Chris@0:      */
Chris@0:     private function cleanup($value)
Chris@0:     {
Chris@0:         $value = str_replace(array("\r\n", "\r"), "\n", $value);
Chris@0: 
Chris@0:         // strip YAML header
Chris@0:         $count = 0;
Chris@0:         $value = preg_replace('#^\%YAML[: ][\d\.]+.*\n#u', '', $value, -1, $count);
Chris@0:         $this->offset += $count;
Chris@0: 
Chris@0:         // remove leading comments
Chris@0:         $trimmedValue = preg_replace('#^(\#.*?\n)+#s', '', $value, -1, $count);
Chris@0:         if ($count == 1) {
Chris@0:             // items have been removed, update the offset
Chris@0:             $this->offset += substr_count($value, "\n") - substr_count($trimmedValue, "\n");
Chris@0:             $value = $trimmedValue;
Chris@0:         }
Chris@0: 
Chris@0:         // remove start of the document marker (---)
Chris@0:         $trimmedValue = preg_replace('#^\-\-\-.*?\n#s', '', $value, -1, $count);
Chris@0:         if ($count == 1) {
Chris@0:             // items have been removed, update the offset
Chris@0:             $this->offset += substr_count($value, "\n") - substr_count($trimmedValue, "\n");
Chris@0:             $value = $trimmedValue;
Chris@0: 
Chris@0:             // remove end of the document marker (...)
Chris@0:             $value = preg_replace('#\.\.\.\s*$#', '', $value);
Chris@0:         }
Chris@0: 
Chris@0:         return $value;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the next line starts unindented collection.
Chris@0:      *
Chris@0:      * @return bool Returns true if the next line starts unindented collection, false otherwise
Chris@0:      */
Chris@0:     private function isNextLineUnIndentedCollection()
Chris@0:     {
Chris@0:         $currentIndentation = $this->getCurrentLineIndentation();
Chris@0:         $notEOF = $this->moveToNextLine();
Chris@0: 
Chris@0:         while ($notEOF && $this->isCurrentLineEmpty()) {
Chris@0:             $notEOF = $this->moveToNextLine();
Chris@0:         }
Chris@0: 
Chris@0:         if (false === $notEOF) {
Chris@0:             return false;
Chris@0:         }
Chris@0: 
Chris@0:         $ret = $this->getCurrentLineIndentation() === $currentIndentation && $this->isStringUnIndentedCollectionItem();
Chris@0: 
Chris@0:         $this->moveToPreviousLine();
Chris@0: 
Chris@0:         return $ret;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the string is un-indented collection item.
Chris@0:      *
Chris@0:      * @return bool Returns true if the string is un-indented collection item, false otherwise
Chris@0:      */
Chris@0:     private function isStringUnIndentedCollectionItem()
Chris@0:     {
Chris@0:         return '-' === rtrim($this->currentLine) || 0 === strpos($this->currentLine, '- ');
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Tests whether or not the current line is the header of a block scalar.
Chris@0:      *
Chris@0:      * @return bool
Chris@0:      */
Chris@0:     private function isBlockScalarHeader()
Chris@0:     {
Chris@0:         return (bool) self::preg_match('~'.self::BLOCK_SCALAR_HEADER_PATTERN.'$~', $this->currentLine);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * A local wrapper for `preg_match` which will throw a ParseException if there
Chris@0:      * is an internal error in the PCRE engine.
Chris@0:      *
Chris@0:      * This avoids us needing to check for "false" every time PCRE is used
Chris@0:      * in the YAML engine
Chris@0:      *
Chris@0:      * @throws ParseException on a PCRE internal error
Chris@0:      *
Chris@0:      * @see preg_last_error()
Chris@0:      *
Chris@0:      * @internal
Chris@0:      */
Chris@0:     public static function preg_match($pattern, $subject, &$matches = null, $flags = 0, $offset = 0)
Chris@0:     {
Chris@0:         if (false === $ret = preg_match($pattern, $subject, $matches, $flags, $offset)) {
Chris@0:             switch (preg_last_error()) {
Chris@0:                 case PREG_INTERNAL_ERROR:
Chris@0:                     $error = 'Internal PCRE error.';
Chris@0:                     break;
Chris@0:                 case PREG_BACKTRACK_LIMIT_ERROR:
Chris@0:                     $error = 'pcre.backtrack_limit reached.';
Chris@0:                     break;
Chris@0:                 case PREG_RECURSION_LIMIT_ERROR:
Chris@0:                     $error = 'pcre.recursion_limit reached.';
Chris@0:                     break;
Chris@0:                 case PREG_BAD_UTF8_ERROR:
Chris@0:                     $error = 'Malformed UTF-8 data.';
Chris@0:                     break;
Chris@0:                 case PREG_BAD_UTF8_OFFSET_ERROR:
Chris@0:                     $error = 'Offset doesn\'t correspond to the begin of a valid UTF-8 code point.';
Chris@0:                     break;
Chris@0:                 default:
Chris@0:                     $error = 'Error.';
Chris@0:             }
Chris@0: 
Chris@0:             throw new ParseException($error);
Chris@0:         }
Chris@0: 
Chris@0:         return $ret;
Chris@0:     }
Chris@0: }