Chris@0: <?php
Chris@0: 
Chris@0: namespace Drupal\Component\Gettext;
Chris@0: 
Chris@0: /**
Chris@0:  * Gettext PO header handler.
Chris@0:  *
Chris@0:  * Possible Gettext PO header elements are explained in
Chris@0:  * http://www.gnu.org/software/gettext/manual/gettext.html#Header-Entry,
Chris@0:  * but we only support a subset of these directly.
Chris@0:  *
Chris@0:  * Example header:
Chris@0:  *
Chris@0:  * "Project-Id-Version: Drupal core (7.11)\n"
Chris@0:  * "POT-Creation-Date: 2012-02-12 22:59+0000\n"
Chris@0:  * "PO-Revision-Date: YYYY-mm-DD HH:MM+ZZZZ\n"
Chris@0:  * "Language-Team: Catalan\n"
Chris@0:  * "MIME-Version: 1.0\n"
Chris@0:  * "Content-Type: text/plain; charset=utf-8\n"
Chris@0:  * "Content-Transfer-Encoding: 8bit\n"
Chris@0:  * "Plural-Forms: nplurals=2; plural=(n>1);\n"
Chris@0:  */
Chris@0: class PoHeader {
Chris@0: 
Chris@0:   /**
Chris@0:    * Language code.
Chris@0:    *
Chris@0:    * @var string
Chris@0:    */
Chris@17:   protected $langcode;
Chris@0: 
Chris@0:   /**
Chris@0:    * Formula for the plural form.
Chris@0:    *
Chris@0:    * @var string
Chris@0:    */
Chris@17:   protected $pluralForms;
Chris@0: 
Chris@0:   /**
Chris@0:    * Author(s) of the file.
Chris@0:    *
Chris@0:    * @var string
Chris@0:    */
Chris@17:   protected $authors;
Chris@0: 
Chris@0:   /**
Chris@0:    * Date the po file got created.
Chris@0:    *
Chris@0:    * @var string
Chris@0:    */
Chris@17:   protected $poDate;
Chris@0: 
Chris@0:   /**
Chris@0:    * Human readable language name.
Chris@0:    *
Chris@0:    * @var string
Chris@0:    */
Chris@17:   protected $languageName;
Chris@0: 
Chris@0:   /**
Chris@0:    * Name of the project the translation belongs to.
Chris@0:    *
Chris@0:    * @var string
Chris@0:    */
Chris@17:   protected $projectName;
Chris@0: 
Chris@0:   /**
Chris@0:    * Constructor, creates a PoHeader with default values.
Chris@0:    *
Chris@0:    * @param string $langcode
Chris@0:    *   Language code.
Chris@0:    */
Chris@0:   public function __construct($langcode = NULL) {
Chris@17:     $this->langcode = $langcode;
Chris@0:     // Ignore errors when run during site installation before
Chris@0:     // date_default_timezone_set() is called.
Chris@17:     $this->poDate = @date("Y-m-d H:iO");
Chris@17:     $this->pluralForms = 'nplurals=2; plural=(n > 1);';
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Gets the plural form.
Chris@0:    *
Chris@0:    * @return string
Chris@0:    *   Plural form component from the header, for example:
Chris@0:    *   'nplurals=2; plural=(n > 1);'.
Chris@0:    */
Chris@0:   public function getPluralForms() {
Chris@17:     return $this->pluralForms;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Set the human readable language name.
Chris@0:    *
Chris@0:    * @param string $languageName
Chris@0:    *   Human readable language name.
Chris@0:    */
Chris@0:   public function setLanguageName($languageName) {
Chris@17:     $this->languageName = $languageName;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Gets the human readable language name.
Chris@0:    *
Chris@0:    * @return string
Chris@0:    *   The human readable language name.
Chris@0:    */
Chris@0:   public function getLanguageName() {
Chris@17:     return $this->languageName;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Set the project name.
Chris@0:    *
Chris@0:    * @param string $projectName
Chris@0:    *   Human readable project name.
Chris@0:    */
Chris@0:   public function setProjectName($projectName) {
Chris@17:     $this->projectName = $projectName;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Gets the project name.
Chris@0:    *
Chris@0:    * @return string
Chris@0:    *   The human readable project name.
Chris@0:    */
Chris@0:   public function getProjectName() {
Chris@17:     return $this->projectName;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Populate internal values from a string.
Chris@0:    *
Chris@0:    * @param string $header
Chris@0:    *   Full header string with key-value pairs.
Chris@0:    */
Chris@0:   public function setFromString($header) {
Chris@0:     // Get an array of all header values for processing.
Chris@0:     $values = $this->parseHeader($header);
Chris@0: 
Chris@0:     // There is only one value relevant for our header implementation when
Chris@0:     // reading, and that is the plural formula.
Chris@0:     if (!empty($values['Plural-Forms'])) {
Chris@17:       $this->pluralForms = $values['Plural-Forms'];
Chris@0:     }
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Generate a Gettext PO formatted header string based on data set earlier.
Chris@0:    */
Chris@0:   public function __toString() {
Chris@0:     $output = '';
Chris@0: 
Chris@17:     $isTemplate = empty($this->languageName);
Chris@0: 
Chris@17:     $output .= '# ' . ($isTemplate ? 'LANGUAGE' : $this->languageName) . ' translation of ' . ($isTemplate ? 'PROJECT' : $this->projectName) . "\n";
Chris@17:     if (!empty($this->authors)) {
Chris@17:       $output .= '# Generated by ' . implode("\n# ", $this->authors) . "\n";
Chris@0:     }
Chris@0:     $output .= "#\n";
Chris@0: 
Chris@0:     // Add the actual header information.
Chris@0:     $output .= "msgid \"\"\n";
Chris@0:     $output .= "msgstr \"\"\n";
Chris@0:     $output .= "\"Project-Id-Version: PROJECT VERSION\\n\"\n";
Chris@17:     $output .= "\"POT-Creation-Date: " . $this->poDate . "\\n\"\n";
Chris@17:     $output .= "\"PO-Revision-Date: " . $this->poDate . "\\n\"\n";
Chris@0:     $output .= "\"Last-Translator: NAME <EMAIL@ADDRESS>\\n\"\n";
Chris@0:     $output .= "\"Language-Team: LANGUAGE <EMAIL@ADDRESS>\\n\"\n";
Chris@0:     $output .= "\"MIME-Version: 1.0\\n\"\n";
Chris@0:     $output .= "\"Content-Type: text/plain; charset=utf-8\\n\"\n";
Chris@0:     $output .= "\"Content-Transfer-Encoding: 8bit\\n\"\n";
Chris@17:     $output .= "\"Plural-Forms: " . $this->pluralForms . "\\n\"\n";
Chris@0:     $output .= "\n";
Chris@0: 
Chris@0:     return $output;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Parses a Plural-Forms entry from a Gettext Portable Object file header.
Chris@0:    *
Chris@0:    * @param string $pluralforms
Chris@0:    *   The Plural-Forms entry value.
Chris@0:    *
Chris@0:    * @return
Chris@0:    *   An indexed array of parsed plural formula data. Containing:
Chris@0:    *   - 'nplurals': The number of plural forms defined by the plural formula.
Chris@0:    *   - 'plurals': Array of plural positions keyed by plural value.
Chris@0:    *
Chris@14:    * @throws \Exception
Chris@0:    */
Chris@0:   public function parsePluralForms($pluralforms) {
Chris@0:     $plurals = [];
Chris@0:     // First, delete all whitespace.
Chris@0:     $pluralforms = strtr($pluralforms, [" " => "", "\t" => ""]);
Chris@0: 
Chris@0:     // Select the parts that define nplurals and plural.
Chris@0:     $nplurals = strstr($pluralforms, "nplurals=");
Chris@0:     if (strpos($nplurals, ";")) {
Chris@0:       // We want the string from the 10th char, because "nplurals=" length is 9.
Chris@0:       $nplurals = substr($nplurals, 9, strpos($nplurals, ";") - 9);
Chris@0:     }
Chris@0:     else {
Chris@0:       return FALSE;
Chris@0:     }
Chris@0:     $plural = strstr($pluralforms, "plural=");
Chris@0:     if (strpos($plural, ";")) {
Chris@0:       // We want the string from the 8th char, because "plural=" length is 7.
Chris@0:       $plural = substr($plural, 7, strpos($plural, ";") - 7);
Chris@0:     }
Chris@0:     else {
Chris@0:       return FALSE;
Chris@0:     }
Chris@0: 
Chris@0:     // If the number of plurals is zero, we return a default result.
Chris@0:     if ($nplurals == 0) {
Chris@0:       return [$nplurals, ['default' => 0]];
Chris@0:     }
Chris@0: 
Chris@0:     // Calculate possible plural positions of different plural values. All known
Chris@0:     // plural formula's are repetitive above 100.
Chris@0:     // For data compression we store the last position the array value
Chris@0:     // changes and store it as default.
Chris@0:     $element_stack = $this->parseArithmetic($plural);
Chris@0:     if ($element_stack !== FALSE) {
Chris@0:       for ($i = 0; $i <= 199; $i++) {
Chris@0:         $plurals[$i] = $this->evaluatePlural($element_stack, $i);
Chris@0:       }
Chris@0:       $default = $plurals[$i - 1];
Chris@0:       $plurals = array_filter($plurals, function ($value) use ($default) {
Chris@0:         return ($value != $default);
Chris@0:       });
Chris@0:       $plurals['default'] = $default;
Chris@0: 
Chris@0:       return [$nplurals, $plurals];
Chris@0:     }
Chris@0:     else {
Chris@0:       throw new \Exception('The plural formula could not be parsed.');
Chris@0:     }
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Parses a Gettext Portable Object file header.
Chris@0:    *
Chris@0:    * @param string $header
Chris@0:    *   A string containing the complete header.
Chris@0:    *
Chris@0:    * @return array
Chris@0:    *   An associative array of key-value pairs.
Chris@0:    */
Chris@0:   private function parseHeader($header) {
Chris@0:     $header_parsed = [];
Chris@0:     $lines = array_map('trim', explode("\n", $header));
Chris@0:     foreach ($lines as $line) {
Chris@0:       if ($line) {
Chris@0:         list($tag, $contents) = explode(":", $line, 2);
Chris@0:         $header_parsed[trim($tag)] = trim($contents);
Chris@0:       }
Chris@0:     }
Chris@0:     return $header_parsed;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Parses and sanitizes an arithmetic formula into a plural element stack.
Chris@0:    *
Chris@0:    * While parsing, we ensure, that the operators have the right
Chris@0:    * precedence and associativity.
Chris@0:    *
Chris@0:    * @param string $string
Chris@0:    *   A string containing the arithmetic formula.
Chris@0:    *
Chris@0:    * @return
Chris@0:    *   A stack of values and operations to be evaluated.
Chris@0:    */
Chris@0:   private function parseArithmetic($string) {
Chris@0:     // Operator precedence table.
Chris@0:     $precedence = ["(" => -1, ")" => -1, "?" => 1, ":" => 1, "||" => 3, "&&" => 4, "==" => 5, "!=" => 5, "<" => 6, ">" => 6, "<=" => 6, ">=" => 6, "+" => 7, "-" => 7, "*" => 8, "/" => 8, "%" => 8];
Chris@0:     // Right associativity.
Chris@0:     $right_associativity = ["?" => 1, ":" => 1];
Chris@0: 
Chris@0:     $tokens = $this->tokenizeFormula($string);
Chris@0: 
Chris@0:     // Parse by converting into infix notation then back into postfix
Chris@0:     // Operator stack - holds math operators and symbols.
Chris@0:     $operator_stack = [];
Chris@0:     // Element Stack - holds data to be operated on.
Chris@0:     $element_stack = [];
Chris@0: 
Chris@0:     foreach ($tokens as $token) {
Chris@0:       $current_token = $token;
Chris@0: 
Chris@0:       // Numbers and the $n variable are simply pushed into $element_stack.
Chris@0:       if (is_numeric($token)) {
Chris@0:         $element_stack[] = $current_token;
Chris@0:       }
Chris@0:       elseif ($current_token == "n") {
Chris@0:         $element_stack[] = '$n';
Chris@0:       }
Chris@0:       elseif ($current_token == "(") {
Chris@0:         $operator_stack[] = $current_token;
Chris@0:       }
Chris@0:       elseif ($current_token == ")") {
Chris@0:         $topop = array_pop($operator_stack);
Chris@0:         while (isset($topop) && ($topop != "(")) {
Chris@0:           $element_stack[] = $topop;
Chris@0:           $topop = array_pop($operator_stack);
Chris@0:         }
Chris@0:       }
Chris@0:       elseif (!empty($precedence[$current_token])) {
Chris@0:         // If it's an operator, then pop from $operator_stack into
Chris@0:         // $element_stack until the precedence in $operator_stack is less
Chris@0:         // than current, then push into $operator_stack.
Chris@0:         $topop = array_pop($operator_stack);
Chris@0:         while (isset($topop) && ($precedence[$topop] >= $precedence[$current_token]) && !(($precedence[$topop] == $precedence[$current_token]) && !empty($right_associativity[$topop]) && !empty($right_associativity[$current_token]))) {
Chris@0:           $element_stack[] = $topop;
Chris@0:           $topop = array_pop($operator_stack);
Chris@0:         }
Chris@0:         if ($topop) {
Chris@0:           // Return element to top.
Chris@0:           $operator_stack[] = $topop;
Chris@0:         }
Chris@0:         // Parentheses are not needed.
Chris@0:         $operator_stack[] = $current_token;
Chris@0:       }
Chris@0:       else {
Chris@0:         return FALSE;
Chris@0:       }
Chris@0:     }
Chris@0: 
Chris@0:     // Flush operator stack.
Chris@0:     $topop = array_pop($operator_stack);
Chris@0:     while ($topop != NULL) {
Chris@0:       $element_stack[] = $topop;
Chris@0:       $topop = array_pop($operator_stack);
Chris@0:     }
Chris@0:     $return = $element_stack;
Chris@0: 
Chris@0:     // Now validate stack.
Chris@0:     $previous_size = count($element_stack) + 1;
Chris@0:     while (count($element_stack) < $previous_size) {
Chris@0:       $previous_size = count($element_stack);
Chris@0:       for ($i = 2; $i < count($element_stack); $i++) {
Chris@0:         $op = $element_stack[$i];
Chris@0:         if (!empty($precedence[$op])) {
Chris@0:           if ($op == ":") {
Chris@0:             $f = $element_stack[$i - 2] . "):" . $element_stack[$i - 1] . ")";
Chris@0:           }
Chris@0:           elseif ($op == "?") {
Chris@0:             $f = "(" . $element_stack[$i - 2] . "?(" . $element_stack[$i - 1];
Chris@0:           }
Chris@0:           else {
Chris@0:             $f = "(" . $element_stack[$i - 2] . $op . $element_stack[$i - 1] . ")";
Chris@0:           }
Chris@0:           array_splice($element_stack, $i - 2, 3, $f);
Chris@0:           break;
Chris@0:         }
Chris@0:       }
Chris@0:     }
Chris@0: 
Chris@0:     // If only one element is left, the number of operators is appropriate.
Chris@0:     return count($element_stack) == 1 ? $return : FALSE;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Tokenize the formula.
Chris@0:    *
Chris@0:    * @param string $formula
Chris@0:    *   A string containing the arithmetic formula.
Chris@0:    *
Chris@0:    * @return array
Chris@0:    *   List of arithmetic tokens identified in the formula.
Chris@0:    */
Chris@0:   private function tokenizeFormula($formula) {
Chris@0:     $formula = str_replace(" ", "", $formula);
Chris@0:     $tokens = [];
Chris@0:     for ($i = 0; $i < strlen($formula); $i++) {
Chris@0:       if (is_numeric($formula[$i])) {
Chris@0:         $num = $formula[$i];
Chris@0:         $j = $i + 1;
Chris@0:         while ($j < strlen($formula) && is_numeric($formula[$j])) {
Chris@0:           $num .= $formula[$j];
Chris@0:           $j++;
Chris@0:         }
Chris@0:         $i = $j - 1;
Chris@0:         $tokens[] = $num;
Chris@0:       }
Chris@0:       elseif ($pos = strpos(" =<>!&|", $formula[$i])) {
Chris@0:         $next = $formula[$i + 1];
Chris@0:         switch ($pos) {
Chris@0:           case 1:
Chris@0:           case 2:
Chris@0:           case 3:
Chris@0:           case 4:
Chris@0:             if ($next == '=') {
Chris@0:               $tokens[] = $formula[$i] . '=';
Chris@0:               $i++;
Chris@0:             }
Chris@0:             else {
Chris@0:               $tokens[] = $formula[$i];
Chris@0:             }
Chris@0:             break;
Chris@0:           case 5:
Chris@0:             if ($next == '&') {
Chris@0:               $tokens[] = '&&';
Chris@0:               $i++;
Chris@0:             }
Chris@0:             else {
Chris@0:               $tokens[] = $formula[$i];
Chris@0:             }
Chris@0:             break;
Chris@0:           case 6:
Chris@0:             if ($next == '|') {
Chris@0:               $tokens[] = '||';
Chris@0:               $i++;
Chris@0:             }
Chris@0:             else {
Chris@0:               $tokens[] = $formula[$i];
Chris@0:             }
Chris@0:             break;
Chris@0:         }
Chris@0:       }
Chris@0:       else {
Chris@0:         $tokens[] = $formula[$i];
Chris@0:       }
Chris@0:     }
Chris@0:     return $tokens;
Chris@0:   }
Chris@0: 
Chris@0:   /**
Chris@0:    * Evaluate the plural element stack using a plural value.
Chris@0:    *
Chris@0:    * Using an element stack, which represents a plural formula, we calculate
Chris@0:    * which plural string should be used for a given plural value.
Chris@0:    *
Chris@0:    * An example of plural formula parting and evaluation:
Chris@0:    *   Plural formula: 'n!=1'
Chris@0:    * This formula is parsed by parseArithmetic() to a stack (array) of elements:
Chris@0:    *   array(
Chris@0:    *     0 => '$n',
Chris@0:    *     1 => '1',
Chris@0:    *     2 => '!=',
Chris@0:    *   );
Chris@0:    * The evaluatePlural() method evaluates the $element_stack using the plural
Chris@0:    * value $n. Before the actual evaluation, the '$n' in the array is replaced
Chris@0:    * by the value of $n.
Chris@0:    *   For example: $n = 2 results in:
Chris@0:    *   array(
Chris@0:    *     0 => '2',
Chris@0:    *     1 => '1',
Chris@0:    *     2 => '!=',
Chris@0:    *   );
Chris@0:    * The stack is processed until only one element is (the result) is left. In
Chris@0:    * every iteration the top elements of the stack, up until the first operator,
Chris@0:    * are evaluated. After evaluation the arguments and the operator itself are
Chris@0:    * removed and replaced by the evaluation result. This is typically 2
Chris@0:    * arguments and 1 element for the operator.
Chris@0:    *   Because the operator is '!=' the example stack is evaluated as:
Chris@0:    *   $f = (int) 2 != 1;
Chris@0:    *   The resulting stack is:
Chris@0:    *   array(
Chris@0:    *     0 => 1,
Chris@0:    *   );
Chris@0:    * With only one element left in the stack (the final result) the loop is
Chris@0:    * terminated and the result is returned.
Chris@0:    *
Chris@0:    * @param array $element_stack
Chris@0:    *   Array of plural formula values and operators create by parseArithmetic().
Chris@0:    * @param int $n
Chris@0:    *   The @count number for which we are determining the right plural position.
Chris@0:    *
Chris@0:    * @return int
Chris@0:    *   Number of the plural string to be used for the given plural value.
Chris@0:    *
Chris@0:    * @see parseArithmetic()
Chris@14:    * @throws \Exception
Chris@0:    */
Chris@0:   protected function evaluatePlural($element_stack, $n) {
Chris@0:     $count = count($element_stack);
Chris@0:     $limit = $count;
Chris@0:     // Replace the '$n' value in the formula by the plural value.
Chris@0:     for ($i = 0; $i < $count; $i++) {
Chris@0:       if ($element_stack[$i] === '$n') {
Chris@0:         $element_stack[$i] = $n;
Chris@0:       }
Chris@0:     }
Chris@0: 
Chris@0:     // We process the stack until only one element is (the result) is left.
Chris@0:     // We limit the number of evaluation cycles to prevent an endless loop in
Chris@0:     // case the stack contains an error.
Chris@0:     while (isset($element_stack[1])) {
Chris@0:       for ($i = 2; $i < $count; $i++) {
Chris@0:         // There's no point in checking non-symbols. Also, switch(TRUE) would
Chris@0:         // match any case and so it would break.
Chris@0:         if (is_bool($element_stack[$i]) || is_numeric($element_stack[$i])) {
Chris@0:           continue;
Chris@0:         }
Chris@0:         $f = NULL;
Chris@0:         $length = 3;
Chris@0:         $delta = 2;
Chris@0:         switch ($element_stack[$i]) {
Chris@0:           case '==':
Chris@0:             $f = $element_stack[$i - 2] == $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '!=':
Chris@0:             $f = $element_stack[$i - 2] != $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '<=':
Chris@0:             $f = $element_stack[$i - 2] <= $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '>=':
Chris@0:             $f = $element_stack[$i - 2] >= $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '<':
Chris@0:             $f = $element_stack[$i - 2] < $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '>':
Chris@0:             $f = $element_stack[$i - 2] > $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '+':
Chris@0:             $f = $element_stack[$i - 2] + $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '-':
Chris@0:             $f = $element_stack[$i - 2] - $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '*':
Chris@0:             $f = $element_stack[$i - 2] * $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '/':
Chris@0:             $f = $element_stack[$i - 2] / $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '%':
Chris@0:             $f = $element_stack[$i - 2] % $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '&&':
Chris@0:             $f = $element_stack[$i - 2] && $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case '||':
Chris@0:             $f = $element_stack[$i - 2] || $element_stack[$i - 1];
Chris@0:             break;
Chris@0:           case ':':
Chris@0:             $f = $element_stack[$i - 3] ? $element_stack[$i - 2] : $element_stack[$i - 1];
Chris@0:             // This operator has 3 preceding elements, instead of the default 2.
Chris@0:             $length = 5;
Chris@0:             $delta = 3;
Chris@0:             break;
Chris@0:         }
Chris@0: 
Chris@0:         // If the element is an operator we remove the processed elements and
Chris@0:         // store the result.
Chris@0:         if (isset($f)) {
Chris@0:           array_splice($element_stack, $i - $delta, $length, $f);
Chris@0:           break;
Chris@0:         }
Chris@0:       }
Chris@0:     }
Chris@0:     if (!$limit) {
Chris@0:       throw new \Exception('The plural formula could not be evaluated.');
Chris@0:     }
Chris@0:     return (int) $element_stack[0];
Chris@0:   }
Chris@0: 
Chris@0: }