Gettext PO header handler.
Possible Gettext PO header elements are explained in http://www.gnu.org/software/gettext/manual/gettext.html#Header-Entry, but we only support a subset of these directly.
Example header:
"Project-Id-Version: Drupal core (7.11)\n" "POT-Creation-Date: 2012-02-12 22:59+0000\n" "PO-Revision-Date: YYYY-mm-DD HH:MM+ZZZZ\n" "Language-Team: Catalan\n" "MIME-Version: 1.0\n" "Content-Type: text/plain; charset=utf-8\n" "Content-Transfer-Encoding: 8bit\n" "Plural-Forms: nplurals=2; plural=(n>1);\n"
Expanded class hierarchy of PoHeader
class PoHeader {
/**
* Language code.
*
* @var string
*/
private $_langcode;
/**
* Formula for the plural form.
*
* @var string
*/
private $_pluralForms;
/**
* Author(s) of the file.
*
* @var string
*/
private $_authors;
/**
* Date the po file got created.
*
* @var string
*/
private $_po_date;
/**
* Human readable language name.
*
* @var string
*/
private $_languageName;
/**
* Name of the project the translation belongs to.
*
* @var string
*/
private $_projectName;
/**
* Constructor, creates a PoHeader with default values.
*
* @param string $langcode
* Language code.
*/
public function __construct($langcode = NULL) {
$this->_langcode = $langcode;
$this->_po_date = date("Y-m-d H:iO");
$this->_pluralForms = 'nplurals=2; plural=(n > 1);';
}
/**
* Get the plural form.
*
* @return string
* Plural form component from the header, for example:
* 'nplurals=2; plural=(n > 1);'.
*/
function getPluralForms() {
return $this->_pluralForms;
}
/**
* Set the human readable language name.
*
* @param string $languageName
* Human readable language name.
*/
function setLanguageName($languageName) {
$this->_languageName = $languageName;
}
/**
* Get the human readable language name.
*
* @return string
* The human readable language name.
*/
function getLanguageName() {
return $this->_languageName;
}
/**
* Set the project name.
*
* @param string $projectName
* Human readable project name.
*/
function setProjectName($projectName) {
$this->_projectName = $projectName;
}
/**
* Get the project name.
*
* @return string
* The human readable project name.
*/
function getProjectName() {
return $this->_projectName;
}
/**
* Populate internal values from a string.
*
* @param string $header
* Full header string with key-value pairs.
*/
public function setFromString($header) {
// Get an array of all header values for processing.
$values = $this
->parseHeader($header);
// There is only one value relevant for our header implementation when
// reading, and that is the plural formula.
if (!empty($values['Plural-Forms'])) {
$this->_pluralForms = $values['Plural-Forms'];
}
}
/**
* Generate a Gettext PO formatted header string based on data set earlier.
*/
public function __toString() {
$output = '';
$isTemplate = empty($this->_languageName);
$output .= '# ' . ($isTemplate ? 'LANGUAGE' : $this->_languageName) . ' translation of ' . ($isTemplate ? 'PROJECT' : $this->_projectName) . "\n";
if (!empty($this->_authors)) {
$output .= '# Generated by ' . implode("\n# ", $this->_authors) . "\n";
}
$output .= "#\n";
// Add the actual header information.
$output .= "msgid \"\"\n";
$output .= "msgstr \"\"\n";
$output .= "\"Project-Id-Version: PROJECT VERSION\\n\"\n";
$output .= "\"POT-Creation-Date: " . $this->_po_date . "\\n\"\n";
$output .= "\"PO-Revision-Date: " . $this->_po_date . "\\n\"\n";
$output .= "\"Last-Translator: NAME <EMAIL@ADDRESS>\\n\"\n";
$output .= "\"Language-Team: LANGUAGE <EMAIL@ADDRESS>\\n\"\n";
$output .= "\"MIME-Version: 1.0\\n\"\n";
$output .= "\"Content-Type: text/plain; charset=utf-8\\n\"\n";
$output .= "\"Content-Transfer-Encoding: 8bit\\n\"\n";
$output .= "\"Plural-Forms: " . $this->_pluralForms . "\\n\"\n";
$output .= "\n";
return $output;
}
/**
* Parses a Plural-Forms entry from a Gettext Portable Object file header.
*
* @param string $pluralforms
* The Plural-Forms entry value.
*
* @return
* An array containing the number of plural forms and the converted version
* of the formula that can be evaluated with PHP later.
*/
function parsePluralForms($pluralforms) {
// First, delete all whitespace.
$pluralforms = strtr($pluralforms, array(
" " => "",
"\t" => "",
));
// Select the parts that define nplurals and plural.
$nplurals = strstr($pluralforms, "nplurals=");
if (strpos($nplurals, ";")) {
// We want the string from the 10th char, because "nplurals=" length is 9.
$nplurals = substr($nplurals, 9, strpos($nplurals, ";") - 9);
}
else {
return FALSE;
}
$plural = strstr($pluralforms, "plural=");
if (strpos($plural, ";")) {
// We want the string from the 8th char, because "plural=" length is 7.
$plural = substr($plural, 7, strpos($plural, ";") - 7);
}
else {
return FALSE;
}
// Get PHP version of the plural formula.
$plural = $this
->parseArithmetic($plural);
if ($plural !== FALSE) {
return array(
$nplurals,
$plural,
);
}
else {
throw new Exception('The plural formula could not be parsed.');
}
}
/**
* Parses a Gettext Portable Object file header.
*
* @param string $header
* A string containing the complete header.
*
* @return array
* An associative array of key-value pairs.
*/
private function parseHeader($header) {
$header_parsed = array();
$lines = array_map('trim', explode("\n", $header));
foreach ($lines as $line) {
if ($line) {
list($tag, $contents) = explode(":", $line, 2);
$header_parsed[trim($tag)] = trim($contents);
}
}
return $header_parsed;
}
/**
* Parses and sanitizes an arithmetic formula into a PHP expression.
*
* While parsing, we ensure, that the operators have the right
* precedence and associativity.
*
* @param string $string
* A string containing the arithmetic formula.
*
* @return
* A version of the formula to evaluate with PHP later.
*/
private function parseArithmetic($string) {
// Operator precedence table.
$precedence = array(
"(" => -1,
")" => -1,
"?" => 1,
":" => 1,
"||" => 3,
"&&" => 4,
"==" => 5,
"!=" => 5,
"<" => 6,
">" => 6,
"<=" => 6,
">=" => 6,
"+" => 7,
"-" => 7,
"*" => 8,
"/" => 8,
"%" => 8,
);
// Right associativity.
$right_associativity = array(
"?" => 1,
":" => 1,
);
$tokens = $this
->tokenizeFormula($string);
// Parse by converting into infix notation then back into postfix
// Operator stack - holds math operators and symbols.
$operator_stack = array();
// Element Stack - holds data to be operated on.
$element_stack = array();
foreach ($tokens as $token) {
$current_token = $token;
// Numbers and the $n variable are simply pushed into $element_stack.
if (is_numeric($token)) {
$element_stack[] = $current_token;
}
elseif ($current_token == "n") {
$element_stack[] = '$n';
}
elseif ($current_token == "(") {
$operator_stack[] = $current_token;
}
elseif ($current_token == ")") {
$topop = array_pop($operator_stack);
while (isset($topop) && $topop != "(") {
$element_stack[] = $topop;
$topop = array_pop($operator_stack);
}
}
elseif (!empty($precedence[$current_token])) {
// If it's an operator, then pop from $operator_stack into
// $element_stack until the precedence in $operator_stack is less
// than current, then push into $operator_stack.
$topop = array_pop($operator_stack);
while (isset($topop) && $precedence[$topop] >= $precedence[$current_token] && !($precedence[$topop] == $precedence[$current_token] && !empty($right_associativity[$topop]) && !empty($right_associativity[$current_token]))) {
$element_stack[] = $topop;
$topop = array_pop($operator_stack);
}
if ($topop) {
// Return element to top.
$operator_stack[] = $topop;
}
// Parentheses are not needed.
$operator_stack[] = $current_token;
}
else {
return FALSE;
}
}
// Flush operator stack.
$topop = array_pop($operator_stack);
while ($topop != NULL) {
$element_stack[] = $topop;
$topop = array_pop($operator_stack);
}
// Now extract formula from stack.
$previous_size = count($element_stack) + 1;
while (count($element_stack) < $previous_size) {
$previous_size = count($element_stack);
for ($i = 2; $i < count($element_stack); $i++) {
$op = $element_stack[$i];
if (!empty($precedence[$op])) {
$f = "";
if ($op == ":") {
$f = $element_stack[$i - 2] . "):" . $element_stack[$i - 1] . ")";
}
elseif ($op == "?") {
$f = "(" . $element_stack[$i - 2] . "?(" . $element_stack[$i - 1];
}
else {
$f = "(" . $element_stack[$i - 2] . $op . $element_stack[$i - 1] . ")";
}
array_splice($element_stack, $i - 2, 3, $f);
break;
}
}
}
// If only one element is left, the number of operators is appropriate.
if (count($element_stack) == 1) {
return $element_stack[0];
}
else {
return FALSE;
}
}
/**
* Tokenize the formula.
*
* @param string $formula
* A string containing the arithmetic formula.
*
* @return array
* List of arithmetic tokens identified in the formula.
*/
private function tokenizeFormula($formula) {
$formula = str_replace(" ", "", $formula);
$tokens = array();
for ($i = 0; $i < strlen($formula); $i++) {
if (is_numeric($formula[$i])) {
$num = $formula[$i];
$j = $i + 1;
while ($j < strlen($formula) && is_numeric($formula[$j])) {
$num .= $formula[$j];
$j++;
}
$i = $j - 1;
$tokens[] = $num;
}
elseif ($pos = strpos(" =<>!&|", $formula[$i])) {
$next = $formula[$i + 1];
switch ($pos) {
case 1:
case 2:
case 3:
case 4:
if ($next == '=') {
$tokens[] = $formula[$i] . '=';
$i++;
}
else {
$tokens[] = $formula[$i];
}
break;
case 5:
if ($next == '&') {
$tokens[] = '&&';
$i++;
}
else {
$tokens[] = $formula[$i];
}
break;
case 6:
if ($next == '|') {
$tokens[] = '||';
$i++;
}
else {
$tokens[] = $formula[$i];
}
break;
}
}
else {
$tokens[] = $formula[$i];
}
}
return $tokens;
}
}
Name | Modifiers | Type | Description | Overrides |
---|---|---|---|---|
PoHeader:: |
private | property | Author(s) of the file. | |
PoHeader:: |
private | property | Language code. | |
PoHeader:: |
private | property | Human readable language name. | |
PoHeader:: |
private | property | Formula for the plural form. | |
PoHeader:: |
private | property | Date the po file got created. | |
PoHeader:: |
private | property | Name of the project the translation belongs to. | |
PoHeader:: |
function | Get the human readable language name. | ||
PoHeader:: |
function | Get the plural form. | ||
PoHeader:: |
function | Get the project name. | ||
PoHeader:: |
private | function | Parses and sanitizes an arithmetic formula into a PHP expression. | |
PoHeader:: |
private | function | Parses a Gettext Portable Object file header. | |
PoHeader:: |
function | Parses a Plural-Forms entry from a Gettext Portable Object file header. | ||
PoHeader:: |
public | function | Populate internal values from a string. | |
PoHeader:: |
function | Set the human readable language name. | ||
PoHeader:: |
function | Set the project name. | ||
PoHeader:: |
private | function | Tokenize the formula. | |
PoHeader:: |
public | function | Constructor, creates a PoHeader with default values. | |
PoHeader:: |
public | function | Generate a Gettext PO formatted header string based on data set earlier. |