bugfix for mga#14899 (proper handling of gettext files)

details in bug report

1 files changed, 112 insertions, 47 deletions

diff --git a/php-mo.php b/php-mo.php
index 470ce80..6f635c2 100644
--- a/php-mo.php
+++ b/php-mo.php
@@ -1,29 +1,68 @@
 <?php
 /**
- * based on php.mo 0.1 by Joss Crowcroft (http://www.josscrowcroft.com)
- * added case for commented unused string
- * fixed "fuzzy flag first line" bug which didn't saved previous proper string at all
- * removed unedeed phpmo_convert and phpmo_write_mo_file
- *
- * Converts gettext translation '.po' files to binary '.mo' files in PHP.
- *
- * Usage:
- * <?php require('php-mo.php'); phpmo_convert( 'input.po', [ 'output.mo' ] ); ?>
- *
- * NB:
- * - If no $output_file specified, output filename is same as $input_file (but .mo)
- * - Returns true/false for success/failure
- * - No warranty, but if it breaks, please let me know
- *
- * More info:
- * https://github.com/josscrowcroft/php.mo
- *
- * Based on php-msgfmt by Matthias Bauer (Copyright © 2007), a command-line PHP tool
- * for converting .po files to .mo.
- * (http://wordpress-soc-2007.googlecode.com/svn/trunk/moeffju/php-msgfmt/msgfmt.php)
- *
- * License: GPL v3 http://www.opensource.org/licenses/gpl-3.0.html
- */
+* Changes from php.mo:
+* supress the error if there's no file
+* fixed "fuzzy flag first line" bug which didn't saved previous proper string at all
+* fix the warning if list finds no $data
+* added case for commented unused string
+* proper handling of msgctxt
+* better storing of same msgid
+* corrected data structure for correct counting of plural msgstr
+* don't add last translation if it's fuzzy
+* removed unedeed phpmo_convert and phpmo_write_mo_file
+* more info on https://bugs.mageia.org/show_bug.cgi?id=14899
+* 
+* it returns array of translations:
+* $dictionary = phpmo_parse_po_file('input.po');
+*
+* $dictionary = array (
+*	'' => array (
+*		'msgid' => '',
+*		'msgstr' => array (
+*		   0 => 'Project-Id-Version: ...',
+*		),
+*	),
+*	'string for translation 1' => array (
+*		'msgid' => 'string for translation 1',
+*		'msgstr' => array (
+*			0 => 'translated string 1',
+*		),
+*	),
+*	'string for translation 2' => array (
+*		'msgid' => 'string for translation 2',
+*		'msgstr' => array (
+*			0 => 'translated string 2',
+*			1 => 'translated string 2 but different context',
+*			2 => 'another translated string 2 with different context',
+*		),
+*		'msgctxt' => array (
+*			1 => 'context explanation for msgstr[1]',
+*			2 => 'context explanation for msgstr[2]',
+*		),
+*	),
+*	'string for translation 3' => array (
+*		'msgid' => 'string for translation 3',
+*		'msgid_plural' => 'plural string for translation 3',
+*		'msgstr' => array (
+*			0 => array (
+*				0 => 'translated string 3 for nplurals 0',
+*				1 => 'translated string 3 for nplurals 1',
+*				2 => 'translated string 3 for nplurals 2',
+*			),
+*		),
+*	),
+* );
+*
+****************************************************************************************
+*
+* based on php.mo 0.1 by Joss Crowcroft (http://www.josscrowcroft.com)
+* which is
+* Based on php-msgfmt by Matthias Bauer (Copyright © 2007), a command-line PHP tool
+* for converting .po files to .mo.
+* (http://wordpress-soc-2007.googlecode.com/svn/trunk/moeffju/php-msgfmt/msgfmt.php)
+*
+* License: GPL v3 http://www.opensource.org/licenses/gpl-3.0.html
+*/
 
 function phpmo_clean_helper($x) {
 	if (is_array($x)) {
@@ -60,34 +99,29 @@ function phpmo_parse_po_file($in) {
 	// iterate over lines
 	while(($line = fgets($fh, 65536)) !== false) {
 		$line = trim($line);
-		if ($line === '') {
-			// save stored entry on empty line
-			// block moved to fix "fuzzy flag first line" bug which didn't saved previous proper string at all
-			if (sizeof($temp) && array_key_exists('msgid', $temp) && array_key_exists('msgstr', $temp)) {
-				if (!$fuzzy)
-					$hash[] = $temp;
-				$temp = array ();
-				$state = null;
-				$fuzzy = false;
-			}
+		if ($line === '')
 			continue;
-		}
+
 		$array_of_splited_string = preg_split('/\s/', $line, 2);
 		$key = $array_of_splited_string[0];
 		$data = (isset($array_of_splited_string[1]) ? $array_of_splited_string[1] : '');
-
 		switch ($key) {
 			case '#,' : // flag...
-				$fuzzy = in_array('fuzzy', preg_split('/,\s*/', $data));
 			case '#' : // translator-comments
 			case '#.' : // extracted-comments
 			case '#:' : // reference...
 			case '#|' : // msgid previous-untranslated-string
-				break;
 			case '#~' : // commented-unused-string
-				$temp = array ();
-				$state = null;
-				$fuzzy = false;
+				// start a new entry
+				if (sizeof($temp) && array_key_exists('msgid', $temp) && array_key_exists('msgstr', $temp)) {
+					if (!$fuzzy)
+						$hash[] = $temp;
+					$temp = array ();
+					$state = null;
+					$fuzzy = false;
+				}
+				if ($key == '#,') 
+					$fuzzy = in_array('fuzzy', preg_split('/,\s*/', $data));
 				break;
 			case 'msgctxt' :
 				// context
@@ -96,7 +130,23 @@ function phpmo_parse_po_file($in) {
 			case 'msgid_plural' :
 				// untranslated-string-plural
 				$state = $key;
-				$temp[$state] = $data;
+				// store previous entry if it exists
+				if (sizeof($temp) && array_key_exists('msgid', $temp) && array_key_exists('msgstr', $temp)) {
+					if (!$fuzzy)
+						$hash[] = $temp;
+					$temp = array ();
+					$fuzzy = false;
+				}
+				if ($key == 'msgctxt') {
+						if (array_key_exists('msgstr', $temp)) {
+							$msgctxt_index = count($temp['msgstr']);
+						} else {
+							$msgctxt_index = 0;
+						}
+						$temp['msgctxt'][$msgctxt_index] = $data;
+				} else {
+					$temp[$state] = $data;
+				}
 				break;
 			case 'msgstr' :
 				// translated-string
@@ -111,11 +161,11 @@ function phpmo_parse_po_file($in) {
 				} else {
 					// continued lines
 					switch ($state) {
-						case 'msgctxt' :
 						case 'msgid' :
 						case 'msgid_plural' :
 							$temp[$state] .= "\n" . $line;
 							break;
+						case 'msgctxt' :
 						case 'msgstr' :
 							$temp[$state][sizeof($temp[$state]) - 1] .= "\n" . $line;
 							break;
@@ -132,7 +182,8 @@ function phpmo_parse_po_file($in) {
 
 	// add final entry
 	if ($state == 'msgstr')
-		$hash[] = $temp;
+		if (!$fuzzy)
+			$hash[] = $temp;
 
 	// Cleanup data, merge multiline entries, reindex hash for ksort
 	$temp = $hash;
@@ -145,10 +196,24 @@ function phpmo_parse_po_file($in) {
 				return FALSE;
 			}
 		}
-		$hash[$entry['msgid']] = $entry;
+		// if there's already same msgid we need to store it there
+		if (array_key_exists($entry['msgid'], $hash)) {
+			$index = count($hash[$entry['msgid']]['msgstr']);
+			$hash[$entry['msgid']]['msgstr'][$index] = $entry['msgstr'][0];
+			// is different msgctxt present also?
+			if (array_key_exists('msgctxt', $entry)) {
+				$hash[$entry['msgid']]['msgctxt'][$index] = $entry['msgctxt'][0];
+			}
+		} else {
+			// if msgid_plural exists we need to store msgstr to subarray for proper counting of l10n completion
+			if (array_key_exists('msgid_plural', $entry)) {
+				$msgstr_plural   = $entry['msgstr'];
+				$entry['msgstr']   = array();
+				$entry['msgstr'][] = $msgstr_plural;
+			}
+			$hash[$entry['msgid']] = $entry;
+		}
 	}
 
 	return $hash;
 }
-
-?>
\ No newline at end of file