add capabilities to properly process l10n *.ts files like mageiaSync_en.ts

2 files changed, 91 insertions, 6 deletions

diff --git a/langs/lib.php b/langs/lib.php
index 42183c027..5fd5f50d3 100644
--- a/langs/lib.php
+++ b/langs/lib.php
@@ -145,6 +145,63 @@ function _po_diff($locale, $resource, $source_l = NULL, $path = NULL)
     );
 }
 
+/**
+ * Diff English and translated ts files, to get:
+ * - source (ts) strings count
+ * - missing strings in target
+ * - untranslated strings in target
+ * - empty array for extra and duplicate strings for backward compatibility
+ *
+ * @param string $locale locale name ('sl')
+ * @param string $resource file name ('mageiaSync')
+ * @param array  $source_l array with source file strings (to avoid duplicated parsing)
+ * @param string $path directly passed path for nonlocal files
+ *
+ * @return array
+*/
+function _ts_diff($locale, $resource, $source_l = NULL, $path = NULL)
+{
+    $source_path_filename = sprintf('%s%s_%s.ts', $path, $resource, $locale); // mageiaSync_sl.ts
+    $source_strings       = array();
+    $untranslated_strings = array();
+    $obsoleted_strings    = array();
+    // read .ts file
+    $file_handle = @fopen($source_path_filename, 'r');
+    if ($file_handle === false) {
+        // Could not open file resource
+        return false;
+    }
+
+    // iterate over lines
+    while(($line = fgets($file_handle, 65536)) !== false) {
+        // count <source>???</source> lines
+        if (false !== strpos($line, '</source>')) {
+            preg_match_all("/(<source>)([a-z_A-Z@-]+)(<\/source>)/", $line, $source_string);
+            $source_strings[] = $source_string[2];
+        }
+        // count <translation type="unfinished"></translation> lines
+        if (false !== strpos($line, '<translation type="unfinished">')) {
+            $untranslated_strings[] = $source_string[2];
+        }
+        // count <translation type="obsolete"></translation> lines
+        if (false !== strpos($line, '<translation type="obsolete">')) {
+            array_pop($source_strings); // subtract obsoleted source strings
+            $obsoleted_strings[] = $source_string[2];
+        }
+    }
+    fclose($file_handle);
+
+    return array(
+        'a'                => count($source_strings), // # of original strings
+//      'b'                => , // # of target strings
+        'source_strings'   => $source_strings, // array of original strings
+        'fuzzy_or_missing' => array(), // array of fuzzy or missing strings
+        'notrans'          => $untranslated_strings, // array of untranslated strings
+        'extra'            => $obsoleted_strings,
+        'dup_str'          => array(),
+    );
+}
+
 /*function _lang_diff_stats($a, $b)
 {
     $diff = _lang_diff($a, $b);
diff --git a/langs/report_tx_git.php b/langs/report_tx_git.php
index 79bfefb87..bbc14234d 100644
--- a/langs/report_tx_git.php
+++ b/langs/report_tx_git.php
@@ -219,6 +219,11 @@ if ('Documentation' == $resource_type) {
 //            'git_path' => 'https://github.com/papoteur-mga/isodumper/raw/master/po',
             'git_path' => 'http://gitweb.mageia.org/software/isodumper/plain/po',
         ),
+        array(
+            'pot_name' => 'mageiaSync_en.ts',
+            'tx_name'  => 'mageiasync',
+            'git_path' => 'http://gitweb.mageia.org/qa/MageiaSync/plain/translations/',
+        ),
     );
 }
 
@@ -292,7 +297,13 @@ function build_links($git_resource_name, $language_code, $resource_type, $stat_d
             $git_array = array('sr@Latn', 'uz@cyrillic');
             $lang_code = str_replace($tx_array, $git_array, $language_code);
         }
-        $git_link = sprintf('%s/%s.po', $stat_data['git_path'], $lang_code);
+        // treat TS files differently
+        if (false !== strpos($stat_data['pot_name'], '_en.ts')) {
+            $partial_git_resource_name = substr($stat_data['pot_name'], 0, -6); // cuts '_en.ts'
+            $git_link = sprintf('%s%s_%s.ts', $stat_data['git_path'], $partial_git_resource_name, $lang_code); // mageiaSync_sl.ts
+        } else {
+            $git_link = sprintf('%s/%s.po', $stat_data['git_path'], $lang_code);
+        }
     }
     $links_and_num  = build_transifex_link($language_code, 'Tx', $resource_type, $tx_resource_name) . ": ";
     $links_and_num .= $stat_data['tx_untran'];
@@ -361,7 +372,13 @@ function generating_report($language_codes, $resource_names, $path = NULL, $pot_
                 $langF = '../_nav/langs/' . $l . '.po' . (($l == 'en') ? 't' : '');
             }
             if (!is_null($path) || file_exists($langF)) {
-                $stat = _po_diff($l, $resource, $source_strings, $path);
+                // treat TS files differently
+                if (false !== strpos($pot_name, '_en.ts')) {
+                    $partial_pot_name = substr($pot_name, 0, -6); // cuts '_en.ts' from mageiaSync_en.ts
+                    $stat = _ts_diff($l, $partial_pot_name, $source_strings, $path);
+                } else {
+                    $stat = _po_diff($l, $resource, $source_strings, $path);
+                }
                 $num_of_fuzzy_or_missing = count($stat['fuzzy_or_missing']);
             } else { // file $langF doesn't exists in 'Webpages' $resource_type
                 $stat = _po_diff('en', $resource, $source_strings);
@@ -390,6 +407,7 @@ function generating_report($language_codes, $resource_names, $path = NULL, $pot_
                 'web_language_code'      => $web_language_code,
                 'num_of_not_fully_trans' => $num_of_not_fully_trans,
                 'webgit_path'            => $path,
+                'pot_name'               => $pot_name,
             );
         }
     }
@@ -481,7 +499,12 @@ if ('Webpages' == $resource_type) {
             $errors['file_get_contents_failed'] = "Access to $git_path failed.";
         }
         // list all po files from links within $raw_html_dump
-        preg_match_all("/('>)([a-z_A-Z@-]+)(\.po<)/", $raw_html_dump, $language_codes);
+        if (false !== strpos($pot_name, '_en.ts')) { // treat TS files differently
+            $first_part_pot_name = substr($pot_name, 0, -5); // cuts 'en.ts' from mageiaSync_en.ts
+            preg_match_all("/('>$first_part_pot_name)([a-z_A-Z@-]+)(\.ts<)/", $raw_html_dump, $language_codes);
+        } else {
+            preg_match_all("/('>)([a-z_A-Z@-]+)(\.po<)/", $raw_html_dump, $language_codes);
+        }
         $git_language_codes = $language_codes[2];
         if (is_null($wanted_language)) {
             $report_language_codes = $git_language_codes;
@@ -542,6 +565,7 @@ foreach ($tx_resources_info as $tx_resource_info) {
                 $git_resource_num_of_all_strings = $git_resource_info['num_of_all_strings'];
                 $git_resource_untrans_in_lang    = $git_resource_info['num_of_not_fully_trans'];
                 $git_resource_path               = $git_resource_info['webgit_path'];
+                $git_resource_pot_name           = $git_resource_info['pot_name'];
 
                 // compare l10n level
                 if ($tx_resource_untrans_in_lang != $git_resource_untrans_in_lang) {
@@ -549,6 +573,7 @@ foreach ($tx_resources_info as $tx_resource_info) {
                     $tx_git_difference[$tx_resource_language][$git_resource_name]['tx_untranslated']       = $tx_resource_untrans_in_lang;
                     $tx_git_difference[$tx_resource_language][$git_resource_name]['git_untranslated']      = $git_resource_untrans_in_lang;
                     $tx_git_difference[$tx_resource_language][$git_resource_name]['webgit_path']           = $git_resource_path;
+                    $tx_git_difference[$tx_resource_language][$git_resource_name]['pot_name']              = $git_resource_pot_name;
                 }
                 // it seems like a parse error
                 if (0 == $git_resource_num_of_all_strings) {
@@ -655,12 +680,12 @@ if (0 < count($git_only_resources)) {
 }
 
 if (0 < count($parse_errors)) {
-    $errors                       = array();
+    $errors_in_parsing            = array();
     foreach ($parse_errors as $single_git_resource_name => $git_resource_languages) {
-        $errors[]                 = "$single_git_resource_name.po (languages: " . implode(", ", $git_resource_languages) . ")";
+        $errors_in_parsing[]      = "$single_git_resource_name.po (languages: " . implode(", ", $git_resource_languages) . ")";
     }
     $reload                       = "Please reload the page and report this on mailing list if it persist.";
-    $errors['po_file_parse']      = "It seems that parsing of some resources failed: " . implode(", ", $errors) . ". $reload";
+    $errors['po_file_parse']      = "It seems that parsing of some resources failed: " . implode(", ", $errors_in_parsing) . ". $reload";
 }
 
 if (0 < count($nonequal_num_of_all_strings)) {
@@ -719,6 +744,7 @@ if (0 < $num_of_resources_w_difference && $num_for_switch_table_to_list > $num_o
                     $stat_data['tx_untran']  = $one_language_array[$resource_with_tx_git_diff]['tx_untranslated'];
                     $stat_data['git_untran'] = $one_language_array[$resource_with_tx_git_diff]['git_untranslated'];
                     $stat_data['git_path']   = $one_language_array[$resource_with_tx_git_diff]['webgit_path'];
+                    $stat_data['pot_name']   = $one_language_array[$resource_with_tx_git_diff]['pot_name'];
 
                     $cell  = "<span class=\"stat\">";
                     $cell .= build_links($resource_with_tx_git_diff, $one_language, $resource_type, $stat_data);
@@ -758,6 +784,7 @@ if (0 < $num_of_resources_w_difference) {
             $stat_data['tx_untran']  = $one_language_array[$one_language_resource]['tx_untranslated'];
             $stat_data['git_untran'] = $one_language_array[$one_language_resource]['git_untranslated'];
             $stat_data['git_path']   = $one_language_array[$one_language_resource]['webgit_path'];
+            $stat_data['pot_name']   = $one_language_array[$one_language_resource]['pot_name'];
 
             $list_of_lang_diff .= ($miss_first ? '' :', ');
             $list_of_lang_diff .= "<span class=\"italic\">$one_language_resource</span> <span class=\"stat\">(";
@@ -782,6 +809,7 @@ if (0 < $num_of_resources_w_difference) {
                 $stat_data['tx_untran']  = $one_language_array[$resource_with_tx_git_diff]['tx_untranslated'];
                 $stat_data['git_untran'] = $one_language_array[$resource_with_tx_git_diff]['git_untranslated'];
                 $stat_data['git_path']   = $one_language_array[$resource_with_tx_git_diff]['webgit_path'];
+                $stat_data['pot_name']   = $one_language_array[$resource_with_tx_git_diff]['pot_name'];
 
                 $list_of_resource_diff .= ($miss_first ? '' :', ');
                 $language_name = get_language_name($one_language);