l10n report can be limited to only one resource + better management of untranslated lines in constitution

3 files changed, 128 insertions, 34 deletions

diff --git a/langs/diff.php b/langs/diff.php
index 117d38a9e..cdaf6b8f2 100644
--- a/langs/diff.php
+++ b/langs/diff.php
@@ -4,6 +4,8 @@
  * against matching file in language 'l', if it exists.
 */
 
+define('APP_ROOT', realpath(__DIR__ . '/..'));
+
 $source_file = isset($_GET['s']) ? strip_tags(trim($_GET['s'])) : null;
 $target_lang = isset($_GET['l']) ? strip_tags(trim($_GET['l'])) : null;
 
@@ -61,9 +63,21 @@ $s = '<a href="report.php">&laquo; back to langs report</a>';
 $s .= sprintf('<h1>Differences between %s source and %s target</h1>',
     substr($source_file, strpos($source_file, '/langs/')+7), $target_lang);
 
+$diff['untranslated_lines_in_constitution'] = array();
 if($constitution) {
-    $s .= '<h2 style="color: red;">Please translate constitution first</h2>';
+    $s .= '<h2 style="color: red;">Please translate constitution fully into ' . $langs[$target_lang] . ' first</h2>';
     $s .= '<p>You can find it in <a href="//svnweb.mageia.org/org/constitution/">svn</a>. ';
+
+    $constitution_results = aproximate_number_of_untranslated_constitution_lines(APP_ROOT, 'en');
+    $unique_lines_in_eng_constitution = $constitution_results['unique_lines_in_constitution'];
+    $lang_constitution_results = aproximate_number_of_untranslated_constitution_lines(APP_ROOT, $target_lang, $unique_lines_in_eng_constitution);
+    $constitution_readable = $lang_constitution_results['constitution_readable'];
+    $num_of_untranslated_lines = $lang_constitution_results['aproximate_number_of_untranslated_lines'];
+    if ($constitution_readable) {
+        $diff['untranslated_lines_in_constitution'] = $lang_constitution_results['untranslated_lines_in_constitution']; // add untranslated license sentences
+    } else {
+        $diff['untranslated_lines_in_constitution'] = $lang_constitution_results['untranslated_lines_in_constitution']; // add untranslated license sentences
+    }
 }
 
 $num_of_untranslated_strings = 0;
@@ -73,7 +87,7 @@ if($license) {
     $license_numbers = array_pop(read_license_from_vcs($target_lang));
     $num_of_untranslated_strings = $license_numbers["untran"]; // number of all license sentences
     $diff['untranslated_sentences_in_license'] = $license_numbers['untranslated_sentences']; // add untranslated license sentences
-    $s .= '<h2 style="color: red;">Please translate license into ' . $langs[$target_lang] . ' first</h2>';
+    $s .= '<h2 style="color: red;">Please translate license fully into ' . $langs[$target_lang] . ' first</h2>';
     $s .= sprintf('<p>You can find it in <a href="http://gitweb.mageia.org/software/drakx/tree/perl-install/share/po/%s.po">git</a>. ', $target_lang); // git doesn't work yet on https
 }
 
@@ -87,14 +101,30 @@ $s .= '<p>After translation:</p>
             <li>or attach it to a mail to Mageia localization team (<code>mageia-i18n</code>, <a href="irc://irc.freenode.net/#mageia-i18n">#mageia-i18n</a> on Freenode).</li>
         </ul>';
 
-foreach (array('missing' => 'missing strings', 'notrans' => 'untranslated strings', 'extra' => 'unused (old) strings', 'dup_str' => 'duplicate strings', 'untranslated_sentences_in_license' => 'untranslated msgid strings in ' . $num_of_untranslated_strings . ' sentences in separate <a href="http://gitweb.mageia.org/software/drakx/tree/perl-install/share/po/' . $target_lang . '.po">license file</a> (in git)') as $type => $name) {
+$dest_constitution = sprintf('%s/%s_%s.md', 'href="//svnweb.mageia.org/web/www/trunk/en/about/constitution', 'mageia.org_statutes', ($constitution_readable ? $target_lang : 'en'));
+
+$issues = array(
+    'missing' => 'missing strings',
+    'notrans' => 'untranslated strings',
+    'extra' => 'unused (old) strings',
+    'dup_str' => 'duplicate strings',
+    'untranslated_lines_in_constitution' => 'of all untranslated lines in <a ' . $dest_constitution . '">constitution</a> (in svn)',
+    'untranslated_sentences_in_license' => 'untranslated msgid strings in ' . $num_of_untranslated_strings . ' sentences in separate <a href="http://gitweb.mageia.org/software/drakx/tree/perl-install/share/po/' . $target_lang . '.po">license file</a> (in git)',
+);
+foreach ($issues as $type => $name) {
 
     if (count($diff[$type]) > 0) {
         $s .= sprintf('<h2>%d %s:</h2>', count($diff[$type]), $name);
         $s .= '<pre>';
-        $prefix = (($type == 'untranslated_sentences_in_license') ? 'msgid: ' : ';');
+        if ($type == 'untranslated_lines_in_constitution') {
+            $prefix = '';
+        } else if ($type == 'untranslated_sentences_in_license') {
+            $prefix = 'msgid: ';
+        } else {
+            $prefix = ';';
+        }
         foreach ($diff[$type] as $l)
-            $s .= sprintf("%s%s\n\n", $prefix, htmlspecialchars($l));
+            $s .= sprintf("%s%s\n\n", $prefix, rtrim(htmlspecialchars($l)));
         $s .= '</pre>';
     }
 }
@@ -113,7 +143,7 @@ header('Content-Type: text/html;charset=utf-8');
     </style>
 </head>
 <body>
-    <?php echo $s; $test_1 = TRUE; $test = strip_tags($test_1); echo '1. test: ' . $test; ?>
+    <?php echo $s; ?>
 <hr>
 </body>
 </html>
diff --git a/langs/lib.php b/langs/lib.php
index 091883292..f93e53777 100644
--- a/langs/lib.php
+++ b/langs/lib.php
@@ -122,4 +122,40 @@ function get_other_langs()
     }
     array_unshift($re, 'en');
     return $re;
-}
\ No newline at end of file
+}
+
+function aproximate_number_of_untranslated_constitution_lines($app_root, $lang, $unique_lines_in_eng_constitution = array())
+{
+    $constitution_is_readable = FALSE;
+    $dest_constitution = sprintf('%s/%s/%s/%s_%s.md', $app_root, $lang, 'about/constitution', 'mageia.org_statutes', $lang);
+    $number_of_unique_lines_in_eng_constitution = count($unique_lines_in_eng_constitution);
+    if(is_readable($dest_constitution)) {
+        $unique_lines_in_constitution = array_unique(file($dest_constitution));
+        $number_of_unique_lines_in_constitution = count($unique_lines_in_constitution);
+        $constitution_is_readable = TRUE;
+        if ($lang == 'en') {
+            $aproximate_number_of_untranslated_lines = $number_of_unique_lines_in_constitution;
+            $untranslated_lines_in_constitution      = array();
+        } else {
+            $untranslated_lines_in_constitution = array_intersect($unique_lines_in_eng_constitution, $unique_lines_in_constitution);
+            $number_of_nonunique_lines_lang_constitution = count($untranslated_lines_in_constitution);
+            $ratio = $number_of_nonunique_lines_lang_constitution / $number_of_unique_lines_in_eng_constitution;
+            $limit_ratio = 0.15; // limit ratio of "allowed" untranslated lines
+            if ($ratio > $limit_ratio) {
+                // add aproximate number of untranslated constitution lines
+                $aproximate_number_of_untranslated_lines = $number_of_nonunique_lines_lang_constitution - round($limit_ratio * $number_of_unique_lines_in_eng_constitution);
+            }
+        }
+    } else {
+        $unique_lines_in_constitution             = $unique_lines_in_eng_constitution;
+        $aproximate_number_of_untranslated_lines  = $number_of_unique_lines_in_eng_constitution;
+        $untranslated_lines_in_constitution       = $unique_lines_in_eng_constitution;
+    }
+
+    return array(
+        'unique_lines_in_constitution'            => $unique_lines_in_constitution,
+        'constitution_readable'                   => $constitution_is_readable,
+        'untranslated_lines_in_constitution'      => $untranslated_lines_in_constitution,
+        'aproximate_number_of_untranslated_lines' => $aproximate_number_of_untranslated_lines,
+    );
+}
diff --git a/langs/report.php b/langs/report.php
index 6f8fcbb8a..63cfde1e7 100644
--- a/langs/report.php
+++ b/langs/report.php
@@ -38,14 +38,24 @@
     /**
     */
     $one_language = isset($_GET['l']) ? strip_tags(trim($_GET['l'])) : null;
+    $one_resource = isset($_GET['r']) ? strip_tags(trim($_GET['r'])) : null;
     include 'lib.php';
     define('APP_ROOT', realpath(__DIR__ . '/..'));
 
-    $enFiles    = get_lang_references();
-    // added navigation file
-    array_unshift($enFiles, '../_nav/langs/en.lang');
+    $restore_languages = FALSE;
+    $restore_resources = FALSE;
+
+    if (isset($one_resource) && $one_resource != 'all') {
+        $enFiles    = array($one_resource);
+        $restore_resources = TRUE;
+    } else {
+        $enFiles    = get_lang_references();
+        array_unshift($enFiles, '../_nav/langs/en.lang'); // added navigation file
+    }
+
     if (isset($one_language) && $one_language != 'all') {
         $otherLangs = array('en', $one_language);
+        $restore_languages = TRUE;
     } else {
         $otherLangs = get_other_langs();
     }
@@ -59,13 +69,15 @@
     $diff_link      = '<a href="diff.php?s=%s&amp;l=%s" title="see detailed diff">';
 
     $languages      = array();
+    $unique_lines_in_eng_constitution = 0;
+    $number_of_unique_lines_in_eng_constitution = 0;
 
     foreach ($otherLangs as $l) {
 
         $stats['en']['files'] += 1;
         $stats[$l]['files']    = 0;
         $stats[$l]['strings']  = 0;
-        if ($l == 'en') {
+        if ($l == 'en' || $restore_languages) {
             $link_one_language = '%s';
         } else {
             $link_one_language = '<a href="?l=' . $l . '" title="see this language only">%s</a>';
@@ -118,14 +130,13 @@
                 $num_of_duplicated_strings   = count($test['dup_str']);
 
                 if ($link == 'about/constitution') {
-                    $dest_constitution = sprintf('%s/%s/%s/%s_%s.md', APP_ROOT, $l, 'about/constitution', 'mageia.org_statutes', $l);
+                    $constitution_results = aproximate_number_of_untranslated_constitution_lines(APP_ROOT, $l, $unique_lines_in_eng_constitution);
                     if ($l == 'en') {
-                        $strings_en_constitution = count(array_unique(file($dest_constitution)));
-                    }
-                    $test['a'] += $strings_en_constitution; // add aproximate number of strings from constitution to translate
-                    if(!is_readable($dest_constitution) || $l == 'en') {
-                        $num_of_untranslated_strings += $strings_en_constitution; // add all constitution strings
+                        $unique_lines_in_eng_constitution = $constitution_results['unique_lines_in_constitution'];
+                        $number_of_unique_lines_in_eng_constitution = count($unique_lines_in_eng_constitution);
                     }
+                    $test['a'] += $number_of_unique_lines_in_eng_constitution; // add aproximate number of lines from constitution to translate
+                    $num_of_untranslated_strings += $constitution_results['aproximate_number_of_untranslated_lines'];
                 }
                 if ($link == 'about/license') {
                     require_once('../en/about/license/license.php');
@@ -148,17 +159,17 @@
                     );
                 } else if ($link == '2') {
                     $mga = array(
-                        'A' => '2/download_index.php',
-                        'B' => '2/for-pc/index.php',
-                        'C' => '2/for-server/index.php',
-                        'D' => '2/index.php',
+                        'A' => '2/index.php',
+                        'B' => '2/download_index.php',
+                        'C' => '2/for-pc/index.php',
+                        'D' => '2/for-server/index.php',
                     );
                 } else if ($link == '3') {
                     $mga = array(
-                        'A' => '3/download_index.php',
-                        'B' => 'for-pc/index.php',
-                        'C' => 'for-server/index.php',
-                        'D' => '3/index.php',
+                        'A' => '3/index.php',
+                        'B' => '3/download_index.php',
+                        'C' => 'for-pc/index.php',
+                        'D' => 'for-server/index.php',
                     );
                 } else {
                     $mga = array(
@@ -228,7 +239,7 @@
             $num_of_col++;
             if ($num_of_col % $num_clmns_to_repeat == 0 && $num_of_col <= count($enFiles) - ($num_clmns_to_repeat / 3)) {
                 if ($l == 'en') {
-                    $cols .= $lang_coloumn = '<td style="font-weight: bold;"><a href="?l=all" title="see all languages">Language</a></td>';
+                    $cols .= $lang_coloumn = '<td style="font-weight: bold;">Language</td>';
                 } else {
                     $cols .= $lang_coloumn = sprintf('<td style="font-weight: bold; vertical-align: top;">' . $link_one_language . '<br><span style="font-weight: normal; font-size: smaller;">%s</span></td>', $langs[$l], $l);
                 }
@@ -257,33 +268,50 @@
     $en_language = array_shift($languages); // shift English for proper sorting
     krsort($languages, SORT_NUMERIC);
 
-    $enFiles = array_map(function ($e) { return str_replace('en/', '', $e); }, $enFiles);
     // add language coloumn repeating
     $lang_line = array();
     $num_of_h_col = 0;
+    $add_last_coloumn = FALSE;
+    $resources = ($restore_resources ? '&nbsp;' : count($enFiles) . ' resources');
     foreach ($enFiles as $lang_chunk) {
         $num_of_h_col++;
-        $lang_line[] = $lang_chunk;
+        if ($restore_resources) {
+            $lang_line[] = str_replace('en/', '', $lang_chunk);
+        } else {
+            $lang_line[] = '<a href="?r=' . $lang_chunk . '" title="see only this resource">' . str_replace('en/', '', $lang_chunk) . '</a>';
+        }
         if ($num_of_h_col % $num_clmns_to_repeat == 0 && $num_of_h_col <= count($enFiles) - ($num_clmns_to_repeat / 3)) {
-            $lang_line[] = '&nbsp;';
+            $lang_line[] = $resources;
+            $add_last_coloumn = TRUE;
         }
     }
-    $lang_line[] = '&nbsp;';// last h coloumn
-    
+    if ($add_last_coloumn) {
+        $lang_line[] = $resources;// last h coloumn
+    }
+
     $thfiles = '<th>' . implode('</th><th>', $lang_line) . '</th>' . PHP_EOL;
-    $count   = count($otherLangs);
+    if ($restore_languages) {
+        echo '<p>Restore <a href="?l=all" title="see all languages">all languages</a>.</p>';
+        $count = '&nbsp;';
+    } else {
+        $count = count($otherLangs) . ' languages';
+    }
+    if ($restore_resources) {
+        echo '<p>Restore <a href="?r=all" title="see all resources">all resources</a>.</p>';
+    }
     $chunks  = array_chunk($languages, $num_lines_to_repeat = 8); // add header repeating
     $table_body = array();
     foreach ($chunks as $chunk) {
-        $table_body = array_merge($table_body, $chunk, array(count($chunk) > ($num_lines_to_repeat / 2) ? '<tr><th>&nbsp;</th><th>File</th>' . $thfiles : ''));
+        $table_body = array_merge($table_body, $chunk, array(count($chunk) > ($num_lines_to_repeat / 2) ? '<tr><th>' . $count . '</th><th>' . $resources . '</th>' . $thfiles : ''));
     }
     array_unshift($table_body, $en_language); // unshift English back
     $s = implode($table_body);
 
+
     echo <<<S
 <table border="1">
 <thead><tr>
-    <th><a href="?l=all" title="see all languages">{$count} languages</a></th>
+    <th>{$count}</th>
     <th>Progress</th>
     {$thfiles}
 </tr></thead>