[DOC-WEB] [web-doc] master: Show the do-not-translate and broken XML statuses (#68)

Author: Louis-Arnaud (lacatoire)
Committer: GitHub (web-flow)
Pusher: jordikroon
Date: 2026-08-11T10:42:29+02:00

Commit: https://github.com/php/web-doc/commit/eda769a2292ef50f267362e4eb7a4086b9d62c0f
Raw diff: https://github.com/php/web-doc/commit/eda769a2292ef50f267362e4eb7a4086b9d62c0f.diff

Show the do-not-translate and broken XML statuses (#68)

Two pages following the shape of the existing ones, their menu
entries, and the two statuses in $TRANSLATION_STATUSES.

Files marked do not translate are kept out of the completion totals,
matching doc-base, so no translation sees its rate drop for files it
is not expected to have. Their share of a total they are not part of
shows as n/a. Other statuses are displayed as before.

Co-authored-by: lacatoire <[email protected]>

Changed paths:
  M  include/lib_general.inc.php
  M  include/lib_revcheck.inc.php
  M  www/revcheck.php


Diff:

diff --git a/include/lib_general.inc.php b/include/lib_general.inc.php
index 1b7ced9..af48826 100644
--- a/include/lib_general.inc.php
+++ b/include/lib_general.inc.php
@@ -100,6 +100,8 @@ function nav_languages($lang = null)
             $out .= '<li><a href="/revcheck.php?p=misstags&amp;lang='.$lang.'">Missing revision numbers</a></li>';
             $out .= '<li><a href="/revcheck.php?p=missfiles&amp;lang='.$lang.'">Untranslated files</a></li>';
             $out .= '<li><a href="/revcheck.php?p=oldfiles&amp;lang='.$lang.'">Not in EN tree</a></li>';
+            $out .= '<li><a href="/revcheck.php?p=brokenfiles&amp;lang='.$lang.'">Broken XML</a></li>';
+            $out .= '<li><a href="/revcheck.php?p=donottranslate&amp;lang='.$lang.'">Do not translate</a></li>';
             $out .= '</ul>';
         }
         $out .= '</li>';
diff --git a/include/lib_revcheck.inc.php b/include/lib_revcheck.inc.php
index 882452b..7cf8479 100644
--- a/include/lib_revcheck.inc.php
+++ b/include/lib_revcheck.inc.php
@@ -27,8 +27,15 @@
     'RevTagProblem' => 'No revision tag',
     'NotInEnTree'   => 'Not in EN tree',
     'Untranslated'  => 'Available for translation',
+    'XmlBroken'     => 'Broken XML',
+    'DoNotTranslate' => 'Marked do not translate',
 ];
 
+// Files marked do not translate are never expected to be translated, so
+// they are listed but never counted against a translation. This matches
+// the totals of doc-base scripts/revcheck.php and genrevdb.php.
+$TRANSLATION_STATUSES_OFF_TOTAL = [ 'DoNotTranslate' ];
+
 function get_language_intro($idx, $lang) {
     $result = $idx->query("SELECT intro FROM languages WHERE lang = '$lang'");
     $answer = $result->fetchArray();
@@ -161,6 +168,40 @@ function get_oldfiles($idx, $lang)
     return $tmp;
 }
 
+function get_brokenfiles($idx, $lang)
+{
+    $sql = <<<SQL
+        SELECT path AS dir, name AS name, xmlError AS error, size / 1024 AS size
+          FROM files
+         WHERE lang = '{$lang}' AND status = 'XmlBroken'
+         ORDER BY dir, name
+    SQL;
+
+    $result = $idx->query($sql);
+    $tmp = array();
+    while ($r = $result->fetchArray(SQLITE3_ASSOC)) {
+        $tmp[] = $r;
+    }
+    return $tmp;
+}
+
+function get_donottranslate($idx, $lang)
+{
+    $sql = <<<SQL
+        SELECT path AS dir, name AS name, size / 1024 AS size
+          FROM files
+         WHERE lang = '{$lang}' AND status = 'DoNotTranslate'
+         ORDER BY dir, name
+    SQL;
+
+    $result = $idx->query($sql);
+    $tmp = array();
+    while ($r = $result->fetchArray(SQLITE3_ASSOC)) {
+        $tmp[] = $r;
+    }
+    return $tmp;
+}
+
 function get_misstags($idx, $lang)
 {
     $sql = <<<SQL
@@ -221,9 +262,12 @@ function get_lang_stats($idx, $lang) {
     $stats = [];
     $total = [ 'total' => 0, 'size' => 0 ];
 
+    global $TRANSLATION_STATUSES_OFF_TOTAL;
+
     while ($row = $result->fetchArray(SQLITE3_ASSOC)) {
         $stats[$row['status']] = $row;
-        if ($row['status'] != 'NotInEnTree') {
+        if ($row['status'] != 'NotInEnTree'
+            && !in_array($row['status'], $TRANSLATION_STATUSES_OFF_TOTAL)) {
             $total['total'] += $row['total'];
             $total['size'] += $row['size'];
         }
diff --git a/www/revcheck.php b/www/revcheck.php
index 28e2bc0..7d30004 100644
--- a/www/revcheck.php
+++ b/www/revcheck.php
@@ -180,6 +180,63 @@
  echo gen_date($DBLANG);
  break;
 
+ case 'brokenfiles':
+     $brokenfiles = get_brokenfiles($dbhandle, $lang);
+
+     if (!$brokenfiles) {
+         echo '<p>Good, all translated files are valid XML.</p>';
+     } else {
+         $num = count($brokenfiles);
+         echo '<p>These files do not parse as XML. The manual build may fail, or silently ';
+         echo 'drop their contents, so they need a fix before anything else.</p>';
+         echo '<table class="c">';
+         echo '<tr><th>Broken XML files ('.$num.' files):</th><th>Error</th><th>kB</th></tr>';
+
+         $last_dir = false;
+         foreach ($brokenfiles as $row) {
+             if (!$last_dir || $last_dir != $row['dir']) {
+                 echo '<tr><th colspan="3">'.htmlspecialchars($row['dir']).'</th></tr>';
+                 $last_dir = $row['dir'];
+             }
+             echo '<tr>',
+                 '<td>', htmlspecialchars($row['name']), '</td>',
+                 '<td>', htmlspecialchars($row['error']), '</td>',
+                 '<td>', $row['size'], '</td>',
+                 '</tr>';
+         }
+         echo '</table>';
+     }
+     echo gen_date($DBLANG);
+ break;
+
+ case 'donottranslate':
+     $donottranslate = get_donottranslate($dbhandle, $lang);
+
+     if (!$donottranslate) {
+         echo '<p>No source file is marked do not translate.</p>';
+     } else {
+         $num = count($donottranslate);
+         echo '<p>These English files carry a <code>&lt;?do-not-translate?&gt;</code> mark. They are ';
+         echo 'not expected to be translated, and are left out of the totals of this translation.</p>';
+         echo '<table class="c">';
+         echo '<tr><th>Marked do not translate ('.$num.' files):</th><th>kB</th></tr>';
+
+         $last_dir = false;
+         foreach ($donottranslate as $row) {
+             if (!$last_dir || $last_dir != $row['dir']) {
+                 echo '<tr><th colspan="2">'.htmlspecialchars($row['dir']).'</th></tr>';
+                 $last_dir = $row['dir'];
+             }
+             echo '<tr>',
+                 '<td>', htmlspecialchars($row['name']), '</td>',
+                 '<td>', $row['size'], '</td>',
+                 '</tr>';
+         }
+         echo '</table>';
+     }
+     echo gen_date($DBLANG);
+ break;
+
  case 'misstags':
      $misstags = get_misstags($dbhandle, $lang);
 
@@ -211,16 +268,21 @@
      echo '<tr><th>File status type</th><th>Number of files</th><th>Percent of files</th><th>Size of files (kB)</th><th>Percent of size</th></tr>';
 
      foreach ($TRANSLATION_STATUSES as $status => $description) {
+         // A status kept out of the total has no meaningful share of it.
+         $offTotal = in_array($status, $TRANSLATION_STATUSES_OFF_TOTAL);
+
          echo
             '<tr>',
             '<td>', $description, '</td>',
             '<td>', $stats[$status]['total'] ?? 0, '</td>',
             '<td>',
-            sprintf('%.2f%%', 100 * (($stats[$status]['total'] ?? 0) / $stats['total']['total'])),
+            $offTotal ? 'n/a'
+                : sprintf('%.2f%%', 100 * (($stats[$status]['total'] ?? 0) / $stats['total']['total'])),
             '</td>',
             '<td>', $stats[$status]['size'] ?? 0, '</td>',
             '<td>',
-            sprintf('%.2f%%', 100 * (($stats[$status]['size'] ?? 0) / $stats['total']['size'])),
+            $offTotal ? 'n/a'
+                : sprintf('%.2f%%', 100 * (($stats[$status]['size'] ?? 0) / $stats['total']['size'])),
             '</td>',
             '</tr>';
      }