first steps as a preparation of l10n report (only diff.php for now - 3 exceptions) to gettext system

1 files changed, 95 insertions, 4 deletions

diff --git a/langs/lib.php b/langs/lib.php
index f7422167f..05740418d 100644
--- a/langs/lib.php
+++ b/langs/lib.php
@@ -21,13 +21,12 @@ include '../langs.inc.php';
  * - missing strings (in $a, not in $b)
  * - extra strings (in $b, not in $a)
  * - untranslated strings (same in $a and $b, or empty in $b)
+ * - duplicate strings in $b
  *
  * @param string $a file name
  * @param string $b file name
  *
  * @return array
- *
- * @todo some strings may be left untranslated on purpose
 */
 function _lang_diff($a, $b)
 {
@@ -70,7 +69,60 @@ function _lang_diff($a, $b)
     );
 }
 
-function _lang_diff_stats($a, $b)
+/**
+ * Diff pot and po files, to get:
+ * - source (pot) 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 $resource file name ('about/license')
+ *
+ * @return array
+*/
+function _po_diff($locale, $resource)
+{
+    $source_l = read_translation_file('en', $resource);
+    $target_l = read_translation_file($locale, $resource);
+
+    $pot_strings = array();
+    $untrans = array();
+    $missing = array();
+
+    foreach ($source_l as $escaped_string => $subarray) {
+        if ($escaped_string == $subarray["msgid"]) { // filter out header
+            $pot_strings[$escaped_string] = $subarray["msgid"];
+        }
+    }
+
+    foreach ($target_l as $escaped_string => $subarray) {
+        if ($escaped_string == $subarray["msgid"]) { // filter out header
+            $po_strings[$escaped_string] = $subarray["msgstr"][0];
+        }
+    }
+
+    foreach ($pot_strings as $escaped_string => $translated_string) {
+        if (isset($po_strings[$escaped_string])) {
+            if (empty($po_strings[$escaped_string])) {
+                $untrans[] = $escaped_string;
+            }
+        } else {
+            $missing[] = $escaped_string;
+        }
+    }
+
+    return array(
+        'a'       => count($pot_strings), // # of original strings
+//        'b'       => count($po_strings), // # of target strings
+        'missing' => $missing, // probably could be empty array() of missing strings
+        'notrans' => $untrans, // array of untranslated strings
+        'extra'   => array(), // could be empty array() of unused (old) strings
+        'dup_str' => array(), // could be empty array() of duplicate strings
+    );
+}
+
+/*function _lang_diff_stats($a, $b)
 {
     $diff = _lang_diff($a, $b);
 
@@ -81,7 +133,7 @@ function _lang_diff_stats($a, $b)
     $diff['correct'] = $diff['b'] - $diff['notrans'] - $diff['missing'];
 
     return $diff;
-}
+} /**/
 
 if ( ! function_exists('glob_recursive'))
 {
@@ -111,12 +163,51 @@ if ( ! function_exists('glob_recursive'))
     }
 }
 
+/**
+ * Create 'sl/about/license.sl.lang'
+ * from 'en/about/license.en.lang'
+ *
+ * @param string $s file name with path
+ * @param string $l locale name
+ *
+ * @return string
+*/
 function _lang_file_switch($s, $l)
 {
     $s = str_replace('en.lang', $l . '.lang', $s);
     return str_replace('en/', $l . '/', $s);
 }
 
+/**
+ * Create 'sl/about/license.po'
+ * from 'en/about/license.pot'
+ *
+ * @param string $s file name with path
+ * @param string $l locale name
+ *
+ * @return string
+*/
+function _po_file_switch($s, $l)
+{
+    $s = str_replace('.pot', '.po', $s);
+    return str_replace('en/', $l . '/', $s);
+}
+
+/**
+ * Create 'about/license'
+ * from 'en/about/license.pot' or 'en/about/license.en.lang'
+ *
+ * @param string $source_file file name with path
+ * @param string $extension file extension to remove
+ *
+ * @return string
+*/
+function _extract_resource($source_file, $extension = '.pot')
+{
+    $resource = str_replace($extension, '', $source_file);
+    return str_replace('en/', '', $resource);
+}
+
 function get_lang_references()
 {
     return glob_recursive('en/*', GLOB_MARK);