added string comparison between git and TX (currently for one file only)

http://mag/langs/report_tx_git.php?c=Webpages&l=sl&r=en/404.pot
http://mag/langs/report_tx_git.php?c=Documentation&l=ro&r=doc_installer
http://mag/langs/report_tx_git.php?c=Cauldron&l=de&r=drakx_standalone

Note that mageiasync (not in gettext format) and page-nav (different location & filename) are not yet working correctly.
Some links for calling the comparison will be needed too.

1 files changed, 149 insertions, 43 deletions

diff --git a/langs/report_tx_git.php b/langs/report_tx_git.php
index 6bd3732ab..840e381bb 100644
--- a/langs/report_tx_git.php
+++ b/langs/report_tx_git.php
@@ -36,9 +36,9 @@ define('HLANG', TRUE);
 include 'lib.php';
 
 $errors = array(); // stored for error management ;)
-$c = get_sane_string('c');
-if (in_array($c, array('Webpages', 'Documentation', 'Cauldron'))) {
-    $resource_type = $c; // filter only valid resource types
+$wanted_category = get_sane_string('c');
+if (in_array($wanted_category, array('Webpages', 'Documentation', 'Cauldron'))) {
+    $resource_type = $wanted_category; // filter only valid resource types
 } else {
     $resource_type = 'Webpages'; // default
 }
@@ -55,6 +55,10 @@ if (!empty($wanted_language)) {
 //      uz@Cyrl     vs. uz@cyrillic.po
 //      en_GB       vs. en_gb.po        exception: http://gitweb.mageia.org/web/identity/plain/lib/CatDap/I18N
 $wanted_resource = get_sane_string('r');
+$wanted_resource_name = '';
+if (!empty($wanted_resource)) {
+    $wanted_resource_name = str_replace(array('../_nav/langs/en', 'en/', '.pot'), array('nav', '', ''), $wanted_resource);
+}
 
 if ('Documentation' == $resource_type) {
     $git_resources = array(
@@ -280,21 +284,16 @@ function tx_call($tx_request, $project = 'project/mageia/')
  *
  * @return string $resource_name like 'page-about-constitution' (tx name)
 */
-function resource_name_conversion($resource_name, $category = '', $tx_to_git_name_conversion = FALSE)
+function resource_name_conversion($resource_name, $category = '')
 {
-    $tx_names  = array('/', 'mageia-welcome', 'identity-catdap');
-    $git_names = array('-', 'Mageia%20Welcome', 'Identity%20(CatDap)');
-    if ($tx_to_git_name_conversion) {
-        $resource_name = str_replace($tx_names, $git_names, $resource_name);
-    } else {
-        $resource_name = str_replace($git_names, $tx_names, $resource_name);
-    }
     if ('Webpages' == $category) {
-        if ($tx_to_git_name_conversion) {
-            $resource_name = str_replace('page-', '', $resource_name);
-        } else {
-            $resource_name = 'page-' . $resource_name;
-        }
+        $tx_names  = array('nav', '-');
+        $git_names = array('_nav/langs/en', '/');
+        $resource_name = 'page-' . str_replace($git_names, $tx_names, $resource_name);
+    } else if ('Cauldron' == $category) {
+        $tx_names  = array('mageia-welcome', 'identity-catdap');
+        $git_names = array('Mageia%20Welcome', 'Identity%20(CatDap)');
+        $resource_name = str_replace($git_names, $tx_names, $resource_name);
     }
 
     return $resource_name;
@@ -395,7 +394,7 @@ function build_transifex_link($tx_language_code, $link_name = NULL, $tx_category
  *
  * @return array
 */
-function generating_report($language_codes, $resource_names, $path = NULL, $pot_name = NULL)
+function generating_report($language_codes, $resource_names, $path = NULL, $pot_name = NULL, $compared = NULL)
 {
     $report = array();
     foreach ($resource_names as $f) {
@@ -417,7 +416,7 @@ function generating_report($language_codes, $resource_names, $path = NULL, $pot_
                     $partial_pot_name = substr($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);
+                    $stat = _po_diff($l, $resource, $source_strings, $path, $compared);
                 }
                 $num_of_fuzzy_or_missing = count($stat['fuzzy_or_missing']);
             } else { // file $langF doesn't exists in 'Webpages' $resource_type
@@ -443,6 +442,10 @@ function generating_report($language_codes, $resource_names, $path = NULL, $pot_
             $num_of_not_fully_trans = $num_of_fuzzy_or_missing + $num_of_untranslated;
             $report[] = array(
                 'num_of_all_strings'     => $stat['a'],
+//                 'fuzzy_or_missing_str'   => $stat['fuzzy_or_missing'],
+//                 'untranslated_strings'   => $stat['notrans'],
+//                 'source_strings'         => $stat['source_strings'],
+                'differences'            => $stat['differences'],
                 'resource_name'          => $resource_name,
                 'web_language_code'      => $web_language_code,
                 'num_of_not_fully_trans' => $num_of_not_fully_trans,
@@ -486,36 +489,74 @@ function get_language_name($language_code)
 // get resources data from TX
 $tx_resources = tx_call("resources");
 
+$details_wanted = FALSE;
+if (!empty($wanted_language) && !empty($wanted_resource_name)) {
+    $details_wanted = TRUE;
+}
+
+if ($details_wanted) {
+    $prefixed_resource_name = resource_name_conversion($wanted_resource_name, $wanted_category);
+    $tx_file = tx_call("resource/$prefixed_resource_name/translation/$wanted_language/");
+    $tx_file_content = $tx_file['content'];
+    $tx_file_content = explode("\n", $tx_file_content);
+    $parsed_tx_file_content = phpmo_parse_po_file($tx_file_content, FALSE);
+} else {
+    $parsed_tx_file_content = NULL;
+}
+
 // create separate array ($tx_resources_info) and add statistics to it from TX
 $tx_resources_info     = array();
 $tx_cauldron_resources = array();
 $tx_stat_for_resource  = array();
 foreach ($tx_resources as $one_resource) {
-    $tx_resource_name     = $one_resource['slug'];
-    $category             = $one_resource['categories'][0];
+    $tx_resource_name = $one_resource['slug'];
+    $category         = $one_resource['categories'][0];
+    $add_data         = FALSE;
     if ('Cauldron' == $category) {
         $tx_cauldron_resources[] = $tx_resource_name;
     }
-    // limit resource type only on one
+    // limit resource type to one
     if ($category == $resource_type) {
-        if (empty($wanted_resource) || $tx_resource_name == $wanted_resource) {
-            // add statistic
-            if (empty($wanted_language)) {
-                $tx_stat_for_resource = tx_call("resource/$tx_resource_name/stats");
-            } else {
-                $tx_stat_for_resource[$wanted_language] = tx_call("resource/$tx_resource_name/stats/$wanted_language");
+        if ('Cauldron' == $category) { // add data for Cauldron category
+            // if name is required
+            if (!empty($wanted_resource) && $tx_resource_name == resource_name_conversion($wanted_resource_name, $category)) {
+                $add_data = TRUE;
+            // if only language is required
+            } else if (!empty($wanted_language) && empty($wanted_resource)) {
+                $add_data = TRUE;
             }
-            $tx_resource_name     = str_replace('page-', '', $tx_resource_name); // unify resource names
-            $one_resource['tx_resource_name'] = $tx_resource_name;
-            $one_resource['statistic']        = $tx_stat_for_resource;
-            $tx_resources_info[$tx_resource_name] = $one_resource;
+        } else { // add data for Webpages and Documentation category
+            if (empty($wanted_resource) && empty($wanted_language)) {
+                $add_data = TRUE;
+            }
+        }
+        // add data if details are needed
+        if ($details_wanted && $tx_resource_name == resource_name_conversion($wanted_resource_name, $category)) {
+            $add_data = TRUE;
+        }
+    }
+    if ($add_data) {
+        // add statistic
+        if (empty($wanted_language)) {
+            $tx_stat_for_resource = tx_call("resource/$tx_resource_name/stats");
+        } else {
+            $tx_stat_for_resource[$wanted_language] = tx_call("resource/$tx_resource_name/stats/$wanted_language");
         }
+        $tx_resource_name = str_replace('page-', '', $tx_resource_name); // unify resource names
+        $one_resource['tx_resource_name'] = $tx_resource_name;
+        $one_resource['statistic']        = $tx_stat_for_resource;
+        $tx_resources_info[$tx_resource_name] = $one_resource;
     }
 }
 $tx_languages = array();
 if (empty($wanted_language)) {
-    foreach ($tx_stat_for_resource as $tx_resource_language => $tx_resource_language_stat) {
-        $tx_languages[] = $tx_resource_language;
+    if ('Webpages' == $resource_type || 'Documentation' == $resource_type) {
+        $tx_stat_for_resource = tx_call("resource/page-index/stats");
+        $tx_languages = array_keys($tx_stat_for_resource);
+    } else {
+        foreach ($tx_stat_for_resource as $tx_resource_language => $tx_resource_language_stat) {
+            $tx_languages[] = $tx_resource_language;
+        }
     }
 } else {
     $tx_stat_for_resource = tx_call("resource/$tx_resource_name/stats");
@@ -525,8 +566,13 @@ if (empty($wanted_language)) {
 // create array ($report) with statistics from git
 $report = array();
 if ('Webpages' == $resource_type) {
-    $enFiles = array_merge(array('../_nav/langs/en.pot'), get_lang_references('*.pot')); // added navigation file
-    $report = generating_report(get_other_langs(), $enFiles);
+    if (empty($wanted_resource) && empty($wanted_language)) {
+        $enFiles = array_merge(array('../_nav/langs/en.pot'), get_lang_references('*.pot')); // added navigation file
+        $report = generating_report(get_other_langs(), $enFiles);
+    } else {
+        $enFiles = array($wanted_resource);
+        $report = generating_report(array($wanted_language), $enFiles, NULL, NULL, $parsed_tx_file_content);
+    }
 } else {
     $tx_resources_not_yet_in_git = array();
     $translated_in_tx = ' translated in Transifex';
@@ -564,7 +610,7 @@ if ('Webpages' == $resource_type) {
                 $errors['lang_not_in_git'] = "Language $wanted_lang_name is not present in git for $resources_not_yet_in_git resource.";
             }
         }
-        $one_report = generating_report(array_values($report_language_codes), array($tx_name), $git_path, $pot_name);
+        $one_report = generating_report(array_values($report_language_codes), array($tx_name), $git_path, $pot_name, $parsed_tx_file_content);
         $report = array_merge($one_report, $report);
     }
 }
@@ -614,14 +660,16 @@ foreach ($tx_resources_info as $tx_resource_info) {
                 $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'];
+                $git_resource_differences        = $git_resource_info['differences'];
 
-                // compare l10n level
-                if ($tx_resource_untrans_in_lang != $git_resource_untrans_in_lang) {
+                // compare l10n level or differences
+                if ($tx_resource_untrans_in_lang != $git_resource_untrans_in_lang || count($git_resource_differences) != 0) {
                     $tx_git_difference[$tx_resource_language][$git_resource_name]['tx_num_of_all_strings'] = $tx_resource_all_strings_in_lang;
                     $tx_git_difference[$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;
+                    $tx_git_difference[$tx_resource_language][$git_resource_name]['differences']           = $git_resource_differences;
                 }
                 // it seems like a parse error
                 if (0 == $git_resource_num_of_all_strings) {
@@ -743,6 +791,58 @@ $all_resources_with_tx_git_diff = array_unique($all_resources_with_tx_git_diff);
 sort($all_resources_with_tx_git_diff, SORT_STRING);
 $num_of_resources_w_difference = count($all_resources_with_tx_git_diff);
 
+// preparing a detailed strings difference html table represantation between git and tx
+if ($details_wanted) {
+    $table_rows   = array();
+    $table_rows[] = "<table>" . PHP_EOL;
+    $table_rows[] = "<thead>" . PHP_EOL;
+    $table_rows[] = "<tr class=\"add\"><th>Original string</th><th>Translated string from git</th><th>And translated string from Transifex</th></tr>" . PHP_EOL; // add header row
+    $table_rows[] = "</thead>" . PHP_EOL;
+    $table_rows[] = "<tbody>" . PHP_EOL;
+    $num_of_differences = 0;
+    $col_repeat = 8;
+    $i = 0;
+    foreach ($tx_git_difference as $one_language => $one_language_array) {
+        $differences = $one_language_array[$wanted_resource_name]['differences'];
+        $num_of_differences = count($differences);
+        foreach ($differences as $curr_msgid => $curr_context_array) {
+            foreach ($curr_context_array as $curr_context_or_num => $array_for_compare) {
+                if (is_string($curr_context_or_num)) {
+                    $curr_msgid .= " (Context: $curr_context_or_num)";
+                }
+                $curr_msgid = htmlentities($curr_msgid, null, 'UTF-8');
+                $num_of_tr_str = max(count($array_for_compare['target_l']), count($array_for_compare['compared'])) - 1;
+                for ($i = 0; $i <= $num_of_tr_str; $i++) {
+                    if (empty($array_for_compare['target_l'][$i])) {
+                        $target_l_str = '&nbsp;';
+                    } else {
+                        $target_l_str = htmlentities($array_for_compare['target_l'][$i], null, 'UTF-8');
+                    }
+                    if (empty($array_for_compare['compared'][$i])) {
+                        $compared_str = '&nbsp;';
+                    } else {
+                        $compared_str = htmlentities($array_for_compare['compared'][$i], null, 'UTF-8');
+                    }
+                    $row  = "<tr><td class=\"add\">$curr_msgid</td>";
+                    $row .= "<td>$target_l_str</td>";
+                    $row .= "<td>$compared_str</td></tr>";
+                    $table_rows[] = $row . PHP_EOL;
+                    // instead of msgid use msgid_plural for plural table lines
+                    if (is_array($array_for_compare['target_l'])) {
+                        if (array_key_exists('msgid_plural', $array_for_compare['target_l'])) {
+                            $curr_msgid = str_replace($curr_msgid, $array_for_compare['target_l']['msgid_plural'], $curr_msgid);
+                        }
+                    }
+                }
+            }
+        }
+    }
+    $table_rows[] = "</tbody>" . PHP_EOL;
+    $table_rows[] = "</table>" . PHP_EOL;
+
+    $table_w_details = (0 != $num_of_differences ? implode($table_rows) : '');
+}
+
 // preparing a html table represantation of git/tx difference in number of untranslated strings
 $table_difference = '';
 $num_for_switch_table_to_list = 15;
@@ -755,12 +855,13 @@ if (0 < $num_of_resources_w_difference && $num_for_switch_table_to_list > $num_o
     $table_rows[]   = "<table>" . PHP_EOL;
     $table_rows[]   = "<thead>" . PHP_EOL;
     $first_row_pass = TRUE;
+    $col_repeat = 8;
     foreach ($tx_git_difference as $one_language => $one_language_array) {
         $j   = 0;
         $row = array();
         $first_col_pass = TRUE;
         if ($first_row_pass || 0 == $i % 12) {
-            $array_chunks = array_chunk($all_resources_with_tx_git_diff, $col_repeat = 8, TRUE);
+            $array_chunks = array_chunk($all_resources_with_tx_git_diff, $col_repeat, TRUE);
             foreach ($array_chunks as $array_chunk) {
                 $row = array_merge($row, array('&nbsp;'), $array_chunk);
             }
@@ -808,7 +909,7 @@ if (0 < $num_of_resources_w_difference && $num_for_switch_table_to_list > $num_o
 $list_of_lang_diff     = '';
 $list_of_resource_diff = '';
 $list_num_diff = FALSE;
-if (0 < $num_of_resources_w_difference && !$table_num_diff) {
+if (0 < $num_of_resources_w_difference && !$details_wanted && !$table_num_diff) {
     $list_num_diff = TRUE;
     // preparing a list represantation of Tx/git differences by language
     $list_of_lang_diff .= '<ul>' . PHP_EOL;
@@ -881,15 +982,17 @@ foreach ($tx_languages as $tx_language) {
     $languages_links[] = "<a href=\"./report_tx_git.php?c=Cauldron&amp;l=$tx_language\" title=\"$language_name\">$tx_language</a>";
 }
 sort($languages_links, SORT_STRING);
+$num_of_tx_langs = count($languages_links);
 $languages_link  = implode(", ", $languages_links);
 $resources_links = array();
 foreach ($tx_cauldron_resources as $tx_resource_name) {
     $resources_links[] = "<a href=\"./report_tx_git.php?c=Cauldron&amp;r=$tx_resource_name\">$tx_resource_name</a>";
 }
+$num_of_tx_cauldron_resources = count($resources_links);
 $resources_link = implode(", ", $resources_links);
 
 $reports_links  = '<p>You can also see reports about differences in ' . implode(" or ", $links) . '.</p>';
-$reports_links .= "<p>For software translations please choose language ($languages_link) or resource ($resources_link).</p>";
+$reports_links .= "<p>For software translations please choose one of it's $num_of_tx_cauldron_resources resources ($resources_link) or one of a $num_of_tx_langs languages ($languages_link).</p>";
 $reports_links .= '<p>There is a nice ';
 $reports_links .= '<a href="https://wiki.mageia.org/en/Transifex_and_git_syncing">wiki page about Transifex and git synchronising</a>.</p>';
 echo $reports_links . PHP_EOL;
@@ -923,8 +1026,11 @@ if (0 < $num_of_resources_w_difference) {
     echo $diff_title;
     echo '<p>To avoid further alienation and confusion please synchronise them.</p>';
     echo '<p>They differs on number of untranslated strings in resources:</p>';
-    // print table if there are only a few resources with differences otherwise print a list
-    if ($table_num_diff) {
+    // print a detailed strings difference table if wanted
+    if ($details_wanted) {
+        echo $table_w_details . PHP_EOL;
+    // else print table if there are only a few resources with differences otherwise print a list
+    } else if ($table_num_diff) {
         echo $table_difference . PHP_EOL;
     } else if ($list_num_diff) {
         echo $list_of_lang_diff . PHP_EOL;