Allow edition of simplified/traditional scripts in Chinese sentences

src/Lib/Autotranscription.php

@@ -163,6 +163,37 @@ private function _unpack_grouped_furigana(&$group)
         return $formatted;
     }
 
+    private function _errorIfNonEqual(&$errors, $sentence, $transcr) {
+        if ($sentence !== $transcr) {
+            /* Find the first character that differs */
+            $character = mb_substr(
+                mb_strcut(
+                    $transcr,
+                    strspn($transcr ^ $sentence, "\0")
+                ),
+                0,
+                1
+            );
+            if ($character) {
+                $errors[] = format(
+                    __(
+                        'The provided sentence differs from the original one '.
+                        'near “{character}”.',
+                        true),
+                    compact('character')
+                );
+            } else {
+                $errors[] = format(
+                    __(
+                        'The provided sentence is shorter than the '.
+                        'original one.',
+                        true),
+                    compact('character')
+                );
+            }
+        }
+    }
+
     /**
      * Convert Japanese text into furigana.
      */
@@ -221,34 +252,7 @@ public function jpn_Jpan_to_Hrkt_validate($sentenceText, $transcr, &$errors) {
         $tokenizeFuriRegex = '/\[([^|]+)\|([\p{Hiragana}\p{Katakana}ー|]*)\]/u';
 
         $withoutFuri = preg_replace($tokenizeFuriRegex, '$1', $transcr);
-        if ($sentenceText !== $withoutFuri) {
-            /* Find the first character that differs */
-            $character = mb_substr(
-                mb_strcut(
-                    $withoutFuri,
-                    strspn($withoutFuri ^ $sentenceText, "\0")
-                ),
-                0,
-                1
-            );
-            if ($character) {
-                $errors[] = format(
-                    __(
-                        'The provided sentence differs from the original one '.
-                        'near “{character}”.',
-                        true),
-                    compact('character')
-                );
-            } else {
-                $errors[] = format(
-                    __(
-                        'The provided sentence is shorter than the '.
-                        'original one.',
-                        true),
-                    compact('character')
-                );
-            }
-        }
+        $this->_errorIfNonEqual($errors, $sentenceText, $withoutFuri);
 
         $withFuri = preg_replace('/\[([^|]+)\|+\]/u', '$1', $transcr);
         $withFuri = preg_replace($tokenizeFuriRegex, '$2', $withFuri);
@@ -333,6 +337,37 @@ private function _basic_pinyin_cleanup($text) {
         return $text;
     }
 
+    public function cmn_Hans_to_Hant_validate($sentence, $transcr, &$errors) {
+        if (mb_strlen($sentence) < mb_strlen($transcr)) {
+            $errors[] = __('The provided sentence is longer than '
+                          .'the original one.');
+            return false;
+        }
+
+        // Compare $sentence with $transcr while ignoring Han chars
+        $sentenceA = preg_split("//u", $sentence, -1, PREG_SPLIT_NO_EMPTY);
+        $transcrA  = preg_split("//u", $transcr,  -1, PREG_SPLIT_NO_EMPTY);
+        $transcrComp = '';
+        for ($i = 0; $i < count($transcrA) && $i < count($sentenceA); $i++) {
+            $charS = $sentenceA[$i];
+            $charT = $transcrA[$i];
+            $charS_isHan = preg_match('/\p{Han}/u', $charS) === 1;
+            $charT_isHan = preg_match('/\p{Han}/u', $charT) === 1;
+            if ($charS_isHan && $charT_isHan) {
+                $transcrComp .= $charS;
+            } else {
+                $transcrComp .= $charT;
+            }
+        }
+        $this->_errorIfNonEqual($errors, $sentence, $transcrComp);
+
+        return count($errors) == 0;
+    }
+
+    public function cmn_Hant_to_Hans_validate($sentence, $transcr, &$errors) {
+        return $this->cmn_Hans_to_Hant_validate($sentence, $transcr, $errors);
+    }
+
     public function cmn_Hant_to_Latn_generate($text, &$needsReview) {
         $pinyin = $this->_call_sinoparserd('pinyin', $text);
         $pinyin = $this->_basic_pinyin_cleanup($pinyin);

src/Model/Entity/Transcription.php

@@ -37,15 +37,13 @@ class Transcription extends Entity
         'cmn-Hans' => array(
             'Hant' => array(
                 'type' => 'altscript',
-                'readonly' => true,
             ),
             'Latn' => array(
             ),
         ),
         'cmn-Hant' => array(
             'Hans' => array(
                 'type' => 'altscript',
-                'readonly' => true,
             ),
             'Latn' => array(
             ),

src/Model/Table/TranscriptionsTable.php

@@ -47,15 +47,13 @@ class TranscriptionsTable extends Table
         'cmn-Hans' => array(
             'Hant' => array(
                 'type' => 'altscript',
-                'readonly' => true,
             ),
             'Latn' => array(
             ),
         ),
         'cmn-Hant' => array(
             'Hans' => array(
                 'type' => 'altscript',
-                'readonly' => true,
             ),
             'Latn' => array(
             ),

tests/Fixture/SentencesFixture.php

@@ -765,5 +765,18 @@ class SentencesFixture extends TestFixture {
 			'license' => 'CC BY 2.0 FR',
 			'based_on_id' => '55',
 		),
+		array(
+			'id' => '58',
+			'lang' => 'uzb',
+			'text' => 'Ишингни қил!',
+			'correctness' => '0',
+			'user_id' => '7',
+			'created' => '2020-01-22 22:22:22',
+			'modified' => '2020-01-22 22:22:22',
+			'script' => 'Cyrl',
+			'hash' => "rjskda\0\0\0\0\0\0\0\0\0\0",
+			'license' => 'CC BY 2.0 FR',
+			'based_on_id' => '0',
+		),
 	);
 }

tests/Fixture/TranscriptionsFixture.php

@@ -51,5 +51,15 @@ class TranscriptionsFixture extends TestFixture {
 			'created' => '2014-10-18 17:43:32',
 			'modified' => '2014-10-18 17:43:32'
 		),
+		array(
+			'id' => 4,
+			'sentence_id' => 58,
+			'script' => 'Latn',
+			'text' => 'Ishingni qil!',
+			'user_id' => null,
+			'needsReview' => 0,
+			'created' => '2020-01-22 22:22:22',
+			'modified' => '2020-01-22 22:22:22'
+		),
 	);
 }

tests/TestCase/Lib/AutotranscriptionTest.php

@@ -109,6 +109,28 @@ function testPinyin() {
         $this->assertInvalidTranscriptions('cmn', 'Hant', 'Latn', $testBad);
     }
 
+    function testHansHantValidation() {
+        $testGood = array(
+            '門開著嗎？' => array(
+                '门开着吗？',
+                '門開著嗎？',
+            ),
+        );
+        $testBad = array(
+            '門開著嗎？' => array(
+                '门开着',
+                '门开着吗',
+                '门开着吗吗',
+                '门开着吗?',
+                '门开着吗？啊',
+            ),
+        );
+        foreach (array('Hans' => 'Hant', 'Hant' => 'Hans') as $script => $oppositeScript) {
+            $this->assertValidTranscriptions('cmn', $script, $oppositeScript, $testGood);
+            $this->assertInvalidTranscriptions('cmn', $script, $oppositeScript, $testBad);
+        }
+    }
+
     function _mockHttpClient($body) {
         $response = $this->getMockBuilder(Cake\Http\Response::class)
                        ->setMethods(['isOk', 'getStringBody'])

tests/TestCase/Model/Table/TranscriptionsTableTest.php

@@ -426,10 +426,10 @@ function testCanCreateReadonlyTranscriptions() {
 
     function testCannotUpdateReadonlyTranscriptions() {
         $result = (bool)$this->Transcription->saveTranscription(array(
-            'id' => 2,
-            'sentence_id' => 2,
-            'script' => 'Hant',
-            'text' => '問題的根源是，在當今世界，愚人充滿了自信，而智者充滿了懷疑。',
+            'id' => 4,
+            'sentence_id' => 58,
+            'script' => 'Latn',
+            'text' => 'Ishingni qqq!',
         ));
         $this->assertFalse($result);
     }