Merge pull request #52 from chineseboost/hg/better-quote-handling

Improve quote handling
chineseboost · Oct 30, 2021 · 668ef20 · 668ef20
2 parents 1c9d2e2 + b5e0ade
commit 668ef20
Show file tree

Hide file tree

Showing 7 changed files with 67 additions and 3 deletions.
diff --git a/data/02_pinyin.data b/data/02_pinyin.data
diff --git a/scripts/build_data_files.php b/scripts/build_data_files.php
@@ -143,6 +143,7 @@
     '亦说'  => 'yi4 yue4',
     '亦說'  => 'yi4 yue4',
     '睡着'  => 'shui4zhao2',
+    '还要'  => 'hai2yao4',
 ];
 $pronouns = [
     '我' => 'wo3',

diff --git a/src/Hanzi/Conversion/FurthestForwardMatching.php b/src/Hanzi/Conversion/FurthestForwardMatching.php
@@ -39,6 +39,17 @@ static function (string $section): string {
         $pinyin = preg_replace('/\s+/u', ' ', $pinyin);
         $pinyin = preg_replace('/\s([,!?.:)])/u', '$1', $pinyin);
         $pinyin = preg_replace('/([(])\s/u', '$1', $pinyin);
+        $pinyin = preg_replace('/(["\'“‘])\s+/u', '$1', $pinyin);
+        $pinyin = preg_replace('/\s+([”’"\'])/u', '$1', $pinyin);
+        $pinyin = preg_replace('/([”’])(\S)/u', '$1 $2', $pinyin);
+        $pinyin = preg_replace('/(["\']){2}/u', '$1 $1', $pinyin);
+        $pinyin = preg_replace_callback(
+            '/(^["\']|[\s"]["\']|[“‘])([a-zA-Z])/u',
+            static function (array $matches): string {
+                return $matches[1].mb_strtoupper($matches[2]);
+            },
+            $pinyin
+        );
         $pinyin = PinyinYear::replaceYears($pinyin);
         $firstChar = mb_strtoupper(mb_substr($pinyin, 0, 1));
         $rest = mb_substr($pinyin, 1);
@@ -64,7 +75,10 @@ private static function furthestForwardMatching(string $subject): string
                     }
                     continue;
                 }
-                if ($pos === 1 && $pinyin && ($furthestForward === '儿' || $furthestForward === '兒')) {
+                if ($pos === 1 && $pinyin
+                    && ($furthestForward === '儿'
+                        || $furthestForward
+                           === '兒')) {
                     $pinyin = preg_replace('/([0-5]?)\s*$/u', 'r$1 ', $pinyin, 1);
                     $subject = mb_substr($subject, 1);
                     break;

diff --git a/src/PinyinSentence.php b/src/PinyinSentence.php
@@ -198,7 +198,7 @@ static function (HtmlAble $element): string {
 
         return trim(
             <<<HTML
-<span class="pinyin sentence" lang="zh-Latn-CN-pinyin">{$elementsHtml}</span>
+<span class="pinyin sentence" lang="zh-Latn-CN-pinyin">$elementsHtml</span>
 HTML
         );
     }

diff --git a/tests/Unit/Hanzi/Conversion/FurthestForwardMatchingTest.php b/tests/Unit/Hanzi/Conversion/FurthestForwardMatchingTest.php
@@ -128,6 +128,7 @@ public function furthestForwardMatchingProvider(): array
             ['母亲安抚着她在哭的婴儿', "Mu3qin1 an1fu3 zhe5 ta1 zai4 ku1 de5 ying1'er2"],
             ['一个人藏，十个人找。', 'Yi1 ge5 ren2 cang2, shi2 ge5 ren2 zhao3.'],
             ['他这么快地把工作做完', 'Ta1 zhe4me5 kuai4 de5 ba3 gong1zuo4 zuo4 wan2'],
+            ['“还要咖啡吗？”“不了，谢谢。”', '“Hai2yao4 ka1fei1 ma5?” “Bu4le5, xie4xie5.”'],
             [
                 '科学家的工作就是对理论加以检验。',
                 'Ke1xue2jia1 de5 gong1zuo4 jiu4shi4 dui4 li3lun4 jia1yi3 jian3yan4.',

diff --git a/tests/Unit/Hanzi/HanziSentence/HanziSentenceElementsTest.php b/tests/Unit/Hanzi/HanziSentence/HanziSentenceElementsTest.php
@@ -76,6 +76,10 @@ public function elementsProvider(): array
                 '如果你喝速溶咖啡，就不需要咖啡壶了。',
                 ['如果', '你', '喝', '速溶', '咖啡', '，', '就', '不', '需要', '咖啡', '壶', '了', '。'],
             ],
+            [
+                '“还要咖啡吗？”“不了，谢谢。”',
+                ['“', '还要', '咖啡', '吗', '？”', '“', '不了', '，', '谢谢', '。”'],
+            ],
         ];
     }
 }
diff --git a/tests/Unit/Hanzi/HanziSentence/HanziSentenceQuoteMarksTest.php b/tests/Unit/Hanzi/HanziSentence/HanziSentenceQuoteMarksTest.php
@@ -0,0 +1,44 @@
+<?php
+
+namespace Pinyin\Tests\Unit\Hanzi\HanziSentence;
+
+use PHPUnit\Framework\TestCase;
+use Pinyin\Hanzi\HanziSentence;
+
+class HanziSentenceQuoteMarksTest extends TestCase
+{
+    /**
+     * @param string $sentence
+     * @param string $expectedPinyin
+     *
+     * @dataProvider quoteMarkSpacingProvider
+     */
+    public function testQuoteMarkSpacing(string $sentence, string $expectedPinyin): void
+    {
+        // Given a hanzi sentence;
+        $hanziSentence = new HanziSentence($sentence);
+
+        // When we convert it to pinyin;
+        $pinyin = $hanziSentence->asPinyin();
+
+        // Then the spacing should be correct.
+        self::assertEquals($expectedPinyin, (string) $pinyin);
+    }
+
+    /**
+     * @return array[]
+     */
+    public static function quoteMarkSpacingProvider(): array
+    {
+        return [
+            [
+                '“还要咖啡吗？”“不了，谢谢。”',
+                '“Hai2yao4 ka1fei1 ma5?” “Bu4le5, xie4xie5.”',
+            ],
+            [
+                '"还要咖啡吗？""不了，谢谢。"',
+                '"Hai2yao4 ka1fei1 ma5?" "Bu4le5, xie4xie5."',
+            ],
+        ];
+    }
+}
-Original file line number
+Diff line change
@@ Expand Up / @@ -198,7 +198,7 @@ static function (HtmlAble $element): string { @@
             return trim(
                 <<<HTML
-    <span class="pinyin sentence" lang="zh-Latn-CN-pinyin">{$elementsHtml}</span>
+    <span class="pinyin sentence" lang="zh-Latn-CN-pinyin">$elementsHtml</span>
     HTML
             );
         }
@@ Expand Down @@