fix(splitter): when merge across punctuation on section stage the lan…

…g of substring inside is wrong

split_lang/split/splitter.py

@@ -684,6 +684,37 @@ def _special_merge_for_zh_ja(
         new_substrings = self._merge_substrings(substrings=new_substrings)
         return new_substrings
 
+    # MARK: _merge_substrings_across_newline
+    def _merge_substrings_across_newline(
+        self,
+        substrings: List[SubString],
+    ) -> List[SubString]:
+        new_substrings: List[SubString] = []
+        last_lang = ""
+
+        for _, substring in enumerate(substrings):
+            if new_substrings:
+                if substring.lang == "newline":
+                    # If the last substring is also a newline, merge them
+                    new_substrings[-1].text += substring.text
+                    new_substrings[-1].length += substring.length
+                else:
+                    if substring.lang == last_lang or last_lang == "":
+                        new_substrings[-1].text += substring.text
+                        new_substrings[-1].length += substring.length
+                        new_substrings[-1].lang = (
+                            substring.lang
+                            if new_substrings[-1].lang == "newline"
+                            else new_substrings[-1].lang
+                        )
+                    else:
+                        new_substrings.append(substring)
+                last_lang = substring.lang
+            else:
+                new_substrings.append(substring)
+
+        return new_substrings
+
     # MARK: _merge_substrings_across_newline_based_on_sections
     def _merge_substrings_across_newline_based_on_sections(
         self,
@@ -769,6 +800,11 @@ def _merge_substrings_across_newline_based_on_sections(
                             section.substrings[substr_index - 1].index
                             + section.substrings[substr_index - 1].length
                         )
+        # NOTE: 合并 sections 中的 substrings 里面的 text
+        for section in new_sections_merged:
+            section.substrings = self._merge_substrings_across_newline(
+                substrings=section.substrings
+            )
         if self.debug:
             logger.debug(
                 "---------------------------------after_merge_newline_sections:"
@@ -788,12 +824,8 @@ def _merge_substrings_across_digit(
         for _, substring in enumerate(substrings):
             if new_substrings:
                 if substring.lang == "digit":
-                    if new_substrings[-1].lang == "digit":
-                        new_substrings[-1].text += substring.text
-                        new_substrings[-1].length += substring.length
-                    else:
-                        new_substrings[-1].text += substring.text
-                        new_substrings[-1].length += substring.length
+                    new_substrings[-1].text += substring.text
+                    new_substrings[-1].length += substring.length
                 else:
                     if substring.lang == last_lang or last_lang == "":
                         new_substrings[-1].text += substring.text
@@ -907,24 +939,29 @@ def _merge_substrings_across_punctuation(
         substrings: List[SubString],
     ) -> List[SubString]:
         new_substrings: List[SubString] = []
-        lang = ""
-        for substring in substrings:
-            if (
-                substring.lang == "punctuation"
-                and substring.text.strip() not in self.not_merge_punctuation
-            ):
-                if new_substrings and new_substrings[-1].lang == lang:
+        last_lang = ""  # Changed from 'lang' to 'last_lang' for consistency
+
+        for _, substring in enumerate(substrings):
+            if new_substrings:
+                if substring.lang == "punctuation":
+                    # If the last substring is also a punctuation, merge them
                     new_substrings[-1].text += substring.text
                     new_substrings[-1].length += substring.length
                 else:
-                    new_substrings.append(substring)
+                    if substring.lang == last_lang or last_lang == "":
+                        new_substrings[-1].text += substring.text
+                        new_substrings[-1].length += substring.length
+                        new_substrings[-1].lang = (
+                            substring.lang
+                            if new_substrings[-1].lang == "punctuation"
+                            else new_substrings[-1].lang
+                        )
+                    else:
+                        new_substrings.append(substring)
+                last_lang = substring.lang
             else:
-                if substring.lang != lang:
-                    new_substrings.append(substring)
-                else:
-                    new_substrings[-1].text += substring.text
-                    new_substrings[-1].length += substring.length
-            lang = substring.lang if substring.lang != "punctuation" else lang
+                new_substrings.append(substring)
+
         return new_substrings
 
     # MARK: _merge_substrings_across_punctuation based on sections