fix parser error

Author: shinonomeow

backend/src/module/parser/analyser/meta_parser.py

@@ -92,7 +92,7 @@ def parser(self, title: str) -> Episode:
         if episode is None:
             episode = self.get_untrusted_episode()
 
-        if not self.episode_trusted:
+        if not self.season_trusted:
             season, season_raw = self.get_untrusted_season()
 
 
@@ -104,6 +104,7 @@ def parser(self, title: str) -> Episode:
             # 暂时没有哪个组把集数放前面
             if len(parts) > 1:
                 temp_title = "[]".join(parts[:-1])
+        print(temp_title)
         self.token = re.split(r"[\[\]]", temp_title)
         (
             name_en,
@@ -116,7 +117,7 @@ def parser(self, title: str) -> Episode:
         source = source_info[0] if source_info else ""
         sub = sub_language
         resolution = resolution_info[0] if resolution_info else ""
-        logger.debug(f"[meta parser] {self.raw_title} >> S{season}E{episode} {name_zh}/{name_en}/{name_jp} {sub} {sub_type} {group} {year} {resolution} {source} {audio_info} {video_info}")
+        # logger.debug(f"[meta parser] {self.raw_title} >> S{season}E{episode} {name_zh}/{name_en}/{name_jp} {sub} {sub_type} {group} {year} {resolution} {source} {audio_info} {video_info}")
 
         return Episode(
             title_en=name_en,
@@ -235,18 +236,21 @@ def get_untrusted_episode(self) -> int:
     def get_trusted_season(self) -> tuple[int, str]:
         """获取可信的季度信息"""
         season_info = self.findall_sub_title(p.SEASON_PATTERN_TRUEST, sym="/[]")
+        print(f"trusted season info: {season_info}")
         if not season_info:
             season_info = self.findall_sub_title(p.SEASON_PATTERN, sym="/[]")
 
+
         if season_info:
             self.season_trusted = True
             return self.parse_season(season_info, True)
             # return season_info[0][0], season
-        return 0, ""
+        return 1, ""
 
     def get_untrusted_season(self) -> tuple[int, str]:
         """获取不可信的季度信息"""
         season_info = re.findall(p.SEASON_PATTERN_UNTRUSTED, self.title)
+        print(f"untrusted season info: {season_info}")
         if season_info:
             return self.parse_season(season_info, False)
         return 1, ""
@@ -290,6 +294,7 @@ def name_process(self) -> tuple[str, str, str]:
         self.token = [token.strip() for token in self.token[:max_len] if len(token.strip()) > 1]
 
         self.token = self.token[:5]
+        print(f"tokens: {self.token}")
         token_priority = [len(s) for s in self.token]
         if len(self.token) == 1:
             anime_title = self.token[0]
@@ -307,10 +312,11 @@ def name_process(self) -> tuple[str, str, str]:
                     token_priority[idx] -= 90
                 if re.search(r"[a-zA-Z]{3,}", token):
                     token_priority[idx] += 2
-                if re.search(r"[\u0800-\u4e00]{2,}", token):
-                    token_priority[idx] += 2
-                if re.search(r"[\u4e00-\u9fa5]{2,}", token):
-                    token_priority[idx] += 2
+                if l:=re.search(r"[\u0800-\u4e00]{2,}", token):
+                    token_priority[idx] += len(l.group(0))*2
+                if l:=re.search(r"[\u4e00-\u9fa5]{2,}", token):
+                    token_priority[idx] += len(l.group(0))*2
+            print(token_priority)
             idx = token_priority.index(max(token_priority))
             anime_title = self.token[idx]
             anime_title = anime_title.strip()

backend/src/module/parser/analyser/patterns.py

@@ -59,8 +59,8 @@
     (第(.{1,3})季       # 匹配"第...季"格式
     |第(.{1,3})期        # 匹配"第...期"格式
     |第.{1,3}部分      # 匹配"第...部分"格式
-    |[Ss]eason\s?(\d{{1,2}})  # 匹配"Season X"格式
-    |SEASON\s?(\d{{1,2}})  # 匹配"SEASON X"格式
+    |[Ss]eason\s?(\d{1,2})  # 匹配"Season X"格式
+    |SEASON\s?(\d{1,2})  # 匹配"SEASON X"格式
     )
     """,
     re.VERBOSE

backend/src/test/test_raw_parser.py

@@ -6,7 +6,8 @@ def test_raw_parser():
     info = TitleMetaParser().parser(content)
     assert info is not None
     assert info.group == "幻樱字幕组"
-    assert info.title_en == "Komi-san wa, Komyushou Desu."
+    assert info.title_zh == "古见同学有交流障碍症"
+    # assert info.title_en == "Komi-san wa, Komyushou Desu."
     assert info.resolution == "1920X1080"
     assert info.episode == 22
     assert info.season == 2
@@ -145,6 +146,7 @@ def test_raw_parser():
     assert info.group == "桜都字幕组&7³ACG"
     assert info.title_jp == "ゆるキャン△"
     assert info.sub == "简繁"
+    assert info.episode == -1
 
     content = "[ANi] Grand Blue Dreaming /  GRAND BLUE 碧蓝之海 2 - 04 [1080P][Baha][WEB-DL][AAC AVC][CHT][MP4]"
     info = TitleMetaParser().parser(content)