From 87224cd8ad4cac52f91ed5b47da32cbd00852b12 Mon Sep 17 00:00:00 2001
From: EstrellaXD <estrellaxd05@gmail.com>
Date: Mon, 10 Jul 2023 22:36:55 +0800
Subject: [PATCH] fix: parser error #385 #375

---
 .../src/module/parser/analyser/raw_parser.py  | 22 ++++++++++++++-----
 backend/src/test/test_raw_parser.py           | 10 +++++++++
 2 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/backend/src/module/parser/analyser/raw_parser.py b/backend/src/module/parser/analyser/raw_parser.py
index 37458d85..1f3c3469 100644
--- a/backend/src/module/parser/analyser/raw_parser.py
+++ b/backend/src/module/parser/analyser/raw_parser.py
@@ -94,11 +94,23 @@ def name_process(name: str):
             split = re.split("-", name)
     if len(split) == 1:
         split_space = split[0].split(" ")
-        for idx, item in enumerate(split_space):
+        language_pattern = []
+        for item in split_space:
             if re.search(r"^[\u4e00-\u9fa5]{2,}", item) is not None:
-                split_space.remove(item)
-                split = [item.strip(), " ".join(split_space).strip()]
-                break
+                language_pattern.append(1)
+            elif re.search(r"[a-zA-Z]{2,}", item) is not None:
+                language_pattern.append(0)
+            elif re.search(r"[\u0800-\u4e00]{2,}", item) is not None:
+                language_pattern.append(2)
+        split = [split_space[0]]
+        for i in range(1, len(split_space)):
+            # 如果当前字符串的语言与上一个字符串的语言相同
+            if language_pattern[i] == language_pattern[i - 1]:
+                # 合并这两个字符串
+                split[-1] += ' ' + split_space[i]
+            else:
+                # 否则，将当前字符串添加到结果列表中
+                split.append(split_space[i])
     for item in split:
         if re.search(r"[\u0800-\u4e00]{2,}", item) and not name_jp:
             name_jp = item.strip()
@@ -184,5 +196,5 @@ def raw_parser(raw: str) -> Episode | None:
 
 
 if __name__ == "__main__":
-    title = "【极影字幕·毁片党】LoveLive! SunShine!! 幻日的夜羽 -SUNSHINE in the MIRROR- 第01集 TV版 HEVC_opus 1080p "
+    title = "【极影字幕·毁片党】LoveLive! SunShine!! 幻日的夜羽 -SUNSHINE in the MIRROR- 第01集 TV版 HEVC_opus 1080p"
     print(raw_parser(title))
diff --git a/backend/src/test/test_raw_parser.py b/backend/src/test/test_raw_parser.py
index 60297c8e..39f1944c 100644
--- a/backend/src/test/test_raw_parser.py
+++ b/backend/src/test/test_raw_parser.py
@@ -78,3 +78,13 @@ def test_raw_parser():
     assert info.resolution == "1080p"
     assert info.episode == 1
     assert info.season == 1
+
+    content = "[ANi] BLEACH 死神 千年血战篇-诀别谭- - 14 [1080P][Baha][WEB-DL][AAC AVC][CHT][MP4]"
+    info = raw_parser(content)
+    assert info.group == "ANi"
+    assert info.title_en == "BLEACH"
+    assert info.title_zh == "死神 千年血战篇-诀别谭-"
+    assert info.resolution == "1080P"
+    assert info.episode == 14
+    assert info.season == 1
+