Fix #268: 修复紧跟影片番号的CDx被误识别为内嵌字幕的问题

(cherry picked from commit 729f652)

core/avid.py

@@ -97,11 +97,12 @@ def get_id(filepath: str) -> str:
     return ''
 
 
+CD_POSTFIX = re.compile(r'([-_]\w|cd\d)$')
 def get_cid(filepath: str) -> str:
     """尝试将给定的文件名匹配为CID（Content ID）"""
     basename = os.path.splitext(os.path.basename(filepath))[0]
     # 移除末尾可能带有的分段影片序号
-    possible = re.sub(r'[-_]\w$', '', basename)
+    possible = CD_POSTFIX.sub('', basename)
     # cid只由数字、小写字母和下划线组成
     match = re.match(r'^([a-z\d_]+)$', possible, re.A)
     if match:

core/lib.py

@@ -59,7 +59,7 @@ def detect_special_attr(filepath: str, avid: str = None) -> str:
     if postfix in ('U', 'C', 'UC'):
         result += postfix
     elif avid:
-        pattern_str = re.sub(r'[_-]', '[_-]*', avid) + '(UC|U|C)'
+        pattern_str = re.sub(r'[_-]', '[_-]*', avid) + r'(UC|U|C)\b'
         match = re.search(pattern_str, base, flags=re.I)
         if match:
             result += match.group(1)
@@ -69,4 +69,4 @@ def detect_special_attr(filepath: str, avid: str = None) -> str:
 
 
 if __name__ == "__main__":
-    print(detect_special_attr('STARS225Uc.mp4', 'STARS-225'))
+    print(detect_special_attr('ipx-177cd1.mp4', 'IPX-177'))

unittest/test_avid.py

@@ -54,6 +54,7 @@ def test_cid_valid():
     assert '1234wvr00001rp' == get_cid('1234wvr00001rp.mp4')
     assert '402abc_hello000089' == get_cid('402abc_hello000089.mp4')
     assert 'h_826zizd021' == get_cid('h_826zizd021.mp4')
+    assert '403abcd56789' == get_cid('403abcd56789cd1.mp4')
 
 
 def test_from_file():

unittest/test_file.py

@@ -97,6 +97,17 @@ def test_scan_movies__cdx(prepare_files):
     assert basenames[2] == 'ABC-123.CD3.mp4'
 
 
+@pytest.mark.parametrize('files', [('abc123cd1.mp4','abc123cd2.mp4')])
+def test_scan_movies__cdx_without_delimeter(prepare_files):
+    movies = scan_movies(tmp_folder)
+    assert len(movies) == 1
+    assert movies[0].dvdid == 'abc-123'
+    assert len(movies[0].files) == 2
+    basenames = [os.path.basename(i) for i in movies[0].files]
+    assert basenames[0] == 'abc123cd1.mp4'
+    assert basenames[1] == 'abc123cd2.mp4'
+
+
 # 文件夹以番号命名，分片位于文件夹内且无番号信息
 @pytest.mark.parametrize('files', [('ABC-123/CD1.mp4','ABC-123/CD2 .mp4','ABC-123/CD3.mp4')])
 def test_scan_movies__from_folder(prepare_files):

unittest/test_lib.py

@@ -1,6 +1,5 @@
 import os
 import sys
-import random
 
 sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
 from core.lib import * 
@@ -23,3 +22,5 @@ def test_detect_special_attr():
     assert run('STARS225u.mp4', 'STARS-225') == 'U'
     assert run('STARS225C.mp4', 'STARS-225') == 'C'
     assert run('STARS225uC.mp4', 'STARS-225') == 'UC'
+    assert run('STARS-225CD1.mp4', 'STARS-225') == ''
+    assert run('stars225cd2.mp4', 'STARS-225') == ''