curlkun commited on
Commit
8d99b6b
·
verified ·
1 Parent(s): 3ea877f

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +14 -0
app.py CHANGED
@@ -192,6 +192,20 @@ def apply_highlight(text, ranges):
192
  result.append(text[prev:])
193
  return ''.join(result)
194
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
195
  test_cases = [
196
  ("TEST-08", "吾輩は、猫である。「名前」は(まだ)ない!", "吾輩は猫である名前はまだない",
197
  {'ignore_yakumono': True}),
 
192
  result.append(text[prev:])
193
  return ''.join(result)
194
 
195
+ # test_cases ループの先頭に追加
196
+ for name, src, dst, settings in test_cases:
197
+ src_flags = build_active_flags(src, settings)
198
+ dst_flags = build_active_flags(dst, settings)
199
+ src_tokens = tokenize_with_flags(src, src_flags)
200
+ dst_tokens = tokenize_with_flags(dst, dst_flags)
201
+ if name in ("TEST-30", "TEST-35"):
202
+ print(f"\n=== {name} tokens ===")
203
+ print(f" src_tokens: {src_tokens}")
204
+ print(f" dst_tokens: {dst_tokens}")
205
+ matcher = difflib.SequenceMatcher(None, [t[0] for t in src_tokens], [t[0] for t in dst_tokens], autojunk=False)
206
+ for tag, i1, i2, j1, j2 in matcher.get_opcodes():
207
+ print(f" {tag:10} src={repr(''.join(t[0] for t in src_tokens[i1:i2]))} dst={repr(''.join(t[0] for t in dst_tokens[j1:j2]))}")
208
+
209
  test_cases = [
210
  ("TEST-08", "吾輩は、猫である。「名前」は(まだ)ない!", "吾輩は猫である名前はまだない",
211
  {'ignore_yakumono': True}),