Upload app.py

app.py

@@ -69,7 +69,9 @@ def ali(ref, hyp, col=80, remove_punc=False, file=None):
 def process_ours(path):
     with open(path) as f:
         ours = f.read()
+    print(ours)
     ours = [i.split('\u3000', maxsplit=1)[-1] for i in ours.splitlines()]
+    print(ours)
     ours = ' '.join(ours)
     return ours
 
@@ -82,18 +84,18 @@ def process_theirs(path):
         if theirs[i].startswith('會議記錄：'):
             break
         i+=1
-    theirs = theirs[i:]
-    transcript = ''
-    for i in range(0, len(theirs), 4):
-        text = (theirs[i+2])
-        transcript += text
+    theirs = theirs[i+1:]
+    transcript = ' '.join(theirs)
+    # for i in range(0, len(theirs), 4):
+    #     text = (theirs[i+2])
+    #     transcript += text
     return transcript
 
 def compare_transcripts(ours_file, theirs_file, remove_punc, number_box):
     ours = process_ours(ours_file)
     theirs = process_theirs(theirs_file)
     output = io.StringIO()
-    ali(theirs, ours, remove_punc=remove_punc, file=output, col=int(number_box))
+    ali(theirs.replace(' ', ''), ours.replace(' ', ''), remove_punc=remove_punc, file=output, col=int(number_box))
     return output.getvalue()
 
 custom_css = """
@@ -104,7 +106,7 @@ textarea[data-testid="textbox"] {
 
 with gr.Blocks(title="Transcript Alignment Viewer",css=custom_css) as demo:
     gr.Markdown("## Transcript Alignment Viewer")
-    gr.Markdown("Upload two transcripts and compare their alignment visually using jiwer.")
+    gr.Markdown("上傳請確認你用的是 `不分段會議紀錄`")
 
     with gr.Row(equal_height=True):
         ours_file = gr.File(label="Our Transcript", file_types=[".txt"], scale=1)