From 8a56450b797c0176228621b62e7e5935269d9818 Mon Sep 17 00:00:00 2001
From: Misaka_Company <pengqiangqiang0@gmail.com>
Date: Mon, 15 Jun 2026 14:11:33 +0800
Subject: [PATCH] Add progress display for S3 upload and transcription polling

- Add real-time progress bar with percentage for S3 upload (file size/MB)
- Replace repetitive polling output with single-line refreshing status
- Show elapsed time and poll count during transcription polling
- Add .claude/ to .gitignore

Co-Authored-By: Claude <noreply@anthropic.com>
---
 .gitignore           |  4 +++-
 transcribe_legacy.py | 32 +++++++++++++++++++++++++++-----
 2 files changed, 30 insertions(+), 6 deletions(-)

diff --git a/.gitignore b/.gitignore
index 6d21059..d37ea01 100644
--- a/.gitignore
+++ b/.gitignore
@@ -37,4 +37,6 @@ output/
 Archive/
 
 # Tools
-tools/
\ No newline at end of file
+tools/
+
+.claude/
\ No newline at end of file
diff --git a/transcribe_legacy.py b/transcribe_legacy.py
index c59e782..18e3a84 100644
--- a/transcribe_legacy.py
+++ b/transcribe_legacy.py
@@ -100,8 +100,23 @@ audio_ext = os.path.splitext(audio_file)[1].lstrip(".")
 audio_name = os.path.splitext(os.path.basename(audio_file))[0]
 
 obj_name = f"audio/{uuid.uuid4().hex[:8]}_{os.path.basename(audio_file)}"
-print(f"Uploading {audio_file} ...")
-s3.upload_file(audio_file, S3_BUCKET, obj_name)
+file_size = os.path.getsize(audio_file)
+uploaded = [0]  # Use list to allow modification in callback
+
+def upload_callback(bytes_transferred):
+    uploaded[0] += bytes_transferred
+    percent = min(100, int(uploaded[0] * 100 / file_size))
+    uploaded_mb = uploaded[0] / (1024 * 1024)
+    total_mb = file_size / (1024 * 1024)
+    bar_width = 30
+    filled = int(bar_width * percent / 100)
+    bar = "█" * filled + "░" * (bar_width - filled)
+    print(f"\rUploading [{bar}] {percent}% ({uploaded_mb:.1f}MB / {total_mb:.1f}MB)", end="", flush=True)
+
+print(f"Uploading {audio_file} ({file_size / (1024 * 1024):.1f}MB) ...")
+s3.upload_file(audio_file, S3_BUCKET, obj_name, Callback=upload_callback)
+print()  # New line after upload completes
+
 url = s3.generate_presigned_url(
     "get_object",
     Params={"Bucket": S3_BUCKET, "Key": obj_name},
@@ -131,6 +146,9 @@ print(f"Submit: {status}")
 if status != "20000000":
     print(f"FAILED: {resp.headers.get('X-Api-Message', '')}")
 else:
+    start_time = time.time()
+    poll_count = 0
+
     while True:
         resp = session.post(
             QUERY_URL,
@@ -156,14 +174,18 @@ else:
             with open(md_path, "w", encoding="utf-8") as f:
                 f.write(build_markdown(data, audio_file))
 
-            print(f"Done ({len(text)} chars, {len(utterances)} utterances)")
+            elapsed = int(time.time() - start_time)
+            print(f"\rDone ({len(text)} chars, {len(utterances)} utterances, elapsed {elapsed}s)")
             print(f"Saved: {json_path}, {md_path}")
             break
         elif code in ("20000001", "20000002"):
-            print(f"  Processing... ({code})")
+            poll_count += 1
+            elapsed = int(time.time() - start_time)
+            status_msg = "Processing" if code == "20000001" else "Queued"
+            print(f"\r  {status_msg}... (poll {poll_count}, elapsed {elapsed}s)    ", end="", flush=True)
             time.sleep(5)
         else:
-            print(f"FAILED: {code} {resp.headers.get('X-Api-Message', '')}")
+            print(f"\rFAILED: {code} {resp.headers.get('X-Api-Message', '')}")
             break
 
 print("\nAll done.")