improve app.py

app.py

@@ -1,7 +1,6 @@
 import streamlit as st
 from pathlib import Path
-import tempfile
-import subprocess
+import tempfile, subprocess, threading, queue
 import textwrap
 
 st.set_page_config(page_title="Lec2Note2 – Lecture-to-Notes", layout="wide")
@@ -31,27 +30,50 @@ if run_btn and video_file:
     output_md = vid_path.with_suffix(".md")
 
     st.info("Processing started. This may take several minutes depending on video length …")
-    progress_txt = st.empty()
+    # container for live log streaming
+    log_container = st.container()
 
     # Run pipeline via subprocess to avoid blocking UI; capture logs
     with st.spinner("Running Lec2Note2 pipeline …"):
-        result = subprocess.run(
-            [
-                "python",
-                "-m",
-                "lec2note.scripts.run_pipeline",
-                "--video",
-                str(vid_path),
-                "--output",
-                str(output_md),
-            ],
-            text=True,
-            capture_output=True,
-        )
-    if result.returncode != 0:
+        # launch pipeline in subprocess with unbuffered output
+        cmd = [
+            "python",
+            "-u",  # unbuffer stdout
+            "-m",
+            "lec2note.scripts.run_pipeline",
+            "--video",
+            str(vid_path),
+            "--output",
+            str(output_md),
+        ]
+        proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True)
+
+        log_queue: "queue.Queue[str]" = queue.Queue()
+
+        def _enqueue_output(pipe, q):
+            for line in iter(pipe.readline, ""):
+                q.put(line)
+            pipe.close()
+
+        threading.Thread(target=_enqueue_output, args=(proc.stdout, log_queue), daemon=True).start()
+
+        logs = ""
+        while True:
+            try:
+                line = log_queue.get(timeout=0.1)
+            except queue.Empty:
+                if proc.poll() is not None:
+                    # process finished and queue empty
+                    break
+                continue
+            logs += line
+            log_container.code(logs, language="bash")
+
+        result_code = proc.wait()
+    if result_code != 0:
         st.error("❌ Pipeline failed. See logs below.")
         with st.expander("Show logs"):
-            st.code(result.stderr + "\n" + result.stdout)
+            st.code(logs)
     else:
         st.success("✅ Notes generated!")
         md_content = output_md.read_text()

lec2note/segmentation/semantic_segmenter.py

@@ -39,8 +39,8 @@ class SemanticSegmenter:  # noqa: D101
         buf_emb = embeddings[0]
         for i in range(1, len(slide_chunks)):
             sim = float(util.cos_sim(buf_emb, embeddings[i]))
-            print("semantic sim:",sim)
-            if sim > 0.7:  # too similar => merge
+            # print("semantic sim:",sim)
+            if sim > 0.55:  # too similar => merge
                 buffer["end"] = slide_chunks[i]["end"]
             else:
                 refined.append(buffer)

lec2note/segmentation/visual_merger.py

@@ -51,7 +51,7 @@ class VisualMerger:  # noqa: D101
             except Exception as exc:  # noqa: BLE001
                 logger.warning("[VisualMerger] similarity calc failed: %s", exc)
                 sim = 0.0  # force split
-            print("visual sim:",sim)
+            # print("visual sim:",sim)
             if sim >= sim_threshold:
                 # merge: extend end and replace keyframe/path to current (last)
                 buffer["end"] = mc["end"]