Spaces:
Sleeping
Sleeping
Update streamlit_app.py
Browse files- streamlit_app.py +119 -22
streamlit_app.py
CHANGED
|
@@ -44,13 +44,11 @@ MODEL_OPTIONS = [
|
|
| 44 |
# Helper utilities
|
| 45 |
# ----------------------------------------------------------------------
|
| 46 |
def _sanitize_filename(url: str) -> str:
|
| 47 |
-
"""Make a safe filename from a URL."""
|
| 48 |
name = Path(url).name.lower()
|
| 49 |
return name.translate(str.maketrans("", "", string.punctuation)).replace(" ", "_")
|
| 50 |
|
| 51 |
|
| 52 |
def _file_sha256(path: Path) -> Optional[str]:
|
| 53 |
-
"""Return SHA‑256 hex digest of *path* or None on error."""
|
| 54 |
try:
|
| 55 |
h = hashlib.sha256()
|
| 56 |
with path.open("rb") as f:
|
|
@@ -62,7 +60,6 @@ def _file_sha256(path: Path) -> Optional[str]:
|
|
| 62 |
|
| 63 |
|
| 64 |
def _convert_to_mp4(src: Path) -> Path:
|
| 65 |
-
"""Convert *src* to MP4 (ffmpeg) and delete the original."""
|
| 66 |
dst = src.with_suffix(".mp4")
|
| 67 |
if dst.exists():
|
| 68 |
return dst
|
|
@@ -79,7 +76,6 @@ def _convert_to_mp4(src: Path) -> Path:
|
|
| 79 |
|
| 80 |
|
| 81 |
def _compress_video(inp: Path, crf: int = 28, preset: str = "fast") -> Path:
|
| 82 |
-
"""Compress *inp* with libx264; return the new file."""
|
| 83 |
out = inp.with_name(f"{inp.stem}_compressed.mp4")
|
| 84 |
try:
|
| 85 |
ffmpeg.input(str(inp)).output(
|
|
@@ -91,7 +87,6 @@ def _compress_video(inp: Path, crf: int = 28, preset: str = "fast") -> Path:
|
|
| 91 |
|
| 92 |
|
| 93 |
def _maybe_compress(path: Path, limit_mb: int) -> Tuple[Path, bool]:
|
| 94 |
-
"""Compress *path* if it exceeds *limit_mb*."""
|
| 95 |
size_mb = path.stat().st_size / (1024 * 1024)
|
| 96 |
if size_mb <= limit_mb:
|
| 97 |
return path, False
|
|
@@ -99,7 +94,6 @@ def _maybe_compress(path: Path, limit_mb: int) -> Tuple[Path, bool]:
|
|
| 99 |
|
| 100 |
|
| 101 |
def _download_direct(url: str, dst: Path) -> Path:
|
| 102 |
-
"""Simple HTTP GET download."""
|
| 103 |
r = requests.get(url, stream=True, timeout=30)
|
| 104 |
r.raise_for_status()
|
| 105 |
out = dst / _sanitize_filename(url.split("/")[-1])
|
|
@@ -111,7 +105,6 @@ def _download_direct(url: str, dst: Path) -> Path:
|
|
| 111 |
|
| 112 |
|
| 113 |
def _download_with_yt_dlp(url: str, dst: Path, password: str = "") -> Path:
|
| 114 |
-
"""Download via yt‑dlp, returning the newest MP4."""
|
| 115 |
tmpl = str(dst / "%(id)s.%(ext)s")
|
| 116 |
fmt = "best[ext=mp4]/best"
|
| 117 |
|
|
@@ -156,7 +149,6 @@ def _download_with_yt_dlp(url: str, dst: Path, password: str = "") -> Path:
|
|
| 156 |
raise RuntimeError("No MP4 file was created.")
|
| 157 |
newest = max(mp4_files, key=lambda p: p.stat().st_mtime)
|
| 158 |
|
| 159 |
-
# Deduplicate via SHA‑256 cache
|
| 160 |
sha = _file_sha256(newest)
|
| 161 |
if sha:
|
| 162 |
for existing in dst.iterdir():
|
|
@@ -167,7 +159,6 @@ def _download_with_yt_dlp(url: str, dst: Path, password: str = "") -> Path:
|
|
| 167 |
|
| 168 |
|
| 169 |
def download_video(url: str, dst: Path, password: str = "") -> Path:
|
| 170 |
-
"""Unified download entry point."""
|
| 171 |
video_exts = (".mp4", ".mov", ".webm", ".mkv", ".avi", ".flv")
|
| 172 |
|
| 173 |
if url.lower().endswith(video_exts):
|
|
@@ -188,12 +179,10 @@ def download_video(url: str, dst: Path, password: str = "") -> Path:
|
|
| 188 |
|
| 189 |
|
| 190 |
def _encode_video_b64(path: Path) -> str:
|
| 191 |
-
"""Base64‑encode a file."""
|
| 192 |
return base64.b64encode(path.read_bytes()).decode()
|
| 193 |
|
| 194 |
|
| 195 |
def generate_report(video_path: Path, prompt: str, model_id: str, timeout: int = 300) -> str:
|
| 196 |
-
"""Send video + prompt to Gemini and return the response text."""
|
| 197 |
b64 = _encode_video_b64(video_path)
|
| 198 |
video_part = {"inline_data": {"mime_type": "video/mp4", "data": b64}}
|
| 199 |
model = genai.GenerativeModel(model_name=model_id)
|
|
@@ -207,7 +196,6 @@ def generate_report(video_path: Path, prompt: str, model_id: str, timeout: int =
|
|
| 207 |
|
| 208 |
|
| 209 |
def _strip_prompt_echo(prompt: str, text: str, threshold: float = 0.68) -> str:
|
| 210 |
-
"""Trim the prompt if the model repeats it at the start."""
|
| 211 |
if not prompt or not text:
|
| 212 |
return text
|
| 213 |
clean_prompt = " ".join(prompt.lower().split())
|
|
@@ -222,7 +210,6 @@ def _strip_prompt_echo(prompt: str, text: str, threshold: float = 0.68) -> str:
|
|
| 222 |
# UI helpers
|
| 223 |
# ----------------------------------------------------------------------
|
| 224 |
def _expand_sidebar(width: int = 380) -> None:
|
| 225 |
-
"""Make the Streamlit sidebar wider."""
|
| 226 |
st.markdown(
|
| 227 |
f"""
|
| 228 |
<style>
|
|
@@ -235,11 +222,128 @@ def _expand_sidebar(width: int = 380) -> None:
|
|
| 235 |
unsafe_allow_html=True,
|
| 236 |
)
|
| 237 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 238 |
# ----------------------------------------------------------------------
|
| 239 |
# Session‑state defaults
|
| 240 |
# ----------------------------------------------------------------------
|
| 241 |
def _init_state() -> None:
|
| 242 |
-
"""Populate Streamlit's session_state with sensible defaults."""
|
| 243 |
defaults = {
|
| 244 |
"url": "",
|
| 245 |
"video_path": "",
|
|
@@ -253,23 +357,16 @@ def _init_state() -> None:
|
|
| 253 |
"analysis_out": "",
|
| 254 |
"raw_output": "",
|
| 255 |
"last_error_detail": "",
|
| 256 |
-
"show_raw_on_error": False,
|
| 257 |
-
"show_analysis": False,
|
| 258 |
}
|
| 259 |
for k, v in defaults.items():
|
| 260 |
st.session_state.setdefault(k, v)
|
| 261 |
|
| 262 |
|
| 263 |
# ----------------------------------------------------------------------
|
| 264 |
-
#
|
| 265 |
# ----------------------------------------------------------------------
|
| 266 |
if __name__ == "__main__":
|
| 267 |
-
# Initialise session state before any UI code runs
|
| 268 |
_init_state()
|
| 269 |
-
|
| 270 |
-
# Initialise the Gemini API – the key can be supplied via the sidebar or env var
|
| 271 |
if st.session_state["api_key"]:
|
| 272 |
genai.configure(api_key=st.session_state["api_key"])
|
| 273 |
-
|
| 274 |
-
# Run the Streamlit app
|
| 275 |
main()
|
|
|
|
| 44 |
# Helper utilities
|
| 45 |
# ----------------------------------------------------------------------
|
| 46 |
def _sanitize_filename(url: str) -> str:
|
|
|
|
| 47 |
name = Path(url).name.lower()
|
| 48 |
return name.translate(str.maketrans("", "", string.punctuation)).replace(" ", "_")
|
| 49 |
|
| 50 |
|
| 51 |
def _file_sha256(path: Path) -> Optional[str]:
|
|
|
|
| 52 |
try:
|
| 53 |
h = hashlib.sha256()
|
| 54 |
with path.open("rb") as f:
|
|
|
|
| 60 |
|
| 61 |
|
| 62 |
def _convert_to_mp4(src: Path) -> Path:
|
|
|
|
| 63 |
dst = src.with_suffix(".mp4")
|
| 64 |
if dst.exists():
|
| 65 |
return dst
|
|
|
|
| 76 |
|
| 77 |
|
| 78 |
def _compress_video(inp: Path, crf: int = 28, preset: str = "fast") -> Path:
|
|
|
|
| 79 |
out = inp.with_name(f"{inp.stem}_compressed.mp4")
|
| 80 |
try:
|
| 81 |
ffmpeg.input(str(inp)).output(
|
|
|
|
| 87 |
|
| 88 |
|
| 89 |
def _maybe_compress(path: Path, limit_mb: int) -> Tuple[Path, bool]:
|
|
|
|
| 90 |
size_mb = path.stat().st_size / (1024 * 1024)
|
| 91 |
if size_mb <= limit_mb:
|
| 92 |
return path, False
|
|
|
|
| 94 |
|
| 95 |
|
| 96 |
def _download_direct(url: str, dst: Path) -> Path:
|
|
|
|
| 97 |
r = requests.get(url, stream=True, timeout=30)
|
| 98 |
r.raise_for_status()
|
| 99 |
out = dst / _sanitize_filename(url.split("/")[-1])
|
|
|
|
| 105 |
|
| 106 |
|
| 107 |
def _download_with_yt_dlp(url: str, dst: Path, password: str = "") -> Path:
|
|
|
|
| 108 |
tmpl = str(dst / "%(id)s.%(ext)s")
|
| 109 |
fmt = "best[ext=mp4]/best"
|
| 110 |
|
|
|
|
| 149 |
raise RuntimeError("No MP4 file was created.")
|
| 150 |
newest = max(mp4_files, key=lambda p: p.stat().st_mtime)
|
| 151 |
|
|
|
|
| 152 |
sha = _file_sha256(newest)
|
| 153 |
if sha:
|
| 154 |
for existing in dst.iterdir():
|
|
|
|
| 159 |
|
| 160 |
|
| 161 |
def download_video(url: str, dst: Path, password: str = "") -> Path:
|
|
|
|
| 162 |
video_exts = (".mp4", ".mov", ".webm", ".mkv", ".avi", ".flv")
|
| 163 |
|
| 164 |
if url.lower().endswith(video_exts):
|
|
|
|
| 179 |
|
| 180 |
|
| 181 |
def _encode_video_b64(path: Path) -> str:
|
|
|
|
| 182 |
return base64.b64encode(path.read_bytes()).decode()
|
| 183 |
|
| 184 |
|
| 185 |
def generate_report(video_path: Path, prompt: str, model_id: str, timeout: int = 300) -> str:
|
|
|
|
| 186 |
b64 = _encode_video_b64(video_path)
|
| 187 |
video_part = {"inline_data": {"mime_type": "video/mp4", "data": b64}}
|
| 188 |
model = genai.GenerativeModel(model_name=model_id)
|
|
|
|
| 196 |
|
| 197 |
|
| 198 |
def _strip_prompt_echo(prompt: str, text: str, threshold: float = 0.68) -> str:
|
|
|
|
| 199 |
if not prompt or not text:
|
| 200 |
return text
|
| 201 |
clean_prompt = " ".join(prompt.lower().split())
|
|
|
|
| 210 |
# UI helpers
|
| 211 |
# ----------------------------------------------------------------------
|
| 212 |
def _expand_sidebar(width: int = 380) -> None:
|
|
|
|
| 213 |
st.markdown(
|
| 214 |
f"""
|
| 215 |
<style>
|
|
|
|
| 222 |
unsafe_allow_html=True,
|
| 223 |
)
|
| 224 |
|
| 225 |
+
|
| 226 |
+
# ----------------------------------------------------------------------
|
| 227 |
+
# Streamlit UI
|
| 228 |
+
# ----------------------------------------------------------------------
|
| 229 |
+
def main() -> None:
|
| 230 |
+
st.set_page_config(page_title="Video Analysis", layout="wide")
|
| 231 |
+
_expand_sidebar()
|
| 232 |
+
|
| 233 |
+
# ---------- Sidebar ----------
|
| 234 |
+
st.sidebar.header("Video Input")
|
| 235 |
+
st.sidebar.text_input("Video URL", key="url", placeholder="https://")
|
| 236 |
+
|
| 237 |
+
if st.sidebar.button("Load Video"):
|
| 238 |
+
try:
|
| 239 |
+
with st.spinner("Downloading video…"):
|
| 240 |
+
raw_path = download_video(
|
| 241 |
+
st.session_state["url"], DATA_DIR, st.session_state["video_password"]
|
| 242 |
+
)
|
| 243 |
+
mp4_path = _convert_to_mp4(Path(raw_path))
|
| 244 |
+
# Optional compression (if size exceeds user‑defined limit)
|
| 245 |
+
mp4_path, _ = _maybe_compress(mp4_path, st.session_state["compress_mb"])
|
| 246 |
+
st.session_state["video_path"] = str(mp4_path)
|
| 247 |
+
st.session_state["last_error"] = ""
|
| 248 |
+
st.toast("Video ready")
|
| 249 |
+
st.experimental_rerun()
|
| 250 |
+
except Exception as e:
|
| 251 |
+
st.session_state["last_error"] = f"Download failed: {e}"
|
| 252 |
+
st.sidebar.error(st.session_state["last_error"])
|
| 253 |
+
|
| 254 |
+
# ---------- Settings ----------
|
| 255 |
+
with st.sidebar.expander("Settings", expanded=False):
|
| 256 |
+
model = st.selectbox(
|
| 257 |
+
"Model", MODEL_OPTIONS, index=MODEL_OPTIONS.index(DEFAULT_MODEL)
|
| 258 |
+
)
|
| 259 |
+
if model == "custom":
|
| 260 |
+
model = st.text_input("Custom model ID", value=DEFAULT_MODEL, key="custom_model")
|
| 261 |
+
st.session_state["model_input"] = model
|
| 262 |
+
|
| 263 |
+
# API key handling
|
| 264 |
+
secret_key = os.getenv("GOOGLE_API_KEY", "")
|
| 265 |
+
if secret_key:
|
| 266 |
+
st.session_state["api_key"] = secret_key
|
| 267 |
+
st.text_input("Google API Key", key="api_key", type="password")
|
| 268 |
+
|
| 269 |
+
st.text_area(
|
| 270 |
+
"Analysis prompt",
|
| 271 |
+
value=DEFAULT_PROMPT,
|
| 272 |
+
key="prompt",
|
| 273 |
+
height=140,
|
| 274 |
+
)
|
| 275 |
+
st.text_input(
|
| 276 |
+
"Video password (if needed)",
|
| 277 |
+
key="video_password",
|
| 278 |
+
type="password",
|
| 279 |
+
)
|
| 280 |
+
st.number_input(
|
| 281 |
+
"Compress if > (MB)",
|
| 282 |
+
min_value=10,
|
| 283 |
+
max_value=2000,
|
| 284 |
+
value=st.session_state.get("compress_mb", 200),
|
| 285 |
+
step=10,
|
| 286 |
+
key="compress_mb",
|
| 287 |
+
)
|
| 288 |
+
|
| 289 |
+
if st.sidebar.button("Clear Video"):
|
| 290 |
+
for f in DATA_DIR.iterdir():
|
| 291 |
+
try:
|
| 292 |
+
f.unlink()
|
| 293 |
+
except Exception:
|
| 294 |
+
pass
|
| 295 |
+
st.session_state["video_path"] = ""
|
| 296 |
+
st.session_state["analysis_out"] = ""
|
| 297 |
+
st.session_state["raw_output"] = ""
|
| 298 |
+
st.toast("All cached videos cleared")
|
| 299 |
+
st.experimental_rerun()
|
| 300 |
+
|
| 301 |
+
# ---------- Main panel ----------
|
| 302 |
+
if st.session_state["last_error"]:
|
| 303 |
+
st.error(st.session_state["last_error"])
|
| 304 |
+
|
| 305 |
+
if st.session_state.get("video_path"):
|
| 306 |
+
st.video(st.session_state["video_path"])
|
| 307 |
+
|
| 308 |
+
if st.button("Run Analysis"):
|
| 309 |
+
st.session_state["busy"] = True
|
| 310 |
+
st.session_state["analysis_out"] = ""
|
| 311 |
+
st.session_state["raw_output"] = ""
|
| 312 |
+
try:
|
| 313 |
+
with st.spinner("Generating report…"):
|
| 314 |
+
raw = generate_report(
|
| 315 |
+
Path(st.session_state["video_path"]),
|
| 316 |
+
st.session_state["prompt"],
|
| 317 |
+
st.session_state["model_input"],
|
| 318 |
+
)
|
| 319 |
+
cleaned = _strip_prompt_echo(st.session_state["prompt"], raw)
|
| 320 |
+
st.session_state["analysis_out"] = cleaned
|
| 321 |
+
st.session_state["raw_output"] = raw
|
| 322 |
+
except Exception as e:
|
| 323 |
+
st.session_state["last_error"] = f"Analysis failed: {e}"
|
| 324 |
+
st.session_state["last_error_detail"] = traceback.format_exc()
|
| 325 |
+
finally:
|
| 326 |
+
st.session_state["busy"] = False
|
| 327 |
+
st.experimental_rerun()
|
| 328 |
+
|
| 329 |
+
# Show results
|
| 330 |
+
if st.session_state.get("analysis_out"):
|
| 331 |
+
st.subheader("📝 Analysis")
|
| 332 |
+
st.write(st.session_state["analysis_out"])
|
| 333 |
+
|
| 334 |
+
with st.expander("Show raw model output"):
|
| 335 |
+
st.code(st.session_state["raw_output"], language="text")
|
| 336 |
+
|
| 337 |
+
# Optional error details
|
| 338 |
+
if st.session_state.get("last_error_detail"):
|
| 339 |
+
with st.expander("Show error details"):
|
| 340 |
+
st.code(st.session_state["last_error_detail"], language="text")
|
| 341 |
+
|
| 342 |
+
|
| 343 |
# ----------------------------------------------------------------------
|
| 344 |
# Session‑state defaults
|
| 345 |
# ----------------------------------------------------------------------
|
| 346 |
def _init_state() -> None:
|
|
|
|
| 347 |
defaults = {
|
| 348 |
"url": "",
|
| 349 |
"video_path": "",
|
|
|
|
| 357 |
"analysis_out": "",
|
| 358 |
"raw_output": "",
|
| 359 |
"last_error_detail": "",
|
|
|
|
|
|
|
| 360 |
}
|
| 361 |
for k, v in defaults.items():
|
| 362 |
st.session_state.setdefault(k, v)
|
| 363 |
|
| 364 |
|
| 365 |
# ----------------------------------------------------------------------
|
| 366 |
+
# Entry point
|
| 367 |
# ----------------------------------------------------------------------
|
| 368 |
if __name__ == "__main__":
|
|
|
|
| 369 |
_init_state()
|
|
|
|
|
|
|
| 370 |
if st.session_state["api_key"]:
|
| 371 |
genai.configure(api_key=st.session_state["api_key"])
|
|
|
|
|
|
|
| 372 |
main()
|