# streamlit_app.py import os import time import string import hashlib import traceback import base64 from glob import glob from pathlib import Path import yt_dlp import ffmpeg # ffmpeg-python import streamlit as st from dotenv import load_dotenv load_dotenv() # Optional PHI integration (kept guarded) try: from phi.agent import Agent from phi.model.google import Gemini from phi.tools.duckduckgo import DuckDuckGo HAS_PHI = True except Exception: Agent = Gemini = DuckDuckGo = None HAS_PHI = False # google.generativeai SDK (guarded) try: import google.generativeai as genai HAS_GENAI = True except Exception: genai = None HAS_GENAI = False st.set_page_config(page_title="Generate the story of videos", layout="wide") DATA_DIR = Path("./data") DATA_DIR.mkdir(exist_ok=True) # ---- Defaults & constants ---- MODEL_OPTIONS = [ "gemini-2.5-flash", "gemini-2.5-flash-lite", "gemini-2.0-flash", "gemini-2.0-flash-lite", "custom", ] DEFAULT_MODEL = "gemini-2.0-flash-lite" DEFAULT_PROMPT = ( "Watch the video and provide a detailed behavioral report focusing on human actions, interactions, posture, movement, and apparent intent. " "Keep language professional. Include a list of observations for notable events." ) # ---- Session defaults ---- st.session_state.setdefault("url", "") # 'current_video_path' will store the path to the video file currently being displayed/analyzed. # This could be a downloaded file, a converted file, or a compressed file. st.session_state.setdefault("current_video_path", "") st.session_state.setdefault("loop_video", False) st.session_state.setdefault("uploaded_file", None) # Kept for consistency if file upload is added later st.session_state.setdefault("processed_file", None) # Kept for consistency if file upload is added later st.session_state.setdefault("busy", False) st.session_state.setdefault("last_url", "") st.session_state.setdefault("analysis_out", "") st.session_state.setdefault("last_error", "") st.session_state.setdefault("file_hash", None) st.session_state.setdefault("api_key", os.getenv("GOOGLE_API_KEY", "")) st.session_state.setdefault("last_model", "") st.session_state.setdefault("processing_timeout", 900) st.session_state.setdefault("generation_timeout", 300) st.session_state.setdefault("compress_threshold_mb", 200) # ---- Helpers ---- def sanitize_filename(path_str: str): name = Path(path_str).name # Remove file extension before sanitizing name_without_ext = Path(name).stem sanitized = name_without_ext.lower().translate(str.maketrans("", "", string.punctuation)).replace(" ", "_") return f"{sanitized}.mp4" # Ensure it's an mp4 extension for download consistency def file_sha256(path: str, block_size: int = 65536) -> str: try: h = hashlib.sha256() with open(path, "rb") as f: for chunk in iter(lambda: f.read(block_size), b""): h.update(chunk) return h.hexdigest() except Exception: return None def convert_video_to_mp4(video_path: str) -> str: target_path = str(Path(video_path).with_suffix(".mp4")) if os.path.exists(target_path) and os.path.getsize(target_path) > 0: if str(Path(video_path).resolve()) != str(Path(target_path).resolve()): try: os.remove(video_path) # Clean up original if different and conversion was successful except Exception: pass return target_path # Only convert if the target doesn't exist or is empty with st.status(f"Converting video to MP4: {Path(video_path).name}...", expanded=True) as status: try: ffmpeg.input(video_path).output(target_path).run(overwrite_output=True, quiet=True) if os.path.exists(target_path) and os.path.getsize(target_path) > 0: status.update(label=f"Conversion successful: {Path(target_path).name}", state="complete", expanded=False) if str(Path(video_path).resolve()) != str(Path(target_path).resolve()): try: os.remove(video_path) except Exception: pass return target_path else: status.update(label=f"Conversion failed, target file empty: {Path(target_path).name}", state="error", expanded=True) raise RuntimeError("Converted MP4 file is empty.") except ffmpeg.Error as e: status.update(label=f"FFmpeg conversion failed: {e.stderr.decode()}", state="error", expanded=True) raise except Exception as e: status.update(label=f"Video conversion failed: {e}", state="error", expanded=True) raise def compress_video(input_path: str, target_path: str, crf: int = 28, preset: str = "fast"): with st.status(f"Compressing video: {Path(input_path).name}...", expanded=True) as status: try: ffmpeg.input(input_path).output( target_path, vcodec="libx264", crf=crf, preset=preset, movflags="+faststart" # Optimize for web streaming ).run(overwrite_output=True, quiet=True) if os.path.exists(target_path) and os.path.getsize(target_path) > 0: status.update(label=f"Compression successful: {Path(target_path).name}", state="complete", expanded=False) return target_path else: status.update(label=f"Compression failed, target file empty: {Path(target_path).name}", state="error", expanded=True) return input_path # Return original if compressed is empty except ffmpeg.Error as e: status.update(label=f"FFmpeg compression failed: {e.stderr.decode()}", state="error", expanded=True) st.session_state["last_error"] = f"FFmpeg compression failed: {e.stderr.decode()}" return input_path except Exception as e: status.update(label=f"Video compression failed: {e}", state="error", expanded=True) st.session_state["last_error"] = f"Video compression failed: {e}" return input_path def download_video_ytdlp(url: str, save_dir: str, video_password: str = None) -> str: if not url: raise ValueError("No URL provided") with st.status(f"Downloading video from {url}...", expanded=True) as status: try: # Use %(title)s for a more descriptive filename, but ensure it's safe outtmpl_base = Path(save_dir) / "%(title)s.%(ext)s" # yt_dlp handles sanitization for filenames, so directly use %(title)s ydl_opts = { "outtmpl": str(outtmpl_base), "format": "bestvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best", # Prefer mp4 if possible "noplaylist": True, # Ensure only single video is downloaded for direct URLs "writethumbnail": False, "quiet": True, "noprogress": True, "geo_bypass": True, "retries": 5 } if video_password: ydl_opts["videopassword"] = video_password with yt_dlp.YoutubeDL(ydl_opts) as ydl: info = ydl.extract_info(url, download=True) # Find the actual downloaded file. yt_dlp typically gives the full path in 'filepath' or 'requested_downloads' downloaded_filepath = None if info: if 'filepath' in info: downloaded_filepath = info['filepath'] elif 'requested_downloads' in info and isinstance(info['requested_downloads'], list): for dl in info['requested_downloads']: if 'filepath' in dl: downloaded_filepath = dl['filepath'] break elif 'id' in info and 'ext' in info: # Fallback if no specific filepath, based on id and ext filename_pattern = Path(save_dir) / f"{info['id']}.*" candidates = glob(str(filename_pattern)) if candidates: downloaded_filepath = sorted(candidates, key=os.path.getmtime, reverse=True)[0] if not downloaded_filepath or not os.path.exists(downloaded_filepath): # Final fallback: scan directory for recently created files all_files = glob(os.path.join(save_dir, "*")) if not all_files: raise FileNotFoundError("Downloaded video not found in expected location.") downloaded_filepath = sorted(all_files, key=os.path.getmtime, reverse=True)[0] status.update(label=f"Download complete: {Path(downloaded_filepath).name}", state="complete", expanded=False) return downloaded_filepath except yt_dlp.DownloadError as e: status.update(label=f"Download failed: {e}", state="error", expanded=True) raise ValueError(f"Failed to download video: {e}") except Exception as e: status.update(label=f"An unexpected error occurred during download: {e}", state="error", expanded=True) raise def remove_prompt_echo(prompt: str, text: str, check_len: int = 600, ratio_threshold: float = 0.68): if not prompt or not text: return text a = " ".join(prompt.strip().lower().split()) b_full = text.strip() b = " ".join(b_full[:check_len].lower().split()) try: from difflib import SequenceMatcher ratio = SequenceMatcher(None, a, b).ratio() except Exception: ratio = 0.0 if ratio >= ratio_threshold: # Cut based on prompt length, ensuring we don't cut too much cut_point = min(len(b_full), len(b_full) - len(b) + len(prompt)) new_text = b_full[cut_point:].lstrip(" \n:-") if len(new_text) >= 3: # Ensure we don't return an empty or almost empty string return new_text placeholders = ["enter analysis", "enter your analysis", "enter analysis here", "please enter analysis"] low = b_full.strip().lower() for ph in placeholders: if low.startswith(ph): return b_full[len(ph):].lstrip(" \n:-") return text def compress_video_if_large(local_path: str, threshold_mb: int = 200): try: file_size_mb = os.path.getsize(local_path) / (1024 * 1024) except Exception: st.session_state["last_error"] = "Failed to stat file before compression." return local_path, False if file_size_mb <= threshold_mb: return local_path, False p = Path(local_path) compressed_name = f"{p.stem}_compressed.mp4" compressed_path = str(p.with_name(compressed_name)) st.toast(f"Compressing video {p.name} (size: {file_size_mb:.1f}MB)...", icon="đŸ—œī¸") result = compress_video(local_path, compressed_path, crf=28, preset="fast") if result and os.path.exists(result) and os.path.getsize(result) > 0: if result != local_path: # Compression was successful and produced a new file try: os.remove(local_path) # Remove original uncompressed file except Exception as e: st.session_state["last_error"] = f"Failed to remove original video after compression: {e}" st.toast(f"Video compressed to {os.path.getsize(result) / (1024 * 1024):.1f}MB.", icon="✅") return result, True else: # Result is the same as input_path, meaning compression failed or didn't reduce size st.session_state["last_error"] = "Video compression did not produce a smaller file or failed." return local_path, False return local_path, False # Fallback # ---- Inline-video generation via base64 (bypass upload) ---- def generate_with_inline_video(local_path: str, prompt: str, model_used: str, timeout: int = 300): if not HAS_GENAI: raise RuntimeError("Google Generative AI SDK not available.") if not Path(local_path).exists(): raise FileNotFoundError(f"Video file not found: {local_path}") # Read the video bytes with open(local_path, "rb") as f: video_bytes = f.read() b64 = base64.b64encode(video_bytes).decode("utf-8") video_part = { "inline_data": { "mime_type": "video/mp4", "data": b64 } } contents = [prompt, video_part] try: # Use genai.GenerativeModel directly for consistent behavior across SDK versions model = genai.GenerativeModel(model_used) resp = model.generate_content( contents, generation_config={"max_output_tokens": 2048}, # Increased max_output_tokens request_options={"timeout": timeout}, ) text = getattr(resp, "text", None) or getattr(resp, "output_text", None) or str(resp) return text except Exception as e: st.session_state["last_error"] = f"Generation failed: {e}\n{traceback.format_exc()}" raise def _init_genai(): """Initializes genai with the API key if not already configured or if key changes.""" current_key = st.session_state.get("api_key") or os.getenv("GOOGLE_API_KEY") if not current_key: return False # Check if genai is already configured with this key # (This is a heuristic, actual SDK doesn't expose configured key easily) if hasattr(st.session_state, '_genai_configured_key') and st.session_state._genai_configured_key == current_key: return True try: genai.configure(api_key=current_key) st.session_state._genai_configured_key = current_key return True except Exception as e: st.session_state["last_error"] = f"Failed to configure Google Generative AI: {e}" return False def _clear_video_state(): """Clears all video-related session state and deletes local video files.""" st.session_state.pop("url", None) st.session_state.pop("current_video_path", None) st.session_state.pop("uploaded_file", None) st.session_state.pop("processed_file", None) st.session_state["analysis_out"] = "" st.session_state["last_error"] = "" st.session_state["file_hash"] = None st.session_state["last_url"] = "" # Clear last_url as well to prevent re-triggering for f in glob(str(DATA_DIR / "*")): try: os.remove(f) except Exception: pass st.toast("Video cleared and local files removed.", icon="đŸ—‘ī¸") # ---- Main UI & logic ---- # Check for URL change and clear state if it's a new URL current_url = st.session_state.get("url", "") if current_url != st.session_state.get("last_url"): if st.session_state.get("last_url"): # Only clear if a previous URL was set _clear_video_state() st.session_state["url"] = current_url # Re-set the new URL after clearing st.session_state["last_url"] = current_url # Sidebar UI st.sidebar.header("Video Input") st.sidebar.text_input("Video URL", key="url", placeholder="https://", on_change=_clear_video_state if st.session_state.get("url") else None) settings_exp = st.sidebar.expander("Settings", expanded=False) model_choice = settings_exp.selectbox("Select model", options=MODEL_OPTIONS, index=MODEL_OPTIONS.index(DEFAULT_MODEL) if DEFAULT_MODEL in MODEL_OPTIONS else 0) if model_choice == "custom": model_input = settings_exp.text_input("Custom model id", value=DEFAULT_MODEL, key="model_input") model_selected = model_input.strip() or DEFAULT_MODEL else: st.session_state["model_input"] = model_choice # Update for custom model name model_selected = model_choice settings_exp.text_input("Google API Key", key="api_key", value=st.session_state.get("api_key"), type="password") analysis_prompt = settings_exp.text_area("Analysis prompt", value=DEFAULT_PROMPT, height=140) settings_exp.text_input("Video Password (if needed)", key="video-password", placeholder="password", type="password") settings_exp.number_input( "Processing timeout (s)", min_value=60, max_value=3600, value=st.session_state.get("processing_timeout", 900), step=30, key="processing_timeout", ) settings_exp.number_input( "Generation timeout (s)", min_value=30, max_value=1800, value=st.session_state.get("generation_timeout", 300), step=10, key="generation_timeout", ) settings_exp.number_input( "Optional compression threshold (MB)", min_value=10, max_value=2000, value=st.session_state.get("compress_threshold_mb", 200), step=10, key="compress_threshold_mb", ) key_source = "session" if st.session_state.get("api_key") else ".env" if os.getenv("GOOGLE_API_KEY") else "none" settings_exp.caption(f"Using API key from: {key_source}") if not st.session_state.get("api_key") and not os.getenv("GOOGLE_API_KEY"): settings_exp.warning("No Google API key provided; generation disabled.", icon="âš ī¸") # Buttons & layout col1, col2 = st.columns([1, 3]) with col1: generate_now = st.button("Generate the story", type="primary", disabled=st.session_state.get("busy") or not st.session_state.get("current_video_path")) with col2: if not st.session_state.get("current_video_path"): st.info("Load a video first (sidebar) to enable generation.", icon="â„šī¸") elif st.session_state.get("busy"): st.info("Processing in progress...", icon="âŗ") else: st.write("") # Placeholder for alignment if st.sidebar.button("Load Video", use_container_width=True, disabled=st.session_state.get("busy")): st.session_state["last_error"] = "" # Clear previous error st.session_state["busy"] = True try: url = st.session_state.get("url", "").strip() if not url: raise ValueError("Please enter a video URL.") vpw = st.session_state.get("video-password", "") downloaded_path = download_video_ytdlp(url, str(DATA_DIR), vpw) # Ensure it's an MP4 and compress if needed converted_path = convert_video_to_mp4(downloaded_path) final_path, was_compressed = compress_video_if_large( converted_path, st.session_state.get("compress_threshold_mb", 200) ) st.session_state["current_video_path"] = final_path st.session_state["file_hash"] = file_sha256(final_path) st.toast("Video loaded and ready!", icon="✅") except Exception as e: st.session_state["last_error"] = f"Failed to load video: {e}\n{traceback.format_exc()}" st.sidebar.error("Failed to load video. Check the error log in the main area.") finally: st.session_state["busy"] = False # Display video and related info if a video is loaded if st.session_state["current_video_path"]: path_to_display = st.session_state["current_video_path"] try: with open(path_to_display, "rb") as vf: video_bytes = vf.read() st.sidebar.video(video_bytes, format="video/mp4", start_time=0, loop=st.session_state.get("loop_video", False)) except Exception: st.sidebar.write("Couldn't preview video.") st.session_state["last_error"] = f"Couldn't preview video from {path_to_display}. File might be corrupt or inaccessible." with st.sidebar.expander("Options", expanded=False): st.session_state["loop_video"] = st.checkbox("Enable Loop", value=st.session_state.get("loop_video", False), key="sidebar_loop_checkbox") if st.button("Clear Video(s)", key="clear_video_button", disabled=st.session_state.get("busy")): _clear_video_state() st.rerun() # Rerun to clear video display immediately try: with open(path_to_display, "rb") as vf: st.download_button( "Download Video", data=vf, file_name=sanitize_filename(path_to_display), # Use sanitized name mime="video/mp4", use_container_width=True, key="download_button" ) except Exception: st.sidebar.error("Failed to prepare download.") st.sidebar.write("Title:", Path(path_to_display).name) try: file_size_mb = os.path.getsize(path_to_display) / (1024 * 1024) st.sidebar.caption(f"File size: {file_size_mb:.1f} MB") if file_size_mb > st.session_state.get("compress_threshold_mb", 200) * 1.5: # Warn if significantly larger than threshold st.sidebar.warning( f"Large file (>{st.session_state.get('compress_threshold_mb')} MB) might exceed inline limits.", icon="âš ī¸" ) except Exception: pass # File might not exist or be accessible for size check # Generation / analysis if generate_now and not st.session_state.get("busy"): st.session_state["last_error"] = "" # Clear previous error if not st.session_state.get("current_video_path"): st.error("No video loaded. Use 'Load Video' in the sidebar.") else: if not _init_genai(): st.error("Google API key not set or failed to configure.") else: st.session_state["busy"] = True try: model_id = (st.session_state.get("model_input") or model_selected or DEFAULT_MODEL).strip() prompt_text = (analysis_prompt.strip() or DEFAULT_PROMPT).strip() st.subheader("Analysis Result") with st.status("Generating analysis (inline video)...", expanded=True, state="running") as status: out = generate_with_inline_video( st.session_state["current_video_path"], prompt_text, model_id, timeout=st.session_state.get("generation_timeout", 300) ) status.update(label="Analysis generation complete.", state="complete", expanded=False) out = remove_prompt_echo(prompt_text, out) st.session_state["analysis_out"] = out st.markdown(out or "No analysis returned.") st.toast("Analysis complete!", icon="✨") except Exception as e: st.session_state["last_error"] = f"An error occurred during analysis: {e}\n{traceback.format_exc()}" st.error("An error occurred while generating the story. Please check the error log below.") finally: st.session_state["busy"] = False # Display last error if any if st.session_state.get("last_error"): st.error("An error occurred:") st.code(st.session_state["last_error"], language="text") # Display previous analysis if available and no new error elif st.session_state.get("analysis_out") and not st.session_state.get("busy"): st.subheader("Analysis Result (Previous)") st.markdown(st.session_state["analysis_out"])