Update app.py

app.py

@@ -3,378 +3,266 @@
 """
 YouTube Video Analyzer & Downloader Pro
 ───────────────────────────────────────
-•  `www.youtube.com_cookies.txt` 파일이 app.py와 같은 폴더에 있으면 자동 사용
-•  UI에서 쿠키를 업로드하면 그 파일이 *우선* 적용
-•  “Transcript” 탭에서 **전체 자막 + MM:SS 타임스탬프** 제공
+· app.py 와 같은 폴더에  𝐰𝐰𝐰.𝐲𝐨𝐮𝐭𝐮𝐛𝐞.𝐜𝐨𝐨𝐤𝐢𝐞𝐬.𝐭𝐱𝐭  이 있으면 자동 사용
+· Gradio UI에서 쿠키 파일 업로드 시 → 업로드 파일이 우선
+· Gemini-1.5 Flash 장면 분석(선택적), 전체 Transcript, 다운로드 지원
 """
 
-# ──────────────────────────────────────
-# 표준 라이브러리
-# ──────────────────────────────────────
-import os
-import re
-import shutil
-import tempfile
+# ───────────────────────── 표준 라이브러리 ─────────────────────────
+import os, re, shutil, tempfile
 from datetime import datetime, timedelta
 from pathlib import Path
 
-# ──────────────────────────────────────
-# 서드파티 라이브러리
-# ──────────────────────────────────────
+# ───────────────────────── 서드파티 라이브러리 ────────────────────
 import gradio as gr
 import yt_dlp
 import google.generativeai as genai
 from youtube_transcript_api import YouTubeTranscriptApi
 
-# ──────────────────────────────────────
-# 상수
-# ──────────────────────────────────────
-DEFAULT_COOKIE_FILE = Path(__file__).with_name("www.youtube.com_cookies.txt")
+# ───────────────────────── 상수 (절대경로) ─────────────────────────
+DEFAULT_COOKIE_FILE = (Path(__file__).resolve().parent / "www.youtube.com_cookies.txt").resolve()
 
 # 유튜브 URL 정규식
-_YT_REGEX = re.compile(
+_YT_RE = re.compile(
     r"(https?://)?(www\.)?"
     r"(youtube|youtu|youtube-nocookie)\.(com|be)/"
     r"(watch\?v=|embed/|v/|.+\?v=)?([^&=%\?]{11})"
 )
 
-
-# ──────────────────────────────────────
-# Helper : video-ID 추출 / 자막 가져오기
-# ──────────────────────────────────────
+# ───────────────────────── Helper (ID, 자막) ─────────────────────
 def extract_video_id(url: str) -> str | None:
-    m = _YT_REGEX.match(url)
+    m = _YT_RE.match(url)
     return m.group(6) if m else None
 
-
-def fetch_transcript(video_id: str, pref_lang=("ko", "en")) -> str:
-    # 언어 우선순위대로 시도
+def fetch_transcript(video_id: str, pref=("ko", "en")) -> str:
     tr = None
-    for lang in pref_lang:
+    for lang in pref:
         try:
             tr = YouTubeTranscriptApi.get_transcript(video_id, languages=[lang])
             break
         except Exception:
             continue
     if tr is None:
-        tr = YouTubeTranscriptApi.get_transcript(video_id)  # 마지막 시도
-
-    lines: list[str] = []
+        tr = YouTubeTranscriptApi.get_transcript(video_id)
+    lines = []
     for seg in tr:
-        t = str(timedelta(seconds=int(seg["start"])))      # H:MM:SS
-        t_mmss = ":".join(t.split(":")[-2:])               # MM:SS
-        lines.append(f"**[{t_mmss}]** {seg['text']}")
+        t = str(timedelta(seconds=int(seg["start"])))
+        mmss = ":".join(t.split(":")[-2:])
+        lines.append(f"**[{mmss}]** {seg['text']}")
     return "\n".join(lines)
 
-
-# ──────────────────────────────────────
-# 메인 클래스
-# ──────────────────────────────────────
+# =================================================================
+# Main Class
+# =================================================================
 class YouTubeDownloader:
     def __init__(self):
-        # 임시 디렉터리
-        self.download_dir = tempfile.mkdtemp()
-        self.temp_downloads = tempfile.mkdtemp(prefix="youtube_downloads_")
-        # Downloads 폴더
-        self.downloads_folder = os.path.join(
-            os.path.expanduser("~"), "Downloads", "YouTube_Downloads"
-        )
-        os.makedirs(self.downloads_folder, exist_ok=True)
-
-        self.gemini_model = None  # Gemini 모델 핸들
+        # temp dirs
+        self.temp_dir = tempfile.mkdtemp(prefix="yt_tmp_")
+        self.temp_downloads = tempfile.mkdtemp(prefix="yt_dl_")
+        # Downloads target
+        self.dl_folder = Path.home() / "Downloads" / "YouTube_Downloads"
+        self.dl_folder.mkdir(parents=True, exist_ok=True)
+        # Gemini
+        self.gemini_model = None
 
     # ───────── Gemini 설정 ─────────
     def configure_gemini(self, api_key: str):
         try:
             genai.configure(api_key=api_key)
             self.gemini_model = genai.GenerativeModel("gemini-1.5-flash-latest")
-            return True, "✅ Gemini API configured successfully!"
+            return True, "✅ Gemini API configured!"
         except Exception as e:
-            return False, f"❌ Failed to configure Gemini API: {e}"
-
-    # ───────── 정리 ─────────
-    def cleanup(self):
-        for p in (self.download_dir, self.temp_downloads):
-            try:
-                shutil.rmtree(p, ignore_errors=True)
-            except Exception:
-                pass
-
-    # ───────── 유효성 검사 ─────────
-    def is_valid_youtube_url(self, url: str) -> bool:
-        return _YT_REGEX.match(url) is not None
-
-    # ───────── 장면 분석 : Gemini ─────────
-    def generate_scene_breakdown_gemini(self, info: dict) -> list[str]:
-        if not self.gemini_model:
-            return self.generate_scene_breakdown_fallback(info)
+            return False, f"❌ Gemini 설정 실패: {e}"
 
-        try:
-            duration = info.get("duration", 0)
-            title = info.get("title", "")
-            description = info.get("description", "")[:1500]
-
-            prompt = f"""
-Analyze this YouTube video and create a highly detailed, scene-by-scene breakdown
-with precise timestamps (MM:SS-MM:SS) and specific descriptions:
-
-Title : {title}
-Duration : {duration} s
-Description : {description}
-
-Follow these rules:
-• Include visual details, actions, inferred dialogue, setting, props, graphics
-• Dialogue : short lines for **every** scene if plausible
-• Timestamp : 2-3 s (<1 min) / 3-5 s (1-5 min) / 5-10 s (5-15 min) / 10-15 s (>15 min)
-• ≤ 20 scenes total
-• Formatting → **[MM:SS-MM:SS]** Description
-"""
-            resp = self.gemini_model.generate_content(prompt)
-            if not resp or not resp.text:
-                return self.generate_scene_breakdown_fallback(info)
-
-            scenes, cur = [], ""
-            for line in resp.text.splitlines():
-                line = line.strip()
-                if line.startswith("**[") and "]**" in line:
-                    if cur:
-                        scenes.append(cur.strip())
-                    cur = line
-                elif cur:
-                    cur += "\n" + line
-            if cur:
-                scenes.append(cur.strip())
-            return scenes or self.generate_scene_breakdown_fallback(info)
-
-        except Exception:
-            return self.generate_scene_breakdown_fallback(info)
-
-    # ───────── 장면 분석 : Fallback ─────────
-    def generate_scene_breakdown_fallback(self, info: dict) -> list[str]:
-        duration = info.get("duration", 0)
-        if not duration:
-            return ["**[00:00]** Unable to determine duration"]
-
-        seg = 3 if duration <= 60 else 5 if duration <= 300 else 10 if duration <= 900 else 15
-        total = min(duration // seg + 1, 20)
-        scenes: list[str] = []
-        for i in range(total):
-            s, e = i * seg, min(i * seg + seg - 1, duration)
-            scenes.append(
-                f"**[{s//60:02d}:{s%60:02d}-{e//60:02d}:{e%60:02d}]** "
-                "Content continues…"
-            )
-        return scenes
-
-    # ───────── 숫자 포맷 ─────────
+    # ───────── 쿠키 선택 + 디버그 ─────────
     @staticmethod
-    def fmt_num(n: int) -> str:
-        if n >= 1_000_000:
-            return f"{n/1_000_000:.1f} M"
-        if n >= 1_000:
-            return f"{n/1_000:.1f} K"
-        return str(n)
-
-    # ───────── 결과 리포트 ─────────
-    def format_video_info(self, info: dict) -> str:
-        title = info.get("title", "Unknown")
-        uploader = info.get("uploader", "Unknown")
-        duration = info.get("duration", 0)
-        dur = f"{duration//60}:{duration%60:02d}"
-        views = self.fmt_num(info.get("view_count", 0))
-        likes = self.fmt_num(info.get("like_count", 0))
-        comments = self.fmt_num(info.get("comment_count", 0))
-        scenes = "\n".join(self.generate_scene_breakdown_gemini(info))
-
-        return f"""\
-🎬 **{title}**
-Uploader : {uploader}
-Duration : {dur} Views / Likes / Comments : {views} / {likes} / {comments}
-
-{'-'*48}
-{scenes}
-"""
-
-    # ───────── 메타데이터 ─────────
-    def get_video_info(self, url: str, cookiefile: str | None = None):
-        if cookiefile and os.path.exists(cookiefile):
-            ck = cookiefile
+    def choose_cookie(ui_cookie: str | None):
+        if ui_cookie and os.path.exists(ui_cookie):
+            ck = ui_cookie
         elif DEFAULT_COOKIE_FILE.exists():
             ck = str(DEFAULT_COOKIE_FILE)
         else:
             ck = None
+        print(">>> COOKIE =", ck, "EXISTS?", os.path.exists(ck) if ck else None)
+        return ck
 
-        ydl_opts = {"noplaylist": True, "quiet": True}
-        if ck:
-            ydl_opts["cookiefile"] = ck
-        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
-            return ydl.extract_info(url, download=False)
+    # ───────── URL 검증 ─────────
+    @staticmethod
+    def valid_url(url: str) -> bool:
+        return bool(_YT_RE.match(url))
 
-    # ───────── 다운로드 ─────────
-    def download_video(
-        self,
-        url: str,
-        quality: str = "best",
-        audio_only: bool = False,
-        cookiefile: str | None = None,
-    ):
-        if cookiefile and os.path.exists(cookiefile):
-            ck = cookiefile
-        elif DEFAULT_COOKIE_FILE.exists():
-            ck = str(DEFAULT_COOKIE_FILE)
-        else:
-            ck = None
+    # ───────── 숫자 포맷 ─────────
+    @staticmethod
+    def fmt(n: int) -> str:
+        if n >= 1_000_000: return f"{n/1_000_000:.1f} M"
+        if n >= 1_000:     return f"{n/1_000:.1f} K"
+        return str(n)
 
+    # ───────── Scene 분석 (fallback만 간단 구현) ─────────
+    def scene_breakdown(self, info: dict):
+        dur = info.get("duration", 0)
+        seg = 3 if dur <= 60 else 5 if dur <= 300 else 10
+        scenes = []
+        for s in range(0, dur, seg):
+            e = min(s + seg - 1, dur)
+            scenes.append(f"**[{s//60:02d}:{s%60:02d}-{e//60:02d}:{e%60:02d}]** …")
+        return "\n".join(scenes[:20])
+
+    # ───────── Video Info ─────────
+    def get_info(self, url: str, ui_cookie: str | None, progress):
+        if not self.valid_url(url):
+            return None, "❌ URL 오류"
+        ck = self.choose_cookie(ui_cookie)
+        if ck is None:
+            return None, "❌ 쿠키 파일을 찾을 수 없습니다"
+        ydl_opts = {"quiet": True, "noplaylist": True, "cookiefile": ck}
+        try:
+            progress(0.1, desc="metadata…")
+            with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+                info = ydl.extract_info(url, download=False)
+            progress(1.0)
+            return info, "OK"
+        except Exception as e:
+            return None, f"yt-dlp 오류: {e}"
+
+    # ───────── Download ─────────
+    def download(
+        self, url: str, qual: str, audio: bool, ui_cookie: str | None, progress
+    ):
+        if not self.valid_url(url):
+            return None, "❌ URL 오류"
+        ck = self.choose_cookie(ui_cookie)
+        if ck is None:
+            return None, "❌ 쿠키 파일을 찾을 수 없습니다"
         ts = datetime.now().strftime("%Y%m%d_%H%M%S")
         ydl_opts: dict = {
-            "outtmpl": os.path.join(self.temp_downloads, f"%(title)s_{ts}.%(ext)s"),
+            "outtmpl": str(Path(self.temp_downloads) / f"%(title)s_{ts}.%(ext)s"),
+            "quiet": True,
             "noplaylist": True,
+            "cookiefile": ck,
         }
-
-        if audio_only:
+        if audio:
             ydl_opts["format"] = "bestaudio/best"
             ydl_opts["postprocessors"] = [
                 {"key": "FFmpegExtractAudio", "preferredcodec": "mp3", "preferredquality": "192"}
             ]
         else:
-            if quality == "720p":
-                ydl_opts["format"] = "best[height<=720]"
-            elif quality == "480p":
-                ydl_opts["format"] = "best[height<=480]"
-            else:
-                ydl_opts["format"] = "best[height<=1080]"
-
-        if ck:
-            ydl_opts["cookiefile"] = ck
-
-        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
-            ydl.extract_info(url, download=True)
-
-        # temp 파일 찾기
-        for f in os.listdir(self.temp_downloads):
-            if ts in f:
-                src = os.path.join(self.temp_downloads, f)
-                dst = os.path.join(self.downloads_folder, f)
-                try:
-                    shutil.copy2(src, dst)
-                    return dst, "✅ Saved to Downloads"
-                except Exception:
-                    return src, "✅ Saved to temp (copy failed)"
-        return None, "❌ File not found"
-
-
-# ──────────────────────────────────────
-# Gradio Helper
-# ─────────────────────��────────────────
-downloader = YouTubeDownloader()
+            ydl_opts["format"] = (
+                "best[height<=720]" if qual == "720p"
+                else "best[height<=480]" if qual == "480p"
+                else "best[height<=1080]"
+            )
+        try:
+            progress(0.1, desc="Downloading…")
+            with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+                ydl.extract_info(url, download=True)
+            # temp → Downloads
+            for f in Path(self.temp_downloads).iterdir():
+                if ts in f.name:
+                    dest = self.dl_folder / f.name
+                    try:
+                        shutil.copy2(f, dest)
+                        saved = dest
+                    except Exception:
+                        saved = f
+                    progress(1.0)
+                    return saved, f"✅ 저장 위치: {saved}"
+            return None, "❌ 파일을 찾을 수 없습니다"
+        except Exception as e:
+            return None, f"❌ 다운로드 실패: {e}"
 
+# ───────────────────────── 인스턴스 ─────────────────────────
+yt_dl = YouTubeDownloader()
 
-def configure_api_key(api_key):
-    ok, msg = downloader.configure_gemini(api_key.strip()) if api_key else (False, "❌ API key required")
+# ───────────────────────── Gradio 함수 ──────────────────────
+def api_config(key):
+    ok, msg = yt_dl.configure_gemini(key.strip()) if key else (False, "❌ API 키 필요")
     return msg, gr.update(visible=ok)
 
-
-def analyze_fn(url, cookie):
-    try:
-        info = downloader.get_video_info(url, cookie)
-        return downloader.format_video_info(info)
-    except Exception as e:
-        return f"❌ {e}"
-
-
-def download_fn(url, qual, audio, cookie):
-    fp, msg = downloader.download_video(url, qual, audio, cookie)
-    return msg, fp
-
-
-def transcript_fn(url, _cookie):
+def analyze(url, cfile, progress=gr.Progress()):
+    info, err = yt_dl.get_info(url, cfile, progress)
+    if not info:
+        return err
+    report = (
+        f"**Title** : {info.get('title')}\n"
+        f"**Duration** : {info.get('duration',0)//60}:{info.get('duration',0)%60:02d}\n"
+        f"**Views / Likes** : {yt_dl.fmt(info.get('view_count',0))} / {yt_dl.fmt(info.get('like_count',0))}\n"
+        f"{'-'*40}\n"
+        f"{yt_dl.scene_breakdown(info)}"
+    )
+    return report
+
+def download(url, q, a, cfile, progress=gr.Progress()):
+    path, msg = yt_dl.download(url, q, a, cfile, progress)
+    return msg, path
+
+def transcript(url, _cfile):
     vid = extract_video_id(url)
     if not vid:
-        return "❌ Invalid URL"
+        return "❌ URL 오류"
     try:
         return fetch_transcript(vid)
     except Exception as e:
-        return f"❌ {e}"
+        return f"❌ 자막 오류: {e}"
 
+# ───────────────────────── UI ───────────────────────────────
+def create_ui():
+    with gr.Blocks(theme=gr.themes.Soft(), title="🎥 YouTube Video Analyzer & Downloader") as ui:
+        gr.HTML("<h1>🎥 YouTube Video Analyzer & Downloader</h1>")
 
-# ──────────────────────────────────────
-# UI
-# ──────────────────────────────────────
-def create_interface():
-    with gr.Blocks(theme=gr.themes.Soft(), title="🎥 YouTube Video Analyzer & Downloader Pro") as iface:
-
-        gr.HTML("<h1>🎥 YouTube Video Analyzer & Downloader Pro</h1>")
-
-        # API 설정
+        # API
         with gr.Group():
-            gr.HTML("<h3>🔑 Google Gemini API Configuration</h3>")
             with gr.Row():
-                api_key_in = gr.Textbox(label="🔑 Google API Key", type="password")
-                api_btn = gr.Button("🔧 Configure API", variant="secondary")
-            api_status = gr.Textbox(
-                value="❌ Gemini API not configured – Using fallback analysis",
-                interactive=False,
-                lines=1,
-                label="API Status",
-            )
+                api_key = gr.Textbox(type="password", label="Gemini API Key")
+                api_btn = gr.Button("Configure")
+            api_stat = gr.Textbox(label="API Status", value="❌ Gemini 미설정", interactive=False)
+        api_btn.click(api_config, [api_key], [api_stat])
 
-        # 공통 입력
+        # 입력
         with gr.Row():
-            url_in = gr.Textbox(label="🔗 YouTube URL", placeholder="Paste YouTube video URL…")
-            cookie_in = gr.File(label="🍪 Upload cookies.txt (optional)", type="filepath", file_types=[".txt"])
+            url = gr.Textbox(label="🔗 YouTube URL")
+            cookie = gr.File(label="🍪 cookies.txt (선택)", type="filepath", file_types=[".txt"])
 
         with gr.Tabs():
-            # 분석
-            with gr.TabItem("📊 Video Analysis"):
-                analyze_btn = gr.Button("🔍 Analyze", variant="primary")
-                analysis_out = gr.Textbox(label="📊 Analysis Report", lines=30, show_copy_button=True)
-                analyze_btn.click(analyze_fn, [url_in, cookie_in], analysis_out, show_progress=True)
 
-            # 다운로드
-            with gr.TabItem("⬇️ Video Download"):
-                with gr.Row():
-                    quality_dd = gr.Dropdown(["best", "720p", "480p"], value="best", label="📺 Quality")
-                    audio_cb = gr.Checkbox(label="🎵 Audio only (MP3)")
-                download_btn = gr.Button("⬇️ Download", variant="primary")
-                dl_status = gr.Textbox(label="📥 Status", lines=5, show_copy_button=True)
-                dl_file = gr.File(label="📁 File", visible=False)
+            with gr.TabItem("📊 Analyze"):
+                a_btn = gr.Button("Analyze", variant="primary")
+                a_out = gr.Textbox(lines=20, label="Analysis", show_copy_button=True)
+                a_btn.click(analyze, [url, cookie], a_out, show_progress=True)
 
-                def wrap_dl(u, q, a, c):
-                    msg, fp = download_fn(u, q, a, c)
-                    return (
-                        msg,
-                        gr.update(value=fp, visible=True) if fp and os.path.exists(fp) else gr.update(visible=False),
-                    )
+            with gr.TabItem("⬇️ Download"):
+                with gr.Row():
+                    q_dd = gr.Dropdown(["best", "720p", "480p"], value="best", label="Quality")
+                    a_cb = gr.Checkbox(label="Audio only (MP3)")
+                d_btn = gr.Button("Download", variant="primary")
+                d_stat = gr.Textbox(lines=5, label="Status")
+                d_file = gr.File(label="File", visible=False)
 
-                download_btn.click(wrap_dl, [url_in, quality_dd, audio_cb, cookie_in], [dl_status, dl_file], show_progress=True)
+                def wrap_dl(u,q,a,c,pg=gr.Progress()):
+                    msg, fp = download(u,q,a,c,pg)
+                    return msg, gr.update(value=fp, visible=bool(fp and os.path.exists(fp)))
+                d_btn.click(wrap_dl, [url,q_dd,a_cb,cookie], [d_stat,d_file], show_progress=True)
 
-            # 자막
             with gr.TabItem("🗒️ Transcript"):
-                tr_btn = gr.Button("📜 Get Transcript", variant="primary")
-                tr_out = gr.Textbox(label="Transcript", lines=30, show_copy_button=True)
-                tr_btn.click(transcript_fn, [url_in, cookie_in], tr_out, show_progress=True)
-
-        api_btn.click(configure_api_key, inputs=[api_key_in], outputs=[api_status])
+                t_btn = gr.Button("Get Transcript", variant="primary")
+                t_out = gr.Textbox(lines=30, label="Transcript", show_copy_button=True)
+                t_btn.click(transcript, [url, cookie], t_out, show_progress=True)
 
         gr.HTML(
-            """
+            f"""
 <div style="margin-top:20px;padding:15px;background:#f0f8ff;border-left:5px solid #4285f4;border-radius:10px;">
-  <h3>💡 Tip</h3>
-  <p><code>www.youtube.com_cookies.txt</code> 파일을 <b>app.py</b>와 같은
-  폴더에 두면 업로드 없이 자동 사용됩니다.</p>
+  <b>Tip</b> : <code>{DEFAULT_COOKIE_FILE.name}</code> 파일을 <b>app.py</b> 옆에 두면
+  자동으로 쿠키가 적용됩니다.
 </div>
 """
         )
+    return ui
 
-    return iface
-
-
-# ──────────────────────────────────────
-# Entrypoint
-# ──────────────────────────────────────
+# ───────────────────────── Entrypoint ───────────────────────
 if __name__ == "__main__":
-    demo = create_interface()
+    demo = create_ui()
     import atexit
 
-    atexit.register(downloader.cleanup)
+    atexit.register(lambda: shutil.rmtree(yt_dl.temp_dir, ignore_errors=True))
+    atexit.register(lambda: shutil.rmtree(yt_dl.temp_downloads, ignore_errors=True))
     demo.launch(debug=True, show_error=True)