ff_plugins/anime_downloader
1
0
Fork 0
Code Issues 1 Pull Requests Packages Projects Releases Wiki Activity

feat: Enhance yt-dlp downloader with auto-installation, adaptive HLS download strategies, CDN-specific headers, and improved ffmpeg progress tracking.

Browse Source
This commit is contained in:
projectdx
2025-12-28 23:22:36 +09:00
parent 028562ea18
commit bb4f2797c1
7 changed files with 524 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
lib/camoufox_anilife.py
View File

@@ -31,7 +31,12 @@ def extract_aldata(detail_url: str, episode_num: str) -> dict:
try:
# Camoufox 시작 (자동 fingerprint 생성)
with Camoufox(headless=False) as browser:
# Docker/서버 환경에서는 DISPLAY가 없으므로 headless 모드 사용
import os
has_display = os.environ.get('DISPLAY') is not None
use_headless = not has_display
with Camoufox(headless=use_headless) as browser:
page = browser.new_page()
try:

10
lib/ffmpeg_queue_v1.py
View File

@@ -88,7 +88,15 @@ class FfmpegQueueEntity(abc.ABCMeta("ABC", (object,), {"__slots__": ()})):
tmp["callback_id"] = getattr(self, 'name', 'anilife') if hasattr(self, 'name') else 'anilife'
tmp["start_time"] = self.created_time
tmp["status_kor"] = self.ffmpeg_status_kor
tmp["status_str"] = str(self.ffmpeg_status) if self.ffmpeg_status != -1 else "WAITING"
# status_str: 템플릿에서 문자열 비교에 사용 (DOWNLOADING, COMPLETED, WAITING)
status_map = {
0: "WAITING",
1: "STARTED",
5: "DOWNLOADING",
7: "COMPLETED",
-1: "FAILED"
}
tmp["status_str"] = status_map.get(self.ffmpeg_status, "WAITING")
tmp["percent"] = self.ffmpeg_percent
tmp["duration_str"] = ""
tmp["duration"] = ""

151
lib/ytdlp_downloader.py
View File

@@ -26,6 +26,7 @@ class YtdlpDownloader:
self.cancelled = False
self.process = None
self.error_output = [] # 에러 메시지 저장
self.total_duration_seconds = 0 # 전체 영상 길이 (초)
# 속도 및 시간 계산용
self.start_time = None
@@ -59,9 +60,53 @@ class YtdlpDownloader:
else:
return f"{bytes_per_sec / (1024 * 1024):.2f} MB/s"
def time_to_seconds(self, time_str):
"""HH:MM:SS.ms 형식을 초로 변환"""
try:
if not time_str:
return 0
parts = time_str.split(':')
if len(parts) != 3:
return 0
h = float(parts[0])
m = float(parts[1])
s = float(parts[2])
return h * 3600 + m * 60 + s
except Exception:
return 0
def _ensure_ytdlp_installed(self):
"""yt-dlp가 설치되어 있는지 확인하고, 없으면 자동 설치"""
import shutil
# yt-dlp binary가 PATH에 있는지 확인
if shutil.which('yt-dlp') is not None:
return True
logger.info("yt-dlp not found in PATH. Installing via pip...")
try:
result = subprocess.run(
[sys.executable, "-m", "pip", "install", "yt-dlp", "-q"],
capture_output=True,
text=True,
timeout=120
)
if result.returncode != 0:
logger.error(f"Failed to install yt-dlp: {result.stderr}")
return False
logger.info("yt-dlp installed successfully")
return True
except Exception as e:
logger.error(f"yt-dlp installation error: {e}")
return False
def download(self):
"""yt-dlp CLI를 통한 브라우저 흉내(Impersonate) 방식 다운로드 수행"""
try:
# yt-dlp 설치 확인
if not self._ensure_ytdlp_installed():
return False, "yt-dlp installation failed"
self.start_time = time.time()
# 출력 디렉토리 생성
@@ -76,14 +121,26 @@ class YtdlpDownloader:
concat_char = '&' if '?' in current_url else '?'
current_url = f"{current_url}{concat_char}dummy=.m3u8"
# 1. 기본 명령어 구성 (Impersonate & HLS 강제)
# 1. 기본 명령어 구성 (Impersonate & HLS 옵션)
# hlz CDN (linkkf)은 .jpg 확장자로 위장된 TS 세그먼트를 사용
# ffmpeg 8.0에서 이를 인식하지 못하므로 native HLS 다운로더 사용
use_native_hls = 'hlz' in current_url and '.top/' in current_url
cmd = [
'yt-dlp',
'--newline',
'--no-playlist',
'--no-part',
'--hls-prefer-ffmpeg',
'--hls-use-mpegts',
]
if use_native_hls:
# hlz CDN: native HLS 다운로더 사용 (ffmpeg의 확장자 제한 우회)
cmd += ['--hls-prefer-native']
else:
# 기타 CDN: ffmpeg 사용 (더 안정적)
cmd += ['--hls-prefer-ffmpeg', '--hls-use-mpegts']
cmd += [
'--no-check-certificate',
'--progress',
'--verbose', # 디버깅용 상세 로그
@@ -121,6 +178,17 @@ class YtdlpDownloader:
cmd += ['--referer', 'https://cdndania.com/']
cmd += ['--add-header', 'Origin:https://cdndania.com']
cmd += ['--add-header', 'X-Requested-With:XMLHttpRequest']
# linkkf CDN (hlz3.top, hlz2.top 등) 헤더 보강
if 'hlz' in current_url and '.top/' in current_url:
# hlz CDN은 자체 도메인을 Referer로 요구함
from urllib.parse import urlparse
parsed = urlparse(current_url)
cdn_origin = f"{parsed.scheme}://{parsed.netloc}"
if not has_referer:
cmd += ['--referer', cdn_origin + '/']
cmd += ['--add-header', f'Origin:{cdn_origin}']
cmd += ['--add-header', 'Accept:*/*']
cmd.append(current_url)
@@ -136,13 +204,23 @@ class YtdlpDownloader:
)
# 여러 진행률 형식 매칭
# [download] 10.5% of ~100.00MiB at 2.45MiB/s
# [download] 10.5% of 100.00MiB at 2.45MiB/s ETA 00:30
# [download] 100% of 100.00MiB
# yt-dlp native: [download] 10.5% of ~100.00MiB at 2.45MiB/s
# yt-dlp native: [download] 10.5% of 100.00MiB at 2.45MiB/s ETA 00:30
# yt-dlp native: [download] 100% of 100.00MiB
# ffmpeg: frame= 1234 fps= 30 size= 12345kB time=00:01:23.45 bitrate=1234.5kbits/s
# ffmpeg: size= 123456kB time=00:01:23.45
prog_patterns = [
re.compile(r'\[download\]\s+(?P<percent>[\d\.]+)%\s+of\s+.*?(?:\s+at\s+(?P<speed>[\d\.]+\s*\w+/s))?'),
re.compile(r'\[download\]\s+(?P<percent>[\d\.]+)%'),
# ffmpeg time 출력 파싱 (time=HH:MM:SS.ms)
re.compile(r'time=(?P<time>\d+:\d+:\d+\.\d+)'),
# ffmpeg size 출력 파싱
re.compile(r'size=\s*(?P<size>\d+)kB'),
]
# ffmpeg time-based progress tracking
last_time_str = ""
ffmpeg_progress_count = 0
for line in self.process.stdout:
if self.cancelled:
@@ -152,11 +230,60 @@ class YtdlpDownloader:
line = line.strip()
if not line: continue
# 디버깅: 모든 출력 로깅 (너무 많으면 주석 해제)
if '[download]' in line or 'fragment' in line.lower():
logger.debug(f"yt-dlp: {line}")
# ffmpeg Duration 파싱 (전체 길이 확인용)
if 'Duration:' in line and self.total_duration_seconds == 0:
dur_match = re.search(r'Duration:\s*(?P<duration>\d+:\d+:\d+\.\d+)', line)
if dur_match:
self.total_duration_seconds = self.time_to_seconds(dur_match.group('duration'))
logger.info(f"[ffmpeg] Total duration detected: {dur_match.group('duration')} ({self.total_duration_seconds}s)")
# ffmpeg time/size 출력 특별 처리
# ffmpeg는 [download] X% 형식을 사용하지 않으므로 time으로 진행 상황 추정
if 'time=' in line:
ffmpeg_progress_count += 1
# 매 5번째 출력마다 UI 업데이트 (너무 자주 업데이트 방지)
if ffmpeg_progress_count % 5 == 0 and self.callback:
# time= 파싱
time_match = re.search(r'time=(?P<time>\d+:\d+:\d+\.\d+)', line)
speed_match = re.search(r'bitrate=\s*([\d\.]+\w+)', line)
time_str = time_match.group('time') if time_match else ""
bitrate = speed_match.group(1) if speed_match else ""
if self.start_time:
elapsed = time.time() - self.start_time
self.elapsed_time = self.format_time(elapsed)
# 비디오 시간 위치 표시 (시:분:초)
current_seconds = self.time_to_seconds(time_str)
if time_str:
# "00:05:30.45" -> "5분 30초"
parts = time_str.split(':')
hours = int(parts[0])
mins = int(parts[1])
secs = int(float(parts[2]))
if hours > 0:
video_time = f"{hours}시간 {mins}분"
else:
video_time = f"{mins}분 {secs}초"
else:
video_time = ""
self.current_speed = bitrate if bitrate else ""
# % 계산 (전체 길이를 알면 정확하게, 모르면 카운터 기반 99% 제한)
if self.total_duration_seconds > 0:
self.percent = (current_seconds / self.total_duration_seconds) * 100
self.percent = min(100.0, self.percent)
else:
self.percent = min(99.0, ffmpeg_progress_count)
logger.info(f"[ffmpeg progress] {self.percent:.1f}% time={video_time} bitrate={bitrate}")
self.callback(percent=int(self.percent), current=int(current_seconds), total=int(self.total_duration_seconds), speed=video_time, elapsed=self.elapsed_time)
continue
for prog_re in prog_patterns:
# 일반 [download] X% 형식 처리 (yt-dlp native 다운로더용)
for prog_re in prog_patterns[:2]: # 첫 두 패턴만 사용 (download 형식)
match = prog_re.search(line)
if match:
try:
@@ -168,8 +295,10 @@ class YtdlpDownloader:
elapsed = time.time() - self.start_time
self.elapsed_time = self.format_time(elapsed)
if self.callback:
logger.info(f"[yt-dlp progress] Calling callback: {int(self.percent)}% speed={self.current_speed}")
self.callback(percent=int(self.percent), current=int(self.percent), total=100, speed=self.current_speed, elapsed=self.elapsed_time)
except: pass
except Exception as cb_err:
logger.warning(f"Callback error: {cb_err}")
break # 한 패턴이 매칭되면 중단
if 'error' in line.lower() or 'security' in line.lower() or 'unable' in line.lower():