128 lines
5.4 KiB
YAML
128 lines
5.4 KiB
YAML
# Policy: Queue-based Authentication Simulation via Direct Docker Exec
|
||
#
|
||
# This policy simulates a continuous stream of info.json fetch requests using
|
||
# the 'direct_docker_cli' mode. It pulls URLs from a Redis queue, creates a
|
||
# temporary batch file, and then calls a yt-dlp command inside a running
|
||
# Docker container.
|
||
#
|
||
name: 12_queue_auth_simulation
|
||
|
||
settings:
|
||
mode: fetch_only
|
||
orchestration_mode: direct_docker_cli
|
||
profile_mode: from_pool_with_lock
|
||
# The save directory MUST be inside the docker_host_mount_path.
|
||
save_info_json_dir: "run/docker_mount/fetched_info_jsons/queue_simulation"
|
||
|
||
execution_control:
|
||
workers: 1
|
||
# How long a worker should pause if it cannot find an available profile to lock.
|
||
worker_polling_interval_seconds: 1
|
||
# No sleep between tasks; throughput is controlled by yt-dlp performance and profile availability.
|
||
|
||
info_json_generation_policy:
|
||
profile_prefix: "user1"
|
||
|
||
direct_docker_cli_policy:
|
||
# Which simulation environment's profiles to use for locking.
|
||
use_profile_env: "auth"
|
||
|
||
# If true, a worker will try to lock a different profile than the one it just used.
|
||
avoid_immediate_profile_reuse: true
|
||
# How long the worker should wait for a different profile before re-using the same one.
|
||
avoid_reuse_max_wait_seconds: 5
|
||
|
||
# NOTE on Rate Limits: With the default yt-dlp settings, the rate limit for guest
|
||
# sessions is ~300 videos/hour (~1000 webpage/player requests per hour).
|
||
# For accounts, it is ~2000 videos/hour (~4000 webpage/player requests per hour).
|
||
# The enforcer policy (e.g., 8_unified_simulation_enforcer.yaml) should be
|
||
# configured to respect these limits via rotation and rest periods.
|
||
|
||
# If true, extract the visitor_id from yt-dlp logs, save it per-profile,
|
||
# and inject it into subsequent requests for that profile.
|
||
#track_visitor_id: true
|
||
|
||
# --- Docker Execution Settings ---
|
||
docker_image_name: "ytops/yt-dlp" # Image to use for `docker run`
|
||
docker_network_name: "airflow_proxynet"
|
||
# IMPORTANT: This path on the HOST will be mounted into the container at `docker_container_mount_path`.
|
||
docker_host_mount_path: "run/docker_mount"
|
||
docker_container_mount_path: "/config" # The mount point inside the container
|
||
|
||
# Host path for persisting cache data (e.g., cookies, sigfuncs) between runs.
|
||
docker_host_cache_path: ".cache/queue_auth_simulation"
|
||
# Path inside the container where the cache is mounted. Should match HOME/.cache
|
||
docker_container_cache_path: "/config/.cache"
|
||
|
||
# If true, create and use a persistent cookie jar per profile inside the cache dir.
|
||
# use_cookies: true
|
||
|
||
# --- User-Agent Generation ---
|
||
# Template for generating User-Agent strings for new profiles.
|
||
# The '{major_version}' will be replaced by a version string.
|
||
user_agent_template: "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/{major_version}.0.0.0 Safari/537.36"
|
||
# Range of Chrome major versions to use for the template.
|
||
# See CHROME_MAJOR_VERSION_RANGE in yt-dlp's random_user_agent():
|
||
# https://github.com/yt-dlp/yt-dlp/blob/master/yt_dlp/utils/networking.py
|
||
user_agent_version_range: [137, 143]
|
||
|
||
# A base config file can be used, with overrides applied from the policy.
|
||
# The orchestrator will inject 'proxy', 'batch-file', and 'output' keys into the overrides.
|
||
ytdlp_config_file: "cli.auth.config"
|
||
ytdlp_config_overrides:
|
||
skip-download: true
|
||
write-info-json: true
|
||
no-write-subs: true
|
||
no-color: true
|
||
ignore-errors: true
|
||
use-extractors: ["youtube"]
|
||
|
||
ytdlp_raw_args:
|
||
- '--extractor-args "youtube:formats=duplicate;jsc_trace=true;player_client=tv_simply;pot_trace=true;skip=translated_subs,hls"'
|
||
- '--extractor-args "youtubepot-bgutilhttp:base_url=http://172.17.0.1:4416"'
|
||
- '--sleep-requests 0.75'
|
||
# --retry-sleep linear=1::2'
|
||
|
||
# --- Live Error Parsing Rules ---
|
||
# These regex patterns are checked against yt-dlp's stderr in real-time.
|
||
# If a fatal error is detected, immediately ban the profile to stop the container
|
||
# and prevent further errors in the same batch.
|
||
ban_on_fatal_error_in_batch: true
|
||
fatal_error_patterns:
|
||
- "Sign in to confirm you’re not a bot"
|
||
- "rate-limited by YouTube"
|
||
- "This content isn't available, try again later"
|
||
- "HTTP Error 502"
|
||
|
||
tolerated_error_patterns:
|
||
- "HTTP Error 429"
|
||
- "The uploader has not made this video available in your country"
|
||
- "This video has been removed by the uploader"
|
||
- "Private video"
|
||
- "This is a private video"
|
||
- "Video is private"
|
||
- "Video unavailable"
|
||
- "account associated with this video has been terminated"
|
||
- "members-only content"
|
||
- "Sign in to confirm your age"
|
||
|
||
# Template for renaming the final info.json.
|
||
rename_file_template: "{video_id}-{profile_name}-{proxy}.info.json"
|
||
|
||
queue_policy:
|
||
# Set to false to use legacy, unprefixed queue names (e.g., 'queue2_auth_inbox').
|
||
# Set to true (or omit) to use environment-prefixed names (e.g., 'sim_auth_queue2_auth_inbox').
|
||
use_env_prefix: false
|
||
|
||
# If specified, create download tasks for these formats
|
||
# Can be "all", a specific format ID, or a list of format IDs
|
||
formats_to_download: "140-dashy/140-dashy-0/140,299-dashy/298-dashy/137-dashy/136-dashy/135-dashy/134-dashy/133-dashy"
|
||
|
||
# How many tasks a worker should pull from the queue at once.
|
||
# This will become the batch size for the docker run.
|
||
batch_size: 25
|
||
|
||
simulation_parameters:
|
||
auth_env: "sim_auth"
|
||
download_env: "sim_download"
|