forks/archivebox
1
0
Fork 0
Code Releases Activity

add archivebox config command and move config into sections

Browse source
This commit is contained in:
Nick Sweeting 2019-04-25 19:03:38 -04:00
parent 583d77bc31
commit 4d6ad7a65d
4 changed files with 394 additions and 122 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

117
archivebox/cli/archivebox_config.py Normal file
View file

@ -0,0 +1,117 @@
#!/usr/bin/env python3
__package__ = 'archivebox.cli'
__command__ = 'archivebox config'
__description__ = 'Get and set your ArchiveBox project configuration values'
import sys
import argparse
from typing import Optional, List
from ..legacy.util import SmartFormatter
from ..legacy.config import (
check_data_folder,
OUTPUT_DIR,
write_config_file,
CONFIG,
ConfigDict,
stderr,
)
def main(args: List[str]=None, stdin: Optional[str]=None) -> None:
check_data_folder()
args = sys.argv[1:] if args is None else args
parser = argparse.ArgumentParser(
prog=__command__,
description=__description__,
add_help=True,
formatter_class=SmartFormatter,
)
group = parser.add_mutually_exclusive_group()
group.add_argument(
'--get', #'-g',
action='store_true',
help="Get the value for the given config KEYs",
)
group.add_argument(
'--set', #'-s',
action='store_true',
help="Set the given KEY=VALUE config values",
)
parser.add_argument(
'config_options',
nargs='*',
type=str,
help='KEY or KEY=VALUE formatted config values to get or set',
)
command = parser.parse_args(args)
if stdin or not sys.stdin.isatty():
stdin_raw_text = stdin or sys.stdin.read()
if stdin_raw_text and command.config_options:
stderr(
'[X] You should either pass config values as an arguments '
'or via stdin, but not both.\n',
color='red',
)
raise SystemExit(1)
config_options = stdin_raw_text.split('\n')
else:
config_options = command.config_options
no_args = not (command.get or command.set or command.config_options)
matching_config: ConfigDict = {}
if command.get or no_args:
if config_options:
matching_config = {key: CONFIG[key] for key in config_options if key in CONFIG}
failed_config = [key for key in config_options if key not in CONFIG]
if failed_config:
stderr()
stderr('[X] These options failed to get', color='red')
stderr(' {}'.format('\n '.join(config_options)))
raise SystemExit(1)
else:
matching_config = CONFIG
print('\n'.join(f'{key}={val}' for key, val in matching_config.items()))
raise SystemExit(not matching_config)
elif command.set:
new_config = {}
failed_options = []
for line in config_options:
if line.startswith('#') or not line.strip():
continue
if '=' not in line:
stderr('[X] Config KEY=VALUE must have an = sign in it', color='red')
stderr(f' {line}')
raise SystemExit(2)
key, val = line.split('=')
if key.upper().strip() in CONFIG:
new_config[key.upper().strip()] = val.strip()
else:
failed_options.append(line)
if new_config:
matching_config = write_config_file(new_config, out_dir=OUTPUT_DIR)
print('\n'.join(f'{key}={val}' for key, val in matching_config.items()))
if failed_options:
stderr()
stderr('[X] These options failed to set:', color='red')
stderr(' {}'.format('\n '.join(failed_options)))
raise SystemExit(bool(failed_options))
else:
stderr('[X] You must pass either --get or --set, or no arguments to get the whole config.', color='red')
stderr(' archivebox config')
stderr(' archivebox config --get SOME_KEY')
stderr(' archivebox config --set SOME_KEY=SOME_VALUE')
raise SystemExit(2)
if __name__ == '__main__':
main()

58
archivebox/legacy/ArchiveBox.conf Normal file
View file

@ -0,0 +1,58 @@
# This is the example default configiration file for ArchiveBox.
#
# Copy example config from here into your project's ArchiveBox.conf file,
# DO NOT EDIT THIS FILE DIRECTLY!
#
# See the list of all the possible options. documentation, and examples here:
# https://github.com/pirate/ArchiveBox/wiki/Configuration
[GENERAL_CONFIG]
OUTPUT_PERMISSIONS = 755
ONLY_NEW = False
TIMEOUT = 60
MEDIA_TIMEOUT = 3600
ACTIVE_THEME = default
FOOTER_INFO = Content is hosted for personal archiving purposes only. Contact server owner for any takedown requests.
URL_BLACKLIST = (://(.*\.)?facebook\.com)|(://(.*\.)?ebay\.com)|(.*\.exe$)
[ARCHIVE_METHOD_TOGGLES]
SAVE_TITLE = True
SAVE_FAVICON = True
SAVE_WGET = True
SAVE_WGET_REQUISITES = True
SAVE_WARC = True
SAVE_PDF = True
SAVE_SCREENSHOT = True
SAVE_DOM = True
SAVE_GIT = True
SAVE_MEDIA = False
SAVE_ARCHIVE_DOT_ORG = True
[ARCHIVE_METHOD_OPTIONS]
CHECK_SSL_VALIDITY = True
RESOLUTION = 1440,900
GIT_DOMAINS = github.com,bitbucket.org,gitlab.com
CROME_HEADLESS = True
CROME_SANDBOX = True
COOKIES_FILE = path/to/cookies.txt
CHROME_USER_DATA_DIR = ~/.config/google-chrome/Default
WGET_USER_AGENT = Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36
CHROME_USER_AGENT = Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36
[DEPENDENCY_CONFIG]
USE_CURL = True
USE_WGET = True
USE_CHROME = True
USE_YOUTUBEDL = True
USE_GIT = True
CURL_BINARY = curl
GIT_BINARY = git"
WGET_BINARY = wget
YOUTUBEDL_BINARY = youtube-dl
CHROME_BINARY = chromium

233
archivebox/legacy/config.py
View file

@ -9,8 +9,9 @@ import getpass
import shutil import shutil
from hashlib import md5 from hashlib import md5
from typing import Optional, Type, Tuple from typing import Optional, Type, Tuple, Dict
from subprocess import run, PIPE, DEVNULL from subprocess import run, PIPE, DEVNULL
from configparser import ConfigParser
from .config_stubs import ( from .config_stubs import (
SimpleConfigValueDict, SimpleConfigValueDict,
@ -29,63 +30,66 @@ from .config_stubs import (
################################# User Config ################################## ################################# User Config ##################################
SHELL_CONFIG_DEFAULTS: ConfigDefaultDict = { CONFIG_DEFAULTS: Dict[str, ConfigDefaultDict] = {
'IS_TTY': {'type': bool, 'default': lambda _: sys.stdout.isatty()}, 'SHELL_CONFIG': {
'USE_COLOR': {'type': bool, 'default': lambda c: c['IS_TTY']}, 'IS_TTY': {'type': bool, 'default': lambda _: sys.stdout.isatty()},
'SHOW_PROGRESS': {'type': bool, 'default': lambda c: c['IS_TTY']}, 'USE_COLOR': {'type': bool, 'default': lambda c: c['IS_TTY']},
} 'SHOW_PROGRESS': {'type': bool, 'default': lambda c: c['IS_TTY']},
},
ARCHIVE_CONFIG_DEFAULTS: ConfigDefaultDict = { 'GENERAL_CONFIG': {
'OUTPUT_DIR': {'type': str, 'default': None}, 'OUTPUT_DIR': {'type': str, 'default': None},
'ONLY_NEW': {'type': bool, 'default': False}, 'CONFIG_FILE': {'type': str, 'default': None},
'TIMEOUT': {'type': int, 'default': 60}, 'ONLY_NEW': {'type': bool, 'default': False},
'MEDIA_TIMEOUT': {'type': int, 'default': 3600}, 'TIMEOUT': {'type': int, 'default': 60},
'OUTPUT_PERMISSIONS': {'type': str, 'default': '755'}, 'MEDIA_TIMEOUT': {'type': int, 'default': 3600},
'FOOTER_INFO': {'type': str, 'default': 'Content is hosted for personal archiving purposes only. Contact server owner for any takedown requests.'}, 'OUTPUT_PERMISSIONS': {'type': str, 'default': '755'},
'URL_BLACKLIST': {'type': str, 'default': None}, 'FOOTER_INFO': {'type': str, 'default': 'Content is hosted for personal archiving purposes only. Contact server owner for any takedown requests.'},
} 'URL_BLACKLIST': {'type': str, 'default': None},
},
ARCHIVE_METHOD_TOGGLES_DEFAULTS: ConfigDefaultDict = { 'ARCHIVE_METHOD_TOGGLES': {
'SAVE_TITLE': {'type': bool, 'default': True, 'aliases': ('FETCH_TITLE',)}, 'SAVE_TITLE': {'type': bool, 'default': True, 'aliases': ('FETCH_TITLE',)},
'SAVE_FAVICON': {'type': bool, 'default': True, 'aliases': ('FETCH_FAVICON',)}, 'SAVE_FAVICON': {'type': bool, 'default': True, 'aliases': ('FETCH_FAVICON',)},
'SAVE_WGET': {'type': bool, 'default': True, 'aliases': ('FETCH_WGET',)}, 'SAVE_WGET': {'type': bool, 'default': True, 'aliases': ('FETCH_WGET',)},
'SAVE_WGET_REQUISITES': {'type': bool, 'default': True, 'aliases': ('FETCH_WGET_REQUISITES',)}, 'SAVE_WGET_REQUISITES': {'type': bool, 'default': True, 'aliases': ('FETCH_WGET_REQUISITES',)},
'SAVE_PDF': {'type': bool, 'default': True, 'aliases': ('FETCH_PDF',)}, 'SAVE_PDF': {'type': bool, 'default': True, 'aliases': ('FETCH_PDF',)},
'SAVE_SCREENSHOT': {'type': bool, 'default': True, 'aliases': ('FETCH_SCREENSHOT',)}, 'SAVE_SCREENSHOT': {'type': bool, 'default': True, 'aliases': ('FETCH_SCREENSHOT',)},
'SAVE_DOM': {'type': bool, 'default': True, 'aliases': ('FETCH_DOM',)}, 'SAVE_DOM': {'type': bool, 'default': True, 'aliases': ('FETCH_DOM',)},
'SAVE_WARC': {'type': bool, 'default': True, 'aliases': ('FETCH_WARC',)}, 'SAVE_WARC': {'type': bool, 'default': True, 'aliases': ('FETCH_WARC',)},
'SAVE_GIT': {'type': bool, 'default': True, 'aliases': ('FETCH_GIT',)}, 'SAVE_GIT': {'type': bool, 'default': True, 'aliases': ('FETCH_GIT',)},
'SAVE_MEDIA': {'type': bool, 'default': True, 'aliases': ('FETCH_MEDIA',)}, 'SAVE_MEDIA': {'type': bool, 'default': True, 'aliases': ('FETCH_MEDIA',)},
'SAVE_ARCHIVE_DOT_ORG': {'type': bool, 'default': True, 'aliases': ('SUBMIT_ARCHIVE_DOT_ORG',)}, 'SAVE_ARCHIVE_DOT_ORG': {'type': bool, 'default': True, 'aliases': ('SUBMIT_ARCHIVE_DOT_ORG',)},
} },
ARCHIVE_METHOD_OPTIONS_DEFAULTS: ConfigDefaultDict = { 'ARCHIVE_METHOD_OPTIONS': {
'RESOLUTION': {'type': str, 'default': '1440,2000', 'aliases': ('SCREENSHOT_RESOLUTION',)}, 'RESOLUTION': {'type': str, 'default': '1440,2000', 'aliases': ('SCREENSHOT_RESOLUTION',)},
'GIT_DOMAINS': {'type': str, 'default': 'github.com,bitbucket.org,gitlab.com'}, 'GIT_DOMAINS': {'type': str, 'default': 'github.com,bitbucket.org,gitlab.com'},
'CHECK_SSL_VALIDITY': {'type': bool, 'default': True}, 'CHECK_SSL_VALIDITY': {'type': bool, 'default': True},
'WGET_USER_AGENT': {'type': str, 'default': 'ArchiveBox/{VERSION} (+https://github.com/pirate/ArchiveBox/) wget/{WGET_VERSION}'}, 'WGET_USER_AGENT': {'type': str, 'default': 'ArchiveBox/{VERSION} (+https://github.com/pirate/ArchiveBox/) wget/{WGET_VERSION}'},
'CHROME_USER_AGENT': {'type': str, 'default': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36'}, 'CHROME_USER_AGENT': {'type': str, 'default': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36'},
'COOKIES_FILE': {'type': str, 'default': None}, 'COOKIES_FILE': {'type': str, 'default': None},
'CHROME_USER_DATA_DIR': {'type': str, 'default': None}, 'CHROME_USER_DATA_DIR': {'type': str, 'default': None},
'CHROME_HEADLESS': {'type': bool, 'default': True}, 'CHROME_HEADLESS': {'type': bool, 'default': True},
'CHROME_SANDBOX': {'type': bool, 'default': True}, 'CHROME_SANDBOX': {'type': bool, 'default': True},
} },
DEPENDENCY_CONFIG_DEFAULTS: ConfigDefaultDict = { 'DEPENDENCY_CONFIG': {
'USE_CURL': {'type': bool, 'default': True}, 'USE_CURL': {'type': bool, 'default': True},
'USE_WGET': {'type': bool, 'default': True}, 'USE_WGET': {'type': bool, 'default': True},
'USE_GIT': {'type': bool, 'default': True}, 'USE_GIT': {'type': bool, 'default': True},
'USE_CHROME': {'type': bool, 'default': True}, 'USE_CHROME': {'type': bool, 'default': True},
'USE_YOUTUBEDL': {'type': bool, 'default': True}, 'USE_YOUTUBEDL': {'type': bool, 'default': True},
'CURL_BINARY': {'type': str, 'default': 'curl'}, 'CURL_BINARY': {'type': str, 'default': 'curl'},
'GIT_BINARY': {'type': str, 'default': 'git'}, 'GIT_BINARY': {'type': str, 'default': 'git'},
'WGET_BINARY': {'type': str, 'default': 'wget'}, 'WGET_BINARY': {'type': str, 'default': 'wget'},
'YOUTUBEDL_BINARY': {'type': str, 'default': 'youtube-dl'}, 'YOUTUBEDL_BINARY': {'type': str, 'default': 'youtube-dl'},
'CHROME_BINARY': {'type': str, 'default': None}, 'CHROME_BINARY': {'type': str, 'default': None},
},
} }
############################## Derived Config ############################## ############################## Derived Config ##############################
@ -120,7 +124,21 @@ JSON_INDEX_FILENAME = 'index.json'
HTML_INDEX_FILENAME = 'index.html' HTML_INDEX_FILENAME = 'index.html'
ROBOTS_TXT_FILENAME = 'robots.txt' ROBOTS_TXT_FILENAME = 'robots.txt'
FAVICON_FILENAME = 'favicon.ico' FAVICON_FILENAME = 'favicon.ico'
CONFIG_FILENAME = 'ArchiveBox.conf'
CONFIG_HEADER = """
# This is the default config file for new ArchiveBox projects.
# Add your archive collection config here in INI format.
#
# After updating your config, make sure to update your archive by running:
# archivebox init
#
# The example default configuration file can be found at:
# ArchiveBox/etc/Archivebox.conf.default
#
# See the list of all the possible options. documentation, and examples here:
# https://github.com/pirate/ArchiveBox/wiki/Configuration
"""
DERIVED_CONFIG_DEFAULTS: ConfigDefaultDict = { DERIVED_CONFIG_DEFAULTS: ConfigDefaultDict = {
@ -137,6 +155,7 @@ DERIVED_CONFIG_DEFAULTS: ConfigDefaultDict = {
'ARCHIVE_DIR': {'default': lambda c: os.path.join(c['OUTPUT_DIR'], ARCHIVE_DIR_NAME)}, 'ARCHIVE_DIR': {'default': lambda c: os.path.join(c['OUTPUT_DIR'], ARCHIVE_DIR_NAME)},
'SOURCES_DIR': {'default': lambda c: os.path.join(c['OUTPUT_DIR'], SOURCES_DIR_NAME)}, 'SOURCES_DIR': {'default': lambda c: os.path.join(c['OUTPUT_DIR'], SOURCES_DIR_NAME)},
'LOGS_DIR': {'default': lambda c: os.path.join(c['OUTPUT_DIR'], LOGS_DIR_NAME)}, 'LOGS_DIR': {'default': lambda c: os.path.join(c['OUTPUT_DIR'], LOGS_DIR_NAME)},
'CONFIG_FILE': {'default': lambda c: os.path.abspath(os.path.expanduser(c['CONFIG_FILE'])) if c['CONFIG_FILE'] else os.path.join(c['OUTPUT_DIR'], CONFIG_FILENAME)},
'COOKIES_FILE': {'default': lambda c: c['COOKIES_FILE'] and os.path.abspath(os.path.expanduser(c['COOKIES_FILE']))}, 'COOKIES_FILE': {'default': lambda c: c['COOKIES_FILE'] and os.path.abspath(os.path.expanduser(c['COOKIES_FILE']))},
'CHROME_USER_DATA_DIR': {'default': lambda c: find_chrome_data_dir() if c['CHROME_USER_DATA_DIR'] is None else (os.path.abspath(os.path.expanduser(c['CHROME_USER_DATA_DIR'])) or None)}, 'CHROME_USER_DATA_DIR': {'default': lambda c: find_chrome_data_dir() if c['CHROME_USER_DATA_DIR'] is None else (os.path.abspath(os.path.expanduser(c['CHROME_USER_DATA_DIR'])) or None)},
'URL_BLACKLIST_PTN': {'default': lambda c: c['URL_BLACKLIST'] and re.compile(c['URL_BLACKLIST'], re.IGNORECASE)}, 'URL_BLACKLIST_PTN': {'default': lambda c: c['URL_BLACKLIST'] and re.compile(c['URL_BLACKLIST'], re.IGNORECASE)},
@ -194,14 +213,20 @@ def load_config_val(key: str,
default: ConfigDefaultValue=None, default: ConfigDefaultValue=None,
type: Optional[Type]=None, type: Optional[Type]=None,
aliases: Optional[Tuple[str, ...]]=None, aliases: Optional[Tuple[str, ...]]=None,
config: Optional[ConfigDict]=None) -> ConfigValue: config: Optional[ConfigDict]=None,
env_vars: Optional[os._Environ]=None,
config_file_vars: Optional[Dict[str, str]]=None) -> ConfigValue:
# check the canonical option name first, then check any older aliases config_keys_to_check = (key, *(aliases or ()))
possible_env_keys = (key, *(aliases or ())) for key in config_keys_to_check:
for key in possible_env_keys: if env_vars:
val = os.getenv(key, None) val = env_vars.get(key)
if val: if val:
break break
if config_file_vars:
val = config_file_vars.get(key)
if val:
break
if type is None or val is None: if type is None or val is None:
if callable(default): if callable(default):
@ -230,7 +255,84 @@ def load_config_val(key: str,
raise Exception('Config values can only be str, bool, or int') raise Exception('Config values can only be str, bool, or int')
def load_config(defaults: ConfigDefaultDict, config: Optional[ConfigDict]=None) -> ConfigDict: def load_config_file(out_dir: str=None) -> Optional[Dict[str, str]]:
"""load the ini-formatted config file from OUTPUT_DIR/Archivebox.conf"""
out_dir = out_dir or os.path.abspath(os.getenv('OUTPUT_DIR', '.'))
config_path = os.path.join(out_dir, CONFIG_FILENAME)
if os.path.exists(config_path):
config_file = ConfigParser()
config_file.optionxform = str
config_file.read(config_path)
# flatten into one namespace
config_file_vars = {
key.upper(): val
for section, options in config_file.items()
for key, val in options.items()
}
# print('[i] Loaded config file', os.path.abspath(config_path))
# print(config_file_vars)
return config_file_vars
return None
def write_config_file(config: Dict[str, str], out_dir: str=None) -> Optional[Dict[str, str]]:
"""load the ini-formatted config file from OUTPUT_DIR/Archivebox.conf"""
out_dir = out_dir or os.path.abspath(os.getenv('OUTPUT_DIR', '.'))
config_path = os.path.join(out_dir, CONFIG_FILENAME)
if not os.path.exists(config_path):
with open(config_path, 'w+') as f:
f.write(CONFIG_HEADER)
config_file = ConfigParser()
config_file.optionxform = str
config_file.read(config_path)
find_section = lambda key: [name for name, opts in CONFIG_DEFAULTS.items() if key in opts][0]
with open(f'{config_path}.old', 'w+') as old:
with open(config_path, 'r') as new:
old.write(new.read())
with open(config_path, 'w+') as f:
for key, val in config.items():
section = find_section(key)
if section in config_file:
existing_config = dict(config_file[section])
else:
existing_config = {}
config_file[section] = {**existing_config, key: val}
config_file.write(f)
try:
CONFIG = load_all_config()
return {
key.upper(): CONFIG.get(key.upper())
for key in config.keys()
}
except:
with open(f'{config_path}.old', 'r') as old:
with open(config_path, 'w+') as new:
new.write(old.read())
if os.path.exists(f'{config_path}.old'):
os.remove(f'{config_path}.old')
return {}
def load_config(defaults: ConfigDefaultDict,
config: Optional[ConfigDict]=None,
out_dir: Optional[str]=None,
env_vars: Optional[os._Environ]=None,
config_file_vars: Optional[Dict[str, str]]=None) -> ConfigDict:
env_vars = env_vars or os.environ
config_file_vars = config_file_vars or load_config_file(out_dir=out_dir)
extended_config: ConfigDict = config.copy() if config else {} extended_config: ConfigDict = config.copy() if config else {}
for key, default in defaults.items(): for key, default in defaults.items():
try: try:
@ -240,6 +342,8 @@ def load_config(defaults: ConfigDefaultDict, config: Optional[ConfigDict]=None)
type=default.get('type'), type=default.get('type'),
aliases=default.get('aliases'), aliases=default.get('aliases'),
config=extended_config, config=extended_config,
env_vars=env_vars,
config_file_vars=config_file_vars,
) )
except KeyboardInterrupt: except KeyboardInterrupt:
raise SystemExit(0) raise SystemExit(0)
@ -253,10 +357,16 @@ def load_config(defaults: ConfigDefaultDict, config: Optional[ConfigDict]=None)
stderr(' For config documentation and examples see:') stderr(' For config documentation and examples see:')
stderr(' https://github.com/pirate/ArchiveBox/wiki/Configuration') stderr(' https://github.com/pirate/ArchiveBox/wiki/Configuration')
stderr() stderr()
raise SystemExit(1) raise SystemExit(2)
return extended_config return extended_config
# def write_config(config: ConfigDict):
# with open(os.path.join(config['OUTPUT_DIR'], CONFIG_FILENAME), 'w+') as f:
def stderr(*args, color: Optional[str]=None, config: Optional[ConfigDict]=None) -> None: def stderr(*args, color: Optional[str]=None, config: Optional[ConfigDict]=None) -> None:
ansi = DEFAULT_CLI_COLORS if (config or {}).get('USE_COLOR') else ANSI ansi = DEFAULT_CLI_COLORS if (config or {}).get('USE_COLOR') else ANSI
@ -391,6 +501,11 @@ def get_code_locations(config: ConfigDict) -> SimpleConfigValueDict:
def get_config_locations(config: ConfigDict) -> ConfigValue: def get_config_locations(config: ConfigDict) -> ConfigValue:
abspath = lambda path: None if path is None else os.path.abspath(path) abspath = lambda path: None if path is None else os.path.abspath(path)
return { return {
'CONFIG_FILE': {
'path': abspath(config['CHROME_USER_DATA_DIR']),
'enabled': config['USE_CHROME'] and config['CHROME_USER_DATA_DIR'],
'is_valid': False if config['CHROME_USER_DATA_DIR'] is None else os.path.exists(os.path.join(config['CHROME_USER_DATA_DIR'], 'Default')),
},
'CHROME_USER_DATA_DIR': { 'CHROME_USER_DATA_DIR': {
'path': abspath(config['CHROME_USER_DATA_DIR']), 'path': abspath(config['CHROME_USER_DATA_DIR']),
'enabled': config['USE_CHROME'] and config['CHROME_USER_DATA_DIR'], 'enabled': config['USE_CHROME'] and config['CHROME_USER_DATA_DIR'],

108
etc/ArchiveBox.conf.default
View file

@ -1,74 +1,56 @@
# Example config file for ArchiveBox: The self-hosted internet archive. # This is the default config file for new ArchiveBox projects.
# Copy this file to ~/.ArchiveBox.conf before editing it. # Edit values below using INI syntax, then update your archive by running:
# Config file is in both Python and .env syntax (all strings must be quoted). # archivebox init
# For documentation, see: # For more options, example setups, and documentation, see:
# https://github.com/pirate/ArchiveBox/wiki/Configuration # https://github.com/pirate/ArchiveBox/wiki/Configuration
################################################################################ [GENERAL_CONFIG]
## General Settings # OUTPUT_PERMISSIONS = 755
################################################################################ # ONLY_NEW = False
# TIMEOUT = 60
# MEDIA_TIMEOUT = 3600
# ACTIVE_THEME = default
# FOOTER_INFO = Content is hosted for personal archiving purposes only. Contact server owner for any takedown requests.
# URL_BLACKLIST = (://(.*\.)?facebook\.com)|(://(.*\.)?ebay\.com)|(.*\.exe$)
#OUTPUT_DIR="output" [ARCHIVE_METHOD_TOGGLES]
#OUTPUT_PERMISSIONS=755 # SAVE_TITLE = True
#ONLY_NEW=False # SAVE_FAVICON = True
#TIMEOUT=60 # SAVE_WGET = True
#MEDIA_TIMEOUT=3600 # SAVE_WGET_REQUISITES = True
#TEMPLATES_DIR="archivebox/templates" # SAVE_WARC = True
#FOOTER_INFO="Content is hosted for personal archiving purposes only. Contact server owner for any takedown requests." # SAVE_PDF = True
#URL_BLACKLIST="(://(.*\.)?youtube\.com)|(://(.*\.)?amazon\.com)|(.*\.exe$)" # SAVE_SCREENSHOT = True
# SAVE_DOM = True
################################################################################ # SAVE_GIT = True
## Archive Method Toggles # SAVE_MEDIA = False
################################################################################ # SAVE_ARCHIVE_DOT_ORG = True
#SAVE_TITLE=True
#SAVE_FAVICON=True
#SAVE_WGET=True
#SAVE_WGET_REQUISITES=True
#SAVE_WARC=True
#SAVE_PDF=True
#SAVE_SCREENSHOT=True
#SAVE_DOM=True
#SAVE_GIT=True
#SAVE_MEDIA=False
#SAVE_ARCHIVE_DOT_ORG=True
################################################################################ [ARCHIVE_METHOD_OPTIONS]
## Archive Method Options # CHECK_SSL_VALIDITY = True
################################################################################ # RESOLUTION = 1440,900
# GIT_DOMAINS = github.com,bitbucket.org,gitlab.com
#CHECK_SSL_VALIDITY=True # CROME_HEADLESS = True
#RESOLUTION="1440,900" # CROME_SANDBOX = True
#GIT_DOMAINS="github.com,bitbucket.org,gitlab.com"
#CROME_HEADLESS=True # COOKIES_FILE = path/to/cookies.txt
#CROME_SANDBOX=True # CHROME_USER_DATA_DIR = ~/.config/google-chrome/Default
#COOKIES_FILE="path/to/cookies.txt" # WGET_USER_AGENT = Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36
#CHROME_USER_DATA_DIR="~/.config/google-chrome/Default" # CHROME_USER_AGENT = Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36
#WGET_USER_AGENT="Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36"
#CHROME_USER_AGENT="Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.75 Safari/537.36"
################################################################################
## Shell Options
################################################################################
#USE_COLOR=True
#SHOW_PROGRESS=True
################################################################################ [DEPENDENCY_CONFIG]
## Dependency Options # USE_CURL = True
################################################################################ # USE_WGET = True
# USE_CHROME = True
# USE_YOUTUBEDL = True
# USE_GIT = True
#USE_CURL=True # CURL_BINARY = curl
#USE_WGET=True # GIT_BINARY = git"
#USE_CHROME=True # WGET_BINARY = wget
# YOUTUBEDL_BINARY = youtube-dl
#CURL_BINARY="curl" # CHROME_BINARY = chromium
#GIT_BINARY="git"
#WGET_BINARY="wget"
#YOUTUBEDL_BINARY="youtube-dl"
#CHROME_BINARY="chromium-browser"