mirror of
https://github.com/bellingcat/auto-archiver.git
synced 2026-06-13 05:38:29 +03:00
wayback fix
This commit is contained in:
@@ -5,18 +5,15 @@ from bs4 import BeautifulSoup
|
|||||||
from storages import Storage
|
from storages import Storage
|
||||||
from .base_archiver import Archiver, ArchiveResult
|
from .base_archiver import Archiver, ArchiveResult
|
||||||
|
|
||||||
# @dataclass
|
|
||||||
# class WaybackConfig:
|
|
||||||
# key: str
|
|
||||||
# secret: str
|
|
||||||
from configs import WaybackConfig
|
from configs import WaybackConfig
|
||||||
|
|
||||||
# TODO: use WaybackConfig
|
# TODO: use WaybackConfig
|
||||||
class WaybackArchiver(Archiver):
|
class WaybackArchiver(Archiver):
|
||||||
name = "wayback"
|
name = "wayback"
|
||||||
|
|
||||||
def __init__(self, storage: Storage, driver):
|
def __init__(self, storage: Storage, driver, config: WaybackConfig):
|
||||||
super(WaybackArchiver, self).__init__(storage, driver)
|
super(WaybackArchiver, self).__init__(storage, driver)
|
||||||
|
self.config = config
|
||||||
self.seen_urls = {}
|
self.seen_urls = {}
|
||||||
|
|
||||||
def download(self, url, check_if_exists=False):
|
def download(self, url, check_if_exists=False):
|
||||||
@@ -25,7 +22,7 @@ class WaybackArchiver(Archiver):
|
|||||||
|
|
||||||
ia_headers = {
|
ia_headers = {
|
||||||
"Accept": "application/json",
|
"Accept": "application/json",
|
||||||
"Authorization": "LOW " + os.getenv('INTERNET_ARCHIVE_S3_KEY') + ":" + os.getenv('INTERNET_ARCHIVE_S3_SECRET')
|
"Authorization": f"LOW {self.config.key}:{self.config.secret}"
|
||||||
}
|
}
|
||||||
|
|
||||||
r = requests.post(
|
r = requests.post(
|
||||||
|
|||||||
@@ -95,7 +95,7 @@ def process_sheet(c: Config, sheet, header=1, columns=GWorksheet.COLUMN_NAMES):
|
|||||||
TiktokArchiver(storage, c.webdriver),
|
TiktokArchiver(storage, c.webdriver),
|
||||||
YoutubeDLArchiver(storage, c.webdriver, c.facebook_cookie),
|
YoutubeDLArchiver(storage, c.webdriver, c.facebook_cookie),
|
||||||
TwitterArchiver(storage, c.webdriver),
|
TwitterArchiver(storage, c.webdriver),
|
||||||
WaybackArchiver(storage, c.webdriver)
|
WaybackArchiver(storage, c.webdriver, c.wayback_config)
|
||||||
]
|
]
|
||||||
|
|
||||||
for archiver in active_archivers:
|
for archiver in active_archivers:
|
||||||
|
|||||||
Reference in New Issue
Block a user