diff options
author | Bryan Newbold <bnewbold@robocracy.org> | 2019-10-08 16:11:09 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@robocracy.org> | 2019-10-08 16:11:09 -0700 |
commit | b9279efacdee9bf8038203d6efe9dc105cc4dce3 (patch) | |
tree | e1c40cebb0ea688182b5d8f1dcf5fbd02c3fa4c0 /python/fatcat_tools/importers/wayback_static.py | |
parent | 4b7c3c7b317cf4793f5ba5ad0d96102f103b66a3 (diff) | |
download | fatcat-b9279efacdee9bf8038203d6efe9dc105cc4dce3.tar.gz fatcat-b9279efacdee9bf8038203d6efe9dc105cc4dce3.zip |
refactor duplicated b32_hex function in importers
Diffstat (limited to 'python/fatcat_tools/importers/wayback_static.py')
-rwxr-xr-x | python/fatcat_tools/importers/wayback_static.py | 11 |
1 files changed, 1 insertions, 10 deletions
diff --git a/python/fatcat_tools/importers/wayback_static.py b/python/fatcat_tools/importers/wayback_static.py index fa0bd4ef..5b23347f 100755 --- a/python/fatcat_tools/importers/wayback_static.py +++ b/python/fatcat_tools/importers/wayback_static.py @@ -8,7 +8,6 @@ Works as a stand-alone script (for debugging) or as library routines. import sys import json -import base64 import hashlib import requests import datetime @@ -17,21 +16,13 @@ import subprocess from bs4 import BeautifulSoup from fatcat_openapi_client import * +from .common import b32_hex CDX_API_BASE = "https://web.archive.org/cdx/search/cdx" GWB_URL_BASE = "https://web.archive.org/web" REQ_SESSION = requests.Session() -def b32_hex(s): - """copy/pasta from elsewhere""" - s = s.strip().split()[0].lower() - if s.startswith("sha1:"): - s = s[5:] - if len(s) != 32: - return s - return base64.b16encode(base64.b32decode(s.upper())).lower().decode('utf-8') - def parse_wbm_url(url): """Takes a wayback machine URL, and returns a tuple: |