diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-14 00:12:23 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2020-02-14 00:12:23 -0800 |
| commit | 016d6d28c24f616897bdb7587205cfe2cc32ec89 (patch) | |
| tree | 9245e8501418abbb8dc13abe553c103dae27bfaa | |
| parent | fd86c5ac74512a92bc5cbe25f8ec1f963d6cc1e6 (diff) | |
| download | fatcat-016d6d28c24f616897bdb7587205cfe2cc32ec89.tar.gz fatcat-016d6d28c24f616897bdb7587205cfe2cc32ec89.zip | |
remove arabesque short wayback URL hack
| -rw-r--r-- | python/fatcat_tools/importers/shadow.py | 6 |
1 files changed, 0 insertions, 6 deletions
diff --git a/python/fatcat_tools/importers/shadow.py b/python/fatcat_tools/importers/shadow.py index 1a76299e..4cd22775 100644 --- a/python/fatcat_tools/importers/shadow.py +++ b/python/fatcat_tools/importers/shadow.py @@ -155,12 +155,6 @@ class ShadowLibraryImporter(EntityImporter): if u.rel == 'social': u.rel = 'academicsocial' - # new wayback URLs, could replace bad old short wayback URLs (from arabesque bug) - new_wb_urls = [u.url for u in fe.urls] - new_short_wb_urls = ['https://web.archive.org/web/{}/{}'.format( - u.split('/')[4][:12], '/'.join(u.split('/')[5:])) for u in new_wb_urls] - existing.urls = [u for u in existing.urls if not u.url in new_short_wb_urls] - # merge the existing into this one and update merged_urls = {} for u in fe.urls + existing.urls: |