make the ignored urls smaller

This commit is contained in:
TheTechRobo 2022-05-23 21:43:17 -04:00
parent dd3f82e623
commit 79164e2c46
2 changed files with 7 additions and 1 deletions

View File

@ -40,6 +40,12 @@ end
wget.callbacks.httploop_result = function(url, err, http_stat)
io.stderr:write(http_stat["statcode"] .. url["url"] .. "\n")
if string.match(url.url, "/fetch-comments/") then
return
end
if string.match(url.url, "/r$") then
return
end
if http_stat.statcode == 200 then
table.insert(ign, url.url)
end

View File

@ -164,7 +164,7 @@ class WgetArgs(object):
'--tries', '10',
'--span-hosts',
'--waitretry', '0',
'-w', '1',
'-w', '0.1',
'--random-wait',
'--warc-file', ItemInterpolation('%(item_dir)s/%(warc_file_base)s'),
'--warc-header', 'operator: TheTechRobo <thetechrobo@protonmail.ch>',