2016-09-17 13:37:13 -05:00
|
|
|
module ArchiveAPI
|
|
|
|
|
|
2017-03-15 17:08:40 -04:00
|
|
|
def get_raw_list_from_api url, page_index
|
2020-05-27 20:28:06 -04:00
|
|
|
request_url = "https://web.archive.org/cdx/search/xd?url="
|
2017-03-15 17:08:40 -04:00
|
|
|
request_url += url
|
|
|
|
|
request_url += parameters_for_api page_index
|
2016-10-31 17:46:35 +01:00
|
|
|
|
2020-05-27 20:28:06 -04:00
|
|
|
URI.open(request_url).read
|
2017-03-15 17:08:40 -04:00
|
|
|
end
|
2016-09-17 13:37:13 -05:00
|
|
|
|
2017-03-15 17:08:40 -04:00
|
|
|
def parameters_for_api page_index
|
|
|
|
|
parameters = "&fl=timestamp,original&collapse=digest&gzip=false"
|
2016-09-18 12:23:33 -05:00
|
|
|
if @all
|
2016-09-24 10:04:57 -07:00
|
|
|
parameters += ""
|
2016-09-18 12:23:33 -05:00
|
|
|
else
|
2016-09-24 10:04:57 -07:00
|
|
|
parameters += "&filter=statuscode:200"
|
2016-09-17 13:37:13 -05:00
|
|
|
end
|
|
|
|
|
if @from_timestamp and @from_timestamp != 0
|
|
|
|
|
parameters += "&from=" + @from_timestamp.to_s
|
|
|
|
|
end
|
|
|
|
|
if @to_timestamp and @to_timestamp != 0
|
|
|
|
|
parameters += "&to=" + @to_timestamp.to_s
|
|
|
|
|
end
|
2016-09-24 10:04:57 -07:00
|
|
|
if page_index
|
|
|
|
|
parameters += "&page=#{page_index}"
|
|
|
|
|
end
|
2016-09-17 13:37:13 -05:00
|
|
|
parameters
|
|
|
|
|
end
|
|
|
|
|
|
2016-10-31 17:46:35 +01:00
|
|
|
end
|