diff --git a/lib/wayback_machine_downloader/archive_api.rb b/lib/wayback_machine_downloader/archive_api.rb index 4a33a45..70e8ff2 100644 --- a/lib/wayback_machine_downloader/archive_api.rb +++ b/lib/wayback_machine_downloader/archive_api.rb @@ -1,20 +1,20 @@ module ArchiveAPI - def get_raw_list_from_api url + def get_raw_list_from_api url, page_index request_url = "http://web.archive.org/cdx/search/xd?url=" request_url += url - request_url += parameters_for_api - request_uri = URI.parse request_url - response = Net::HTTP.get_response request_uri - response.body + request_url += parameters_for_api page_index + request_uri = URI.parse request_url + response = Net::HTTP.get_response request_uri + response.body end - def parameters_for_api - parameters = "&fl=timestamp,original&gzip=false" + def parameters_for_api page_index + parameters = "&fl=timestamp,original&collapse=digest&gzip=false" if @all - parameters += "&collapse=digest" + parameters += "" else - parameters += "&filter=statuscode:200&collapse=original" + parameters += "&filter=statuscode:200" end if @from_timestamp and @from_timestamp != 0 parameters += "&from=" + @from_timestamp.to_s @@ -22,6 +22,9 @@ module ArchiveAPI if @to_timestamp and @to_timestamp != 0 parameters += "&to=" + @to_timestamp.to_s end + if page_index + parameters += "&page=#{page_index}" + end parameters end