From fa366c14a9c7af6346fef1aa6d14fec6ef6a7e6d Mon Sep 17 00:00:00 2001 From: hartator Date: Sun, 9 Aug 2015 22:33:32 -0500 Subject: [PATCH] Make file_list_curated aware of timestamp option --- lib/wayback_machine_downloader.rb | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/lib/wayback_machine_downloader.rb b/lib/wayback_machine_downloader.rb index 9812ec0..aea82ff 100644 --- a/lib/wayback_machine_downloader.rb +++ b/lib/wayback_machine_downloader.rb @@ -7,7 +7,7 @@ class WaybackMachineDownloader def initialize params @base_url = params[:base_url] - @timestamp = params[:timestamp] + @timestamp = params[:timestamp].to_i end def backup_name @@ -23,16 +23,18 @@ class WaybackMachineDownloader file_list_curated = Hash.new file_list_raw.each_line do |line| line = line.split(' ') - timestamp = line[1].to_i + file_timestamp = line[1].to_i file_url = line[2] file_id = file_url.split('/')[3..-1].join('/') file_id = URI.unescape file_id - if file_list_curated[file_id] - unless file_list_curated[file_id][:timestamp] > timestamp - file_list_curated[file_id] = {file_url: file_url, timestamp: timestamp} + if @timestamp == 0 or file_timestamp <= @timestamp + if file_list_curated[file_id] + unless file_list_curated[file_id][:timestamp] > file_timestamp + file_list_curated[file_id] = {file_url: file_url, timestamp: file_timestamp} + end + else + file_list_curated[file_id] = {file_url: file_url, timestamp: file_timestamp} end - else - file_list_curated[file_id] = {file_url: file_url, timestamp: timestamp} end end file_list_curated @@ -43,7 +45,6 @@ class WaybackMachineDownloader puts file_list_curated = get_file_list_curated file_list_curated.each do |file_id, file_remote_info| - timestamp = file_remote_info[:timestamp] file_url = file_remote_info[:file_url] file_path_elements = file_id.split('/') if file_id == "" @@ -74,7 +75,7 @@ class WaybackMachineDownloader end end puts - puts "Download complete, saved in #{backup_path}. (#{file_list_curated.size} files downloaded.)" + puts "Download complete, saved in #{backup_path}. (#{file_list_curated.size} files downloaded)" end end