bin/sup-recover-sources

bin/sup-recover-sources (2673B) - raw
      1 #!/usr/bin/env ruby
      2 
      3 $:.unshift File.join(File.dirname(__FILE__), *%w[.. lib])
      4 
      5 require 'optparse'
      6 
      7 $opts = {
      8   :unusual => false,
      9   :archive => false,
     10   :scan_num => 10,
     11 }
     12 
     13 
     14 OPTIONPARSERSUCKS = "\n" + " " * 38
     15 OptionParser.new do |opts|
     16   opts.banner = <<EOS
     17 Usage: sup-recover-sources [options] <source>+
     18 
     19 Rebuilds a lost sources.yaml file by reading messages from a list of
     20 sources and determining, for each source, the most prevalent
     21 'source_id' field of messages from that source in the index.
     22 
     23 The only non-deterministic component to this is that if the same
     24 message appears in multiple sources, those sources may be
     25 mis-diagnosed by this program.
     26 
     27 If the first N messages (--scan-num below) all have the same source_id
     28 in the index, the source will be added to sources.yaml. Otherwise, the
     29 distribution will be printed, and you will have to add it by hand.
     30 
     31 The offset pointer into the sources will be set to the end of the source,
     32 so you will have to run sup-import --rebuild for each new source after
     33 doing this.
     34 
     35 Options include:
     36 EOS
     37 
     38   opts.on("--unusual", "Mark sources as 'unusual'. Only usual#{OPTIONPARSERSUCKS}sources will be polled by hand. Default:#{OPTIONPARSERSUCKS}#{$opts[:unusual]}.") { $opts[:unusual] = true }
     39 
     40   opts.on("--archive", "Mark sources as 'archive'. New messages#{OPTIONPARSERSUCKS}from these sources will not appear in#{OPTIONPARSERSUCKS}the inbox. Default: #{$opts[:archive]}.") { $opts[:archive] = true }
     41 
     42   opts.on("--scan-num N", Integer, "Number of messages to scan per source.#{OPTIONPARSERSUCKS}Default: #{$opts[:scan_num]}.") do |n|
     43     $opts[:scan_num] = n
     44   end
     45 
     46   opts.on_tail("-h", "--help", "Show this message") do
     47     puts opts
     48     exit
     49   end
     50 end.parse(ARGV)
     51 
     52 require "sup"
     53 Redwood::start
     54 puts "loading index..."
     55 index = Redwood::Index.init
     56 index.load
     57 puts "loaded index of #{index.size} messages"
     58 
     59 ARGV.each do |fn|
     60   next if Redwood::SourceManager.source_for fn
     61 
     62   ## TODO: merge this code with the same snippet in import
     63   source = Redwood::MBox.new(fn, nil, !$opts[:unusual], $opts[:archive])
     64 
     65   source_ids = Hash.new 0
     66   count = 0
     67   source.each do |offset, labels|
     68     m = Redwood::Message.new :source => source, :source_info => offset
     69     m.load_from_source!
     70     source_id = Redwood::SourceManager.source_for_id m.id
     71     next unless source_id
     72     source_ids[source_id] += 1
     73     count += 1
     74     break if count == $opts[:scan_num]
     75   end
     76 
     77   if source_ids.size == 1
     78     id = source_ids.keys.first.to_i
     79     puts "assigned #{source} to #{source_ids.keys.first}"
     80     source.id = id
     81     Redwood::SourceManager.add_source source
     82   else
     83     puts ">> unable to determine #{source}: #{source_ids.inspect}"
     84   end
     85 end
     86 
     87 index.save
sup.git
