X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/636d833ed4206ffe44e5e9b9b9100ac9d82bc603..f71634c065eff38970b178958349ca9381aee996:/services/api/script/salvage_collection.rb

diff --git a/services/api/script/salvage_collection.rb b/services/api/script/salvage_collection.rb
index bac033e214..c5f798bd21 100755
--- a/services/api/script/salvage_collection.rb
+++ b/services/api/script/salvage_collection.rb
@@ -1,4 +1,7 @@
 #!/usr/bin/env ruby
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: AGPL-3.0
 
 # Take two input parameters: a collection uuid and reason
 # Get "src_collection" with the given uuid
@@ -10,104 +13,18 @@
 #   Append to src_collection.name: " (reason; salvaged data at new_collection.uuid)"
 #   Set portable_data_hash to "d41d8cd98f00b204e9800998ecf8427e+0"
 
-require 'trollop'
+require 'optimist'
+require './lib/salvage_collection'
+include SalvageCollection
 
-opts = Trollop::options do
+opts = Optimist::options do
   banner ''
   banner "Usage: salvage_collection.rb " +
     "{uuid} {reason}"
   banner ''
-  opt :uuid, <<-eos
-uuid of the collection to be salvaged.
-  eos
-  opt :reason, <<-eos
-Reason for salvaging.
-  eos
-end
-
-if ARGV.count < 1
-  Trollop::die "required uuid argument is missing"
-end
-
-uuid, reason = ARGV
-
-require File.dirname(__FILE__) + '/../config/environment'
-require 'arvados/keep'
-include ApplicationHelper
-require 'tempfile'
-
-def salvage_collection uuid, reason
-  act_as_system_user do
-    src_collection = Collection.find_by_uuid uuid
-    if !src_collection
-      $stderr.puts "No collection found for #{uuid}. Returning."
-      return
-    end
-
-    begin
-      src_manifest = src_collection.manifest_text || ''
-
-      # Get all the locators from the original manifest
-      locators = []
-      src_manifest.each_line do |line|
-        line.split(' ').each do |word|
-          if match = Keep::Locator::LOCATOR_REGEXP.match(word)
-            word = word.split('+')[0..1].join('+')  # get rid of any hints
-            locators << word
-          end
-        end
-      end
-      locators << 'd41d8cd98f00b204e9800998ecf8427e+0' if !locators.any?
-
-      # create new collection using 'arv-put' with original manifest_text as the data
-      temp_file = Tempfile.new('temp')
-      temp_file.write(src_manifest)
-      temp_file.close
-
-      created = %x(arv-put --use-filename invalid_manifest_text.txt #{temp_file.path})
-
-      temp_file.unlink
-
-      created.rstrip!
-      match = created.match HasUuid::UUID_REGEX
-      raise "uuid not found" if !match
-
-      # update this new collection manifest to reference all locators from the original manifest
-      new_collection = Collection.find_by_uuid created
-
-      new_manifest = new_collection['manifest_text']
-      new_manifest = new_manifest.gsub(/\+A[^+]*/, '')
-      total_size = 0
-      locators.each do |locator|
-        total_size += locator.split('+')[1].to_i
-      end
-      new_manifest += (". #{locators.join(' ')} 0:#{total_size}:salvaged_data\n")
-
-      new_collection.name = "salvaged from #{src_collection.uuid}, #{src_collection.portable_data_hash}"
-      new_collection.manifest_text = new_manifest
-      new_collection.portable_data_hash = Digest::MD5.hexdigest(new_manifest)
-
-      new_collection.save!
-
-      $stderr.puts "Salvaged manifest and data for #{uuid} are in #{new_collection.uuid}."
-      puts "Created new collection #{created}"
-    rescue => error
-      $stderr.puts "Error creating collection for #{uuid}: #{error}"
-      return
-    end
-
-    begin
-      # update src_collection collection name, pdh, and manifest_text
-      src_collection.name = (src_collection.name || '') + ' (' + (reason || '') + '; salvaged data at ' + created + ')'
-      src_collection.manifest_text = ''
-      src_collection.portable_data_hash = 'd41d8cd98f00b204e9800998ecf8427e+0'
-      src_collection.save!
-      $stderr.puts "Collection #{uuid} emptied and renamed to #{src_collection.name.inspect}."
-    rescue => error
-      $stderr.puts "Error salvaging collection #{uuid}: #{error}"
-    end
-  end
+  opt :uuid, "uuid of the collection to be salvaged.", type: :string, required: true
+  opt :reason, "Reason for salvaging.", type: :string, required: false
 end
 
 # Salvage the collection with the given uuid
-salvage_collection uuid, reason
+SalvageCollection.salvage_collection opts.uuid, opts.reason