1 module SalvageCollection
2 # Take two input parameters: a collection uuid and reason
3 # Get "src_collection" with the given uuid
4 # Create a new collection with:
5 # src_collection.manifest_text as "invalid_manifest_text.txt"
6 # Locators from src_collection.manifest_text as "salvaged_data"
7 # Update src_collection:
8 # Set src_collection.manifest_text to: ""
9 # Append to src_collection.name: " (reason; salvaged data at new_collection.uuid)"
10 # Set portable_data_hash to "d41d8cd98f00b204e9800998ecf8427e+0"
12 require File.dirname(__FILE__) + '/../config/environment'
13 include ApplicationHelper
17 def self.salvage_collection_arv_put cmd
18 new_manifest = %x(#{cmd})
22 raise "Error during arv-put."
26 LOCATOR_REGEXP = /^([[:xdigit:]]{32})(\+(.*))?\z/
27 def self.salvage_collection_locator_data manifest
28 # Get all the locators from the original manifest
31 manifest.each_line do |line|
32 line.split(' ').each do |word|
33 if match = LOCATOR_REGEXP.match(word)
34 if match.size > 3 and match[3]
35 size_str = match[3].split('+')[0]
36 if size_str.to_i.to_s == size_str
37 word = match[1] + '+' + size_str # get rid of any hints
49 locators << 'd41d8cd98f00b204e9800998ecf8427e+0' if !locators.any?
50 return [locators, size]
53 def self.salvage_collection uuid, reason='salvaged - see #6277, #6859'
55 if !ENV['ARVADOS_API_TOKEN'].present? or !ENV['ARVADOS_API_HOST'].present?
56 raise "ARVADOS environment variables missing. Please set your admin user credentials as ARVADOS environment variables."
60 raise "Collection UUID is required."
63 src_collection = Collection.find_by_uuid uuid
65 raise "No collection found for #{uuid}."
68 src_manifest = src_collection.manifest_text || ''
70 # create new collection using 'arv-put' with original manifest_text as the data
71 temp_file = Tempfile.new('temp')
72 temp_file.write(src_manifest)
75 new_manifest = salvage_collection_arv_put "arv-put --as-stream --use-filename invalid_manifest_text.txt #{Shellwords::shellescape(temp_file.path)}"
79 # Get the locator data in the format [[locators], size] from the original manifest
80 locator_data = salvage_collection_locator_data src_manifest
82 new_manifest += (". #{locator_data[0].join(' ')} 0:#{locator_data[1]}:salvaged_data\n")
84 new_collection = Collection.new
85 new_collection.name = "salvaged from #{src_collection.uuid}, #{src_collection.portable_data_hash}"
86 new_collection.manifest_text = new_manifest
87 new_collection.portable_data_hash = Digest::MD5.hexdigest(new_collection.manifest_text)
89 created = new_collection.save!
90 raise "New collection creation failed." if !created
92 $stderr.puts "Salvaged manifest and data for #{uuid} are in #{new_collection.uuid}."
93 puts "Created new collection #{new_collection.uuid}"
95 # update src_collection collection name, pdh, and manifest_text
96 src_collection.name = (src_collection.name || '') + ' (' + (reason || '') + '; salvaged data at ' + new_collection.uuid + ')'
97 src_collection.manifest_text = ''
98 src_collection.portable_data_hash = 'd41d8cd98f00b204e9800998ecf8427e+0'
100 $stderr.puts "Collection #{uuid} emptied and renamed to #{src_collection.name.inspect}."