X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/d6cc9d6e055974384be446f6da8218ff7da048ae..eb7227693e8847a65798afa7f7e8a4ffe8a199a4:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index df84b09dfb..600c07511b 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -8,6 +8,10 @@ class Collection < ArvadosModel
     t.add :files
   end
 
+  api_accessible :with_data, extend: :user do |t|
+    t.add :manifest_text
+  end
+
   def redundancy_status
     if redundancy_confirmed_as.nil?
       'unconfirmed'
@@ -28,17 +32,19 @@ class Collection < ArvadosModel
     if self.manifest_text.nil? and self.uuid.nil?
       super
     elsif self.manifest_text and self.uuid
-      if self.uuid.gsub(/\+[^,]+/,'') == Digest::MD5.hexdigest(self.manifest_text)
+      self.uuid.gsub! /\+.*/, ''
+      if self.uuid == Digest::MD5.hexdigest(self.manifest_text)
+        self.uuid.gsub! /$/, '+' + self.manifest_text.length.to_s
         true
       else
-        errors.add :uuid, 'uuid does not match checksum of manifest_text'
+        errors.add :uuid, 'does not match checksum of manifest_text'
         false
       end
     elsif self.manifest_text
-      errors.add :uuid, 'checksum for manifest_text not supplied in uuid'
+      errors.add :uuid, 'not supplied (must match checksum of manifest_text)'
       false
     else
-      errors.add :manifest_text, 'manifest_text not supplied'
+      errors.add :manifest_text, 'not supplied'
       false
     end
   end
@@ -59,10 +65,29 @@ class Collection < ArvadosModel
       @files = []
       return
     end
+
+    #normalized_manifest = ""
+    #IO.popen(['arv-normalize'], 'w+b') do |io|
+    #  io.write manifest_text
+    #  io.close_write
+    #  while buf = io.read(2**20)
+    #    normalized_manifest += buf
+    #  end
+    #end
+
     @data_size = 0
-    @files = []
+    tmp = {}
+
     manifest_text.split("\n").each do |stream|
       toks = stream.split(" ")
+
+      stream = toks[0].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
+        case $1
+        when '\\' '\\'
+        else $1.to_i(8).chr
+        end
+      end
+
       toks[1..-1].each do |tok|
         if (re = tok.match /^[0-9a-f]{32}/)
           blocksize = nil
@@ -78,10 +103,48 @@ class Collection < ArvadosModel
           @data_size += blocksize if @data_size
         else
           if (re = tok.match /^(\d+):(\d+):(\S+)$/)
-            @files << [toks[0], re[3], re[2].to_i]
+            filename = re[3].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
+              case $1
+              when '\\' '\\'
+              else $1.to_i(8).chr
+              end
+            end
+            fn = stream + '/' + filename
+            i = re[2].to_i
+            if tmp[fn]
+              tmp[fn] += i
+            else
+              tmp[fn] = i
+            end
           end
         end
       end
     end
+
+    @files = []
+    tmp.each do |k, v|
+      re = k.match(/^(.+)\/(.+)/)
+      @files << [re[1], re[2], v]
+    end
+  end
+
+  def self.uuid_like_pattern
+    "________________________________+%"
+  end
+
+  def self.normalize_uuid uuid
+    hash_part = nil
+    size_part = nil
+    uuid.split('+').each do |token|
+      if token.match /^[0-9a-f]{32,}$/
+        raise "uuid #{uuid} has multiple hash parts" if hash_part
+        hash_part = token
+      elsif token.match /^\d+$/
+        raise "uuid #{uuid} has multiple size parts" if size_part
+        size_part = token
+      end
+    end
+    raise "uuid #{uuid} has no hash part" if !hash_part
+    [hash_part, size_part].compact.join '+'
   end
 end