X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/7024cc159936593350aaf7939d700102f6510787..549f4a1deddb41f6abbc493a660d9fb0976da91a:/services/api/app/models/collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index f6c8b26598..64a6bb0530 100644
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -1,5 +1,5 @@
 class Collection < ArvadosModel
-  include AssignUuid
+  include HasUuid
   include KindAndEtag
   include CommonApiTemplate
 
@@ -65,8 +65,19 @@ 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(" ")
 
@@ -98,11 +109,27 @@ class Collection < ArvadosModel
               else $1.to_i(8).chr
               end
             end
-            @files << [stream, filename, re[2].to_i]
+            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
@@ -120,4 +147,49 @@ class Collection < ArvadosModel
     raise "uuid #{uuid} has no hash part" if !hash_part
     [hash_part, size_part].compact.join '+'
   end
+
+  def self.for_latest_docker_image(search_term, search_tag=nil, readers=nil)
+    readers ||= [Thread.current[:user]]
+    base_search = Link.
+      readable_by(*readers).
+      readable_by(*readers, table_name: "collections").
+      joins("JOIN collections ON links.head_uuid = collections.uuid").
+      order("links.created_at DESC")
+
+    # If the search term is a Collection locator with an associated
+    # Docker image hash link, return that Collection.
+    coll_matches = base_search.
+      where(link_class: "docker_image_hash", collections: {uuid: search_term})
+    if match = coll_matches.first
+      return find_by_uuid(match.head_uuid)
+    end
+
+    # Find Collections with matching Docker image repository+tag pairs.
+    matches = base_search.
+      where(link_class: "docker_image_repo+tag",
+            name: "#{search_term}:#{search_tag || 'latest'}")
+
+    # If that didn't work, find Collections with matching Docker image hashes.
+    if matches.empty?
+      matches = base_search.
+        where("link_class = ? and name LIKE ?",
+              "docker_image_hash", "#{search_term}%")
+    end
+
+    # Select the image that was created most recently.  Note that the
+    # SQL search order and fallback timestamp values are chosen so
+    # that if image timestamps are missing, we use the image with the
+    # newest link.
+    latest_image_link = nil
+    latest_image_timestamp = "1900-01-01T00:00:00Z"
+    matches.find_each do |link|
+      link_timestamp = link.properties.fetch("image_timestamp",
+                                             "1900-01-01T00:00:01Z")
+      if link_timestamp > latest_image_timestamp
+        latest_image_link = link
+        latest_image_timestamp = link_timestamp
+      end
+    end
+    latest_image_link.nil? ? nil : find_by_uuid(latest_image_link.head_uuid)
+  end
 end