Merge pull request #481 from cul/risearch-query-escape-dc-source-lookup

When performing an risearch query for a dc source path, escape colons in the path

Author: elohanlon

lib/hyacinth/utils/fedora_utils.rb

@@ -63,8 +63,12 @@ def self.get_or_create_namespace_object(namespace_string)
     namespace_fedora_object
   end
 
+  def self.escape_path_or_uri_for_risearch_query(path_or_uri)
+    path_or_uri.gsub(%q('), %q(\\\')).gsub(%q(:), %q(\\\:))
+  end
+
   def self.find_object_pid_by_filesystem_path(full_filesystem_path, active_only = true)
-    query = "select $pid from <#ri> where $pid <http://purl.org/dc/elements/1.1/source> '#{full_filesystem_path.gsub(%q('), %q(\\\'))}'"
+    query = "select $pid from <#ri> where $pid <http://purl.org/dc/elements/1.1/source> '#{escape_path_or_uri_for_risearch_query(full_filesystem_path)}'"
     query += " and $pid <info:fedora/fedora-system:def/model#state> <fedora-model:Active>" if active_only
     ri_opts = {
       type: 'tuples',

spec/lib/hyacinth/utils/fedora_utils_spec.rb

@@ -1,8 +1,7 @@
 require 'rails_helper'
 
 RSpec.describe Hyacinth::Utils::FedoraUtils do
-
-  context ".find_object_pid_by_filesystem_path" do
+  describe ".find_object_pid_by_filesystem_path" do
     it "escapes single quotes in file paths" do
       full_filesystem_path = %q(/some/path/cool-o'something-irish-filename.pdf)
       escaped_full_filesystem_path = %q(/some/path/cool-o\\'something-irish-filename.pdf)
@@ -31,4 +30,29 @@
     end
   end
 
+  describe '.escape_path_or_uri_for_risearch_query' do
+    it 'does not modify a path with safe characters' do
+      expect(
+        described_class.escape_path_or_uri_for_risearch_query("/path/to/file.tiff")
+      ).to eq(
+        "/path/to/file.tiff"
+      )
+    end
+
+    it 'escapes a single quote in a path that contains single quote' do
+      expect(
+        described_class.escape_path_or_uri_for_risearch_query("/path/to/someone's-file.tiff")
+      ).to eq(
+        "/path/to/someone\\'s-file.tiff"
+      )
+    end
+
+    it 'escapes a colon in a path that contains a colon' do
+      expect(
+        described_class.escape_path_or_uri_for_risearch_query('s3://bucket_name/path/to/file.tiff')
+      ).to eq(
+        "s3\\://bucket_name/path/to/file.tiff"
+      )
+    end
+  end
 end