[ARVADOS] updated: deaf93e06ddfb5346f0492f99f441aa4734e6bf5

[git at public.curoverse.com]

Summary of changes:
 services/api/lib/salvage_collection.rb            | 37 +++++++++--------------
 services/api/test/unit/salvage_collection_test.rb | 25 +++++++--------
 2 files changed, 25 insertions(+), 37 deletions(-)

       via  deaf93e06ddfb5346f0492f99f441aa4734e6bf5 (commit)
      from  9259c169b6254ea581fcdcb18e1cdbe9b9fbea1e (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.


commit deaf93e06ddfb5346f0492f99f441aa4734e6bf5
Author: Tom Clegg <tom at curoverse.com>
Date:   Thu Aug 27 11:31:14 2015 -0400

    6859: Find locator-like strings even if not space-delimited.

diff --git a/services/api/lib/salvage_collection.rb b/services/api/lib/salvage_collection.rb
index 79a2d58..2011f81 100755
--- a/services/api/lib/salvage_collection.rb
+++ b/services/api/lib/salvage_collection.rb
@@ -23,31 +23,22 @@ module SalvageCollection
     end
   end
 
-  # Get all the locators from the original manifest
-  LOCATOR_REGEXP = /((.*))?([[:xdigit:]]{32})(\+(.*))?\z/
+  # Get all the locators (and perhaps other strings that look a lot
+  # like a locators) from the original manifest, even if they don't
+  # appear in the correct positions with the correct space delimiters.
   def salvage_collection_locator_data manifest
-      locators = []
-      size = 0
-      manifest.each_line do |line|
-        line.split(' ').each do |word|
-          if match = LOCATOR_REGEXP.match(word)
-            if match.size == 6 and match[5]
-              size_str = match[5].split('+')[0]
-              if size_str.to_i.to_s == size_str
-                word = match[3] + '+' + size_str     # get rid of any other hints
-                size += size_str.to_i
-              else
-                word = match[3]
-              end
-            else
-              word = match[3]
-            end
-            locators << word
-          end
-        end
+    locators = []
+    size = 0
+    manifest.scan /(^|[^[:xdigit:]])([[:xdigit:]]{32})((\+\d+)(\+|\b))?/ do |_, hash, _, sizehint, _|
+      if sizehint
+        locators << hash.downcase + sizehint
+        size += sizehint.to_i
+      else
+        locators << hash.downcase
       end
-      locators << 'd41d8cd98f00b204e9800998ecf8427e+0' if !locators.any?
-      return [locators, size]
+    end
+    locators << 'd41d8cd98f00b204e9800998ecf8427e+0' if !locators.any?
+    return [locators, size]
   end
 
   def salvage_collection uuid, reason='salvaged - see #6277, #6859'
diff --git a/services/api/test/unit/salvage_collection_test.rb b/services/api/test/unit/salvage_collection_test.rb
index 47abe5a..a269078 100644
--- a/services/api/test/unit/salvage_collection_test.rb
+++ b/services/api/test/unit/salvage_collection_test.rb
@@ -15,7 +15,7 @@ TEST_MANIFEST_STRIPPED = ". 341dabea2bd78ad0d6fc3f5b926b450e+85626 0:85626:brca2
 #     341dabea2bd78ad0d6fc3f5b926b450e+abc
 #     341dabea2bd78ad0d6fc3f5b926abcdf
 # Expectation: All these locators are preserved in salvaged_data
-BAD_MANIFEST = "faaaafaafaafaabd78ad0d6fc3f5b926b450e+foo bar-baabaabaabd78ad0d6fc3f5b926b450e bad12345dae58ad0d6fc3f5b926b450e+ 341dabea2bd78ad0d6fc3f5b926b450e+abc 341dabea2bd78ad0d6fc3f5b926abcdf 0:85626:brca2-hg19.fa\n. abcdabea2bd78ad0d6fc3f5b926b450e+1000 0:1000:brca-hg19.fa\n. d7321a918923627c972d8f8080c07d29+2000+A22e0a1d9b9bc85c848379d98bedc64238b0b1532 at 55e076ce 0:2000:brca1-hg19.fa\n"
+BAD_MANIFEST = "faafaafaabd78ad0d6fc3f5b926b450e+foo bar-baabaabaabd78ad0d6fc3f5b926b450e_bad12345dae58ad0d6fc3f5b926b450e+ 341dabea2bd78ad0d6fc3f5b926b450e+abc 341dabea2bd78ad0d6fc3f5b926abcdf 0:85626:brca2-hg19.fa\n. abcdabea2bd78ad0d6fc3f5b926b450e+1000 0:1000:brca-hg19.fa\n. d7321a918923627c972d8f8080c07d29+2000+A22e0a1d9b9bc85c848379d98bedc64238b0b1532 at 55e076ce 0:2000:brca1-hg19.fa\n"
 
 class SalvageCollectionTest < ActiveSupport::TestCase
   include SalvageCollection
@@ -121,19 +121,16 @@ class SalvageCollectionTest < ActiveSupport::TestCase
   # Expectation: All these locators are preserved in salvaged_data
   test "invalid locators preserved during salvaging" do
     locator_data = salvage_collection_locator_data BAD_MANIFEST
-    assert_equal 7, locator_data[0].size
-    assert_equal false, locator_data[0].include?("foo-faafaafaabd78ad0d6fc3f5b926b450e+foo")
-    assert_equal true,  locator_data[0].include?("faafaafaabd78ad0d6fc3f5b926b450e")
-    assert_equal false, locator_data[0].include?("bar-baabaabaabd78ad0d6fc3f5b926b450e")
-    assert_equal true,  locator_data[0].include?("baabaabaabd78ad0d6fc3f5b926b450e")
-    assert_equal false, locator_data[0].include?("bad12345dae58ad0d6fc3f5b926b450e+")
-    assert_equal true,  locator_data[0].include?("bad12345dae58ad0d6fc3f5b926b450e")
-    assert_equal false, locator_data[0].include?("341dabea2bd78ad0d6fc3f5b926b450e+abc")
-    assert_equal true,  locator_data[0].include?("341dabea2bd78ad0d6fc3f5b926b450e")
-    assert_equal true,  locator_data[0].include?("341dabea2bd78ad0d6fc3f5b926abcdf")
-    assert_equal true,  locator_data[0].include?("abcdabea2bd78ad0d6fc3f5b926b450e+1000")
-    assert_equal true,  locator_data[0].include?("d7321a918923627c972d8f8080c07d29+2000")
-    assert_equal true,  locator_data[1].eql?(1000 + 2000)   # size
+    assert_equal \
+    ["faafaafaabd78ad0d6fc3f5b926b450e",
+     "baabaabaabd78ad0d6fc3f5b926b450e",
+     "bad12345dae58ad0d6fc3f5b926b450e",
+     "341dabea2bd78ad0d6fc3f5b926b450e",
+     "341dabea2bd78ad0d6fc3f5b926abcdf",
+     "abcdabea2bd78ad0d6fc3f5b926b450e+1000",
+     "d7321a918923627c972d8f8080c07d29+2000",
+    ], locator_data[0]
+    assert_equal 1000+2000, locator_data[1]
   end
 
   test "salvage a collection with invalid manifest text" do

-----------------------------------------------------------------------


hooks/post-receive
--