X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/b92d4bfa6f4378756df5d28360bb708e9251917a..d14421423f2bd72b2c3eb95bdebe85a210972a12:/sdk/ruby/test/test_keep_manifest.rb

diff --git a/sdk/ruby/test/test_keep_manifest.rb b/sdk/ruby/test/test_keep_manifest.rb
index 18a2b0688f..71a24a28c0 100644
--- a/sdk/ruby/test/test_keep_manifest.rb
+++ b/sdk/ruby/test/test_keep_manifest.rb
@@ -1,3 +1,7 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 require "arvados/keep"
 require "minitest/autorun"
 require "sdk_fixtures"
@@ -266,6 +270,8 @@ class ManifestTest < Minitest::Test
    [true, 'd41d8cd98f00b204e9800998ecf8427e+0', '+0','0',nil],
    [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Fizz+Buzz','+0','0','+Fizz+Buzz'],
    [true, 'd41d8cd98f00b204e9800998ecf8427e+Fizz+Buzz', nil,nil,'+Fizz+Buzz'],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo', '+0','0','+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo'],
+   [true, 'd41d8cd98f00b204e9800998ecf8427e+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo', nil,nil,'+Ad41d8cd98f00b204e9800998ecf8427e00000000+Foo'],
    [true, 'd41d8cd98f00b204e9800998ecf8427e+0+Z', '+0','0','+Z'],
    [true, 'd41d8cd98f00b204e9800998ecf8427e+Z', nil,nil,'+Z'],
   ].each do |ok, locator, match2, match3, match4|
@@ -278,22 +284,80 @@ class ManifestTest < Minitest::Test
         assert_equal match4, match[4]
       end
     end
+    define_method "test_parse_method_on_#{locator.inspect}" do
+      loc = Keep::Locator.parse locator
+      if !ok
+        assert_nil loc
+      else
+        refute_nil loc
+        assert loc.is_a?(Keep::Locator)
+        #assert loc.hash
+        #assert loc.size
+        #assert loc.hints.is_a?(Array)
+      end
+    end
   end
 
   [
+    [false, nil, "No manifest found"],
+    [true, ""],
+    [false, " ", "Invalid manifest: does not end with newline"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"], # 2 locators
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.foo.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.foo\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:...\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:.../.foo./.../bar\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/...\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/.../bar\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/.bar/baz.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar./baz.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 000000000000000000000000000000:0777:foo.txt\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:0:0\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\040\n"],
+    [true, ". 00000000000000000000000000000000+0 0:0:0\n"],
+    [true, ". 00000000000000000000000000000000+0 0:0:d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff\n"],
+    [true, ". d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff 0:0:empty.txt\n"],
     [false, '. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt',
-      "Invalid manifest: does not end with new line"],
+      "Invalid manifest: does not end with newline"],
     [false, "abc d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
       "invalid stream name \"abc\""],
+    [false, "abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
+      "invalid stream name \"abc/./foo\""],
+    [false, "./abc/../foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
+      "invalid stream name \"./abc/../foo\""],
+    [false, "./abc/. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
+      "invalid stream name \"./abc/.\""],
+    [false, "./abc/.. d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
+      "invalid stream name \"./abc/..\""],
+    [false, "./abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
+      "invalid stream name \"./abc/./foo\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:.\n",
+      "invalid file token \"0:0:.\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:..\n",
+      "invalid file token \"0:0:..\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:./abc.txt\n",
+      "invalid file token \"0:0:./abc.txt\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:../abc.txt\n",
+      "invalid file token \"0:0:../abc.txt\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt/.\n",
+      "invalid file token \"0:0:abc.txt/.\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt/..\n",
+      "invalid file token \"0:0:abc.txt/..\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:a/./bc.txt\n",
+      "invalid file token \"0:0:a/./bc.txt\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e 0:0:a/../bc.txt\n",
+      "invalid file token \"0:0:a/../bc.txt\""],
+    [false, "./abc/./foo d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n",
+      "invalid stream name \"./abc/./foo\""],
     [false, "d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n",
       "invalid stream name \"d41d8cd98f00b204e9800998ecf8427e+0\""],
     [false, ". d41d8cd98f00b204e9800998ecf8427 0:0:abc.txt\n",
       "invalid locator \"d41d8cd98f00b204e9800998ecf8427\""],
     [false, ". d41d8cd98f00b204e9800998ecf8427e\n",
       "Manifest invalid for stream 1: no file tokens"],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e 0:0:abc.txt\n"],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n"], # 2 locators
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n/dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt\n",
       "Manifest invalid for stream 2: missing or invalid stream name \"/dir1\""],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 d41d8cd98f00b204e9800998ecf842 0:0:abc.txt\n",
@@ -304,12 +368,6 @@ class ManifestTest < Minitest::Test
       "Manifest invalid for stream 2: invalid file token \"0:abc.txt\""],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt\n./dir1 a41d8cd98f00b204e9800998ecf8427e+0 0:0:abc.txt xyz.txt\n",
       "Manifest invalid for stream 2: invalid file token \"xyz.txt\""],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/bar.txt\n"],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 000000000000000000000000000000:0777:foo.txt\n"],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:0:0\n"],
-    [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\040\n"],
-    [true, ". 00000000000000000000000000000000+0 0:0:0\n"],
-    [true, ". 00000000000000000000000000000000+0 0:0:d41d8cd98f00b204e9800998ecf8427e+0+Ad41d8cd98f00b204e9800998ecf8427e00000000@ffffffff\n"],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt d41d8cd98f00b204e9800998ecf8427e+0\n",
       "Manifest invalid for stream 1: invalid file token \"d41d8cd98f00b204e9800998ecf8427e+0\""],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\n",
@@ -318,28 +376,45 @@ class ManifestTest < Minitest::Test
       "Manifest invalid for stream 1: no file tokens"],
     [false, ". 0:0:foo.txt d41d8cd98f00b204e9800998ecf8427e+0\n",
       "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""],
-    [false, ". 0:0:foo.txt\n", "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""],
+    [false, ". 0:0:foo.txt\n",
+      "Manifest invalid for stream 1: missing or invalid locator \"0:0:foo.txt\""],
     [false, ".\n", "Manifest invalid for stream 1: missing or invalid locator"],
-    [false, ".", "Invalid manifest: does not end with new line"],
+    [false, ".", "Invalid manifest: does not end with newline"],
     [false, ". \n", "Manifest invalid for stream 1: missing or invalid locator"],
     [false, ".  \n", "Manifest invalid for stream 1: missing or invalid locator"],
-# This does not fail since \t is interpreted as one white space character and hence is split upon
-#    [false, ".\td41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", "Manifest invalid for stream 1: no file tokens"],
-    [false, " . d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n", "Manifest invalid for stream 1: missing or invalid stream name"],
-# This does not fail since the white space after foo.txt is used to split upon
-#    [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt \n", "Manifest invalid for stream 1: no file tokens"],
+    [false, " . d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
+      "Manifest invalid for stream 1: missing or invalid stream name"],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt \n",
+      "stream 1: trailing space"],
+   # TAB and other tricky whitespace characters:
+    [false, "\v. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
+      "Manifest invalid for stream 1: missing or invalid stream name \"\\v."],
+    [false, "./foo\vbar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
+      "Manifest invalid for stream 1: missing or invalid stream name \"./foo\\vbar"],
+    [false, "\t. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
+      "Manifest invalid for stream 1: missing or invalid stream name \"\\t"],
+    [false, ".\td41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
+      "Manifest invalid for stream 1: missing or invalid stream name \".\\t"],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\t\n",
+      "stream 1: invalid file token \"0:0:foo.txt\\t\""],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e+0\t 0:0:foo.txt\n",
+      "stream 1: missing or invalid locator \"d41d8cd98f00b204e9800998ecf8427e+0\\t\""],
+    [false, "./foo\tbar d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
+      "stream 1: missing or invalid stream name \"./foo\\tbar\""],
+    # other whitespace errors:
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0  0:0:foo.txt\n",
-      "Manifest invalid for stream 1: invalid file token"],
+      "Manifest invalid for stream 1: invalid file token \"\""],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n \n",
       "Manifest invalid for stream 2: missing stream name"],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n\n",
       "Manifest invalid for stream 2: missing stream name"],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n ",
-      "Invalid manifest: does not end with new line"],
+      "Invalid manifest: does not end with newline"],
     [false, "\n. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
       "Manifest invalid for stream 1: missing stream name"],
     [false, " \n. d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
       "Manifest invalid for stream 1: missing stream name"],
+    # empty file and stream name components:
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:/foo.txt\n",
       "Manifest invalid for stream 1: invalid file token \"0:0:/foo.txt\""],
     [false, "./ d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo.txt\n",
@@ -355,12 +430,13 @@ class ManifestTest < Minitest::Test
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:foo/\n",
       "Manifest invalid for stream 1: invalid file token \"0:0:foo/\""],
   ].each do |ok, manifest, expected_error=nil|
-    define_method "test_manifest_valid_#{ok}_#{manifest}_and_expect_error_#{expected_error}" do
+    define_method "test_validate manifest #{manifest.inspect}" do
+      assert_equal ok, Keep::Manifest.valid?(manifest)
       if ok
-        assert Keep::Manifest.valid? manifest
+        assert Keep::Manifest.validate! manifest
       else
         begin
-          Keep::Manifest.valid? manifest
+          Keep::Manifest.validate! manifest
         rescue ArgumentError => e
           msg = e.message
         end