RUBY-1152 Implement BSON corpus tests

spec/bson/corpus_spec.rb

@@ -0,0 +1,50 @@
+require 'spec_helper'
+
+describe 'Driver BSON Corpus spec tests' do
+
+  specs = BSON_CORPUS_TESTS.map { |file| BSON::Corpus::Spec.new(file) }
+
+  specs.each do |spec|
+
+    context(spec.description) do
+
+      spec.valid_tests.each do |test|
+
+        context(test.description) do
+
+          it 'roundtrips the bson correctly' do
+            expect(test.reencoded_bson).to eq(test.correct_bson)
+          end
+
+          context 'when the bson provided results from an incorrect encoder', if: test.test_canonical_bson? do
+
+            it 'encodes the canonical bson correctly' do
+              expect(test.reencoded_canonical_bson).to eq(test.correct_bson)
+            end
+          end
+
+          context 'when the document can be represented as extended json', if: test.test_extjson? do
+
+            it 'decodes from bson, then encodes the document as extended json correctly' do
+              expect(test.extjson_from_bson).to eq(test.correct_extjson)
+              expect(test.extjson_from_bson[test.test_key]).to eq(test.correct_extjson[test.test_key])
+            end
+
+            it 'decodes from extended json, then encodes the document as extended json correctly' do
+              expect(test.extjson_from_encoded_extjson).to eq(test.correct_extjson)
+              expect(test.extjson_from_encoded_extjson[test.test_key]).to eq(test.correct_extjson[test.test_key])
+            end
+
+            context 'when the canonical bson can be represented as extended json', if: (test.test_canonical_bson? && test.test_extjson?) do
+
+              it 'encodes the canonical bson correctly as extended json' do
+                expect(test.extjson_from_canonical_bson).to eq(test.correct_extjson)
+                expect(test.extjson_from_canonical_bson[test.test_key]).to eq(test.correct_extjson[test.test_key])
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

spec/spec_helper.rb

@@ -14,6 +14,7 @@
 
 CURRENT_PATH = File.expand_path(File.dirname(__FILE__))
 DRIVER_COMMON_BSON_TESTS = Dir.glob("#{CURRENT_PATH}/support/driver-spec-tests/**/*.json")
+BSON_CORPUS_TESTS = Dir.glob("#{CURRENT_PATH}/support/corpus-tests/*.json")
 
 $LOAD_PATH.unshift(File.dirname(__FILE__))
 $LOAD_PATH.unshift(File.join(File.dirname(__FILE__), "..", "lib"))

spec/support/corpus-tests/array.json

@@ -0,0 +1,43 @@
+{
+    "description": "Array",
+    "bson_type": "0x04",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty",
+            "bson": "0D000000046100050000000000",
+            "extjson": "{\"a\" : []}"
+        },
+        {
+            "description": "Single Element Array",
+            "bson": "140000000461000C0000001030000A0000000000",
+            "extjson": "{\"a\" : [10]}"
+        },
+        {
+            "description": "Single Element Array with index set incorrectly",
+            "bson": "130000000461000B00000010000A0000000000",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "extjson": "{\"a\" : [10]}"
+        },
+        {
+            "description": "Single Element Array with index set incorrectly",
+            "bson": "150000000461000D000000106162000A0000000000",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "extjson": "{\"a\" : [10]}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Array length too long: eats outer terminator",
+            "bson": "140000000461000D0000001030000A0000000000"
+        },
+        {
+            "description": "Array length too short: leaks terminator",
+            "bson": "140000000461000B0000001030000A0000000000"
+        },
+        {
+            "description": "Invalid Array: bad string length in field",
+            "bson": "1A00000004666F6F00100000000230000500000062617A000000"
+        }
+    ]
+}

spec/support/corpus-tests/boolean.json

@@ -0,0 +1,27 @@
+{
+    "description": "Boolean",
+    "bson_type": "0x08",
+    "test_key": "b",
+    "valid": [
+        {
+            "description": "True",
+            "bson": "090000000862000100",
+            "extjson": "{\"b\" : true}"
+        },
+        {
+            "description": "False",
+            "bson": "090000000862000000",
+            "extjson": "{\"b\" : false}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Invalid boolean value of 2",
+            "bson": "090000000862000200"
+        },
+        {
+            "description": "Invalid boolean value of -1",
+            "bson": "09000000086200FF00"
+        }
+    ]
+}

spec/support/corpus-tests/code.json

@@ -0,0 +1,67 @@
+{
+    "description": "Javascript Code",
+    "bson_type": "0x0D",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty string",
+            "bson": "0D0000000D6100010000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\"}}"
+        },
+        {
+            "description": "Single character",
+            "bson": "0E0000000D610002000000620000",
+            "extjson": "{\"a\" : {\"$code\" : \"b\"}}"
+        },
+        {
+            "description": "Multi-character",
+            "bson": "190000000D61000D0000006162616261626162616261620000",
+            "extjson": "{\"a\" : {\"$code\" : \"abababababab\"}}"
+        },
+        {
+            "description": "two-byte UTF-8 (\u00e9)",
+            "bson": "190000000261000D000000C3A9C3A9C3A9C3A9C3A9C3A90000",
+            "extjson": "{\"a\" : \"\\u00e9\\u00e9\\u00e9\\u00e9\\u00e9\\u00e9\"}"
+        },
+        {
+            "description": "three-byte UTF-8 (\u2606)",
+            "bson": "190000000261000D000000E29886E29886E29886E298860000",
+            "extjson": "{\"a\" : \"\\u2606\\u2606\\u2606\\u2606\"}"
+        },
+        {
+            "description": "Embedded nulls",
+            "bson": "190000000261000D0000006162006261620062616261620000",
+            "extjson": "{\"a\" : \"ab\\u0000bab\\u0000babab\"}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "bad code string length: 0 (but no 0x00 either)",
+            "bson": "0C0000000261000000000000"
+        },
+        {
+            "description": "bad code string length: -1",
+            "bson": "0C000000026100FFFFFFFF00"
+        },
+        {
+            "description": "bad code string length: eats terminator",
+            "bson": "10000000026100050000006200620000"
+        },
+        {
+            "description": "bad code string length: longer than rest of document",
+            "bson": "120000000200FFFFFF00666F6F6261720000"
+        },
+        {
+            "description": "code string is not null-terminated",
+            "bson": "1000000002610004000000616263FF00"
+        },
+        {
+            "description": "empty code string, but extra null",
+            "bson": "0E00000002610001000000000000"
+        },
+        {
+            "description": "invalid UTF-8",
+            "bson": "0E00000002610002000000E90000"
+        }
+    ]
+}

spec/support/corpus-tests/code_w_scope.json

@@ -0,0 +1,78 @@
+{
+    "description": "Javascript Code with Scope",
+    "bson_type": "0x0F",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty code string, empty scope",
+            "bson": "160000000F61000E0000000100000000050000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\", \"$scope\" : {}}}"
+        },
+        {
+            "description": "Non-empty code string, empty scope",
+            "bson": "1A0000000F610012000000050000006162636400050000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"abcd\", \"$scope\" : {}}}"
+        },
+        {
+            "description": "Empty code string, non-empty scope",
+            "bson": "1D0000000F61001500000001000000000C000000107800010000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\", \"$scope\" : {\"x\" : 1}}}"
+        },
+        {
+            "description": "Non-empty code string and non-empty scope",
+            "bson": "210000000F6100190000000500000061626364000C000000107800010000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"abcd\", \"$scope\" : {\"x\" : 1}}}"
+        },
+        {
+            "description": "Unicode and embedded null in code string, empty scope",
+            "bson": "1A0000000F61001200000005000000C3A9006400050000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\\u00e9\\u0000d\", \"$scope\" : {}}}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "field length zero",
+            "bson": "280000000F6100000000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length negative",
+            "bson": "280000000F6100FFFFFFFF0500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length too short (less than minimum size)",
+            "bson": "160000000F61000D0000000100000000050000000000"
+        },
+        {
+            "description": "field length too short (truncates scope)",
+            "bson": "280000000F61001F0000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length too long (clips outer doc)",
+            "bson": "280000000F6100210000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length too long (longer than outer doc)",
+            "bson": "280000000F6100FF0000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: length too short",
+            "bson": "280000000F6100200000000400000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: length too long (clips scope)",
+            "bson": "280000000F6100200000000600000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: negative length",
+            "bson": "280000000F610020000000FFFFFFFF61626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: length longer than field",
+            "bson": "280000000F610020000000FF00000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad scope doc (field has bad string length)",
+            "bson": "1C0000000F001500000001000000000C000000020000000000000000"
+        }
+    ]
+}

spec/support/corpus-tests/document.json

@@ -0,0 +1,36 @@
+{
+    "description": "Document type (sub-documents)",
+    "bson_type": "0x03",
+    "test_key": "x",
+    "valid": [
+        {
+            "description": "Empty subdoc",
+            "bson": "0D000000037800050000000000",
+            "extjson": "{\"x\" : {}}"
+        },
+        {
+            "description": "Empty-string key subdoc",
+            "bson": "150000000378000D00000002000200000062000000",
+            "extjson": "{\"x\" : {\"\" : \"b\"}}"
+        },
+        {
+            "description": "Single-character key subdoc",
+            "bson": "160000000378000E0000000261000200000062000000",
+            "extjson": "{\"x\" : {\"a\" : \"b\"}}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Subdocument length too long: eats outer terminator",
+            "bson": "1800000003666F6F000F0000001062617200FFFFFF7F0000"
+        },
+        {
+            "description": "Subdocument length too short: leaks terminator",
+            "bson": "1500000003666F6F000A0000000862617200010000"
+        },
+        {
+            "description": "Invalid subdocument: bad string length in field",
+            "bson": "1C00000003666F6F001200000002626172000500000062617A000000"
+        }
+    ]
+}

spec/support/corpus-tests/double.json

@@ -0,0 +1,69 @@
+{
+    "description": "Double type",
+    "bson_type": "0x01",
+    "test_key": "d",
+    "valid": [
+        {
+            "description": "+1.0",
+            "bson": "10000000016400000000000000F03F00",
+            "extjson": "{\"d\" : 1.0}"
+        },
+        {
+            "description": "-1.0",
+            "bson": "10000000016400000000000000F0BF00",
+            "extjson": "{\"d\" : -1.0}"
+        },
+        {
+            "description": "+1.0001220703125",
+            "bson": "10000000016400000000008000F03F00",
+            "extjson": "{\"d\" : 1.0001220703125}"
+        },
+        {
+            "description": "-1.0001220703125",
+            "bson": "10000000016400000000008000F0BF00",
+            "extjson": "{\"d\" : -1.0001220703125}"
+        },
+        {
+            "description": "+2.0001220703125e10",
+            "bson": "1000000001640000807ca1a9a0124200",
+            "extjson": "{\"d\" : 2.0001220703125e10}"
+        },
+        {
+            "description": "-2.0001220703125e10",
+            "bson": "1000000001640000807ca1a9a012c200",
+            "extjson": "{\"d\" : -2.0001220703125e10}"
+        },
+        {
+            "description": "0.0",
+            "bson": "10000000016400000000000000000000",
+            "extjson": "{\"d\" : 0.0}"
+        },
+        {
+            "description": "-0.0",
+            "bson": "10000000016400000000000000008000",
+            "extjson": "{\"d\" : -0.0}"
+        },
+        {
+            "description": "NaN",
+            "bson": "10000000016400000000000000F87F00"
+        },
+        {
+            "description": "NaN with payload",
+            "bson": "10000000016400120000000000F87F00"
+        },
+        {
+            "description": "Inf",
+            "bson": "10000000016400000000000000F07F00"
+        },
+        {
+            "description": "-Inf",
+            "bson": "10000000016400000000000000F0FF00"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "double truncated",
+            "bson": "0B0000000164000000F03F00"
+        }
+    ]
+}