minor: add bson corpus tests

etc/update-spec-tests.sh

@@ -0,0 +1,24 @@
+#!/usr/bin/env bash
+
+# This script is used to fetch the latest JSON tests for the CRUD spec. It puts the tests in the
+# direcory $reporoot/data/crud. It should be run from the root of the repository.
+
+set -o errexit
+set -o nounset
+
+if [ ! -d ".git" ]; then
+    echo "$0: This script must be run from the root of the repository" >&2
+    exit 1
+fi
+
+if [ $# -ne 1 ]; then
+    echo "$0: This script must be passed exactly one argument for which tests to sync" >&2
+    exit 1
+fi
+
+tmpdir=`perl -MFile::Temp=tempdir -wle 'print tempdir(TMPDIR => 1, CLEANUP => 0)'`
+curl -sL https://github.com/mongodb/specifications/archive/master.zip -o "$tmpdir/specs.zip"
+unzip -d "$tmpdir" "$tmpdir/specs.zip" > /dev/null
+mkdir -p "tests/spec/json/$1"
+rsync -ah "$tmpdir/specifications-master/source/$1/tests/" "tests/spec/json/$1" --delete
+rm -rf "$tmpdir"

tests/spec/json/bson-corpus/array.json

@@ -0,0 +1,49 @@
+{
+    "description": "Array",
+    "bson_type": "0x04",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty",
+            "canonical_bson": "0D000000046100050000000000",
+            "canonical_extjson": "{\"a\" : []}"
+        },
+        {
+            "description": "Single Element Array",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "canonical_extjson": "{\"a\" : [{\"$numberInt\": \"10\"}]}"
+        },
+        {
+            "description": "Single Element Array with index set incorrectly to empty string",
+            "degenerate_bson": "130000000461000B00000010000A0000000000",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "canonical_extjson": "{\"a\" : [{\"$numberInt\": \"10\"}]}"
+        },
+        {
+            "description": "Single Element Array with index set incorrectly to ab",
+            "degenerate_bson": "150000000461000D000000106162000A0000000000",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "canonical_extjson": "{\"a\" : [{\"$numberInt\": \"10\"}]}"
+        },
+        {
+            "description": "Multi Element Array with duplicate indexes",
+            "degenerate_bson": "1b000000046100130000001030000a000000103000140000000000",
+            "canonical_bson": "1b000000046100130000001030000a000000103100140000000000",
+            "canonical_extjson": "{\"a\" : [{\"$numberInt\": \"10\"}, {\"$numberInt\": \"20\"}]}"
+        }
+    ],
+    "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"
+        }
+    ]
+}

tests/spec/json/bson-corpus/binary.json

@@ -0,0 +1,85 @@
+{
+    "description": "Binary type",
+    "bson_type": "0x05",
+    "test_key": "x",
+    "valid": [
+        {
+            "description": "subtype 0x00 (Zero-length)",
+            "canonical_bson": "0D000000057800000000000000",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"\", \"subType\" : \"00\"}}}"
+        },
+        {
+            "description": "subtype 0x00 (Zero-length, keys reversed)",
+            "canonical_bson": "0D000000057800000000000000",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"\", \"subType\" : \"00\"}}}",
+            "degenerate_extjson": "{\"x\" : { \"$binary\" : {\"subType\" : \"00\", \"base64\" : \"\"}}}"
+        },
+        {
+            "description": "subtype 0x00",
+            "canonical_bson": "0F0000000578000200000000FFFF00",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"//8=\", \"subType\" : \"00\"}}}"
+        },
+        {
+            "description": "subtype 0x01",
+            "canonical_bson": "0F0000000578000200000001FFFF00",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"//8=\", \"subType\" : \"01\"}}}"
+        },
+        {
+            "description": "subtype 0x02",
+            "canonical_bson": "13000000057800060000000202000000FFFF00",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"//8=\", \"subType\" : \"02\"}}}"
+        },
+        {
+            "description": "subtype 0x03",
+            "canonical_bson": "1D000000057800100000000373FFD26444B34C6990E8E7D1DFC035D400",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"c//SZESzTGmQ6OfR38A11A==\", \"subType\" : \"03\"}}}"
+        },
+        {
+            "description": "subtype 0x04",
+            "canonical_bson": "1D000000057800100000000473FFD26444B34C6990E8E7D1DFC035D400",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"c//SZESzTGmQ6OfR38A11A==\", \"subType\" : \"04\"}}}"
+        },
+        {
+            "description": "subtype 0x05",
+            "canonical_bson": "1D000000057800100000000573FFD26444B34C6990E8E7D1DFC035D400",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"c//SZESzTGmQ6OfR38A11A==\", \"subType\" : \"05\"}}}"
+        },
+        {
+            "description": "subtype 0x80",
+            "canonical_bson": "0F0000000578000200000080FFFF00",
+            "canonical_extjson": "{\"x\" : { \"$binary\" : {\"base64\" : \"//8=\", \"subType\" : \"80\"}}}"
+        },
+        {
+            "description": "$type query operator (conflicts with legacy $binary form with $type field)",
+            "canonical_bson": "1F000000037800170000000224747970650007000000737472696E67000000",
+            "canonical_extjson": "{\"x\" : { \"$type\" : \"string\"}}"
+        },
+        {
+            "description": "$type query operator (conflicts with legacy $binary form with $type field)",
+            "canonical_bson": "180000000378001000000010247479706500020000000000",
+            "canonical_extjson": "{\"x\" : { \"$type\" : {\"$numberInt\": \"2\"}}}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Length longer than document",
+            "bson": "1D000000057800FF0000000573FFD26444B34C6990E8E7D1DFC035D400"
+        },
+        {
+            "description": "Negative length",
+            "bson": "0D000000057800FFFFFFFF0000"
+        },
+        {
+            "description": "subtype 0x02 length too long ",
+            "bson": "13000000057800060000000203000000FFFF00"
+        },
+        {
+            "description": "subtype 0x02 length too short",
+            "bson": "13000000057800060000000201000000FFFF00"
+        },
+        {
+            "description": "subtype 0x02 length negative one",
+            "bson": "130000000578000600000002FFFFFFFFFFFF00"
+        }
+    ]
+}

tests/spec/json/bson-corpus/boolean.json

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