Make JSON parsing work as a chunked conversion sink.

tests/lib/convert/json_utf8_chunk_test.dart

+// Copyright (c) 2014, the Dart project authors.  Please see the AUTHORS file
+// for details. All rights reserved. Use of this source code is governed by a
+// BSD-style license that can be found in the LICENSE file.
+
+library test;
+
+import "package:expect/expect.dart";
+import "dart:convert";
+
+bool badFormat(e) => e is FormatException;
+
+main() {
+  testNumbers();
+  testStrings();
+  testKeywords();
+  testAll();
+  testMalformed();
+}
+
+// Create an UTF-8 sink from a chunked JSON decoder, then let [action]
+// put data into it, and check that what comes out is equal to [expect].
+void jsonTest(name, expect, action(sink), [bool allowMalformed = false]) {

[Søren Gjesse, 2014/10/24 11:12:24]

name -> testName or nameOfTest

[Lasse Reichstein Nielsen, 2014/10/27 12:42:34]

Done.

+  jsonParse(name, (value) {
+    Expect.equals(expect, value, "$name$value");

[Søren Gjesse, 2014/10/24 11:12:25]

Separator before $value.

[Lasse Reichstein Nielsen, 2014/10/27 12:42:34]

Done.

+  }, action, allowMalformed);
+}
+
+void jsonParse(name, check, action, [bool allowMalformed = false]) {
+  var sink = new ChunkedConversionSink.withCallback((values) {
+    var value = values[0];
+    check(value);
+  });
+  var decoderSink = JSON.decoder.startChunkedConversion(sink)
+                                .asUtf8Sink(allowMalformed);
+  try {
+    action(decoderSink);
+  } on FormatException catch (e, s) {
+    print("Source: ${e.source} @ ${e.offset}");
+    Expect.fail("Unexpected throw($name): $e\n$s");
+  }
+}
+
+void testStrings() {
+  // String literal containing characters, all escape types,
+  // and a number of UTF-8 encoded characters.
+  var s = r'"abc\f\ndef\r\t\b\"\/\\\u0001\u9999\uffff'
+           '\x7f\xc2\x80\xdf\xbf\xe0\xa0\x80\xef\xbf\xbf'
+           '\xf0\x90\x80\x80\xf4\x8f\xbf\xbf"';  // UTF-8.
+  var expected = "abc\f\ndef\r\t\b\"\/\\\u0001\u9999\uffff"
+                 "\x7f\x80\u07ff\u0800\uffff"
+                 "\u{10000}\u{10ffff}";
+  for (var i = 1; i < s.length - 1; i++) {
+    var s1 = s.substring(0, i);
+    var s2 = s.substring(i);
+    jsonTest("$s1|$s2-$i", expected, (sink) {
+      sink.add(s1.codeUnits);
+      sink.add(s2.codeUnits);
+      sink.close();
+    });
+    jsonTest("$s1|$s2-$i-slice", expected, (sink) {
+      sink.addSlice(s.codeUnits, 0, i, false);
+      sink.addSlice(s.codeUnits, i, s.length, true);
+    });
+    for (var j = i; j < s.length - 1; j++) {
+      var s2a = s.substring(i, j);
+      var s2b = s.substring(j);
+      jsonTest("$s1|$s2a|$s2b-$i-$j", expected, (sink) {
+        sink.add(s1.codeUnits);
+        sink.add(s2a.codeUnits);
+        sink.add(s2b.codeUnits);
+        sink.close();
+      });
+    }
+  }
+}
+
+void testNumbers() {
+  for (var number in ["-0.12e-12", "-34.12E+12", "0.0e0", "9.9E9", "0", "9"
+                      "1234.56789123456701418035663664340972900390625",
+                      "1.2345678912345671e-14",
+                      "99999999999999999999"]) {
+    var expected = num.parse(number);
+    for (int i = 1; i < number.length - 1; i++) {
+      var p1 = number.substring(0, i);
+      var p2 = number.substring(i);
+      jsonTest("$p1|$p2", expected, (sink) {
+        sink.add(p1.codeUnits);
+        sink.add(p2.codeUnits);
+        sink.close();
+      });
+
+      jsonTest("$p1|$p2/slice", expected, (sink) {
+        sink.addSlice(number.codeUnits, 0, i, false);
+        sink.addSlice(number.codeUnits, i, number.length, true);
+      });
+
+      for (int j = i; j < number.length - 1; j++) {
+        var p2a = number.substring(i, j);
+        var p2b = number.substring(j);
+        jsonTest("$p1|$p2a|$p2b", expected, (sink) {
+          sink.add(p1.codeUnits);
+          sink.add(p2a.codeUnits);
+          sink.add(p2b.codeUnits);
+          sink.close();
+        });
+      }
+    }
+  }
+}
+
+// Test that `null`, `true`, and `false` keywords are decoded correctly.
+void testKeywords() {
+  for (var expected in [null, true, false]) {
+    var s = "$expected";
+    for (int i = 1; i < s.length - 1; i++) {
+      var s1 = s.substring(0, i);
+      var s2 = s.substring(i);
+      jsonTest("$s1|$s2", expected, (sink) {
+        sink.add(s1.codeUnits);
+        sink.add(s2.codeUnits);
+        sink.close();
+      });
+      jsonTest("$s1|$s2", expected, (sink) {
+        sink.addSlice(s.codeUnits, 0, i, false);
+        sink.addSlice(s.codeUnits, i, s.length, true);
+      });
+      for (var j = i; j < s.length - 1; j++) {
+        var s2a = s.substring(i, j);
+        var s2b = s.substring(j);
+        jsonTest("$s1|$s2a|$s2b", expected, (sink) {
+          sink.add(s1.codeUnits);
+          sink.add(s2a.codeUnits);
+          sink.add(s2b.codeUnits);
+          sink.close();
+        });
+      }
+    }
+  }
+}
+
+void testAll() {
+  var s = r'{"":[true,false,42, -33e-3,null,"\u0080"], "z": 0}';
+  bool check(o) {
+    if (o is Map) {
+      Expect.equals(2, o.length);
+      Expect.equals(0, o["z"]);
+      var v = o[""];
+      if (v is List) {
+        Expect.listEquals([true, false, 42, -33e-3, null, "\u0080"], v);
+      } else {
+        Expect.fail("Expected list, found ${v.runtimeType}");
+      }
+    } else {
+      Expect.fail("Expected map, found ${o.runtimeType}");
+    }
+  }
+  for (var i = 1; i < s.length - 1; i++) {
+  var s1 = s.substring(0, i);
+  var s2 = s.substring(i);
+  jsonParse("$s1|$s2-$i", check, (sink) {
+    sink.add(s1.codeUnits);
+    sink.add(s2.codeUnits);
+    sink.close();
+  });
+  jsonParse("$s1|$s2-$i-slice", check, (sink) {
+    sink.addSlice(s.codeUnits, 0, i, false);
+    sink.addSlice(s.codeUnits, i, s.length, true);
+  });
+  for (var j = i; j < s.length - 1; j++) {
+    var s2a = s.substring(i, j);
+    var s2b = s.substring(j);
+    jsonParse("$s1|$s2a|$s2b-$i-$j", check, (sink) {
+      sink.add(s1.codeUnits);
+      sink.add(s2a.codeUnits);
+      sink.add(s2b.codeUnits);
+      sink.close();
+    });
+  }
+}
+
+}
+
+// Check that [codes] decode to [expect] when allowing malformed UTF-8,
+// and throws otherwise.
+void jsonMalformedTest(name, expect, codes) {
+  // Helper method.
+  void test(name, expect, action(sink)) {
+    var tag = "Malform:$name-$expect";
+    {  // Allowing malformed, expect [expect]
+      var sink = new ChunkedConversionSink.withCallback((values) {
+        var value = values[0];
+        Expect.equals(expect, value, tag);
+      });
+      var decoderSink = JSON.decoder.startChunkedConversion(sink)
+                                    .asUtf8Sink(true);
+      try {
+        action(decoderSink);
+      } catch (e, s) {
+        Expect.fail("Unexpected throw ($tag): $e\n$s");
+      }
+    }
+    {  // Not allowing malformed, expect throw.
+      var sink = new ChunkedConversionSink.withCallback((values) {
+        Expect.unreachable(tag);
+      });
+      var decoderSink = JSON.decoder.startChunkedConversion(sink)
+                                    .asUtf8Sink(false);
+      Expect.throws(() { action(decoderSink); }, null, tag);
+    }
+  }
+
+  // Test all two and three part slices.
+  for (int i = 1; i < codes.length - 1; i++) {
+    test("$name:$i", expect, (sink) {
+      sink.add(codes.sublist(0, i));
+      sink.add(codes.sublist(i));
+      sink.close();
+    });
+    test("$name:$i-slice", expect, (sink) {
+      sink.addSlice(codes, 0, i, false);
+      sink.addSlice(codes, i, codes.length, true);
+    });
+    for (int j = i; j < codes.length - 1; j++) {
+      test("$name:$i|$j", expect, (sink) {
+        sink.add(codes.sublist(0, i));
+        sink.add(codes.sublist(i, j));
+        sink.add(codes.sublist(j));
+        sink.close();
+      });
+    }
+  }
+}
+
+// Test that `codeString.codeUnits` fails to parse as UTF-8 JSON,
+// even with decoder not throwing on malformed encodings.
+void jsonThrows(Strig name, String codeString) {
+  testJsonThrows(tag, action) {
+    // Not allowing malformed, expect throw.
+    var sink = new ChunkedConversionSink.withCallback((values) {
+      Expect.unreachable(tag);
+    });
+    var decoderSink = JSON.decoder.startChunkedConversion(sink)
+                                  .asUtf8Sink(true);
+    Expect.throws(() { action(decoderSink); }, null, tag);
+  }
+
+  var codes = codeString.codeUnits;
+  for (int i = 1; i < codes.length - 1; i++) {
+    testJsonThrows("$name:$i", (sink) {
+      sink.add(codes.sublist(0, i));
+      sink.add(codes.sublist(i));
+      sink.close();
+    });
+    testJsonThrows("$name:$i-slice", (sink) {
+      sink.addSlice(codes, 0, i, false);
+      sink.addSlice(codes, i, codes.length, true);
+    });
+    for (int j = i; j < codes.length - 1; j++) {
+      testJsonThrows("$name:$i|$j", (sink) {
+        sink.add(codes.sublist(0, i));
+        sink.add(codes.sublist(i, j));
+        sink.add(codes.sublist(j));
+        sink.close();
+      });
+    }
+  }
+}
+
+// Malformed UTF-8 encodings.
+void testMalformed() {
+  // Overlong encodings.
+  jsonMalformedTest("overlong-0-2", "@\uFFFD@",
+                    [0x22, 0x40, 0xc0, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("overlong-0-3", "@\uFFFD@",
+                    [0x22, 0x40, 0xe0, 0x80, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("overlong-0-4", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x80, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("overlong-7f-2", "@\uFFFD@",
+                    [0x22, 0x40, 0xc1, 0xbf, 0x40, 0x22]);
+  jsonMalformedTest("overlong-7f-3", "@\uFFFD@",
+                    [0x22, 0x40, 0xe0, 0x81, 0xbf, 0x40, 0x22]);
+  jsonMalformedTest("overlong-7f-4", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x81, 0xbf, 0x40, 0x22]);
+
+  jsonMalformedTest("overlong-80-3", "@\uFFFD@",
+                    [0x22, 0x40, 0xe0, 0x82, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("overlong-80-4", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x82, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("overlong-7ff-3", "@\uFFFD@",
+                    [0x22, 0x40, 0xe0, 0x9f, 0xbf, 0x40, 0x22]);
+  jsonMalformedTest("overlong-7ff-4", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x9f, 0xbf, 0x40, 0x22]);
+
+  jsonMalformedTest("overlong-800-4", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x80, 0xa0, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("overlong-ffff-4", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x8f, 0xbf, 0xbf, 0x40, 0x22]);
+
+  // Unterminated multibyte sequences.
+  jsonMalformedTest("unterminated-2-normal", "@\uFFFD@",
+                    [0x22, 0x40, 0xc0, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-3-normal", "@\uFFFD@",
+                    [0x22, 0x40, 0xe0, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-4-normal", "@\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-2-multi", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xc0, 0xc2, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-3-multi", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xe0, 0x80, 0xc2, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-4-multi", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x80, 0xc2, 0x80, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-2-escape", "@\uFFFD\n@",
+                    [0x22, 0x40, 0xc0, 0x5c, 0x6e, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-3-escape", "@\uFFFD\n@",
+                    [0x22, 0x40, 0xe0, 0x80, 0x5c, 0x6e, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-4-escape", "@\uFFFD\n@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x80, 0x5c, 0x6e, 0x40, 0x22]);
+
+  jsonMalformedTest("unterminated-2-end", "@\uFFFD",
+                    [0x22, 0x40, 0xc0, 0x22]);
+
+  jsonMalformedTest("unterminated-3-end", "@\uFFFD",
+                    [0x22, 0x40, 0xe0, 0x80, 0x22]);
+
+  jsonMalformedTest("unterminated-4-end", "@\uFFFD",
+                    [0x22, 0x40, 0xf0, 0x80, 0x80, 0x22]);
+
+  // Unexpected continuation byte
+  // - after a normal character.
+  jsonMalformedTest("continuation-normal", "@\uFFFD@",
+                    [0x22, 0x40, 0x80, 0x40, 0x22]);
+
+  // - after a valid continuation byte.
+  jsonMalformedTest("continuation-continuation-2", "@\x80\uFFFD@",
+                    [0x22, 0x40, 0xc2, 0x80, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("continuation-continuation-3", "@\u0800\uFFFD@",
+                    [0x22, 0x40, 0xe0, 0xa0, 0x80, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("continuation-continuation-4", "@\u{10000}\uFFFD@",
+                    [0x22, 0x40, 0xf0, 0x90, 0x80, 0x80, 0x80, 0x40, 0x22]);
+
+  // - after another invalid continuation byte
+  jsonMalformedTest("continuation-twice", "@\uFFFD\uFFFD\uFFFD@",
+                    [0x22, 0x40, 0x80, 0x80, 0x80, 0x40, 0x22]);
+  // - at start.
+  jsonMalformedTest("continuation-start", "\uFFFD@",
+                    [0x22, 0x80, 0x40, 0x22]);
+
+  // Unexpected leading byte where continuation byte expected.
+  jsonMalformedTest("leading-2", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xc0, 0xc2, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("leading-3-1", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xe0, 0xc2, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("leading-3-2", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xe0, 0x80, 0xc2, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("leading-4-1", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xf0, 0xc2, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("leading-4-2", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xf0, 0x80, 0xc2, 0x80, 0x40, 0x22]);
+  jsonMalformedTest("leading-4-3", "@\uFFFD\x80@",
+                    [0x22, 0x40, 0xf0, 0x80, 0x80, 0xc2, 0x80, 0x40, 0x22]);
+
+  // Overlong encodings of ASCII outside of strings always fail.
+  // Use Latin-1 strings as argument since most chars are correct,
+  // pass string.codeUnits to decoder as UTF-8.
+  jsonThrows("number-1", "\xc0\xab0.0e-0");  // '-' is 0x2b => \xc0\xab
+  jsonThrows("number-2", "-\xc0\xb0.0e-0");  // '0' is 0x30 => \xc0\xb0
+  jsonThrows("number-3", "-0\xc0\xae0e-0");  // '.' is 0x2e => \xc0\xae
+  jsonThrows("number-4", "-0.\xc0\xb0e-0");
+  jsonThrows("number-5", "-0.0\xc1\xa5-0");  // 'e' is 0x65 => \xc1\xa5
+  jsonThrows("number-6", "-0.0e\xc0\xab0");
+  jsonThrows("number-7", "-0.0e-\xc0\xb0");
+
+  jsonThrows("true-1", "\xc1\xb4rue");  // 't' is 0x74
+  jsonThrows("true-2", "t\xc1\xb2ue");  // 'r' is 0x72
+  jsonThrows("true-3", "tr\xc1\xb5e");  // 'u' is 0x75
+  jsonThrows("true-4", "tru\xc1\xa5");  // 'e' is 0x65
+
+  jsonThrows("false-1", "\xc1\xa6alse");  // 'f' is 0x66
+  jsonThrows("false-2", "f\xc1\xa1lse");  // 'a' is 0x61
+  jsonThrows("false-3", "fa\xc1\xacse");  // 'l' is 0x6c
+  jsonThrows("false-4", "fal\xc1\xb3e");  // 's' is 0x73
+  jsonThrows("false-5", "fals\xc1\xa5");  // 'e' is 0x65
+
+  jsonThrows("null-1", "\xc1\xaeull");  // 'n' is 0x6e
+  jsonThrows("null-2", "n\xc1\xb5ll");  // 'u' is 0x75
+  jsonThrows("null-3", "nu\xc1\xacl");  // 'l' is 0x6c
+  jsonThrows("null-4", "nul\xc1\xac");  // 'l' is 0x6c
+
+  jsonThrows("array-1", "\xc1\x9b0,0]");  // '[' is 0x5b
+  jsonThrows("array-2", "[0,0\xc1\x9d");  // ']' is 0x5d
+  jsonThrows("array-2", "[0\xc0\xac0]");  // ',' is 0x2c
+
+  jsonThrows("object-1", '\xc1\xbb"x":0}');  // '{' is 0x7b
+  jsonThrows("object-2", '{"x":0\xc1\xbd');  // '}' is 0x7d
+  jsonThrows("object-2", '{"x\xc0\xba0}');  // ':' is 0x3a
+
+  jsonThrows("string-1", '\xc0\xa2x"');  // '"' is 0x22
+  jsonThrows("string-1", '"x\xc0\xa2');  // Unterminated string.
+
+  jsonThrows("whitespace-1", "\xc0\xa01");  // ' ' is 0x20
+}