numeric_literal_test.cpp 9.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330
  1. // Part of the Carbon Language project, under the Apache License v2.0 with LLVM
  2. // Exceptions. See /LICENSE for license information.
  3. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
  4. #include "toolchain/lex/numeric_literal.h"
  5. #include <gmock/gmock.h>
  6. #include <gtest/gtest.h>
  7. #include "common/check.h"
  8. #include "toolchain/diagnostics/diagnostic_emitter.h"
  9. #include "toolchain/lex/test_helpers.h"
  10. namespace Carbon::Lex {
  11. namespace {
  12. using ::testing::_;
  13. using ::testing::Field;
  14. using ::testing::Matcher;
  15. using ::testing::Property;
  16. using ::testing::Truly;
  17. using ::testing::VariantWith;
  18. class NumericLiteralTest : public ::testing::Test {
  19. protected:
  20. NumericLiteralTest() : error_tracker(ConsoleDiagnosticConsumer()) {}
  21. auto Lex(llvm::StringRef text) -> NumericLiteral {
  22. std::optional<NumericLiteral> result = NumericLiteral::Lex(text);
  23. CARBON_CHECK(result);
  24. EXPECT_EQ(result->text(), text);
  25. return *result;
  26. }
  27. auto Parse(llvm::StringRef text) -> NumericLiteral::Value {
  28. Testing::SingleTokenDiagnosticTranslator translator(text);
  29. DiagnosticEmitter<const char*> emitter(translator, error_tracker);
  30. return Lex(text).ComputeValue(emitter);
  31. }
  32. ErrorTrackingDiagnosticConsumer error_tracker;
  33. };
  34. // Matcher for signed llvm::APInt.
  35. auto IsSignedInt(int64_t value) -> Matcher<llvm::APInt> {
  36. return Property(&llvm::APInt::getSExtValue, value);
  37. }
  38. // Matcher for unsigned llvm::APInt.
  39. auto IsUnsignedInt(uint64_t value) -> Matcher<llvm::APInt> {
  40. return Property(&llvm::APInt::getZExtValue, value);
  41. }
  42. // Matcher for an integer literal value.
  43. template <typename ValueMatcher>
  44. auto HasIntValue(const ValueMatcher& value_matcher)
  45. -> Matcher<NumericLiteral::Value> {
  46. return VariantWith<NumericLiteral::IntValue>(
  47. Field(&NumericLiteral::IntValue::value, value_matcher));
  48. }
  49. struct RealMatcher {
  50. Matcher<int> radix = _;
  51. Matcher<llvm::APInt> mantissa = _;
  52. Matcher<llvm::APInt> exponent = _;
  53. };
  54. // Matcher for a real literal value.
  55. auto HasRealValue(const RealMatcher& real_matcher)
  56. -> Matcher<NumericLiteral::Value> {
  57. return VariantWith<NumericLiteral::RealValue>(AllOf(
  58. Field(&NumericLiteral::RealValue::radix, real_matcher.radix),
  59. Field(&NumericLiteral::RealValue::mantissa, real_matcher.mantissa),
  60. Field(&NumericLiteral::RealValue::exponent, real_matcher.exponent)));
  61. }
  62. // Matcher for an unrecoverable parse error.
  63. auto HasUnrecoverableError() -> Matcher<NumericLiteral::Value> {
  64. return VariantWith<NumericLiteral::UnrecoverableError>(_);
  65. }
  66. TEST_F(NumericLiteralTest, HandlesIntLiteral) {
  67. struct Testcase {
  68. llvm::StringLiteral token;
  69. uint64_t value;
  70. int radix;
  71. };
  72. Testcase testcases[] = {
  73. {.token = "12", .value = 12, .radix = 10},
  74. {.token = "0x12_3ABC", .value = 0x12'3ABC, .radix = 16},
  75. {.token = "0b10_10_11", .value = 0b10'10'11, .radix = 2},
  76. {.token = "1_234_567", .value = 1'234'567, .radix = 10},
  77. };
  78. for (Testcase testcase : testcases) {
  79. error_tracker.Reset();
  80. EXPECT_THAT(Parse(testcase.token),
  81. HasIntValue(IsUnsignedInt(testcase.value)))
  82. << testcase.token;
  83. EXPECT_FALSE(error_tracker.seen_error()) << testcase.token;
  84. }
  85. }
  86. TEST_F(NumericLiteralTest, ValidatesBaseSpecifier) {
  87. llvm::StringLiteral valid[] = {
  88. // Decimal integer literals.
  89. "0",
  90. "1",
  91. "123456789000000000000000000000000000000000000",
  92. // Hexadecimal integer literals.
  93. "0x0123456789ABCDEF",
  94. "0x0000000000000000000000000000000",
  95. // Binary integer literals.
  96. "0b10110100101001010",
  97. "0b0000000",
  98. };
  99. for (llvm::StringLiteral literal : valid) {
  100. error_tracker.Reset();
  101. EXPECT_THAT(Parse(literal), HasIntValue(_)) << literal;
  102. EXPECT_FALSE(error_tracker.seen_error()) << literal;
  103. }
  104. llvm::StringLiteral invalid[] = {
  105. "00", "0X123", "0o123", "0B1",
  106. "007", "123L", "123456789A", "0x",
  107. "0b", "0x123abc", "0b011101201001", "0b10A",
  108. "0x_", "0b_",
  109. };
  110. for (llvm::StringLiteral literal : invalid) {
  111. error_tracker.Reset();
  112. EXPECT_THAT(Parse(literal), HasUnrecoverableError()) << literal;
  113. EXPECT_TRUE(error_tracker.seen_error()) << literal;
  114. }
  115. }
  116. TEST_F(NumericLiteralTest, ValidatesIntDigitSeparators) {
  117. llvm::StringLiteral valid[] = {
  118. // Decimal literals optionally have digit separators every 3 places.
  119. "1_234",
  120. "123_456",
  121. "1_234_567",
  122. // Hexadecimal literals optionally have digit separators every 4 places.
  123. "0x1_0000",
  124. "0x1000_0000",
  125. "0x1_0000_0000",
  126. // Binary integer literals can have digit separators anywhere..
  127. "0b1_0_1_0_1_0",
  128. "0b111_0000",
  129. };
  130. for (llvm::StringLiteral literal : valid) {
  131. error_tracker.Reset();
  132. EXPECT_THAT(Parse(literal), HasIntValue(_)) << literal;
  133. EXPECT_FALSE(error_tracker.seen_error()) << literal;
  134. }
  135. llvm::StringLiteral invalid[] = {
  136. // Decimal literals.
  137. "12_34",
  138. "123_4_6_789",
  139. "12_3456_789",
  140. "12__345",
  141. "1_",
  142. // Hexadecimal literals.
  143. "0x_1234",
  144. "0x123_",
  145. "0x12_3",
  146. "0x_234_5678",
  147. "0x1234_567",
  148. // Binary literals.
  149. "0b_10101",
  150. "0b1__01",
  151. "0b1011_",
  152. "0b1_01_01_",
  153. };
  154. for (llvm::StringLiteral literal : invalid) {
  155. error_tracker.Reset();
  156. EXPECT_THAT(Parse(literal), HasIntValue(_)) << literal;
  157. EXPECT_TRUE(error_tracker.seen_error()) << literal;
  158. }
  159. }
  160. TEST_F(NumericLiteralTest, HandlesRealLiteral) {
  161. struct Testcase {
  162. llvm::StringLiteral token;
  163. uint64_t mantissa;
  164. int64_t exponent;
  165. unsigned radix;
  166. };
  167. Testcase testcases[] = {
  168. // Decimal real literals.
  169. {.token = "0.0", .mantissa = 0, .exponent = -1, .radix = 10},
  170. {.token = "12.345", .mantissa = 12345, .exponent = -3, .radix = 10},
  171. {.token = "12.345e6", .mantissa = 12345, .exponent = 3, .radix = 10},
  172. {.token = "12.345e+6", .mantissa = 12345, .exponent = 3, .radix = 10},
  173. {.token = "1_234.5e-2", .mantissa = 12345, .exponent = -3, .radix = 10},
  174. {.token = "1.0e-2_000_000",
  175. .mantissa = 10,
  176. .exponent = -2'000'001,
  177. .radix = 10},
  178. // Hexadecimal real literals.
  179. {.token = "0x1_2345_6789.CDEF",
  180. .mantissa = 0x1'2345'6789'CDEF,
  181. .exponent = -16,
  182. .radix = 16},
  183. {.token = "0x0.0001p4", .mantissa = 1, .exponent = -12, .radix = 16},
  184. {.token = "0x0.0001p+4", .mantissa = 1, .exponent = -12, .radix = 16},
  185. {.token = "0x0.0001p-4", .mantissa = 1, .exponent = -20, .radix = 16},
  186. // The exponent here works out as exactly INT64_MIN.
  187. {.token = "0x1.01p-9223372036854775800",
  188. .mantissa = 0x101,
  189. .exponent = -9223372036854775807L - 1L,
  190. .radix = 16},
  191. // The exponent here doesn't fit in a signed 64-bit integer until we
  192. // adjust for the radix point.
  193. {.token = "0x1.01p9223372036854775809",
  194. .mantissa = 0x101,
  195. .exponent = 9223372036854775801L,
  196. .radix = 16},
  197. // Binary real literals. These are invalid, but we accept them for error
  198. // recovery.
  199. {.token = "0b10_11_01.01",
  200. .mantissa = 0b10110101,
  201. .exponent = -2,
  202. .radix = 2},
  203. };
  204. for (Testcase testcase : testcases) {
  205. error_tracker.Reset();
  206. EXPECT_THAT(Parse(testcase.token),
  207. HasRealValue({.radix = (testcase.radix == 10 ? 10 : 2),
  208. .mantissa = IsUnsignedInt(testcase.mantissa),
  209. .exponent = IsSignedInt(testcase.exponent)}))
  210. << testcase.token;
  211. EXPECT_EQ(error_tracker.seen_error(), testcase.radix == 2)
  212. << testcase.token;
  213. }
  214. }
  215. TEST_F(NumericLiteralTest, HandlesRealLiteralOverflow) {
  216. llvm::StringLiteral input = "0x1.000001p-9223372036854775800";
  217. error_tracker.Reset();
  218. EXPECT_THAT(
  219. Parse(input),
  220. HasRealValue({.radix = 2,
  221. .mantissa = IsUnsignedInt(0x1000001),
  222. .exponent = Truly([](llvm::APInt exponent) {
  223. return (exponent + 9223372036854775800).getSExtValue() ==
  224. -24;
  225. })}));
  226. EXPECT_FALSE(error_tracker.seen_error());
  227. }
  228. TEST_F(NumericLiteralTest, ValidatesRealLiterals) {
  229. llvm::StringLiteral invalid_digit_separators[] = {
  230. // Invalid digit separators.
  231. "12_34.5", "123.4_567", "123.456_7", "1_2_3.4",
  232. "123.4e56_78", "0x12_34.5", "0x12.3_4", "0x12.34p5_6",
  233. };
  234. for (llvm::StringLiteral literal : invalid_digit_separators) {
  235. error_tracker.Reset();
  236. EXPECT_THAT(Parse(literal), HasRealValue({})) << literal;
  237. EXPECT_TRUE(error_tracker.seen_error()) << literal;
  238. }
  239. llvm::StringLiteral invalid[] = {
  240. // No digits in integer part.
  241. "0x.0",
  242. "0b.0",
  243. "0x_.0",
  244. "0b_.0",
  245. // No digits in fractional part.
  246. "0.e",
  247. "0.e0",
  248. "0.e+0",
  249. "0x0.p",
  250. "0x0.p-0",
  251. // Invalid digits in mantissa.
  252. "123A.4",
  253. "123.4A",
  254. "123A.4e0",
  255. "123.4Ae0",
  256. "0x123ABCDEFG.0",
  257. "0x123.ABCDEFG",
  258. "0x123ABCDEFG.0p0",
  259. "0x123.ABCDEFGp0",
  260. // Invalid exponent letter.
  261. "0.0f0",
  262. "0.0p0",
  263. "0.0z+0",
  264. "0x0.0e0",
  265. "0x0.0f0",
  266. "0x0.0z-0",
  267. // No digits in exponent part.
  268. "0.0e",
  269. "0x0.0p",
  270. "0.0e_",
  271. "0x0.0p_",
  272. // Invalid digits in exponent part.
  273. "0.0eHELLO",
  274. "0.0eA",
  275. "0.0e+A",
  276. "0x0.0pA",
  277. "0x0.0p-A",
  278. };
  279. for (llvm::StringLiteral literal : invalid) {
  280. error_tracker.Reset();
  281. EXPECT_THAT(Parse(literal), HasUnrecoverableError()) << literal;
  282. EXPECT_TRUE(error_tracker.seen_error()) << literal;
  283. }
  284. }
  285. TEST_F(NumericLiteralTest, TooManyDigits) {
  286. std::string long_number(2000, '1');
  287. EXPECT_THAT(Parse(long_number), HasUnrecoverableError());
  288. EXPECT_TRUE(error_tracker.seen_error());
  289. }
  290. } // namespace
  291. } // namespace Carbon::Lex