1# Advanced GoogleTest Topics 2 3## Introduction 4 5Now that you have read the [GoogleTest Primer](primer.md) and learned how to 6write tests using GoogleTest, it's time to learn some new tricks. This document 7will show you more assertions as well as how to construct complex failure 8messages, propagate fatal failures, reuse and speed up your test fixtures, and 9use various flags with your tests. 10 11## More Assertions 12 13This section covers some less frequently used, but still significant, 14assertions. 15 16### Explicit Success and Failure 17 18See [Explicit Success and Failure](reference/assertions.md#success-failure) in 19the Assertions Reference. 20 21### Exception Assertions 22 23See [Exception Assertions](reference/assertions.md#exceptions) in the Assertions 24Reference. 25 26### Predicate Assertions for Better Error Messages 27 28Even though GoogleTest has a rich set of assertions, they can never be complete, 29as it's impossible (nor a good idea) to anticipate all scenarios a user might 30run into. Therefore, sometimes a user has to use `EXPECT_TRUE()` to check a 31complex expression, for lack of a better macro. This has the problem of not 32showing you the values of the parts of the expression, making it hard to 33understand what went wrong. As a workaround, some users choose to construct the 34failure message by themselves, streaming it into `EXPECT_TRUE()`. However, this 35is awkward especially when the expression has side-effects or is expensive to 36evaluate. 37 38GoogleTest gives you three different options to solve this problem: 39 40#### Using an Existing Boolean Function 41 42If you already have a function or functor that returns `bool` (or a type that 43can be implicitly converted to `bool`), you can use it in a *predicate 44assertion* to get the function arguments printed for free. See 45[`EXPECT_PRED*`](reference/assertions.md#EXPECT_PRED) in the Assertions 46Reference for details. 47 48#### Using a Function That Returns an AssertionResult 49 50While `EXPECT_PRED*()` and friends are handy for a quick job, the syntax is not 51satisfactory: you have to use different macros for different arities, and it 52feels more like Lisp than C++. The `::testing::AssertionResult` class solves 53this problem. 54 55An `AssertionResult` object represents the result of an assertion (whether it's 56a success or a failure, and an associated message). You can create an 57`AssertionResult` using one of these factory functions: 58 59```c++ 60namespace testing { 61 62// Returns an AssertionResult object to indicate that an assertion has 63// succeeded. 64AssertionResult AssertionSuccess(); 65 66// Returns an AssertionResult object to indicate that an assertion has 67// failed. 68AssertionResult AssertionFailure(); 69 70} 71``` 72 73You can then use the `<<` operator to stream messages to the `AssertionResult` 74object. 75 76To provide more readable messages in Boolean assertions (e.g. `EXPECT_TRUE()`), 77write a predicate function that returns `AssertionResult` instead of `bool`. For 78example, if you define `IsEven()` as: 79 80```c++ 81testing::AssertionResult IsEven(int n) { 82 if ((n % 2) == 0) 83 return testing::AssertionSuccess(); 84 else 85 return testing::AssertionFailure() << n << " is odd"; 86} 87``` 88 89instead of: 90 91```c++ 92bool IsEven(int n) { 93 return (n % 2) == 0; 94} 95``` 96 97the failed assertion `EXPECT_TRUE(IsEven(Fib(4)))` will print: 98 99```none 100Value of: IsEven(Fib(4)) 101 Actual: false (3 is odd) 102Expected: true 103``` 104 105instead of a more opaque 106 107```none 108Value of: IsEven(Fib(4)) 109 Actual: false 110Expected: true 111``` 112 113If you want informative messages in `EXPECT_FALSE` and `ASSERT_FALSE` as well 114(one third of Boolean assertions in the Google code base are negative ones), and 115are fine with making the predicate slower in the success case, you can supply a 116success message: 117 118```c++ 119testing::AssertionResult IsEven(int n) { 120 if ((n % 2) == 0) 121 return testing::AssertionSuccess() << n << " is even"; 122 else 123 return testing::AssertionFailure() << n << " is odd"; 124} 125``` 126 127Then the statement `EXPECT_FALSE(IsEven(Fib(6)))` will print 128 129```none 130 Value of: IsEven(Fib(6)) 131 Actual: true (8 is even) 132 Expected: false 133``` 134 135#### Using a Predicate-Formatter 136 137If you find the default message generated by 138[`EXPECT_PRED*`](reference/assertions.md#EXPECT_PRED) and 139[`EXPECT_TRUE`](reference/assertions.md#EXPECT_TRUE) unsatisfactory, or some 140arguments to your predicate do not support streaming to `ostream`, you can 141instead use *predicate-formatter assertions* to *fully* customize how the 142message is formatted. See 143[`EXPECT_PRED_FORMAT*`](reference/assertions.md#EXPECT_PRED_FORMAT) in the 144Assertions Reference for details. 145 146### Floating-Point Comparison 147 148See [Floating-Point Comparison](reference/assertions.md#floating-point) in the 149Assertions Reference. 150 151#### Floating-Point Predicate-Format Functions 152 153Some floating-point operations are useful, but not that often used. In order to 154avoid an explosion of new macros, we provide them as predicate-format functions 155that can be used in the predicate assertion macro 156[`EXPECT_PRED_FORMAT2`](reference/assertions.md#EXPECT_PRED_FORMAT), for 157example: 158 159```c++ 160using ::testing::FloatLE; 161using ::testing::DoubleLE; 162... 163EXPECT_PRED_FORMAT2(FloatLE, val1, val2); 164EXPECT_PRED_FORMAT2(DoubleLE, val1, val2); 165``` 166 167The above code verifies that `val1` is less than, or approximately equal to, 168`val2`. 169 170### Asserting Using gMock Matchers 171 172See [`EXPECT_THAT`](reference/assertions.md#EXPECT_THAT) in the Assertions 173Reference. 174 175### More String Assertions 176 177(Please read the [previous](#asserting-using-gmock-matchers) section first if 178you haven't.) 179 180You can use the gMock [string matchers](reference/matchers.md#string-matchers) 181with [`EXPECT_THAT`](reference/assertions.md#EXPECT_THAT) to do more string 182comparison tricks (sub-string, prefix, suffix, regular expression, and etc). For 183example, 184 185```c++ 186using ::testing::HasSubstr; 187using ::testing::MatchesRegex; 188... 189 ASSERT_THAT(foo_string, HasSubstr("needle")); 190 EXPECT_THAT(bar_string, MatchesRegex("\\w*\\d+")); 191``` 192 193### Windows HRESULT assertions 194 195See [Windows HRESULT Assertions](reference/assertions.md#HRESULT) in the 196Assertions Reference. 197 198### Type Assertions 199 200You can call the function 201 202```c++ 203::testing::StaticAssertTypeEq<T1, T2>(); 204``` 205 206to assert that types `T1` and `T2` are the same. The function does nothing if 207the assertion is satisfied. If the types are different, the function call will 208fail to compile, the compiler error message will say that `T1 and T2 are not the 209same type` and most likely (depending on the compiler) show you the actual 210values of `T1` and `T2`. This is mainly useful inside template code. 211 212**Caveat**: When used inside a member function of a class template or a function 213template, `StaticAssertTypeEq<T1, T2>()` is effective only if the function is 214instantiated. For example, given: 215 216```c++ 217template <typename T> class Foo { 218 public: 219 void Bar() { testing::StaticAssertTypeEq<int, T>(); } 220}; 221``` 222 223the code: 224 225```c++ 226void Test1() { Foo<bool> foo; } 227``` 228 229will not generate a compiler error, as `Foo<bool>::Bar()` is never actually 230instantiated. Instead, you need: 231 232```c++ 233void Test2() { Foo<bool> foo; foo.Bar(); } 234``` 235 236to cause a compiler error. 237 238### Assertion Placement 239 240You can use assertions in any C++ function. In particular, it doesn't have to be 241a method of the test fixture class. The one constraint is that assertions that 242generate a fatal failure (`FAIL*` and `ASSERT_*`) can only be used in 243void-returning functions. This is a consequence of Google's not using 244exceptions. By placing it in a non-void function you'll get a confusing compile 245error like `"error: void value not ignored as it ought to be"` or `"cannot 246initialize return object of type 'bool' with an rvalue of type 'void'"` or 247`"error: no viable conversion from 'void' to 'string'"`. 248 249If you need to use fatal assertions in a function that returns non-void, one 250option is to make the function return the value in an out parameter instead. For 251example, you can rewrite `T2 Foo(T1 x)` to `void Foo(T1 x, T2* result)`. You 252need to make sure that `*result` contains some sensible value even when the 253function returns prematurely. As the function now returns `void`, you can use 254any assertion inside of it. 255 256If changing the function's type is not an option, you should just use assertions 257that generate non-fatal failures, such as `ADD_FAILURE*` and `EXPECT_*`. 258 259{: .callout .note} 260NOTE: Constructors and destructors are not considered void-returning functions, 261according to the C++ language specification, and so you may not use fatal 262assertions in them; you'll get a compilation error if you try. Instead, either 263call `abort` and crash the entire test executable, or put the fatal assertion in 264a `SetUp`/`TearDown` function; see 265[constructor/destructor vs. `SetUp`/`TearDown`](faq.md#CtorVsSetUp) 266 267{: .callout .warning} 268WARNING: A fatal assertion in a helper function (private void-returning method) 269called from a constructor or destructor does not terminate the current test, as 270your intuition might suggest: it merely returns from the constructor or 271destructor early, possibly leaving your object in a partially-constructed or 272partially-destructed state! You almost certainly want to `abort` or use 273`SetUp`/`TearDown` instead. 274 275## Skipping test execution 276 277Related to the assertions `SUCCEED()` and `FAIL()`, you can prevent further test 278execution at runtime with the `GTEST_SKIP()` macro. This is useful when you need 279to check for preconditions of the system under test during runtime and skip 280tests in a meaningful way. 281 282`GTEST_SKIP()` can be used in individual test cases or in the `SetUp()` methods 283of classes derived from either `::testing::Environment` or `::testing::Test`. 284For example: 285 286```c++ 287TEST(SkipTest, DoesSkip) { 288 GTEST_SKIP() << "Skipping single test"; 289 EXPECT_EQ(0, 1); // Won't fail; it won't be executed 290} 291 292class SkipFixture : public ::testing::Test { 293 protected: 294 void SetUp() override { 295 GTEST_SKIP() << "Skipping all tests for this fixture"; 296 } 297}; 298 299// Tests for SkipFixture won't be executed. 300TEST_F(SkipFixture, SkipsOneTest) { 301 EXPECT_EQ(5, 7); // Won't fail 302} 303``` 304 305As with assertion macros, you can stream a custom message into `GTEST_SKIP()`. 306 307## Teaching GoogleTest How to Print Your Values 308 309When a test assertion such as `EXPECT_EQ` fails, GoogleTest prints the argument 310values to help you debug. It does this using a user-extensible value printer. 311 312This printer knows how to print built-in C++ types, native arrays, STL 313containers, and any type that supports the `<<` operator. For other types, it 314prints the raw bytes in the value and hopes that you the user can figure it out. 315 316As mentioned earlier, the printer is *extensible*. That means you can teach it 317to do a better job at printing your particular type than to dump the bytes. To 318do that, define an `AbslStringify()` overload as a `friend` function template 319for your type: 320 321```cpp 322namespace foo { 323 324class Point { // We want GoogleTest to be able to print instances of this. 325 ... 326 // Provide a friend overload. 327 template <typename Sink> 328 friend void AbslStringify(Sink& sink, const Point& point) { 329 absl::Format(&sink, "(%d, %d)", point.x, point.y); 330 } 331 332 int x; 333 int y; 334}; 335 336// If you can't declare the function in the class it's important that the 337// AbslStringify overload is defined in the SAME namespace that defines Point. 338// C++'s look-up rules rely on that. 339enum class EnumWithStringify { kMany = 0, kChoices = 1 }; 340 341template <typename Sink> 342void AbslStringify(Sink& sink, EnumWithStringify e) { 343 absl::Format(&sink, "%s", e == EnumWithStringify::kMany ? "Many" : "Choices"); 344} 345 346} // namespace foo 347``` 348 349{: .callout .note} 350Note: `AbslStringify()` utilizes a generic "sink" buffer to construct its 351string. For more information about supported operations on `AbslStringify()`'s 352sink, see go/abslstringify. 353 354`AbslStringify()` can also use `absl::StrFormat`'s catch-all `%v` type specifier 355within its own format strings to perform type deduction. `Point` above could be 356formatted as `"(%v, %v)"` for example, and deduce the `int` values as `%d`. 357 358Sometimes, `AbslStringify()` might not be an option: your team may wish to print 359types with extra debugging information for testing purposes only. If so, you can 360instead define a `PrintTo()` function like this: 361 362```c++ 363#include <ostream> 364 365namespace foo { 366 367class Point { 368 ... 369 friend void PrintTo(const Point& point, std::ostream* os) { 370 *os << "(" << point.x << "," << point.y << ")"; 371 } 372 373 int x; 374 int y; 375}; 376 377// If you can't declare the function in the class it's important that PrintTo() 378// is defined in the SAME namespace that defines Point. C++'s look-up rules 379// rely on that. 380void PrintTo(const Point& point, std::ostream* os) { 381 *os << "(" << point.x << "," << point.y << ")"; 382} 383 384} // namespace foo 385``` 386 387If you have defined both `AbslStringify()` and `PrintTo()`, the latter will be 388used by GoogleTest. This allows you to customize how the value appears in 389GoogleTest's output without affecting code that relies on the behavior of 390`AbslStringify()`. 391 392If you have an existing `<<` operator and would like to define an 393`AbslStringify()`, the latter will be used for GoogleTest printing. 394 395If you want to print a value `x` using GoogleTest's value printer yourself, just 396call `::testing::PrintToString(x)`, which returns an `std::string`: 397 398```c++ 399vector<pair<Point, int> > point_ints = GetPointIntVector(); 400 401EXPECT_TRUE(IsCorrectPointIntVector(point_ints)) 402 << "point_ints = " << testing::PrintToString(point_ints); 403``` 404 405For more details regarding `AbslStringify()` and its integration with other 406libraries, see go/abslstringify. 407 408## Death Tests 409 410In many applications, there are assertions that can cause application failure if 411a condition is not met. These consistency checks, which ensure that the program 412is in a known good state, are there to fail at the earliest possible time after 413some program state is corrupted. If the assertion checks the wrong condition, 414then the program may proceed in an erroneous state, which could lead to memory 415corruption, security holes, or worse. Hence it is vitally important to test that 416such assertion statements work as expected. 417 418Since these precondition checks cause the processes to die, we call such tests 419_death tests_. More generally, any test that checks that a program terminates 420(except by throwing an exception) in an expected fashion is also a death test. 421 422Note that if a piece of code throws an exception, we don't consider it "death" 423for the purpose of death tests, as the caller of the code could catch the 424exception and avoid the crash. If you want to verify exceptions thrown by your 425code, see [Exception Assertions](#ExceptionAssertions). 426 427If you want to test `EXPECT_*()/ASSERT_*()` failures in your test code, see 428["Catching" Failures](#catching-failures). 429 430### How to Write a Death Test 431 432GoogleTest provides assertion macros to support death tests. See 433[Death Assertions](reference/assertions.md#death) in the Assertions Reference 434for details. 435 436To write a death test, simply use one of the macros inside your test function. 437For example, 438 439```c++ 440TEST(MyDeathTest, Foo) { 441 // This death test uses a compound statement. 442 ASSERT_DEATH({ 443 int n = 5; 444 Foo(&n); 445 }, "Error on line .* of Foo()"); 446} 447 448TEST(MyDeathTest, NormalExit) { 449 EXPECT_EXIT(NormalExit(), testing::ExitedWithCode(0), "Success"); 450} 451 452TEST(MyDeathTest, KillProcess) { 453 EXPECT_EXIT(KillProcess(), testing::KilledBySignal(SIGKILL), 454 "Sending myself unblockable signal"); 455} 456``` 457 458verifies that: 459 460* calling `Foo(5)` causes the process to die with the given error message, 461* calling `NormalExit()` causes the process to print `"Success"` to stderr and 462 exit with exit code 0, and 463* calling `KillProcess()` kills the process with signal `SIGKILL`. 464 465The test function body may contain other assertions and statements as well, if 466necessary. 467 468Note that a death test only cares about three things: 469 4701. does `statement` abort or exit the process? 4712. (in the case of `ASSERT_EXIT` and `EXPECT_EXIT`) does the exit status 472 satisfy `predicate`? Or (in the case of `ASSERT_DEATH` and `EXPECT_DEATH`) 473 is the exit status non-zero? And 4743. does the stderr output match `matcher`? 475 476In particular, if `statement` generates an `ASSERT_*` or `EXPECT_*` failure, it 477will **not** cause the death test to fail, as GoogleTest assertions don't abort 478the process. 479 480### Death Test Naming 481 482{: .callout .important} 483IMPORTANT: We strongly recommend you to follow the convention of naming your 484**test suite** (not test) `*DeathTest` when it contains a death test, as 485demonstrated in the above example. The 486[Death Tests And Threads](#death-tests-and-threads) section below explains why. 487 488If a test fixture class is shared by normal tests and death tests, you can use 489`using` or `typedef` to introduce an alias for the fixture class and avoid 490duplicating its code: 491 492```c++ 493class FooTest : public testing::Test { ... }; 494 495using FooDeathTest = FooTest; 496 497TEST_F(FooTest, DoesThis) { 498 // normal test 499} 500 501TEST_F(FooDeathTest, DoesThat) { 502 // death test 503} 504``` 505 506### Regular Expression Syntax 507 508When built with Bazel and using Abseil, GoogleTest uses the 509[RE2](https://github.com/google/re2/wiki/Syntax) syntax. Otherwise, for POSIX 510systems (Linux, Cygwin, Mac), GoogleTest uses the 511[POSIX extended regular expression](https://www.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap09.html#tag_09_04) 512syntax. To learn about POSIX syntax, you may want to read this 513[Wikipedia entry](https://en.wikipedia.org/wiki/Regular_expression#POSIX_extended). 514 515On Windows, GoogleTest uses its own simple regular expression implementation. It 516lacks many features. For example, we don't support union (`"x|y"`), grouping 517(`"(xy)"`), brackets (`"[xy]"`), and repetition count (`"x{5,7}"`), among 518others. Below is what we do support (`A` denotes a literal character, period 519(`.`), or a single `\\ ` escape sequence; `x` and `y` denote regular 520expressions.): 521 522Expression | Meaning 523---------- | -------------------------------------------------------------- 524`c` | matches any literal character `c` 525`\\d` | matches any decimal digit 526`\\D` | matches any character that's not a decimal digit 527`\\f` | matches `\f` 528`\\n` | matches `\n` 529`\\r` | matches `\r` 530`\\s` | matches any ASCII whitespace, including `\n` 531`\\S` | matches any character that's not a whitespace 532`\\t` | matches `\t` 533`\\v` | matches `\v` 534`\\w` | matches any letter, `_`, or decimal digit 535`\\W` | matches any character that `\\w` doesn't match 536`\\c` | matches any literal character `c`, which must be a punctuation 537`.` | matches any single character except `\n` 538`A?` | matches 0 or 1 occurrences of `A` 539`A*` | matches 0 or many occurrences of `A` 540`A+` | matches 1 or many occurrences of `A` 541`^` | matches the beginning of a string (not that of each line) 542`$` | matches the end of a string (not that of each line) 543`xy` | matches `x` followed by `y` 544 545To help you determine which capability is available on your system, GoogleTest 546defines macros to govern which regular expression it is using. The macros are: 547`GTEST_USES_SIMPLE_RE=1` or `GTEST_USES_POSIX_RE=1`. If you want your death 548tests to work in all cases, you can either `#if` on these macros or use the more 549limited syntax only. 550 551### How It Works 552 553See [Death Assertions](reference/assertions.md#death) in the Assertions 554Reference. 555 556### Death Tests And Threads 557 558The reason for the two death test styles has to do with thread safety. Due to 559well-known problems with forking in the presence of threads, death tests should 560be run in a single-threaded context. Sometimes, however, it isn't feasible to 561arrange that kind of environment. For example, statically-initialized modules 562may start threads before main is ever reached. Once threads have been created, 563it may be difficult or impossible to clean them up. 564 565GoogleTest has three features intended to raise awareness of threading issues. 566 5671. A warning is emitted if multiple threads are running when a death test is 568 encountered. 5692. Test suites with a name ending in "DeathTest" are run before all other 570 tests. 5713. It uses `clone()` instead of `fork()` to spawn the child process on Linux 572 (`clone()` is not available on Cygwin and Mac), as `fork()` is more likely 573 to cause the child to hang when the parent process has multiple threads. 574 575It's perfectly fine to create threads inside a death test statement; they are 576executed in a separate process and cannot affect the parent. 577 578### Death Test Styles 579 580The "threadsafe" death test style was introduced in order to help mitigate the 581risks of testing in a possibly multithreaded environment. It trades increased 582test execution time (potentially dramatically so) for improved thread safety. 583 584The automated testing framework does not set the style flag. You can choose a 585particular style of death tests by setting the flag programmatically: 586 587```c++ 588GTEST_FLAG_SET(death_test_style, "threadsafe"); 589``` 590 591You can do this in `main()` to set the style for all death tests in the binary, 592or in individual tests. Recall that flags are saved before running each test and 593restored afterwards, so you need not do that yourself. For example: 594 595```c++ 596int main(int argc, char** argv) { 597 testing::InitGoogleTest(&argc, argv); 598 GTEST_FLAG_SET(death_test_style, "fast"); 599 return RUN_ALL_TESTS(); 600} 601 602TEST(MyDeathTest, TestOne) { 603 GTEST_FLAG_SET(death_test_style, "threadsafe"); 604 // This test is run in the "threadsafe" style: 605 ASSERT_DEATH(ThisShouldDie(), ""); 606} 607 608TEST(MyDeathTest, TestTwo) { 609 // This test is run in the "fast" style: 610 ASSERT_DEATH(ThisShouldDie(), ""); 611} 612``` 613 614### Caveats 615 616The `statement` argument of `ASSERT_EXIT()` can be any valid C++ statement. If 617it leaves the current function via a `return` statement or by throwing an 618exception, the death test is considered to have failed. Some GoogleTest macros 619may return from the current function (e.g. `ASSERT_TRUE()`), so be sure to avoid 620them in `statement`. 621 622Since `statement` runs in the child process, any in-memory side effect (e.g. 623modifying a variable, releasing memory, etc) it causes will *not* be observable 624in the parent process. In particular, if you release memory in a death test, 625your program will fail the heap check as the parent process will never see the 626memory reclaimed. To solve this problem, you can 627 6281. try not to free memory in a death test; 6292. free the memory again in the parent process; or 6303. do not use the heap checker in your program. 631 632Due to an implementation detail, you cannot place multiple death test assertions 633on the same line; otherwise, compilation will fail with an unobvious error 634message. 635 636Despite the improved thread safety afforded by the "threadsafe" style of death 637test, thread problems such as deadlock are still possible in the presence of 638handlers registered with `pthread_atfork(3)`. 639 640## Using Assertions in Sub-routines 641 642{: .callout .note} 643Note: If you want to put a series of test assertions in a subroutine to check 644for a complex condition, consider using 645[a custom GMock matcher](gmock_cook_book.md#NewMatchers) instead. This lets you 646provide a more readable error message in case of failure and avoid all of the 647issues described below. 648 649### Adding Traces to Assertions 650 651If a test sub-routine is called from several places, when an assertion inside it 652fails, it can be hard to tell which invocation of the sub-routine the failure is 653from. You can alleviate this problem using extra logging or custom failure 654messages, but that usually clutters up your tests. A better solution is to use 655the `SCOPED_TRACE` macro or the `ScopedTrace` utility: 656 657```c++ 658SCOPED_TRACE(message); 659``` 660 661```c++ 662ScopedTrace trace("file_path", line_number, message); 663``` 664 665where `message` can be anything streamable to `std::ostream`. `SCOPED_TRACE` 666macro will cause the current file name, line number, and the given message to be 667added in every failure message. `ScopedTrace` accepts explicit file name and 668line number in arguments, which is useful for writing test helpers. The effect 669will be undone when the control leaves the current lexical scope. 670 671For example, 672 673```c++ 67410: void Sub1(int n) { 67511: EXPECT_EQ(Bar(n), 1); 67612: EXPECT_EQ(Bar(n + 1), 2); 67713: } 67814: 67915: TEST(FooTest, Bar) { 68016: { 68117: SCOPED_TRACE("A"); // This trace point will be included in 68218: // every failure in this scope. 68319: Sub1(1); 68420: } 68521: // Now it won't. 68622: Sub1(9); 68723: } 688``` 689 690could result in messages like these: 691 692```none 693path/to/foo_test.cc:11: Failure 694Value of: Bar(n) 695Expected: 1 696 Actual: 2 697Google Test trace: 698path/to/foo_test.cc:17: A 699 700path/to/foo_test.cc:12: Failure 701Value of: Bar(n + 1) 702Expected: 2 703 Actual: 3 704``` 705 706Without the trace, it would've been difficult to know which invocation of 707`Sub1()` the two failures come from respectively. (You could add an extra 708message to each assertion in `Sub1()` to indicate the value of `n`, but that's 709tedious.) 710 711Some tips on using `SCOPED_TRACE`: 712 7131. With a suitable message, it's often enough to use `SCOPED_TRACE` at the 714 beginning of a sub-routine, instead of at each call site. 7152. When calling sub-routines inside a loop, make the loop iterator part of the 716 message in `SCOPED_TRACE` such that you can know which iteration the failure 717 is from. 7183. Sometimes the line number of the trace point is enough for identifying the 719 particular invocation of a sub-routine. In this case, you don't have to 720 choose a unique message for `SCOPED_TRACE`. You can simply use `""`. 7214. You can use `SCOPED_TRACE` in an inner scope when there is one in the outer 722 scope. In this case, all active trace points will be included in the failure 723 messages, in reverse order they are encountered. 7245. The trace dump is clickable in Emacs - hit `return` on a line number and 725 you'll be taken to that line in the source file! 726 727### Propagating Fatal Failures 728 729A common pitfall when using `ASSERT_*` and `FAIL*` is not understanding that 730when they fail they only abort the _current function_, not the entire test. For 731example, the following test will segfault: 732 733```c++ 734void Subroutine() { 735 // Generates a fatal failure and aborts the current function. 736 ASSERT_EQ(1, 2); 737 738 // The following won't be executed. 739 ... 740} 741 742TEST(FooTest, Bar) { 743 Subroutine(); // The intended behavior is for the fatal failure 744 // in Subroutine() to abort the entire test. 745 746 // The actual behavior: the function goes on after Subroutine() returns. 747 int* p = nullptr; 748 *p = 3; // Segfault! 749} 750``` 751 752To alleviate this, GoogleTest provides three different solutions. You could use 753either exceptions, the `(ASSERT|EXPECT)_NO_FATAL_FAILURE` assertions or the 754`HasFatalFailure()` function. They are described in the following two 755subsections. 756 757#### Asserting on Subroutines with an exception 758 759The following code can turn ASSERT-failure into an exception: 760 761```c++ 762class ThrowListener : public testing::EmptyTestEventListener { 763 void OnTestPartResult(const testing::TestPartResult& result) override { 764 if (result.type() == testing::TestPartResult::kFatalFailure) { 765 throw testing::AssertionException(result); 766 } 767 } 768}; 769int main(int argc, char** argv) { 770 ... 771 testing::UnitTest::GetInstance()->listeners().Append(new ThrowListener); 772 return RUN_ALL_TESTS(); 773} 774``` 775 776This listener should be added after other listeners if you have any, otherwise 777they won't see failed `OnTestPartResult`. 778 779#### Asserting on Subroutines 780 781As shown above, if your test calls a subroutine that has an `ASSERT_*` failure 782in it, the test will continue after the subroutine returns. This may not be what 783you want. 784 785Often people want fatal failures to propagate like exceptions. For that 786GoogleTest offers the following macros: 787 788Fatal assertion | Nonfatal assertion | Verifies 789------------------------------------- | ------------------------------------- | -------- 790`ASSERT_NO_FATAL_FAILURE(statement);` | `EXPECT_NO_FATAL_FAILURE(statement);` | `statement` doesn't generate any new fatal failures in the current thread. 791 792Only failures in the thread that executes the assertion are checked to determine 793the result of this type of assertions. If `statement` creates new threads, 794failures in these threads are ignored. 795 796Examples: 797 798```c++ 799ASSERT_NO_FATAL_FAILURE(Foo()); 800 801int i; 802EXPECT_NO_FATAL_FAILURE({ 803 i = Bar(); 804}); 805``` 806 807Assertions from multiple threads are currently not supported on Windows. 808 809#### Checking for Failures in the Current Test 810 811`HasFatalFailure()` in the `::testing::Test` class returns `true` if an 812assertion in the current test has suffered a fatal failure. This allows 813functions to catch fatal failures in a sub-routine and return early. 814 815```c++ 816class Test { 817 public: 818 ... 819 static bool HasFatalFailure(); 820}; 821``` 822 823The typical usage, which basically simulates the behavior of a thrown exception, 824is: 825 826```c++ 827TEST(FooTest, Bar) { 828 Subroutine(); 829 // Aborts if Subroutine() had a fatal failure. 830 if (HasFatalFailure()) return; 831 832 // The following won't be executed. 833 ... 834} 835``` 836 837If `HasFatalFailure()` is used outside of `TEST()` , `TEST_F()` , or a test 838fixture, you must add the `::testing::Test::` prefix, as in: 839 840```c++ 841if (testing::Test::HasFatalFailure()) return; 842``` 843 844Similarly, `HasNonfatalFailure()` returns `true` if the current test has at 845least one non-fatal failure, and `HasFailure()` returns `true` if the current 846test has at least one failure of either kind. 847 848## Logging Additional Information 849 850In your test code, you can call `RecordProperty("key", value)` to log additional 851information, where `value` can be either a string or an `int`. The *last* value 852recorded for a key will be emitted to the 853[XML output](#generating-an-xml-report) if you specify one. For example, the 854test 855 856```c++ 857TEST_F(WidgetUsageTest, MinAndMaxWidgets) { 858 RecordProperty("MaximumWidgets", ComputeMaxUsage()); 859 RecordProperty("MinimumWidgets", ComputeMinUsage()); 860} 861``` 862 863will output XML like this: 864 865```xml 866 ... 867 <testcase name="MinAndMaxWidgets" file="test.cpp" line="1" status="run" time="0.006" classname="WidgetUsageTest" MaximumWidgets="12" MinimumWidgets="9" /> 868 ... 869``` 870 871{: .callout .note} 872> NOTE: 873> 874> * `RecordProperty()` is a static member of the `Test` class. Therefore it 875> needs to be prefixed with `::testing::Test::` if used outside of the 876> `TEST` body and the test fixture class. 877> * *`key`* must be a valid XML attribute name, and cannot conflict with the 878> ones already used by GoogleTest (`name`, `status`, `time`, `classname`, 879> `type_param`, and `value_param`). 880> * Calling `RecordProperty()` outside of the lifespan of a test is allowed. 881> If it's called outside of a test but between a test suite's 882> `SetUpTestSuite()` and `TearDownTestSuite()` methods, it will be 883> attributed to the XML element for the test suite. If it's called outside 884> of all test suites (e.g. in a test environment), it will be attributed to 885> the top-level XML element. 886 887## Sharing Resources Between Tests in the Same Test Suite 888 889GoogleTest creates a new test fixture object for each test in order to make 890tests independent and easier to debug. However, sometimes tests use resources 891that are expensive to set up, making the one-copy-per-test model prohibitively 892expensive. 893 894If the tests don't change the resource, there's no harm in their sharing a 895single resource copy. So, in addition to per-test set-up/tear-down, GoogleTest 896also supports per-test-suite set-up/tear-down. To use it: 897 8981. In your test fixture class (say `FooTest` ), declare as `static` some member 899 variables to hold the shared resources. 9002. Outside your test fixture class (typically just below it), define those 901 member variables, optionally giving them initial values. 9023. In the same test fixture class, define a public member function `static void 903 SetUpTestSuite()` (remember not to spell it as **`SetupTestSuite`** with a 904 small `u`!) to set up the shared resources and a `static void 905 TearDownTestSuite()` function to tear them down. 906 907That's it! GoogleTest automatically calls `SetUpTestSuite()` before running the 908*first test* in the `FooTest` test suite (i.e. before creating the first 909`FooTest` object), and calls `TearDownTestSuite()` after running the *last test* 910in it (i.e. after deleting the last `FooTest` object). In between, the tests can 911use the shared resources. 912 913Remember that the test order is undefined, so your code can't depend on a test 914preceding or following another. Also, the tests must either not modify the state 915of any shared resource, or, if they do modify the state, they must restore the 916state to its original value before passing control to the next test. 917 918Note that `SetUpTestSuite()` may be called multiple times for a test fixture 919class that has derived classes, so you should not expect code in the function 920body to be run only once. Also, derived classes still have access to shared 921resources defined as static members, so careful consideration is needed when 922managing shared resources to avoid memory leaks if shared resources are not 923properly cleaned up in `TearDownTestSuite()`. 924 925Here's an example of per-test-suite set-up and tear-down: 926 927```c++ 928class FooTest : public testing::Test { 929 protected: 930 // Per-test-suite set-up. 931 // Called before the first test in this test suite. 932 // Can be omitted if not needed. 933 static void SetUpTestSuite() { 934 shared_resource_ = new ...; 935 936 // If `shared_resource_` is **not deleted** in `TearDownTestSuite()`, 937 // reallocation should be prevented because `SetUpTestSuite()` may be called 938 // in subclasses of FooTest and lead to memory leak. 939 // 940 // if (shared_resource_ == nullptr) { 941 // shared_resource_ = new ...; 942 // } 943 } 944 945 // Per-test-suite tear-down. 946 // Called after the last test in this test suite. 947 // Can be omitted if not needed. 948 static void TearDownTestSuite() { 949 delete shared_resource_; 950 shared_resource_ = nullptr; 951 } 952 953 // You can define per-test set-up logic as usual. 954 void SetUp() override { ... } 955 956 // You can define per-test tear-down logic as usual. 957 void TearDown() override { ... } 958 959 // Some expensive resource shared by all tests. 960 static T* shared_resource_; 961}; 962 963T* FooTest::shared_resource_ = nullptr; 964 965TEST_F(FooTest, Test1) { 966 ... you can refer to shared_resource_ here ... 967} 968 969TEST_F(FooTest, Test2) { 970 ... you can refer to shared_resource_ here ... 971} 972``` 973 974{: .callout .note} 975NOTE: Though the above code declares `SetUpTestSuite()` protected, it may 976sometimes be necessary to declare it public, such as when using it with 977`TEST_P`. 978 979## Global Set-Up and Tear-Down 980 981Just as you can do set-up and tear-down at the test level and the test suite 982level, you can also do it at the test program level. Here's how. 983 984First, you subclass the `::testing::Environment` class to define a test 985environment, which knows how to set-up and tear-down: 986 987```c++ 988class Environment : public ::testing::Environment { 989 public: 990 ~Environment() override {} 991 992 // Override this to define how to set up the environment. 993 void SetUp() override {} 994 995 // Override this to define how to tear down the environment. 996 void TearDown() override {} 997}; 998``` 999 1000Then, you register an instance of your environment class with GoogleTest by 1001calling the `::testing::AddGlobalTestEnvironment()` function: 1002 1003```c++ 1004Environment* AddGlobalTestEnvironment(Environment* env); 1005``` 1006 1007Now, when `RUN_ALL_TESTS()` is invoked, it first calls the `SetUp()` method. The 1008tests are then executed, provided that none of the environments have reported 1009fatal failures and `GTEST_SKIP()` has not been invoked. Finally, `TearDown()` is 1010called. 1011 1012Note that `SetUp()` and `TearDown()` are only invoked if there is at least one 1013test to be performed. Importantly, `TearDown()` is executed even if the test is 1014not run due to a fatal failure or `GTEST_SKIP()`. 1015 1016Calling `SetUp()` and `TearDown()` for each iteration depends on the flag 1017`gtest_recreate_environments_when_repeating`. `SetUp()` and `TearDown()` are 1018called for each environment object when the object is recreated for each 1019iteration. However, if test environments are not recreated for each iteration, 1020`SetUp()` is called only on the first iteration, and `TearDown()` is called only 1021on the last iteration. 1022 1023It's OK to register multiple environment objects. In this suite, their `SetUp()` 1024will be called in the order they are registered, and their `TearDown()` will be 1025called in the reverse order. 1026 1027Note that GoogleTest takes ownership of the registered environment objects. 1028Therefore **do not delete them** by yourself. 1029 1030You should call `AddGlobalTestEnvironment()` before `RUN_ALL_TESTS()` is called, 1031probably in `main()`. If you use `gtest_main`, you need to call this before 1032`main()` starts for it to take effect. One way to do this is to define a global 1033variable like this: 1034 1035```c++ 1036testing::Environment* const foo_env = 1037 testing::AddGlobalTestEnvironment(new FooEnvironment); 1038``` 1039 1040However, we strongly recommend you to write your own `main()` and call 1041`AddGlobalTestEnvironment()` there, as relying on initialization of global 1042variables makes the code harder to read and may cause problems when you register 1043multiple environments from different translation units and the environments have 1044dependencies among them (remember that the compiler doesn't guarantee the order 1045in which global variables from different translation units are initialized). 1046 1047## Value-Parameterized Tests 1048 1049*Value-parameterized tests* allow you to test your code with different 1050parameters without writing multiple copies of the same test. This is useful in a 1051number of situations, for example: 1052 1053* You have a piece of code whose behavior is affected by one or more 1054 command-line flags. You want to make sure your code performs correctly for 1055 various values of those flags. 1056* You want to test different implementations of an OO interface. 1057* You want to test your code over various inputs (a.k.a. data-driven testing). 1058 This feature is easy to abuse, so please exercise your good sense when doing 1059 it! 1060 1061### How to Write Value-Parameterized Tests 1062 1063To write value-parameterized tests, first you should define a fixture class. It 1064must be derived from both `testing::Test` and `testing::WithParamInterface<T>` 1065(the latter is a pure interface), where `T` is the type of your parameter 1066values. For convenience, you can just derive the fixture class from 1067`testing::TestWithParam<T>`, which itself is derived from both `testing::Test` 1068and `testing::WithParamInterface<T>`. `T` can be any copyable type. If it's a 1069raw pointer, you are responsible for managing the lifespan of the pointed 1070values. 1071 1072{: .callout .note} 1073NOTE: If your test fixture defines `SetUpTestSuite()` or `TearDownTestSuite()` 1074they must be declared **public** rather than **protected** in order to use 1075`TEST_P`. 1076 1077```c++ 1078class FooTest : 1079 public testing::TestWithParam<absl::string_view> { 1080 // You can implement all the usual fixture class members here. 1081 // To access the test parameter, call GetParam() from class 1082 // TestWithParam<T>. 1083}; 1084 1085// Or, when you want to add parameters to a pre-existing fixture class: 1086class BaseTest : public testing::Test { 1087 ... 1088}; 1089class BarTest : public BaseTest, 1090 public testing::WithParamInterface<absl::string_view> { 1091 ... 1092}; 1093``` 1094 1095Then, use the `TEST_P` macro to define as many test patterns using this fixture 1096as you want. The `_P` suffix is for "parameterized" or "pattern", whichever you 1097prefer to think. 1098 1099```c++ 1100TEST_P(FooTest, DoesBlah) { 1101 // Inside a test, access the test parameter with the GetParam() method 1102 // of the TestWithParam<T> class: 1103 EXPECT_TRUE(foo.Blah(GetParam())); 1104 ... 1105} 1106 1107TEST_P(FooTest, HasBlahBlah) { 1108 ... 1109} 1110``` 1111 1112Finally, you can use the `INSTANTIATE_TEST_SUITE_P` macro to instantiate the 1113test suite with any set of parameters you want. GoogleTest defines a number of 1114functions for generating test parameters—see details at 1115[`INSTANTIATE_TEST_SUITE_P`](reference/testing.md#INSTANTIATE_TEST_SUITE_P) in 1116the Testing Reference. 1117 1118For example, the following statement will instantiate tests from the `FooTest` 1119test suite each with parameter values `"meeny"`, `"miny"`, and `"moe"` using the 1120[`Values`](reference/testing.md#param-generators) parameter generator: 1121 1122```c++ 1123INSTANTIATE_TEST_SUITE_P(MeenyMinyMoe, 1124 FooTest, 1125 testing::Values("meeny", "miny", "moe")); 1126``` 1127 1128{: .callout .note} 1129NOTE: The code above must be placed at global or namespace scope, not at 1130function scope. 1131 1132The first argument to `INSTANTIATE_TEST_SUITE_P` is a unique name for the 1133instantiation of the test suite. The next argument is the name of the test 1134pattern, and the last is the 1135[parameter generator](reference/testing.md#param-generators). 1136 1137The parameter generator expression is not evaluated until GoogleTest is 1138initialized (via `InitGoogleTest()`). Any prior initialization done in the 1139`main` function will be accessible from the parameter generator, for example, 1140the results of flag parsing. 1141 1142You can instantiate a test pattern more than once, so to distinguish different 1143instances of the pattern, the instantiation name is added as a prefix to the 1144actual test suite name. Remember to pick unique prefixes for different 1145instantiations. The tests from the instantiation above will have these names: 1146 1147* `MeenyMinyMoe/FooTest.DoesBlah/0` for `"meeny"` 1148* `MeenyMinyMoe/FooTest.DoesBlah/1` for `"miny"` 1149* `MeenyMinyMoe/FooTest.DoesBlah/2` for `"moe"` 1150* `MeenyMinyMoe/FooTest.HasBlahBlah/0` for `"meeny"` 1151* `MeenyMinyMoe/FooTest.HasBlahBlah/1` for `"miny"` 1152* `MeenyMinyMoe/FooTest.HasBlahBlah/2` for `"moe"` 1153 1154You can use these names in [`--gtest_filter`](#running-a-subset-of-the-tests). 1155 1156The following statement will instantiate all tests from `FooTest` again, each 1157with parameter values `"cat"` and `"dog"` using the 1158[`ValuesIn`](reference/testing.md#param-generators) parameter generator: 1159 1160```c++ 1161constexpr absl::string_view kPets[] = {"cat", "dog"}; 1162INSTANTIATE_TEST_SUITE_P(Pets, FooTest, testing::ValuesIn(kPets)); 1163``` 1164 1165The tests from the instantiation above will have these names: 1166 1167* `Pets/FooTest.DoesBlah/0` for `"cat"` 1168* `Pets/FooTest.DoesBlah/1` for `"dog"` 1169* `Pets/FooTest.HasBlahBlah/0` for `"cat"` 1170* `Pets/FooTest.HasBlahBlah/1` for `"dog"` 1171 1172Please note that `INSTANTIATE_TEST_SUITE_P` will instantiate *all* tests in the 1173given test suite, whether their definitions come before or *after* the 1174`INSTANTIATE_TEST_SUITE_P` statement. 1175 1176Additionally, by default, every `TEST_P` without a corresponding 1177`INSTANTIATE_TEST_SUITE_P` causes a failing test in test suite 1178`GoogleTestVerification`. If you have a test suite where that omission is not an 1179error, for example it is in a library that may be linked in for other reasons or 1180where the list of test cases is dynamic and may be empty, then this check can be 1181suppressed by tagging the test suite: 1182 1183```c++ 1184GTEST_ALLOW_UNINSTANTIATED_PARAMETERIZED_TEST(FooTest); 1185``` 1186 1187You can see [sample7_unittest.cc] and [sample8_unittest.cc] for more examples. 1188 1189[sample7_unittest.cc]: https://github.com/google/googletest/blob/main/googletest/samples/sample7_unittest.cc "Parameterized Test example" 1190[sample8_unittest.cc]: https://github.com/google/googletest/blob/main/googletest/samples/sample8_unittest.cc "Parameterized Test example with multiple parameters" 1191 1192### Creating Value-Parameterized Abstract Tests 1193 1194In the above, we define and instantiate `FooTest` in the *same* source file. 1195Sometimes you may want to define value-parameterized tests in a library and let 1196other people instantiate them later. This pattern is known as *abstract tests*. 1197As an example of its application, when you are designing an interface you can 1198write a standard suite of abstract tests (perhaps using a factory function as 1199the test parameter) that all implementations of the interface are expected to 1200pass. When someone implements the interface, they can instantiate your suite to 1201get all the interface-conformance tests for free. 1202 1203To define abstract tests, you should organize your code like this: 1204 12051. Put the definition of the parameterized test fixture class (e.g. `FooTest`) 1206 in a header file, say `foo_param_test.h`. Think of this as *declaring* your 1207 abstract tests. 12082. Put the `TEST_P` definitions in `foo_param_test.cc`, which includes 1209 `foo_param_test.h`. Think of this as *implementing* your abstract tests. 1210 1211Once they are defined, you can instantiate them by including `foo_param_test.h`, 1212invoking `INSTANTIATE_TEST_SUITE_P()`, and depending on the library target that 1213contains `foo_param_test.cc`. You can instantiate the same abstract test suite 1214multiple times, possibly in different source files. 1215 1216### Specifying Names for Value-Parameterized Test Parameters 1217 1218The optional last argument to `INSTANTIATE_TEST_SUITE_P()` allows the user to 1219specify a function or functor that generates custom test name suffixes based on 1220the test parameters. The function should accept one argument of type 1221`testing::TestParamInfo<class ParamType>`, and return `std::string`. 1222 1223`testing::PrintToStringParamName` is a builtin test suffix generator that 1224returns the value of `testing::PrintToString(GetParam())`. It does not work for 1225`std::string` or C strings. 1226 1227{: .callout .note} 1228NOTE: test names must be non-empty, unique, and may only contain ASCII 1229alphanumeric characters. In particular, they 1230[should not contain underscores](faq.md#why-should-test-suite-names-and-test-names-not-contain-underscore) 1231 1232```c++ 1233class MyTestSuite : public testing::TestWithParam<int> {}; 1234 1235TEST_P(MyTestSuite, MyTest) 1236{ 1237 std::cout << "Example Test Param: " << GetParam() << std::endl; 1238} 1239 1240INSTANTIATE_TEST_SUITE_P(MyGroup, MyTestSuite, testing::Range(0, 10), 1241 testing::PrintToStringParamName()); 1242``` 1243 1244Providing a custom functor allows for more control over test parameter name 1245generation, especially for types where the automatic conversion does not 1246generate helpful parameter names (e.g. strings as demonstrated above). The 1247following example illustrates this for multiple parameters, an enumeration type 1248and a string, and also demonstrates how to combine generators. It uses a lambda 1249for conciseness: 1250 1251```c++ 1252enum class MyType { MY_FOO = 0, MY_BAR = 1 }; 1253 1254class MyTestSuite : public testing::TestWithParam<std::tuple<MyType, std::string>> { 1255}; 1256 1257INSTANTIATE_TEST_SUITE_P( 1258 MyGroup, MyTestSuite, 1259 testing::Combine( 1260 testing::Values(MyType::MY_FOO, MyType::MY_BAR), 1261 testing::Values("A", "B")), 1262 [](const testing::TestParamInfo<MyTestSuite::ParamType>& info) { 1263 std::string name = absl::StrCat( 1264 std::get<0>(info.param) == MyType::MY_FOO ? "Foo" : "Bar", 1265 std::get<1>(info.param)); 1266 absl::c_replace_if(name, [](char c) { return !std::isalnum(c); }, '_'); 1267 return name; 1268 }); 1269``` 1270 1271## Typed Tests 1272 1273Suppose you have multiple implementations of the same interface and want to make 1274sure that all of them satisfy some common requirements. Or, you may have defined 1275several types that are supposed to conform to the same "concept" and you want to 1276verify it. In both cases, you want the same test logic repeated for different 1277types. 1278 1279While you can write one `TEST` or `TEST_F` for each type you want to test (and 1280you may even factor the test logic into a function template that you invoke from 1281the `TEST`), it's tedious and doesn't scale: if you want `m` tests over `n` 1282types, you'll end up writing `m*n` `TEST`s. 1283 1284*Typed tests* allow you to repeat the same test logic over a list of types. You 1285only need to write the test logic once, although you must know the type list 1286when writing typed tests. Here's how you do it: 1287 1288First, define a fixture class template. It should be parameterized by a type. 1289Remember to derive it from `::testing::Test`: 1290 1291```c++ 1292template <typename T> 1293class FooTest : public testing::Test { 1294 public: 1295 ... 1296 using List = std::list<T>; 1297 static T shared_; 1298 T value_; 1299}; 1300``` 1301 1302Next, associate a list of types with the test suite, which will be repeated for 1303each type in the list: 1304 1305```c++ 1306using MyTypes = ::testing::Types<char, int, unsigned int>; 1307TYPED_TEST_SUITE(FooTest, MyTypes); 1308``` 1309 1310The type alias (`using` or `typedef`) is necessary for the `TYPED_TEST_SUITE` 1311macro to parse correctly. Otherwise the compiler will think that each comma in 1312the type list introduces a new macro argument. 1313 1314Then, use `TYPED_TEST()` instead of `TEST_F()` to define a typed test for this 1315test suite. You can repeat this as many times as you want: 1316 1317```c++ 1318TYPED_TEST(FooTest, DoesBlah) { 1319 // Inside a test, refer to the special name TypeParam to get the type 1320 // parameter. Since we are inside a derived class template, C++ requires 1321 // us to visit the members of FooTest via 'this'. 1322 TypeParam n = this->value_; 1323 1324 // To visit static members of the fixture, add the 'TestFixture::' 1325 // prefix. 1326 n += TestFixture::shared_; 1327 1328 // To refer to typedefs in the fixture, add the 'typename TestFixture::' 1329 // prefix. The 'typename' is required to satisfy the compiler. 1330 typename TestFixture::List values; 1331 1332 values.push_back(n); 1333 ... 1334} 1335 1336TYPED_TEST(FooTest, HasPropertyA) { ... } 1337``` 1338 1339You can see [sample6_unittest.cc] for a complete example. 1340 1341[sample6_unittest.cc]: https://github.com/google/googletest/blob/main/googletest/samples/sample6_unittest.cc "Typed Test example" 1342 1343## Type-Parameterized Tests 1344 1345*Type-parameterized tests* are like typed tests, except that they don't require 1346you to know the list of types ahead of time. Instead, you can define the test 1347logic first and instantiate it with different type lists later. You can even 1348instantiate it more than once in the same program. 1349 1350If you are designing an interface or concept, you can define a suite of 1351type-parameterized tests to verify properties that any valid implementation of 1352the interface/concept should have. Then, the author of each implementation can 1353just instantiate the test suite with their type to verify that it conforms to 1354the requirements, without having to write similar tests repeatedly. Here's an 1355example: 1356 1357First, define a fixture class template, as we did with typed tests: 1358 1359```c++ 1360template <typename T> 1361class FooTest : public testing::Test { 1362 void DoSomethingInteresting(); 1363 ... 1364}; 1365``` 1366 1367Next, declare that you will define a type-parameterized test suite: 1368 1369```c++ 1370TYPED_TEST_SUITE_P(FooTest); 1371``` 1372 1373Then, use `TYPED_TEST_P()` to define a type-parameterized test. You can repeat 1374this as many times as you want: 1375 1376```c++ 1377TYPED_TEST_P(FooTest, DoesBlah) { 1378 // Inside a test, refer to TypeParam to get the type parameter. 1379 TypeParam n = 0; 1380 1381 // You will need to use `this` explicitly to refer to fixture members. 1382 this->DoSomethingInteresting() 1383 ... 1384} 1385 1386TYPED_TEST_P(FooTest, HasPropertyA) { ... } 1387``` 1388 1389Now the tricky part: you need to register all test patterns using the 1390`REGISTER_TYPED_TEST_SUITE_P` macro before you can instantiate them. The first 1391argument of the macro is the test suite name; the rest are the names of the 1392tests in this test suite: 1393 1394```c++ 1395REGISTER_TYPED_TEST_SUITE_P(FooTest, 1396 DoesBlah, HasPropertyA); 1397``` 1398 1399Finally, you are free to instantiate the pattern with the types you want. If you 1400put the above code in a header file, you can `#include` it in multiple C++ 1401source files and instantiate it multiple times. 1402 1403```c++ 1404using MyTypes = ::testing::Types<char, int, unsigned int>; 1405INSTANTIATE_TYPED_TEST_SUITE_P(My, FooTest, MyTypes); 1406``` 1407 1408To distinguish different instances of the pattern, the first argument to the 1409`INSTANTIATE_TYPED_TEST_SUITE_P` macro is a prefix that will be added to the 1410actual test suite name. Remember to pick unique prefixes for different 1411instances. 1412 1413In the special case where the type list contains only one type, you can write 1414that type directly without `::testing::Types<...>`, like this: 1415 1416```c++ 1417INSTANTIATE_TYPED_TEST_SUITE_P(My, FooTest, int); 1418``` 1419 1420You can see [sample6_unittest.cc] for a complete example. 1421 1422## Testing Private Code 1423 1424If you change your software's internal implementation, your tests should not 1425break as long as the change is not observable by users. Therefore, **per the 1426black-box testing principle, most of the time you should test your code through 1427its public interfaces.** 1428 1429**If you still find yourself needing to test internal implementation code, 1430consider if there's a better design.** The desire to test internal 1431implementation is often a sign that the class is doing too much. Consider 1432extracting an implementation class, and testing it. Then use that implementation 1433class in the original class. 1434 1435If you absolutely have to test non-public interface code though, you can. There 1436are two cases to consider: 1437 1438* Static functions ( *not* the same as static member functions!) or unnamed 1439 namespaces, and 1440* Private or protected class members 1441 1442To test them, we use the following special techniques: 1443 1444* Both static functions and definitions/declarations in an unnamed namespace 1445 are only visible within the same translation unit. To test them, you can 1446 `#include` the entire `.cc` file being tested in your `*_test.cc` file. 1447 (#including `.cc` files is not a good way to reuse code - you should not do 1448 this in production code!) 1449 1450 However, a better approach is to move the private code into the 1451 `foo::internal` namespace, where `foo` is the namespace your project 1452 normally uses, and put the private declarations in a `*-internal.h` file. 1453 Your production `.cc` files and your tests are allowed to include this 1454 internal header, but your clients are not. This way, you can fully test your 1455 internal implementation without leaking it to your clients. 1456 1457* Private class members are only accessible from within the class or by 1458 friends. To access a class' private members, you can declare your test 1459 fixture as a friend to the class and define accessors in your fixture. Tests 1460 using the fixture can then access the private members of your production 1461 class via the accessors in the fixture. Note that even though your fixture 1462 is a friend to your production class, your tests are not automatically 1463 friends to it, as they are technically defined in sub-classes of the 1464 fixture. 1465 1466 Another way to test private members is to refactor them into an 1467 implementation class, which is then declared in a `*-internal.h` file. Your 1468 clients aren't allowed to include this header but your tests can. Such is 1469 called the 1470 [Pimpl](https://www.gamedev.net/articles/programming/general-and-gameplay-programming/the-c-pimpl-r1794/) 1471 (Private Implementation) idiom. 1472 1473 Or, you can declare an individual test as a friend of your class by adding 1474 this line in the class body: 1475 1476 ```c++ 1477 FRIEND_TEST(TestSuiteName, TestName); 1478 ``` 1479 1480 For example, 1481 1482 ```c++ 1483 // foo.h 1484 class Foo { 1485 ... 1486 private: 1487 FRIEND_TEST(FooTest, BarReturnsZeroOnNull); 1488 1489 int Bar(void* x); 1490 }; 1491 1492 // foo_test.cc 1493 ... 1494 TEST(FooTest, BarReturnsZeroOnNull) { 1495 Foo foo; 1496 EXPECT_EQ(foo.Bar(NULL), 0); // Uses Foo's private member Bar(). 1497 } 1498 ``` 1499 1500 Pay special attention when your class is defined in a namespace. If you want 1501 your test fixtures and tests to be friends of your class, then they must be 1502 defined in the exact same namespace (no anonymous or inline namespaces). 1503 1504 For example, if the code to be tested looks like: 1505 1506 ```c++ 1507 namespace my_namespace { 1508 1509 class Foo { 1510 friend class FooTest; 1511 FRIEND_TEST(FooTest, Bar); 1512 FRIEND_TEST(FooTest, Baz); 1513 ... definition of the class Foo ... 1514 }; 1515 1516 } // namespace my_namespace 1517 ``` 1518 1519 Your test code should be something like: 1520 1521 ```c++ 1522 namespace my_namespace { 1523 1524 class FooTest : public testing::Test { 1525 protected: 1526 ... 1527 }; 1528 1529 TEST_F(FooTest, Bar) { ... } 1530 TEST_F(FooTest, Baz) { ... } 1531 1532 } // namespace my_namespace 1533 ``` 1534 1535## "Catching" Failures 1536 1537If you are building a testing utility on top of GoogleTest, you'll want to test 1538your utility. What framework would you use to test it? GoogleTest, of course. 1539 1540The challenge is to verify that your testing utility reports failures correctly. 1541In frameworks that report a failure by throwing an exception, you could catch 1542the exception and assert on it. But GoogleTest doesn't use exceptions, so how do 1543we test that a piece of code generates an expected failure? 1544 1545`"gtest/gtest-spi.h"` contains some constructs to do this. 1546After #including this header, you can use 1547 1548```c++ 1549 EXPECT_FATAL_FAILURE(statement, substring); 1550``` 1551 1552to assert that `statement` generates a fatal (e.g. `ASSERT_*`) failure in the 1553current thread whose message contains the given `substring`, or use 1554 1555```c++ 1556 EXPECT_NONFATAL_FAILURE(statement, substring); 1557``` 1558 1559if you are expecting a non-fatal (e.g. `EXPECT_*`) failure. 1560 1561Only failures in the current thread are checked to determine the result of this 1562type of expectations. If `statement` creates new threads, failures in these 1563threads are also ignored. If you want to catch failures in other threads as 1564well, use one of the following macros instead: 1565 1566```c++ 1567 EXPECT_FATAL_FAILURE_ON_ALL_THREADS(statement, substring); 1568 EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(statement, substring); 1569``` 1570 1571{: .callout .note} 1572NOTE: Assertions from multiple threads are currently not supported on Windows. 1573 1574For technical reasons, there are some caveats: 1575 15761. You cannot stream a failure message to either macro. 1577 15782. `statement` in `EXPECT_FATAL_FAILURE{_ON_ALL_THREADS}()` cannot reference 1579 local non-static variables or non-static members of `this` object. 1580 15813. `statement` in `EXPECT_FATAL_FAILURE{_ON_ALL_THREADS}()` cannot return a 1582 value. 1583 1584## Registering tests programmatically 1585 1586The `TEST` macros handle the vast majority of all use cases, but there are few 1587where runtime registration logic is required. For those cases, the framework 1588provides the `::testing::RegisterTest` that allows callers to register arbitrary 1589tests dynamically. 1590 1591This is an advanced API only to be used when the `TEST` macros are insufficient. 1592The macros should be preferred when possible, as they avoid most of the 1593complexity of calling this function. 1594 1595It provides the following signature: 1596 1597```c++ 1598template <typename Factory> 1599TestInfo* RegisterTest(const char* test_suite_name, const char* test_name, 1600 const char* type_param, const char* value_param, 1601 const char* file, int line, Factory factory); 1602``` 1603 1604The `factory` argument is a factory callable (move-constructible) object or 1605function pointer that creates a new instance of the Test object. It handles 1606ownership to the caller. The signature of the callable is `Fixture*()`, where 1607`Fixture` is the test fixture class for the test. All tests registered with the 1608same `test_suite_name` must return the same fixture type. This is checked at 1609runtime. 1610 1611The framework will infer the fixture class from the factory and will call the 1612`SetUpTestSuite` and `TearDownTestSuite` for it. 1613 1614Must be called before `RUN_ALL_TESTS()` is invoked, otherwise behavior is 1615undefined. 1616 1617Use case example: 1618 1619```c++ 1620class MyFixture : public testing::Test { 1621 public: 1622 // All of these optional, just like in regular macro usage. 1623 static void SetUpTestSuite() { ... } 1624 static void TearDownTestSuite() { ... } 1625 void SetUp() override { ... } 1626 void TearDown() override { ... } 1627}; 1628 1629class MyTest : public MyFixture { 1630 public: 1631 explicit MyTest(int data) : data_(data) {} 1632 void TestBody() override { ... } 1633 1634 private: 1635 int data_; 1636}; 1637 1638void RegisterMyTests(const std::vector<int>& values) { 1639 for (int v : values) { 1640 testing::RegisterTest( 1641 "MyFixture", ("Test" + std::to_string(v)).c_str(), nullptr, 1642 std::to_string(v).c_str(), 1643 __FILE__, __LINE__, 1644 // Important to use the fixture type as the return type here. 1645 [=]() -> MyFixture* { return new MyTest(v); }); 1646 } 1647} 1648... 1649int main(int argc, char** argv) { 1650 testing::InitGoogleTest(&argc, argv); 1651 std::vector<int> values_to_test = LoadValuesFromConfig(); 1652 RegisterMyTests(values_to_test); 1653 ... 1654 return RUN_ALL_TESTS(); 1655} 1656``` 1657 1658## Getting the Current Test's Name 1659 1660Sometimes a function may need to know the name of the currently running test. 1661For example, you may be using the `SetUp()` method of your test fixture to set 1662the golden file name based on which test is running. The 1663[`TestInfo`](reference/testing.md#TestInfo) class has this information. 1664 1665To obtain a `TestInfo` object for the currently running test, call 1666`current_test_info()` on the [`UnitTest`](reference/testing.md#UnitTest) 1667singleton object: 1668 1669```c++ 1670 // Gets information about the currently running test. 1671 // Do NOT delete the returned object - it's managed by the UnitTest class. 1672 const testing::TestInfo* const test_info = 1673 testing::UnitTest::GetInstance()->current_test_info(); 1674 1675 printf("We are in test %s of test suite %s.\n", 1676 test_info->name(), 1677 test_info->test_suite_name()); 1678``` 1679 1680`current_test_info()` returns a null pointer if no test is running. In 1681particular, you cannot find the test suite name in `SetUpTestSuite()`, 1682`TearDownTestSuite()` (where you know the test suite name implicitly), or 1683functions called from them. 1684 1685## Extending GoogleTest by Handling Test Events 1686 1687GoogleTest provides an **event listener API** to let you receive notifications 1688about the progress of a test program and test failures. The events you can 1689listen to include the start and end of the test program, a test suite, or a test 1690method, among others. You may use this API to augment or replace the standard 1691console output, replace the XML output, or provide a completely different form 1692of output, such as a GUI or a database. You can also use test events as 1693checkpoints to implement a resource leak checker, for example. 1694 1695### Defining Event Listeners 1696 1697To define a event listener, you subclass either 1698[`testing::TestEventListener`](reference/testing.md#TestEventListener) or 1699[`testing::EmptyTestEventListener`](reference/testing.md#EmptyTestEventListener) 1700The former is an (abstract) interface, where *each pure virtual method can be 1701overridden to handle a test event* (For example, when a test starts, the 1702`OnTestStart()` method will be called.). The latter provides an empty 1703implementation of all methods in the interface, such that a subclass only needs 1704to override the methods it cares about. 1705 1706When an event is fired, its context is passed to the handler function as an 1707argument. The following argument types are used: 1708 1709* UnitTest reflects the state of the entire test program, 1710* TestSuite has information about a test suite, which can contain one or more 1711 tests, 1712* TestInfo contains the state of a test, and 1713* TestPartResult represents the result of a test assertion. 1714 1715An event handler function can examine the argument it receives to find out 1716interesting information about the event and the test program's state. 1717 1718Here's an example: 1719 1720```c++ 1721 class MinimalistPrinter : public testing::EmptyTestEventListener { 1722 // Called before a test starts. 1723 void OnTestStart(const testing::TestInfo& test_info) override { 1724 printf("*** Test %s.%s starting.\n", 1725 test_info.test_suite_name(), test_info.name()); 1726 } 1727 1728 // Called after a failed assertion or a SUCCESS(). 1729 void OnTestPartResult(const testing::TestPartResult& test_part_result) override { 1730 printf("%s in %s:%d\n%s\n", 1731 test_part_result.failed() ? "*** Failure" : "Success", 1732 test_part_result.file_name(), 1733 test_part_result.line_number(), 1734 test_part_result.summary()); 1735 } 1736 1737 // Called after a test ends. 1738 void OnTestEnd(const testing::TestInfo& test_info) override { 1739 printf("*** Test %s.%s ending.\n", 1740 test_info.test_suite_name(), test_info.name()); 1741 } 1742 }; 1743``` 1744 1745### Using Event Listeners 1746 1747To use the event listener you have defined, add an instance of it to the 1748GoogleTest event listener list (represented by class 1749[`TestEventListeners`](reference/testing.md#TestEventListeners) - note the "s" 1750at the end of the name) in your `main()` function, before calling 1751`RUN_ALL_TESTS()`: 1752 1753```c++ 1754int main(int argc, char** argv) { 1755 testing::InitGoogleTest(&argc, argv); 1756 // Gets hold of the event listener list. 1757 testing::TestEventListeners& listeners = 1758 testing::UnitTest::GetInstance()->listeners(); 1759 // Adds a listener to the end. GoogleTest takes the ownership. 1760 listeners.Append(new MinimalistPrinter); 1761 return RUN_ALL_TESTS(); 1762} 1763``` 1764 1765There's only one problem: the default test result printer is still in effect, so 1766its output will mingle with the output from your minimalist printer. To suppress 1767the default printer, just release it from the event listener list and delete it. 1768You can do so by adding one line: 1769 1770```c++ 1771 ... 1772 delete listeners.Release(listeners.default_result_printer()); 1773 listeners.Append(new MinimalistPrinter); 1774 return RUN_ALL_TESTS(); 1775``` 1776 1777Now, sit back and enjoy a completely different output from your tests. For more 1778details, see [sample9_unittest.cc]. 1779 1780[sample9_unittest.cc]: https://github.com/google/googletest/blob/main/googletest/samples/sample9_unittest.cc "Event listener example" 1781 1782You may append more than one listener to the list. When an `On*Start()` or 1783`OnTestPartResult()` event is fired, the listeners will receive it in the order 1784they appear in the list (since new listeners are added to the end of the list, 1785the default text printer and the default XML generator will receive the event 1786first). An `On*End()` event will be received by the listeners in the *reverse* 1787order. This allows output by listeners added later to be framed by output from 1788listeners added earlier. 1789 1790### Generating Failures in Listeners 1791 1792You may use failure-raising macros (`EXPECT_*()`, `ASSERT_*()`, `FAIL()`, etc) 1793when processing an event. There are some restrictions: 1794 17951. You cannot generate any failure in `OnTestPartResult()` (otherwise it will 1796 cause `OnTestPartResult()` to be called recursively). 17972. A listener that handles `OnTestPartResult()` is not allowed to generate any 1798 failure. 1799 1800When you add listeners to the listener list, you should put listeners that 1801handle `OnTestPartResult()` *before* listeners that can generate failures. This 1802ensures that failures generated by the latter are attributed to the right test 1803by the former. 1804 1805See [sample10_unittest.cc] for an example of a failure-raising listener. 1806 1807[sample10_unittest.cc]: https://github.com/google/googletest/blob/main/googletest/samples/sample10_unittest.cc "Failure-raising listener example" 1808 1809## Running Test Programs: Advanced Options 1810 1811GoogleTest test programs are ordinary executables. Once built, you can run them 1812directly and affect their behavior via the following environment variables 1813and/or command line flags. For the flags to work, your programs must call 1814`::testing::InitGoogleTest()` before calling `RUN_ALL_TESTS()`. 1815 1816To see a list of supported flags and their usage, please run your test program 1817with the `--help` flag. 1818 1819If an option is specified both by an environment variable and by a flag, the 1820latter takes precedence. 1821 1822### Selecting Tests 1823 1824#### Listing Test Names 1825 1826Sometimes it is necessary to list the available tests in a program before 1827running them so that a filter may be applied if needed. Including the flag 1828`--gtest_list_tests` overrides all other flags and lists tests in the following 1829format: 1830 1831```none 1832TestSuite1. 1833 TestName1 1834 TestName2 1835TestSuite2. 1836 TestName 1837``` 1838 1839None of the tests listed are actually run if the flag is provided. There is no 1840corresponding environment variable for this flag. 1841 1842#### Running a Subset of the Tests 1843 1844By default, a GoogleTest program runs all tests the user has defined. Sometimes, 1845you want to run only a subset of the tests (e.g. for debugging or quickly 1846verifying a change). If you set the `GTEST_FILTER` environment variable or the 1847`--gtest_filter` flag to a filter string, GoogleTest will only run the tests 1848whose full names (in the form of `TestSuiteName.TestName`) match the filter. 1849 1850The format of a filter is a '`:`'-separated list of wildcard patterns (called 1851the *positive patterns*) optionally followed by a '`-`' and another 1852'`:`'-separated pattern list (called the *negative patterns*). A test matches 1853the filter if and only if it matches any of the positive patterns but does not 1854match any of the negative patterns. 1855 1856A pattern may contain `'*'` (matches any string) or `'?'` (matches any single 1857character). For convenience, the filter `'*-NegativePatterns'` can be also 1858written as `'-NegativePatterns'`. 1859 1860For example: 1861 1862* `./foo_test` Has no flag, and thus runs all its tests. 1863* `./foo_test --gtest_filter=*` Also runs everything, due to the single 1864 match-everything `*` value. 1865* `./foo_test --gtest_filter=FooTest.*` Runs everything in test suite 1866 `FooTest` . 1867* `./foo_test --gtest_filter=*Null*:*Constructor*` Runs any test whose full 1868 name contains either `"Null"` or `"Constructor"` . 1869* `./foo_test --gtest_filter=-*DeathTest.*` Runs all non-death tests. 1870* `./foo_test --gtest_filter=FooTest.*-FooTest.Bar` Runs everything in test 1871 suite `FooTest` except `FooTest.Bar`. 1872* `./foo_test --gtest_filter=FooTest.*:BarTest.*-FooTest.Bar:BarTest.Foo` Runs 1873 everything in test suite `FooTest` except `FooTest.Bar` and everything in 1874 test suite `BarTest` except `BarTest.Foo`. 1875 1876#### Stop test execution upon first failure 1877 1878By default, a GoogleTest program runs all tests the user has defined. In some 1879cases (e.g. iterative test development & execution) it may be desirable stop 1880test execution upon first failure (trading improved latency for completeness). 1881If `GTEST_FAIL_FAST` environment variable or `--gtest_fail_fast` flag is set, 1882the test runner will stop execution as soon as the first test failure is found. 1883 1884#### Temporarily Disabling Tests 1885 1886If you have a broken test that you cannot fix right away, you can add the 1887`DISABLED_` prefix to its name. This will exclude it from execution. This is 1888better than commenting out the code or using `#if 0`, as disabled tests are 1889still compiled (and thus won't rot). 1890 1891If you need to disable all tests in a test suite, you can either add `DISABLED_` 1892to the front of the name of each test, or alternatively add it to the front of 1893the test suite name. 1894 1895For example, the following tests won't be run by GoogleTest, even though they 1896will still be compiled: 1897 1898```c++ 1899// Tests that Foo does Abc. 1900TEST(FooTest, DISABLED_DoesAbc) { ... } 1901 1902class DISABLED_BarTest : public testing::Test { ... }; 1903 1904// Tests that Bar does Xyz. 1905TEST_F(DISABLED_BarTest, DoesXyz) { ... } 1906``` 1907 1908{: .callout .note} 1909NOTE: This feature should only be used for temporary pain-relief. You still have 1910to fix the disabled tests at a later date. As a reminder, GoogleTest will print 1911a banner warning you if a test program contains any disabled tests. 1912 1913{: .callout .tip} 1914TIP: You can easily count the number of disabled tests you have using 1915`grep`. This number can be used as a metric for 1916improving your test quality. 1917 1918#### Temporarily Enabling Disabled Tests 1919 1920To include disabled tests in test execution, just invoke the test program with 1921the `--gtest_also_run_disabled_tests` flag or set the 1922`GTEST_ALSO_RUN_DISABLED_TESTS` environment variable to a value other than `0`. 1923You can combine this with the `--gtest_filter` flag to further select which 1924disabled tests to run. 1925 1926### Repeating the Tests 1927 1928Once in a while you'll run into a test whose result is hit-or-miss. Perhaps it 1929will fail only 1% of the time, making it rather hard to reproduce the bug under 1930a debugger. This can be a major source of frustration. 1931 1932The `--gtest_repeat` flag allows you to repeat all (or selected) test methods in 1933a program many times. Hopefully, a flaky test will eventually fail and give you 1934a chance to debug. Here's how to use it: 1935 1936```none 1937$ foo_test --gtest_repeat=1000 1938Repeat foo_test 1000 times and don't stop at failures. 1939 1940$ foo_test --gtest_repeat=-1 1941A negative count means repeating forever. 1942 1943$ foo_test --gtest_repeat=1000 --gtest_break_on_failure 1944Repeat foo_test 1000 times, stopping at the first failure. This 1945is especially useful when running under a debugger: when the test 1946fails, it will drop into the debugger and you can then inspect 1947variables and stacks. 1948 1949$ foo_test --gtest_repeat=1000 --gtest_filter=FooBar.* 1950Repeat the tests whose name matches the filter 1000 times. 1951``` 1952 1953If your test program contains 1954[global set-up/tear-down](#global-set-up-and-tear-down) code, it will be 1955repeated in each iteration as well, as the flakiness may be in it. To avoid 1956repeating global set-up/tear-down, specify 1957`--gtest_recreate_environments_when_repeating=false`{.nowrap}. 1958 1959You can also specify the repeat count by setting the `GTEST_REPEAT` environment 1960variable. 1961 1962### Shuffling the Tests 1963 1964You can specify the `--gtest_shuffle` flag (or set the `GTEST_SHUFFLE` 1965environment variable to `1`) to run the tests in a program in a random order. 1966This helps to reveal bad dependencies between tests. 1967 1968By default, GoogleTest uses a random seed calculated from the current time. 1969Therefore you'll get a different order every time. The console output includes 1970the random seed value, such that you can reproduce an order-related test failure 1971later. To specify the random seed explicitly, use the `--gtest_random_seed=SEED` 1972flag (or set the `GTEST_RANDOM_SEED` environment variable), where `SEED` is an 1973integer in the range [0, 99999]. The seed value 0 is special: it tells 1974GoogleTest to do the default behavior of calculating the seed from the current 1975time. 1976 1977If you combine this with `--gtest_repeat=N`, GoogleTest will pick a different 1978random seed and re-shuffle the tests in each iteration. 1979 1980### Distributing Test Functions to Multiple Machines 1981 1982If you have more than one machine you can use to run a test program, you might 1983want to run the test functions in parallel and get the result faster. We call 1984this technique *sharding*, where each machine is called a *shard*. 1985 1986GoogleTest is compatible with test sharding. To take advantage of this feature, 1987your test runner (not part of GoogleTest) needs to do the following: 1988 19891. Allocate a number of machines (shards) to run the tests. 19901. On each shard, set the `GTEST_TOTAL_SHARDS` environment variable to the total 1991 number of shards. It must be the same for all shards. 19921. On each shard, set the `GTEST_SHARD_INDEX` environment variable to the index 1993 of the shard. Different shards must be assigned different indices, which 1994 must be in the range `[0, GTEST_TOTAL_SHARDS - 1]`. 19951. Run the same test program on all shards. When GoogleTest sees the above two 1996 environment variables, it will select a subset of the test functions to run. 1997 Across all shards, each test function in the program will be run exactly 1998 once. 19991. Wait for all shards to finish, then collect and report the results. 2000 2001Your project may have tests that were written without GoogleTest and thus don't 2002understand this protocol. In order for your test runner to figure out which test 2003supports sharding, it can set the environment variable `GTEST_SHARD_STATUS_FILE` 2004to a non-existent file path. If a test program supports sharding, it will create 2005this file to acknowledge that fact; otherwise it will not create it. The actual 2006contents of the file are not important at this time, although we may put some 2007useful information in it in the future. 2008 2009Here's an example to make it clear. Suppose you have a test program `foo_test` 2010that contains the following 5 test functions: 2011 2012``` 2013TEST(A, V) 2014TEST(A, W) 2015TEST(B, X) 2016TEST(B, Y) 2017TEST(B, Z) 2018``` 2019 2020Suppose you have 3 machines at your disposal. To run the test functions in 2021parallel, you would set `GTEST_TOTAL_SHARDS` to 3 on all machines, and set 2022`GTEST_SHARD_INDEX` to 0, 1, and 2 on the machines respectively. Then you would 2023run the same `foo_test` on each machine. 2024 2025GoogleTest reserves the right to change how the work is distributed across the 2026shards, but here's one possible scenario: 2027 2028* Machine #0 runs `A.V` and `B.X`. 2029* Machine #1 runs `A.W` and `B.Y`. 2030* Machine #2 runs `B.Z`. 2031 2032### Controlling Test Output 2033 2034#### Colored Terminal Output 2035 2036GoogleTest can use colors in its terminal output to make it easier to spot the 2037important information: 2038 2039<pre>... 2040<font color="green">[----------]</font> 1 test from FooTest 2041<font color="green">[ RUN ]</font> FooTest.DoesAbc 2042<font color="green">[ OK ]</font> FooTest.DoesAbc 2043<font color="green">[----------]</font> 2 tests from BarTest 2044<font color="green">[ RUN ]</font> BarTest.HasXyzProperty 2045<font color="green">[ OK ]</font> BarTest.HasXyzProperty 2046<font color="green">[ RUN ]</font> BarTest.ReturnsTrueOnSuccess 2047... some error messages ... 2048<font color="red">[ FAILED ]</font> BarTest.ReturnsTrueOnSuccess 2049... 2050<font color="green">[==========]</font> 30 tests from 14 test suites ran. 2051<font color="green">[ PASSED ]</font> 28 tests. 2052<font color="red">[ FAILED ]</font> 2 tests, listed below: 2053<font color="red">[ FAILED ]</font> BarTest.ReturnsTrueOnSuccess 2054<font color="red">[ FAILED ]</font> AnotherTest.DoesXyz 2055 2056 2 FAILED TESTS 2057</pre> 2058 2059You can set the `GTEST_COLOR` environment variable or the `--gtest_color` 2060command line flag to `yes`, `no`, or `auto` (the default) to enable colors, 2061disable colors, or let GoogleTest decide. When the value is `auto`, GoogleTest 2062will use colors if and only if the output goes to a terminal and (on non-Windows 2063platforms) the `TERM` environment variable is set to `xterm` or `xterm-color`. 2064 2065#### Suppressing test passes 2066 2067By default, GoogleTest prints 1 line of output for each test, indicating if it 2068passed or failed. To show only test failures, run the test program with 2069`--gtest_brief=1`, or set the GTEST_BRIEF environment variable to `1`. 2070 2071#### Suppressing the Elapsed Time 2072 2073By default, GoogleTest prints the time it takes to run each test. To disable 2074that, run the test program with the `--gtest_print_time=0` command line flag, or 2075set the GTEST_PRINT_TIME environment variable to `0`. 2076 2077#### Suppressing UTF-8 Text Output 2078 2079In case of assertion failures, GoogleTest prints expected and actual values of 2080type `string` both as hex-encoded strings as well as in readable UTF-8 text if 2081they contain valid non-ASCII UTF-8 characters. If you want to suppress the UTF-8 2082text because, for example, you don't have an UTF-8 compatible output medium, run 2083the test program with `--gtest_print_utf8=0` or set the `GTEST_PRINT_UTF8` 2084environment variable to `0`. 2085 2086#### Generating an XML Report 2087 2088GoogleTest can emit a detailed XML report to a file in addition to its normal 2089textual output. The report contains the duration of each test, and thus can help 2090you identify slow tests. 2091 2092To generate the XML report, set the `GTEST_OUTPUT` environment variable or the 2093`--gtest_output` flag to the string `"xml:path_to_output_file"`, which will 2094create the file at the given location. You can also just use the string `"xml"`, 2095in which case the output can be found in the `test_detail.xml` file in the 2096current directory. 2097 2098If you specify a directory (for example, `"xml:output/directory/"` on Linux or 2099`"xml:output\directory\"` on Windows), GoogleTest will create the XML file in 2100that directory, named after the test executable (e.g. `foo_test.xml` for test 2101program `foo_test` or `foo_test.exe`). If the file already exists (perhaps left 2102over from a previous run), GoogleTest will pick a different name (e.g. 2103`foo_test_1.xml`) to avoid overwriting it. 2104 2105The report is based on the `junitreport` Ant task. Since that format was 2106originally intended for Java, a little interpretation is required to make it 2107apply to GoogleTest tests, as shown here: 2108 2109```xml 2110<testsuites name="AllTests" ...> 2111 <testsuite name="test_case_name" ...> 2112 <testcase name="test_name" ...> 2113 <failure message="..."/> 2114 <failure message="..."/> 2115 <failure message="..."/> 2116 </testcase> 2117 </testsuite> 2118</testsuites> 2119``` 2120 2121* The root `<testsuites>` element corresponds to the entire test program. 2122* `<testsuite>` elements correspond to GoogleTest test suites. 2123* `<testcase>` elements correspond to GoogleTest test functions. 2124 2125For instance, the following program 2126 2127```c++ 2128TEST(MathTest, Addition) { ... } 2129TEST(MathTest, Subtraction) { ... } 2130TEST(LogicTest, NonContradiction) { ... } 2131``` 2132 2133could generate this report: 2134 2135```xml 2136<?xml version="1.0" encoding="UTF-8"?> 2137<testsuites tests="3" failures="1" errors="0" time="0.035" timestamp="2011-10-31T18:52:42" name="AllTests"> 2138 <testsuite name="MathTest" tests="2" failures="1" errors="0" time="0.015"> 2139 <testcase name="Addition" file="test.cpp" line="1" status="run" time="0.007" classname=""> 2140 <failure message="Value of: add(1, 1)
 Actual: 3
Expected: 2" type="">...</failure> 2141 <failure message="Value of: add(1, -1)
 Actual: 1
Expected: 0" type="">...</failure> 2142 </testcase> 2143 <testcase name="Subtraction" file="test.cpp" line="2" status="run" time="0.005" classname=""> 2144 </testcase> 2145 </testsuite> 2146 <testsuite name="LogicTest" tests="1" failures="0" errors="0" time="0.005"> 2147 <testcase name="NonContradiction" file="test.cpp" line="3" status="run" time="0.005" classname=""> 2148 </testcase> 2149 </testsuite> 2150</testsuites> 2151``` 2152 2153Things to note: 2154 2155* The `tests` attribute of a `<testsuites>` or `<testsuite>` element tells how 2156 many test functions the GoogleTest program or test suite contains, while the 2157 `failures` attribute tells how many of them failed. 2158 2159* The `time` attribute expresses the duration of the test, test suite, or 2160 entire test program in seconds. 2161 2162* The `timestamp` attribute records the local date and time of the test 2163 execution. 2164 2165* The `file` and `line` attributes record the source file location, where the 2166 test was defined. 2167 2168* Each `<failure>` element corresponds to a single failed GoogleTest 2169 assertion. 2170 2171#### Generating a JSON Report 2172 2173GoogleTest can also emit a JSON report as an alternative format to XML. To 2174generate the JSON report, set the `GTEST_OUTPUT` environment variable or the 2175`--gtest_output` flag to the string `"json:path_to_output_file"`, which will 2176create the file at the given location. You can also just use the string 2177`"json"`, in which case the output can be found in the `test_detail.json` file 2178in the current directory. 2179 2180The report format conforms to the following JSON Schema: 2181 2182```json 2183{ 2184 "$schema": "https://json-schema.org/schema#", 2185 "type": "object", 2186 "definitions": { 2187 "TestCase": { 2188 "type": "object", 2189 "properties": { 2190 "name": { "type": "string" }, 2191 "tests": { "type": "integer" }, 2192 "failures": { "type": "integer" }, 2193 "disabled": { "type": "integer" }, 2194 "time": { "type": "string" }, 2195 "testsuite": { 2196 "type": "array", 2197 "items": { 2198 "$ref": "#/definitions/TestInfo" 2199 } 2200 } 2201 } 2202 }, 2203 "TestInfo": { 2204 "type": "object", 2205 "properties": { 2206 "name": { "type": "string" }, 2207 "file": { "type": "string" }, 2208 "line": { "type": "integer" }, 2209 "status": { 2210 "type": "string", 2211 "enum": ["RUN", "NOTRUN"] 2212 }, 2213 "time": { "type": "string" }, 2214 "classname": { "type": "string" }, 2215 "failures": { 2216 "type": "array", 2217 "items": { 2218 "$ref": "#/definitions/Failure" 2219 } 2220 } 2221 } 2222 }, 2223 "Failure": { 2224 "type": "object", 2225 "properties": { 2226 "failures": { "type": "string" }, 2227 "type": { "type": "string" } 2228 } 2229 } 2230 }, 2231 "properties": { 2232 "tests": { "type": "integer" }, 2233 "failures": { "type": "integer" }, 2234 "disabled": { "type": "integer" }, 2235 "errors": { "type": "integer" }, 2236 "timestamp": { 2237 "type": "string", 2238 "format": "date-time" 2239 }, 2240 "time": { "type": "string" }, 2241 "name": { "type": "string" }, 2242 "testsuites": { 2243 "type": "array", 2244 "items": { 2245 "$ref": "#/definitions/TestCase" 2246 } 2247 } 2248 } 2249} 2250``` 2251 2252The report uses the format that conforms to the following Proto3 using the 2253[JSON encoding](https://developers.google.com/protocol-buffers/docs/proto3#json): 2254 2255```proto 2256syntax = "proto3"; 2257 2258package googletest; 2259 2260import "google/protobuf/timestamp.proto"; 2261import "google/protobuf/duration.proto"; 2262 2263message UnitTest { 2264 int32 tests = 1; 2265 int32 failures = 2; 2266 int32 disabled = 3; 2267 int32 errors = 4; 2268 google.protobuf.Timestamp timestamp = 5; 2269 google.protobuf.Duration time = 6; 2270 string name = 7; 2271 repeated TestCase testsuites = 8; 2272} 2273 2274message TestCase { 2275 string name = 1; 2276 int32 tests = 2; 2277 int32 failures = 3; 2278 int32 disabled = 4; 2279 int32 errors = 5; 2280 google.protobuf.Duration time = 6; 2281 repeated TestInfo testsuite = 7; 2282} 2283 2284message TestInfo { 2285 string name = 1; 2286 string file = 6; 2287 int32 line = 7; 2288 enum Status { 2289 RUN = 0; 2290 NOTRUN = 1; 2291 } 2292 Status status = 2; 2293 google.protobuf.Duration time = 3; 2294 string classname = 4; 2295 message Failure { 2296 string failures = 1; 2297 string type = 2; 2298 } 2299 repeated Failure failures = 5; 2300} 2301``` 2302 2303For instance, the following program 2304 2305```c++ 2306TEST(MathTest, Addition) { ... } 2307TEST(MathTest, Subtraction) { ... } 2308TEST(LogicTest, NonContradiction) { ... } 2309``` 2310 2311could generate this report: 2312 2313```json 2314{ 2315 "tests": 3, 2316 "failures": 1, 2317 "errors": 0, 2318 "time": "0.035s", 2319 "timestamp": "2011-10-31T18:52:42Z", 2320 "name": "AllTests", 2321 "testsuites": [ 2322 { 2323 "name": "MathTest", 2324 "tests": 2, 2325 "failures": 1, 2326 "errors": 0, 2327 "time": "0.015s", 2328 "testsuite": [ 2329 { 2330 "name": "Addition", 2331 "file": "test.cpp", 2332 "line": 1, 2333 "status": "RUN", 2334 "time": "0.007s", 2335 "classname": "", 2336 "failures": [ 2337 { 2338 "message": "Value of: add(1, 1)\n Actual: 3\nExpected: 2", 2339 "type": "" 2340 }, 2341 { 2342 "message": "Value of: add(1, -1)\n Actual: 1\nExpected: 0", 2343 "type": "" 2344 } 2345 ] 2346 }, 2347 { 2348 "name": "Subtraction", 2349 "file": "test.cpp", 2350 "line": 2, 2351 "status": "RUN", 2352 "time": "0.005s", 2353 "classname": "" 2354 } 2355 ] 2356 }, 2357 { 2358 "name": "LogicTest", 2359 "tests": 1, 2360 "failures": 0, 2361 "errors": 0, 2362 "time": "0.005s", 2363 "testsuite": [ 2364 { 2365 "name": "NonContradiction", 2366 "file": "test.cpp", 2367 "line": 3, 2368 "status": "RUN", 2369 "time": "0.005s", 2370 "classname": "" 2371 } 2372 ] 2373 } 2374 ] 2375} 2376``` 2377 2378{: .callout .important} 2379IMPORTANT: The exact format of the JSON document is subject to change. 2380 2381### Controlling How Failures Are Reported 2382 2383#### Detecting Test Premature Exit 2384 2385Google Test implements the _premature-exit-file_ protocol for test runners to 2386catch any kind of unexpected exits of test programs. Upon start, Google Test 2387creates the file which will be automatically deleted after all work has been 2388finished. Then, the test runner can check if this file exists. In case the file 2389remains undeleted, the inspected test has exited prematurely. 2390 2391This feature is enabled only if the `TEST_PREMATURE_EXIT_FILE` environment 2392variable has been set. 2393 2394#### Turning Assertion Failures into Break-Points 2395 2396When running test programs under a debugger, it's very convenient if the 2397debugger can catch an assertion failure and automatically drop into interactive 2398mode. GoogleTest's *break-on-failure* mode supports this behavior. 2399 2400To enable it, set the `GTEST_BREAK_ON_FAILURE` environment variable to a value 2401other than `0`. Alternatively, you can use the `--gtest_break_on_failure` 2402command line flag. 2403 2404#### Disabling Catching Test-Thrown Exceptions 2405 2406GoogleTest can be used either with or without exceptions enabled. If a test 2407throws a C++ exception or (on Windows) a structured exception (SEH), by default 2408GoogleTest catches it, reports it as a test failure, and continues with the next 2409test method. This maximizes the coverage of a test run. Also, on Windows an 2410uncaught exception will cause a pop-up window, so catching the exceptions allows 2411you to run the tests automatically. 2412 2413When debugging the test failures, however, you may instead want the exceptions 2414to be handled by the debugger, such that you can examine the call stack when an 2415exception is thrown. To achieve that, set the `GTEST_CATCH_EXCEPTIONS` 2416environment variable to `0`, or use the `--gtest_catch_exceptions=0` flag when 2417running the tests. 2418 2419### Sanitizer Integration 2420 2421The 2422[Undefined Behavior Sanitizer](https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html), 2423[Address Sanitizer](https://github.com/google/sanitizers/wiki/AddressSanitizer), 2424and 2425[Thread Sanitizer](https://github.com/google/sanitizers/wiki/ThreadSanitizerCppManual) 2426all provide weak functions that you can override to trigger explicit failures 2427when they detect sanitizer errors, such as creating a reference from `nullptr`. 2428To override these functions, place definitions for them in a source file that 2429you compile as part of your main binary: 2430 2431``` 2432extern "C" { 2433void __ubsan_on_report() { 2434 FAIL() << "Encountered an undefined behavior sanitizer error"; 2435} 2436void __asan_on_error() { 2437 FAIL() << "Encountered an address sanitizer error"; 2438} 2439void __tsan_on_report() { 2440 FAIL() << "Encountered a thread sanitizer error"; 2441} 2442} // extern "C" 2443``` 2444 2445After compiling your project with one of the sanitizers enabled, if a particular 2446test triggers a sanitizer error, GoogleTest will report that it failed. 2447