1 2 3Now that you have read [Primer](Primer.md) and learned how to write tests 4using Google Test, it's time to learn some new tricks. This document 5will show you more assertions as well as how to construct complex 6failure messages, propagate fatal failures, reuse and speed up your 7test fixtures, and use various flags with your tests. 8 9# More Assertions # 10 11This section covers some less frequently used, but still significant, 12assertions. 13 14## Explicit Success and Failure ## 15 16These three assertions do not actually test a value or expression. Instead, 17they generate a success or failure directly. Like the macros that actually 18perform a test, you may stream a custom failure message into the them. 19 20| `SUCCEED();` | 21|:-------------| 22 23Generates a success. This does NOT make the overall test succeed. A test is 24considered successful only if none of its assertions fail during its execution. 25 26Note: `SUCCEED()` is purely documentary and currently doesn't generate any 27user-visible output. However, we may add `SUCCEED()` messages to Google Test's 28output in the future. 29 30| `FAIL();` | `ADD_FAILURE();` | `ADD_FAILURE_AT("`_file\_path_`", `_line\_number_`);` | 31|:-----------|:-----------------|:------------------------------------------------------| 32 33`FAIL()` generates a fatal failure, while `ADD_FAILURE()` and `ADD_FAILURE_AT()` generate a nonfatal 34failure. These are useful when control flow, rather than a Boolean expression, 35deteremines the test's success or failure. For example, you might want to write 36something like: 37 38``` 39switch(expression) { 40 case 1: ... some checks ... 41 case 2: ... some other checks 42 ... 43 default: FAIL() << "We shouldn't get here."; 44} 45``` 46 47Note: you can only use `FAIL()` in functions that return `void`. See the [Assertion Placement section](#assertion-placement) for more information. 48 49_Availability_: Linux, Windows, Mac. 50 51## Exception Assertions ## 52 53These are for verifying that a piece of code throws (or does not 54throw) an exception of the given type: 55 56| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 57|:--------------------|:-----------------------|:-------------| 58| `ASSERT_THROW(`_statement_, _exception\_type_`);` | `EXPECT_THROW(`_statement_, _exception\_type_`);` | _statement_ throws an exception of the given type | 59| `ASSERT_ANY_THROW(`_statement_`);` | `EXPECT_ANY_THROW(`_statement_`);` | _statement_ throws an exception of any type | 60| `ASSERT_NO_THROW(`_statement_`);` | `EXPECT_NO_THROW(`_statement_`);` | _statement_ doesn't throw any exception | 61 62Examples: 63 64``` 65ASSERT_THROW(Foo(5), bar_exception); 66 67EXPECT_NO_THROW({ 68 int n = 5; 69 Bar(&n); 70}); 71``` 72 73_Availability_: Linux, Windows, Mac; since version 1.1.0. 74 75## Predicate Assertions for Better Error Messages ## 76 77Even though Google Test has a rich set of assertions, they can never be 78complete, as it's impossible (nor a good idea) to anticipate all the scenarios 79a user might run into. Therefore, sometimes a user has to use `EXPECT_TRUE()` 80to check a complex expression, for lack of a better macro. This has the problem 81of not showing you the values of the parts of the expression, making it hard to 82understand what went wrong. As a workaround, some users choose to construct the 83failure message by themselves, streaming it into `EXPECT_TRUE()`. However, this 84is awkward especially when the expression has side-effects or is expensive to 85evaluate. 86 87Google Test gives you three different options to solve this problem: 88 89### Using an Existing Boolean Function ### 90 91If you already have a function or a functor that returns `bool` (or a type 92that can be implicitly converted to `bool`), you can use it in a _predicate 93assertion_ to get the function arguments printed for free: 94 95| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 96|:--------------------|:-----------------------|:-------------| 97| `ASSERT_PRED1(`_pred1, val1_`);` | `EXPECT_PRED1(`_pred1, val1_`);` | _pred1(val1)_ returns true | 98| `ASSERT_PRED2(`_pred2, val1, val2_`);` | `EXPECT_PRED2(`_pred2, val1, val2_`);` | _pred2(val1, val2)_ returns true | 99| ... | ... | ... | 100 101In the above, _predn_ is an _n_-ary predicate function or functor, where 102_val1_, _val2_, ..., and _valn_ are its arguments. The assertion succeeds 103if the predicate returns `true` when applied to the given arguments, and fails 104otherwise. When the assertion fails, it prints the value of each argument. In 105either case, the arguments are evaluated exactly once. 106 107Here's an example. Given 108 109``` 110// Returns true iff m and n have no common divisors except 1. 111bool MutuallyPrime(int m, int n) { ... } 112const int a = 3; 113const int b = 4; 114const int c = 10; 115``` 116 117the assertion `EXPECT_PRED2(MutuallyPrime, a, b);` will succeed, while the 118assertion `EXPECT_PRED2(MutuallyPrime, b, c);` will fail with the message 119 120<pre> 121!MutuallyPrime(b, c) is false, where<br> 122b is 4<br> 123c is 10<br> 124</pre> 125 126**Notes:** 127 128 1. If you see a compiler error "no matching function to call" when using `ASSERT_PRED*` or `EXPECT_PRED*`, please see [this FAQ](FAQ.md#the-compiler-complains-no-matching-function-to-call-when-i-use-assert_predn-how-do-i-fix-it) for how to resolve it. 129 1. Currently we only provide predicate assertions of arity <= 5. If you need a higher-arity assertion, let us know. 130 131_Availability_: Linux, Windows, Mac 132 133### Using a Function That Returns an AssertionResult ### 134 135While `EXPECT_PRED*()` and friends are handy for a quick job, the 136syntax is not satisfactory: you have to use different macros for 137different arities, and it feels more like Lisp than C++. The 138`::testing::AssertionResult` class solves this problem. 139 140An `AssertionResult` object represents the result of an assertion 141(whether it's a success or a failure, and an associated message). You 142can create an `AssertionResult` using one of these factory 143functions: 144 145``` 146namespace testing { 147 148// Returns an AssertionResult object to indicate that an assertion has 149// succeeded. 150AssertionResult AssertionSuccess(); 151 152// Returns an AssertionResult object to indicate that an assertion has 153// failed. 154AssertionResult AssertionFailure(); 155 156} 157``` 158 159You can then use the `<<` operator to stream messages to the 160`AssertionResult` object. 161 162To provide more readable messages in Boolean assertions 163(e.g. `EXPECT_TRUE()`), write a predicate function that returns 164`AssertionResult` instead of `bool`. For example, if you define 165`IsEven()` as: 166 167``` 168::testing::AssertionResult IsEven(int n) { 169 if ((n % 2) == 0) 170 return ::testing::AssertionSuccess(); 171 else 172 return ::testing::AssertionFailure() << n << " is odd"; 173} 174``` 175 176instead of: 177 178``` 179bool IsEven(int n) { 180 return (n % 2) == 0; 181} 182``` 183 184the failed assertion `EXPECT_TRUE(IsEven(Fib(4)))` will print: 185 186<pre> 187Value of: IsEven(Fib(4))<br> 188Actual: false (*3 is odd*)<br> 189Expected: true<br> 190</pre> 191 192instead of a more opaque 193 194<pre> 195Value of: IsEven(Fib(4))<br> 196Actual: false<br> 197Expected: true<br> 198</pre> 199 200If you want informative messages in `EXPECT_FALSE` and `ASSERT_FALSE` 201as well, and are fine with making the predicate slower in the success 202case, you can supply a success message: 203 204``` 205::testing::AssertionResult IsEven(int n) { 206 if ((n % 2) == 0) 207 return ::testing::AssertionSuccess() << n << " is even"; 208 else 209 return ::testing::AssertionFailure() << n << " is odd"; 210} 211``` 212 213Then the statement `EXPECT_FALSE(IsEven(Fib(6)))` will print 214 215<pre> 216Value of: IsEven(Fib(6))<br> 217Actual: true (8 is even)<br> 218Expected: false<br> 219</pre> 220 221_Availability_: Linux, Windows, Mac; since version 1.4.1. 222 223### Using a Predicate-Formatter ### 224 225If you find the default message generated by `(ASSERT|EXPECT)_PRED*` and 226`(ASSERT|EXPECT)_(TRUE|FALSE)` unsatisfactory, or some arguments to your 227predicate do not support streaming to `ostream`, you can instead use the 228following _predicate-formatter assertions_ to _fully_ customize how the 229message is formatted: 230 231| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 232|:--------------------|:-----------------------|:-------------| 233| `ASSERT_PRED_FORMAT1(`_pred\_format1, val1_`);` | `EXPECT_PRED_FORMAT1(`_pred\_format1, val1_`);` | _pred\_format1(val1)_ is successful | 234| `ASSERT_PRED_FORMAT2(`_pred\_format2, val1, val2_`);` | `EXPECT_PRED_FORMAT2(`_pred\_format2, val1, val2_`);` | _pred\_format2(val1, val2)_ is successful | 235| `...` | `...` | `...` | 236 237The difference between this and the previous two groups of macros is that instead of 238a predicate, `(ASSERT|EXPECT)_PRED_FORMAT*` take a _predicate-formatter_ 239(_pred\_formatn_), which is a function or functor with the signature: 240 241`::testing::AssertionResult PredicateFormattern(const char* `_expr1_`, const char* `_expr2_`, ... const char* `_exprn_`, T1 `_val1_`, T2 `_val2_`, ... Tn `_valn_`);` 242 243where _val1_, _val2_, ..., and _valn_ are the values of the predicate 244arguments, and _expr1_, _expr2_, ..., and _exprn_ are the corresponding 245expressions as they appear in the source code. The types `T1`, `T2`, ..., and 246`Tn` can be either value types or reference types. For example, if an 247argument has type `Foo`, you can declare it as either `Foo` or `const Foo&`, 248whichever is appropriate. 249 250A predicate-formatter returns a `::testing::AssertionResult` object to indicate 251whether the assertion has succeeded or not. The only way to create such an 252object is to call one of these factory functions: 253 254As an example, let's improve the failure message in the previous example, which uses `EXPECT_PRED2()`: 255 256``` 257// Returns the smallest prime common divisor of m and n, 258// or 1 when m and n are mutually prime. 259int SmallestPrimeCommonDivisor(int m, int n) { ... } 260 261// A predicate-formatter for asserting that two integers are mutually prime. 262::testing::AssertionResult AssertMutuallyPrime(const char* m_expr, 263 const char* n_expr, 264 int m, 265 int n) { 266 if (MutuallyPrime(m, n)) 267 return ::testing::AssertionSuccess(); 268 269 return ::testing::AssertionFailure() 270 << m_expr << " and " << n_expr << " (" << m << " and " << n 271 << ") are not mutually prime, " << "as they have a common divisor " 272 << SmallestPrimeCommonDivisor(m, n); 273} 274``` 275 276With this predicate-formatter, we can use 277 278``` 279EXPECT_PRED_FORMAT2(AssertMutuallyPrime, b, c); 280``` 281 282to generate the message 283 284<pre> 285b and c (4 and 10) are not mutually prime, as they have a common divisor 2.<br> 286</pre> 287 288As you may have realized, many of the assertions we introduced earlier are 289special cases of `(EXPECT|ASSERT)_PRED_FORMAT*`. In fact, most of them are 290indeed defined using `(EXPECT|ASSERT)_PRED_FORMAT*`. 291 292_Availability_: Linux, Windows, Mac. 293 294 295## Floating-Point Comparison ## 296 297Comparing floating-point numbers is tricky. Due to round-off errors, it is 298very unlikely that two floating-points will match exactly. Therefore, 299`ASSERT_EQ` 's naive comparison usually doesn't work. And since floating-points 300can have a wide value range, no single fixed error bound works. It's better to 301compare by a fixed relative error bound, except for values close to 0 due to 302the loss of precision there. 303 304In general, for floating-point comparison to make sense, the user needs to 305carefully choose the error bound. If they don't want or care to, comparing in 306terms of Units in the Last Place (ULPs) is a good default, and Google Test 307provides assertions to do this. Full details about ULPs are quite long; if you 308want to learn more, see 309[this article on float comparison](http://www.cygnus-software.com/papers/comparingfloats/comparingfloats.htm). 310 311### Floating-Point Macros ### 312 313| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 314|:--------------------|:-----------------------|:-------------| 315| `ASSERT_FLOAT_EQ(`_val1, val2_`);` | `EXPECT_FLOAT_EQ(`_val1, val2_`);` | the two `float` values are almost equal | 316| `ASSERT_DOUBLE_EQ(`_val1, val2_`);` | `EXPECT_DOUBLE_EQ(`_val1, val2_`);` | the two `double` values are almost equal | 317 318By "almost equal", we mean the two values are within 4 ULP's from each 319other. 320 321The following assertions allow you to choose the acceptable error bound: 322 323| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 324|:--------------------|:-----------------------|:-------------| 325| `ASSERT_NEAR(`_val1, val2, abs\_error_`);` | `EXPECT_NEAR`_(val1, val2, abs\_error_`);` | the difference between _val1_ and _val2_ doesn't exceed the given absolute error | 326 327_Availability_: Linux, Windows, Mac. 328 329### Floating-Point Predicate-Format Functions ### 330 331Some floating-point operations are useful, but not that often used. In order 332to avoid an explosion of new macros, we provide them as predicate-format 333functions that can be used in predicate assertion macros (e.g. 334`EXPECT_PRED_FORMAT2`, etc). 335 336``` 337EXPECT_PRED_FORMAT2(::testing::FloatLE, val1, val2); 338EXPECT_PRED_FORMAT2(::testing::DoubleLE, val1, val2); 339``` 340 341Verifies that _val1_ is less than, or almost equal to, _val2_. You can 342replace `EXPECT_PRED_FORMAT2` in the above table with `ASSERT_PRED_FORMAT2`. 343 344_Availability_: Linux, Windows, Mac. 345 346## Windows HRESULT assertions ## 347 348These assertions test for `HRESULT` success or failure. 349 350| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 351|:--------------------|:-----------------------|:-------------| 352| `ASSERT_HRESULT_SUCCEEDED(`_expression_`);` | `EXPECT_HRESULT_SUCCEEDED(`_expression_`);` | _expression_ is a success `HRESULT` | 353| `ASSERT_HRESULT_FAILED(`_expression_`);` | `EXPECT_HRESULT_FAILED(`_expression_`);` | _expression_ is a failure `HRESULT` | 354 355The generated output contains the human-readable error message 356associated with the `HRESULT` code returned by _expression_. 357 358You might use them like this: 359 360``` 361CComPtr shell; 362ASSERT_HRESULT_SUCCEEDED(shell.CoCreateInstance(L"Shell.Application")); 363CComVariant empty; 364ASSERT_HRESULT_SUCCEEDED(shell->ShellExecute(CComBSTR(url), empty, empty, empty, empty)); 365``` 366 367_Availability_: Windows. 368 369## Type Assertions ## 370 371You can call the function 372``` 373::testing::StaticAssertTypeEq<T1, T2>(); 374``` 375to assert that types `T1` and `T2` are the same. The function does 376nothing if the assertion is satisfied. If the types are different, 377the function call will fail to compile, and the compiler error message 378will likely (depending on the compiler) show you the actual values of 379`T1` and `T2`. This is mainly useful inside template code. 380 381_Caveat:_ When used inside a member function of a class template or a 382function template, `StaticAssertTypeEq<T1, T2>()` is effective _only if_ 383the function is instantiated. For example, given: 384``` 385template <typename T> class Foo { 386 public: 387 void Bar() { ::testing::StaticAssertTypeEq<int, T>(); } 388}; 389``` 390the code: 391``` 392void Test1() { Foo<bool> foo; } 393``` 394will _not_ generate a compiler error, as `Foo<bool>::Bar()` is never 395actually instantiated. Instead, you need: 396``` 397void Test2() { Foo<bool> foo; foo.Bar(); } 398``` 399to cause a compiler error. 400 401_Availability:_ Linux, Windows, Mac; since version 1.3.0. 402 403## Assertion Placement ## 404 405You can use assertions in any C++ function. In particular, it doesn't 406have to be a method of the test fixture class. The one constraint is 407that assertions that generate a fatal failure (`FAIL*` and `ASSERT_*`) 408can only be used in void-returning functions. This is a consequence of 409Google Test not using exceptions. By placing it in a non-void function 410you'll get a confusing compile error like 411`"error: void value not ignored as it ought to be"`. 412 413If you need to use assertions in a function that returns non-void, one option 414is to make the function return the value in an out parameter instead. For 415example, you can rewrite `T2 Foo(T1 x)` to `void Foo(T1 x, T2* result)`. You 416need to make sure that `*result` contains some sensible value even when the 417function returns prematurely. As the function now returns `void`, you can use 418any assertion inside of it. 419 420If changing the function's type is not an option, you should just use 421assertions that generate non-fatal failures, such as `ADD_FAILURE*` and 422`EXPECT_*`. 423 424_Note_: Constructors and destructors are not considered void-returning 425functions, according to the C++ language specification, and so you may not use 426fatal assertions in them. You'll get a compilation error if you try. A simple 427workaround is to transfer the entire body of the constructor or destructor to a 428private void-returning method. However, you should be aware that a fatal 429assertion failure in a constructor does not terminate the current test, as your 430intuition might suggest; it merely returns from the constructor early, possibly 431leaving your object in a partially-constructed state. Likewise, a fatal 432assertion failure in a destructor may leave your object in a 433partially-destructed state. Use assertions carefully in these situations! 434 435# Teaching Google Test How to Print Your Values # 436 437When a test assertion such as `EXPECT_EQ` fails, Google Test prints the 438argument values to help you debug. It does this using a 439user-extensible value printer. 440 441This printer knows how to print built-in C++ types, native arrays, STL 442containers, and any type that supports the `<<` operator. For other 443types, it prints the raw bytes in the value and hopes that you the 444user can figure it out. 445 446As mentioned earlier, the printer is _extensible_. That means 447you can teach it to do a better job at printing your particular type 448than to dump the bytes. To do that, define `<<` for your type: 449 450``` 451#include <iostream> 452 453namespace foo { 454 455class Bar { ... }; // We want Google Test to be able to print instances of this. 456 457// It's important that the << operator is defined in the SAME 458// namespace that defines Bar. C++'s look-up rules rely on that. 459::std::ostream& operator<<(::std::ostream& os, const Bar& bar) { 460 return os << bar.DebugString(); // whatever needed to print bar to os 461} 462 463} // namespace foo 464``` 465 466Sometimes, this might not be an option: your team may consider it bad 467style to have a `<<` operator for `Bar`, or `Bar` may already have a 468`<<` operator that doesn't do what you want (and you cannot change 469it). If so, you can instead define a `PrintTo()` function like this: 470 471``` 472#include <iostream> 473 474namespace foo { 475 476class Bar { ... }; 477 478// It's important that PrintTo() is defined in the SAME 479// namespace that defines Bar. C++'s look-up rules rely on that. 480void PrintTo(const Bar& bar, ::std::ostream* os) { 481 *os << bar.DebugString(); // whatever needed to print bar to os 482} 483 484} // namespace foo 485``` 486 487If you have defined both `<<` and `PrintTo()`, the latter will be used 488when Google Test is concerned. This allows you to customize how the value 489appears in Google Test's output without affecting code that relies on the 490behavior of its `<<` operator. 491 492If you want to print a value `x` using Google Test's value printer 493yourself, just call `::testing::PrintToString(`_x_`)`, which 494returns an `std::string`: 495 496``` 497vector<pair<Bar, int> > bar_ints = GetBarIntVector(); 498 499EXPECT_TRUE(IsCorrectBarIntVector(bar_ints)) 500 << "bar_ints = " << ::testing::PrintToString(bar_ints); 501``` 502 503# Death Tests # 504 505In many applications, there are assertions that can cause application failure 506if a condition is not met. These sanity checks, which ensure that the program 507is in a known good state, are there to fail at the earliest possible time after 508some program state is corrupted. If the assertion checks the wrong condition, 509then the program may proceed in an erroneous state, which could lead to memory 510corruption, security holes, or worse. Hence it is vitally important to test 511that such assertion statements work as expected. 512 513Since these precondition checks cause the processes to die, we call such tests 514_death tests_. More generally, any test that checks that a program terminates 515(except by throwing an exception) in an expected fashion is also a death test. 516 517Note that if a piece of code throws an exception, we don't consider it "death" 518for the purpose of death tests, as the caller of the code could catch the exception 519and avoid the crash. If you want to verify exceptions thrown by your code, 520see [Exception Assertions](#exception-assertions). 521 522If you want to test `EXPECT_*()/ASSERT_*()` failures in your test code, see [Catching Failures](#catching-failures). 523 524## How to Write a Death Test ## 525 526Google Test has the following macros to support death tests: 527 528| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 529|:--------------------|:-----------------------|:-------------| 530| `ASSERT_DEATH(`_statement, regex_`);` | `EXPECT_DEATH(`_statement, regex_`);` | _statement_ crashes with the given error | 531| `ASSERT_DEATH_IF_SUPPORTED(`_statement, regex_`);` | `EXPECT_DEATH_IF_SUPPORTED(`_statement, regex_`);` | if death tests are supported, verifies that _statement_ crashes with the given error; otherwise verifies nothing | 532| `ASSERT_EXIT(`_statement, predicate, regex_`);` | `EXPECT_EXIT(`_statement, predicate, regex_`);` |_statement_ exits with the given error and its exit code matches _predicate_ | 533 534where _statement_ is a statement that is expected to cause the process to 535die, _predicate_ is a function or function object that evaluates an integer 536exit status, and _regex_ is a regular expression that the stderr output of 537_statement_ is expected to match. Note that _statement_ can be _any valid 538statement_ (including _compound statement_) and doesn't have to be an 539expression. 540 541As usual, the `ASSERT` variants abort the current test function, while the 542`EXPECT` variants do not. 543 544**Note:** We use the word "crash" here to mean that the process 545terminates with a _non-zero_ exit status code. There are two 546possibilities: either the process has called `exit()` or `_exit()` 547with a non-zero value, or it may be killed by a signal. 548 549This means that if _statement_ terminates the process with a 0 exit 550code, it is _not_ considered a crash by `EXPECT_DEATH`. Use 551`EXPECT_EXIT` instead if this is the case, or if you want to restrict 552the exit code more precisely. 553 554A predicate here must accept an `int` and return a `bool`. The death test 555succeeds only if the predicate returns `true`. Google Test defines a few 556predicates that handle the most common cases: 557 558``` 559::testing::ExitedWithCode(exit_code) 560``` 561 562This expression is `true` if the program exited normally with the given exit 563code. 564 565``` 566::testing::KilledBySignal(signal_number) // Not available on Windows. 567``` 568 569This expression is `true` if the program was killed by the given signal. 570 571The `*_DEATH` macros are convenient wrappers for `*_EXIT` that use a predicate 572that verifies the process' exit code is non-zero. 573 574Note that a death test only cares about three things: 575 576 1. does _statement_ abort or exit the process? 577 1. (in the case of `ASSERT_EXIT` and `EXPECT_EXIT`) does the exit status satisfy _predicate_? Or (in the case of `ASSERT_DEATH` and `EXPECT_DEATH`) is the exit status non-zero? And 578 1. does the stderr output match _regex_? 579 580In particular, if _statement_ generates an `ASSERT_*` or `EXPECT_*` failure, it will **not** cause the death test to fail, as Google Test assertions don't abort the process. 581 582To write a death test, simply use one of the above macros inside your test 583function. For example, 584 585``` 586TEST(MyDeathTest, Foo) { 587 // This death test uses a compound statement. 588 ASSERT_DEATH({ int n = 5; Foo(&n); }, "Error on line .* of Foo()"); 589} 590TEST(MyDeathTest, NormalExit) { 591 EXPECT_EXIT(NormalExit(), ::testing::ExitedWithCode(0), "Success"); 592} 593TEST(MyDeathTest, KillMyself) { 594 EXPECT_EXIT(KillMyself(), ::testing::KilledBySignal(SIGKILL), "Sending myself unblockable signal"); 595} 596``` 597 598verifies that: 599 600 * calling `Foo(5)` causes the process to die with the given error message, 601 * calling `NormalExit()` causes the process to print `"Success"` to stderr and exit with exit code 0, and 602 * calling `KillMyself()` kills the process with signal `SIGKILL`. 603 604The test function body may contain other assertions and statements as well, if 605necessary. 606 607_Important:_ We strongly recommend you to follow the convention of naming your 608test case (not test) `*DeathTest` when it contains a death test, as 609demonstrated in the above example. The `Death Tests And Threads` section below 610explains why. 611 612If a test fixture class is shared by normal tests and death tests, you 613can use typedef to introduce an alias for the fixture class and avoid 614duplicating its code: 615``` 616class FooTest : public ::testing::Test { ... }; 617 618typedef FooTest FooDeathTest; 619 620TEST_F(FooTest, DoesThis) { 621 // normal test 622} 623 624TEST_F(FooDeathTest, DoesThat) { 625 // death test 626} 627``` 628 629_Availability:_ Linux, Windows (requires MSVC 8.0 or above), Cygwin, and Mac (the latter three are supported since v1.3.0). `(ASSERT|EXPECT)_DEATH_IF_SUPPORTED` are new in v1.4.0. 630 631## Regular Expression Syntax ## 632 633On POSIX systems (e.g. Linux, Cygwin, and Mac), Google Test uses the 634[POSIX extended regular expression](http://www.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap09.html#tag_09_04) 635syntax in death tests. To learn about this syntax, you may want to read this [Wikipedia entry](http://en.wikipedia.org/wiki/Regular_expression#POSIX_Extended_Regular_Expressions). 636 637On Windows, Google Test uses its own simple regular expression 638implementation. It lacks many features you can find in POSIX extended 639regular expressions. For example, we don't support union (`"x|y"`), 640grouping (`"(xy)"`), brackets (`"[xy]"`), and repetition count 641(`"x{5,7}"`), among others. Below is what we do support (Letter `A` denotes a 642literal character, period (`.`), or a single `\\` escape sequence; `x` 643and `y` denote regular expressions.): 644 645| `c` | matches any literal character `c` | 646|:----|:----------------------------------| 647| `\\d` | matches any decimal digit | 648| `\\D` | matches any character that's not a decimal digit | 649| `\\f` | matches `\f` | 650| `\\n` | matches `\n` | 651| `\\r` | matches `\r` | 652| `\\s` | matches any ASCII whitespace, including `\n` | 653| `\\S` | matches any character that's not a whitespace | 654| `\\t` | matches `\t` | 655| `\\v` | matches `\v` | 656| `\\w` | matches any letter, `_`, or decimal digit | 657| `\\W` | matches any character that `\\w` doesn't match | 658| `\\c` | matches any literal character `c`, which must be a punctuation | 659| `\\.` | matches the `.` character | 660| `.` | matches any single character except `\n` | 661| `A?` | matches 0 or 1 occurrences of `A` | 662| `A*` | matches 0 or many occurrences of `A` | 663| `A+` | matches 1 or many occurrences of `A` | 664| `^` | matches the beginning of a string (not that of each line) | 665| `$` | matches the end of a string (not that of each line) | 666| `xy` | matches `x` followed by `y` | 667 668To help you determine which capability is available on your system, 669Google Test defines macro `GTEST_USES_POSIX_RE=1` when it uses POSIX 670extended regular expressions, or `GTEST_USES_SIMPLE_RE=1` when it uses 671the simple version. If you want your death tests to work in both 672cases, you can either `#if` on these macros or use the more limited 673syntax only. 674 675## How It Works ## 676 677Under the hood, `ASSERT_EXIT()` spawns a new process and executes the 678death test statement in that process. The details of of how precisely 679that happens depend on the platform and the variable 680`::testing::GTEST_FLAG(death_test_style)` (which is initialized from the 681command-line flag `--gtest_death_test_style`). 682 683 * On POSIX systems, `fork()` (or `clone()` on Linux) is used to spawn the child, after which: 684 * If the variable's value is `"fast"`, the death test statement is immediately executed. 685 * If the variable's value is `"threadsafe"`, the child process re-executes the unit test binary just as it was originally invoked, but with some extra flags to cause just the single death test under consideration to be run. 686 * On Windows, the child is spawned using the `CreateProcess()` API, and re-executes the binary to cause just the single death test under consideration to be run - much like the `threadsafe` mode on POSIX. 687 688Other values for the variable are illegal and will cause the death test to 689fail. Currently, the flag's default value is `"fast"`. However, we reserve the 690right to change it in the future. Therefore, your tests should not depend on 691this. 692 693In either case, the parent process waits for the child process to complete, and checks that 694 695 1. the child's exit status satisfies the predicate, and 696 1. the child's stderr matches the regular expression. 697 698If the death test statement runs to completion without dying, the child 699process will nonetheless terminate, and the assertion fails. 700 701## Death Tests And Threads ## 702 703The reason for the two death test styles has to do with thread safety. Due to 704well-known problems with forking in the presence of threads, death tests should 705be run in a single-threaded context. Sometimes, however, it isn't feasible to 706arrange that kind of environment. For example, statically-initialized modules 707may start threads before main is ever reached. Once threads have been created, 708it may be difficult or impossible to clean them up. 709 710Google Test has three features intended to raise awareness of threading issues. 711 712 1. A warning is emitted if multiple threads are running when a death test is encountered. 713 1. Test cases with a name ending in "DeathTest" are run before all other tests. 714 1. It uses `clone()` instead of `fork()` to spawn the child process on Linux (`clone()` is not available on Cygwin and Mac), as `fork()` is more likely to cause the child to hang when the parent process has multiple threads. 715 716It's perfectly fine to create threads inside a death test statement; they are 717executed in a separate process and cannot affect the parent. 718 719## Death Test Styles ## 720 721The "threadsafe" death test style was introduced in order to help mitigate the 722risks of testing in a possibly multithreaded environment. It trades increased 723test execution time (potentially dramatically so) for improved thread safety. 724We suggest using the faster, default "fast" style unless your test has specific 725problems with it. 726 727You can choose a particular style of death tests by setting the flag 728programmatically: 729 730``` 731::testing::FLAGS_gtest_death_test_style = "threadsafe"; 732``` 733 734You can do this in `main()` to set the style for all death tests in the 735binary, or in individual tests. Recall that flags are saved before running each 736test and restored afterwards, so you need not do that yourself. For example: 737 738``` 739TEST(MyDeathTest, TestOne) { 740 ::testing::FLAGS_gtest_death_test_style = "threadsafe"; 741 // This test is run in the "threadsafe" style: 742 ASSERT_DEATH(ThisShouldDie(), ""); 743} 744 745TEST(MyDeathTest, TestTwo) { 746 // This test is run in the "fast" style: 747 ASSERT_DEATH(ThisShouldDie(), ""); 748} 749 750int main(int argc, char** argv) { 751 ::testing::InitGoogleTest(&argc, argv); 752 ::testing::FLAGS_gtest_death_test_style = "fast"; 753 return RUN_ALL_TESTS(); 754} 755``` 756 757## Caveats ## 758 759The _statement_ argument of `ASSERT_EXIT()` can be any valid C++ statement. 760If it leaves the current function via a `return` statement or by throwing an exception, 761the death test is considered to have failed. Some Google Test macros may return 762from the current function (e.g. `ASSERT_TRUE()`), so be sure to avoid them in _statement_. 763 764Since _statement_ runs in the child process, any in-memory side effect (e.g. 765modifying a variable, releasing memory, etc) it causes will _not_ be observable 766in the parent process. In particular, if you release memory in a death test, 767your program will fail the heap check as the parent process will never see the 768memory reclaimed. To solve this problem, you can 769 770 1. try not to free memory in a death test; 771 1. free the memory again in the parent process; or 772 1. do not use the heap checker in your program. 773 774Due to an implementation detail, you cannot place multiple death test 775assertions on the same line; otherwise, compilation will fail with an unobvious 776error message. 777 778Despite the improved thread safety afforded by the "threadsafe" style of death 779test, thread problems such as deadlock are still possible in the presence of 780handlers registered with `pthread_atfork(3)`. 781 782# Using Assertions in Sub-routines # 783 784## Adding Traces to Assertions ## 785 786If a test sub-routine is called from several places, when an assertion 787inside it fails, it can be hard to tell which invocation of the 788sub-routine the failure is from. You can alleviate this problem using 789extra logging or custom failure messages, but that usually clutters up 790your tests. A better solution is to use the `SCOPED_TRACE` macro: 791 792| `SCOPED_TRACE(`_message_`);` | 793|:-----------------------------| 794 795where _message_ can be anything streamable to `std::ostream`. This 796macro will cause the current file name, line number, and the given 797message to be added in every failure message. The effect will be 798undone when the control leaves the current lexical scope. 799 800For example, 801 802``` 80310: void Sub1(int n) { 80411: EXPECT_EQ(1, Bar(n)); 80512: EXPECT_EQ(2, Bar(n + 1)); 80613: } 80714: 80815: TEST(FooTest, Bar) { 80916: { 81017: SCOPED_TRACE("A"); // This trace point will be included in 81118: // every failure in this scope. 81219: Sub1(1); 81320: } 81421: // Now it won't. 81522: Sub1(9); 81623: } 817``` 818 819could result in messages like these: 820 821``` 822path/to/foo_test.cc:11: Failure 823Value of: Bar(n) 824Expected: 1 825 Actual: 2 826 Trace: 827path/to/foo_test.cc:17: A 828 829path/to/foo_test.cc:12: Failure 830Value of: Bar(n + 1) 831Expected: 2 832 Actual: 3 833``` 834 835Without the trace, it would've been difficult to know which invocation 836of `Sub1()` the two failures come from respectively. (You could add an 837extra message to each assertion in `Sub1()` to indicate the value of 838`n`, but that's tedious.) 839 840Some tips on using `SCOPED_TRACE`: 841 842 1. With a suitable message, it's often enough to use `SCOPED_TRACE` at the beginning of a sub-routine, instead of at each call site. 843 1. When calling sub-routines inside a loop, make the loop iterator part of the message in `SCOPED_TRACE` such that you can know which iteration the failure is from. 844 1. Sometimes the line number of the trace point is enough for identifying the particular invocation of a sub-routine. In this case, you don't have to choose a unique message for `SCOPED_TRACE`. You can simply use `""`. 845 1. You can use `SCOPED_TRACE` in an inner scope when there is one in the outer scope. In this case, all active trace points will be included in the failure messages, in reverse order they are encountered. 846 1. The trace dump is clickable in Emacs' compilation buffer - hit return on a line number and you'll be taken to that line in the source file! 847 848_Availability:_ Linux, Windows, Mac. 849 850## Propagating Fatal Failures ## 851 852A common pitfall when using `ASSERT_*` and `FAIL*` is not understanding that 853when they fail they only abort the _current function_, not the entire test. For 854example, the following test will segfault: 855``` 856void Subroutine() { 857 // Generates a fatal failure and aborts the current function. 858 ASSERT_EQ(1, 2); 859 // The following won't be executed. 860 ... 861} 862 863TEST(FooTest, Bar) { 864 Subroutine(); 865 // The intended behavior is for the fatal failure 866 // in Subroutine() to abort the entire test. 867 // The actual behavior: the function goes on after Subroutine() returns. 868 int* p = NULL; 869 *p = 3; // Segfault! 870} 871``` 872 873Since we don't use exceptions, it is technically impossible to 874implement the intended behavior here. To alleviate this, Google Test 875provides two solutions. You could use either the 876`(ASSERT|EXPECT)_NO_FATAL_FAILURE` assertions or the 877`HasFatalFailure()` function. They are described in the following two 878subsections. 879 880### Asserting on Subroutines ### 881 882As shown above, if your test calls a subroutine that has an `ASSERT_*` 883failure in it, the test will continue after the subroutine 884returns. This may not be what you want. 885 886Often people want fatal failures to propagate like exceptions. For 887that Google Test offers the following macros: 888 889| **Fatal assertion** | **Nonfatal assertion** | **Verifies** | 890|:--------------------|:-----------------------|:-------------| 891| `ASSERT_NO_FATAL_FAILURE(`_statement_`);` | `EXPECT_NO_FATAL_FAILURE(`_statement_`);` | _statement_ doesn't generate any new fatal failures in the current thread. | 892 893Only failures in the thread that executes the assertion are checked to 894determine the result of this type of assertions. If _statement_ 895creates new threads, failures in these threads are ignored. 896 897Examples: 898 899``` 900ASSERT_NO_FATAL_FAILURE(Foo()); 901 902int i; 903EXPECT_NO_FATAL_FAILURE({ 904 i = Bar(); 905}); 906``` 907 908_Availability:_ Linux, Windows, Mac. Assertions from multiple threads 909are currently not supported. 910 911### Checking for Failures in the Current Test ### 912 913`HasFatalFailure()` in the `::testing::Test` class returns `true` if an 914assertion in the current test has suffered a fatal failure. This 915allows functions to catch fatal failures in a sub-routine and return 916early. 917 918``` 919class Test { 920 public: 921 ... 922 static bool HasFatalFailure(); 923}; 924``` 925 926The typical usage, which basically simulates the behavior of a thrown 927exception, is: 928 929``` 930TEST(FooTest, Bar) { 931 Subroutine(); 932 // Aborts if Subroutine() had a fatal failure. 933 if (HasFatalFailure()) 934 return; 935 // The following won't be executed. 936 ... 937} 938``` 939 940If `HasFatalFailure()` is used outside of `TEST()` , `TEST_F()` , or a test 941fixture, you must add the `::testing::Test::` prefix, as in: 942 943``` 944if (::testing::Test::HasFatalFailure()) 945 return; 946``` 947 948Similarly, `HasNonfatalFailure()` returns `true` if the current test 949has at least one non-fatal failure, and `HasFailure()` returns `true` 950if the current test has at least one failure of either kind. 951 952_Availability:_ Linux, Windows, Mac. `HasNonfatalFailure()` and 953`HasFailure()` are available since version 1.4.0. 954 955# Logging Additional Information # 956 957In your test code, you can call `RecordProperty("key", value)` to log 958additional information, where `value` can be either a string or an `int`. The _last_ value recorded for a key will be emitted to the XML output 959if you specify one. For example, the test 960 961``` 962TEST_F(WidgetUsageTest, MinAndMaxWidgets) { 963 RecordProperty("MaximumWidgets", ComputeMaxUsage()); 964 RecordProperty("MinimumWidgets", ComputeMinUsage()); 965} 966``` 967 968will output XML like this: 969 970``` 971... 972 <testcase name="MinAndMaxWidgets" status="run" time="6" classname="WidgetUsageTest" 973 MaximumWidgets="12" 974 MinimumWidgets="9" /> 975... 976``` 977 978_Note_: 979 * `RecordProperty()` is a static member of the `Test` class. Therefore it needs to be prefixed with `::testing::Test::` if used outside of the `TEST` body and the test fixture class. 980 * `key` must be a valid XML attribute name, and cannot conflict with the ones already used by Google Test (`name`, `status`, `time`, `classname`, `type_param`, and `value_param`). 981 * Calling `RecordProperty()` outside of the lifespan of a test is allowed. If it's called outside of a test but between a test case's `SetUpTestCase()` and `TearDownTestCase()` methods, it will be attributed to the XML element for the test case. If it's called outside of all test cases (e.g. in a test environment), it will be attributed to the top-level XML element. 982 983_Availability_: Linux, Windows, Mac. 984 985# Sharing Resources Between Tests in the Same Test Case # 986 987 988 989Google Test creates a new test fixture object for each test in order to make 990tests independent and easier to debug. However, sometimes tests use resources 991that are expensive to set up, making the one-copy-per-test model prohibitively 992expensive. 993 994If the tests don't change the resource, there's no harm in them sharing a 995single resource copy. So, in addition to per-test set-up/tear-down, Google Test 996also supports per-test-case set-up/tear-down. To use it: 997 998 1. In your test fixture class (say `FooTest` ), define as `static` some member variables to hold the shared resources. 999 1. In the same test fixture class, define a `static void SetUpTestCase()` function (remember not to spell it as **`SetupTestCase`** with a small `u`!) to set up the shared resources and a `static void TearDownTestCase()` function to tear them down. 1000 1001That's it! Google Test automatically calls `SetUpTestCase()` before running the 1002_first test_ in the `FooTest` test case (i.e. before creating the first 1003`FooTest` object), and calls `TearDownTestCase()` after running the _last test_ 1004in it (i.e. after deleting the last `FooTest` object). In between, the tests 1005can use the shared resources. 1006 1007Remember that the test order is undefined, so your code can't depend on a test 1008preceding or following another. Also, the tests must either not modify the 1009state of any shared resource, or, if they do modify the state, they must 1010restore the state to its original value before passing control to the next 1011test. 1012 1013Here's an example of per-test-case set-up and tear-down: 1014``` 1015class FooTest : public ::testing::Test { 1016 protected: 1017 // Per-test-case set-up. 1018 // Called before the first test in this test case. 1019 // Can be omitted if not needed. 1020 static void SetUpTestCase() { 1021 shared_resource_ = new ...; 1022 } 1023 1024 // Per-test-case tear-down. 1025 // Called after the last test in this test case. 1026 // Can be omitted if not needed. 1027 static void TearDownTestCase() { 1028 delete shared_resource_; 1029 shared_resource_ = NULL; 1030 } 1031 1032 // You can define per-test set-up and tear-down logic as usual. 1033 virtual void SetUp() { ... } 1034 virtual void TearDown() { ... } 1035 1036 // Some expensive resource shared by all tests. 1037 static T* shared_resource_; 1038}; 1039 1040T* FooTest::shared_resource_ = NULL; 1041 1042TEST_F(FooTest, Test1) { 1043 ... you can refer to shared_resource here ... 1044} 1045TEST_F(FooTest, Test2) { 1046 ... you can refer to shared_resource here ... 1047} 1048``` 1049 1050_Availability:_ Linux, Windows, Mac. 1051 1052# Global Set-Up and Tear-Down # 1053 1054Just as you can do set-up and tear-down at the test level and the test case 1055level, you can also do it at the test program level. Here's how. 1056 1057First, you subclass the `::testing::Environment` class to define a test 1058environment, which knows how to set-up and tear-down: 1059 1060``` 1061class Environment { 1062 public: 1063 virtual ~Environment() {} 1064 // Override this to define how to set up the environment. 1065 virtual void SetUp() {} 1066 // Override this to define how to tear down the environment. 1067 virtual void TearDown() {} 1068}; 1069``` 1070 1071Then, you register an instance of your environment class with Google Test by 1072calling the `::testing::AddGlobalTestEnvironment()` function: 1073 1074``` 1075Environment* AddGlobalTestEnvironment(Environment* env); 1076``` 1077 1078Now, when `RUN_ALL_TESTS()` is called, it first calls the `SetUp()` method of 1079the environment object, then runs the tests if there was no fatal failures, and 1080finally calls `TearDown()` of the environment object. 1081 1082It's OK to register multiple environment objects. In this case, their `SetUp()` 1083will be called in the order they are registered, and their `TearDown()` will be 1084called in the reverse order. 1085 1086Note that Google Test takes ownership of the registered environment objects. 1087Therefore **do not delete them** by yourself. 1088 1089You should call `AddGlobalTestEnvironment()` before `RUN_ALL_TESTS()` is 1090called, probably in `main()`. If you use `gtest_main`, you need to call 1091this before `main()` starts for it to take effect. One way to do this is to 1092define a global variable like this: 1093 1094``` 1095::testing::Environment* const foo_env = ::testing::AddGlobalTestEnvironment(new FooEnvironment); 1096``` 1097 1098However, we strongly recommend you to write your own `main()` and call 1099`AddGlobalTestEnvironment()` there, as relying on initialization of global 1100variables makes the code harder to read and may cause problems when you 1101register multiple environments from different translation units and the 1102environments have dependencies among them (remember that the compiler doesn't 1103guarantee the order in which global variables from different translation units 1104are initialized). 1105 1106_Availability:_ Linux, Windows, Mac. 1107 1108 1109# Value Parameterized Tests # 1110 1111_Value-parameterized tests_ allow you to test your code with different 1112parameters without writing multiple copies of the same test. 1113 1114Suppose you write a test for your code and then realize that your code is affected by a presence of a Boolean command line flag. 1115 1116``` 1117TEST(MyCodeTest, TestFoo) { 1118 // A code to test foo(). 1119} 1120``` 1121 1122Usually people factor their test code into a function with a Boolean parameter in such situations. The function sets the flag, then executes the testing code. 1123 1124``` 1125void TestFooHelper(bool flag_value) { 1126 flag = flag_value; 1127 // A code to test foo(). 1128} 1129 1130TEST(MyCodeTest, TestFoo) { 1131 TestFooHelper(false); 1132 TestFooHelper(true); 1133} 1134``` 1135 1136But this setup has serious drawbacks. First, when a test assertion fails in your tests, it becomes unclear what value of the parameter caused it to fail. You can stream a clarifying message into your `EXPECT`/`ASSERT` statements, but it you'll have to do it with all of them. Second, you have to add one such helper function per test. What if you have ten tests? Twenty? A hundred? 1137 1138Value-parameterized tests will let you write your test only once and then easily instantiate and run it with an arbitrary number of parameter values. 1139 1140Here are some other situations when value-parameterized tests come handy: 1141 1142 * You want to test different implementations of an OO interface. 1143 * You want to test your code over various inputs (a.k.a. data-driven testing). This feature is easy to abuse, so please exercise your good sense when doing it! 1144 1145## How to Write Value-Parameterized Tests ## 1146 1147To write value-parameterized tests, first you should define a fixture 1148class. It must be derived from both `::testing::Test` and 1149`::testing::WithParamInterface<T>` (the latter is a pure interface), 1150where `T` is the type of your parameter values. For convenience, you 1151can just derive the fixture class from `::testing::TestWithParam<T>`, 1152which itself is derived from both `::testing::Test` and 1153`::testing::WithParamInterface<T>`. `T` can be any copyable type. If 1154it's a raw pointer, you are responsible for managing the lifespan of 1155the pointed values. 1156 1157``` 1158class FooTest : public ::testing::TestWithParam<const char*> { 1159 // You can implement all the usual fixture class members here. 1160 // To access the test parameter, call GetParam() from class 1161 // TestWithParam<T>. 1162}; 1163 1164// Or, when you want to add parameters to a pre-existing fixture class: 1165class BaseTest : public ::testing::Test { 1166 ... 1167}; 1168class BarTest : public BaseTest, 1169 public ::testing::WithParamInterface<const char*> { 1170 ... 1171}; 1172``` 1173 1174Then, use the `TEST_P` macro to define as many test patterns using 1175this fixture as you want. The `_P` suffix is for "parameterized" or 1176"pattern", whichever you prefer to think. 1177 1178``` 1179TEST_P(FooTest, DoesBlah) { 1180 // Inside a test, access the test parameter with the GetParam() method 1181 // of the TestWithParam<T> class: 1182 EXPECT_TRUE(foo.Blah(GetParam())); 1183 ... 1184} 1185 1186TEST_P(FooTest, HasBlahBlah) { 1187 ... 1188} 1189``` 1190 1191Finally, you can use `INSTANTIATE_TEST_CASE_P` to instantiate the test 1192case with any set of parameters you want. Google Test defines a number of 1193functions for generating test parameters. They return what we call 1194(surprise!) _parameter generators_. Here is a summary of them, 1195which are all in the `testing` namespace: 1196 1197| `Range(begin, end[, step])` | Yields values `{begin, begin+step, begin+step+step, ...}`. The values do not include `end`. `step` defaults to 1. | 1198|:----------------------------|:------------------------------------------------------------------------------------------------------------------| 1199| `Values(v1, v2, ..., vN)` | Yields values `{v1, v2, ..., vN}`. | 1200| `ValuesIn(container)` and `ValuesIn(begin, end)` | Yields values from a C-style array, an STL-style container, or an iterator range `[begin, end)`. `container`, `begin`, and `end` can be expressions whose values are determined at run time. | 1201| `Bool()` | Yields sequence `{false, true}`. | 1202| `Combine(g1, g2, ..., gN)` | Yields all combinations (the Cartesian product for the math savvy) of the values generated by the `N` generators. This is only available if your system provides the `<tr1/tuple>` header. If you are sure your system does, and Google Test disagrees, you can override it by defining `GTEST_HAS_TR1_TUPLE=1`. See comments in [include/gtest/internal/gtest-port.h](../include/gtest/internal/gtest-port.h) for more information. | 1203 1204For more details, see the comments at the definitions of these functions in the [source code](../include/gtest/gtest-param-test.h). 1205 1206The following statement will instantiate tests from the `FooTest` test case 1207each with parameter values `"meeny"`, `"miny"`, and `"moe"`. 1208 1209``` 1210INSTANTIATE_TEST_CASE_P(InstantiationName, 1211 FooTest, 1212 ::testing::Values("meeny", "miny", "moe")); 1213``` 1214 1215To distinguish different instances of the pattern (yes, you can 1216instantiate it more than once), the first argument to 1217`INSTANTIATE_TEST_CASE_P` is a prefix that will be added to the actual 1218test case name. Remember to pick unique prefixes for different 1219instantiations. The tests from the instantiation above will have these 1220names: 1221 1222 * `InstantiationName/FooTest.DoesBlah/0` for `"meeny"` 1223 * `InstantiationName/FooTest.DoesBlah/1` for `"miny"` 1224 * `InstantiationName/FooTest.DoesBlah/2` for `"moe"` 1225 * `InstantiationName/FooTest.HasBlahBlah/0` for `"meeny"` 1226 * `InstantiationName/FooTest.HasBlahBlah/1` for `"miny"` 1227 * `InstantiationName/FooTest.HasBlahBlah/2` for `"moe"` 1228 1229You can use these names in [--gtest\_filter](#running-a-subset-of-the-tests). 1230 1231This statement will instantiate all tests from `FooTest` again, each 1232with parameter values `"cat"` and `"dog"`: 1233 1234``` 1235const char* pets[] = {"cat", "dog"}; 1236INSTANTIATE_TEST_CASE_P(AnotherInstantiationName, FooTest, 1237 ::testing::ValuesIn(pets)); 1238``` 1239 1240The tests from the instantiation above will have these names: 1241 1242 * `AnotherInstantiationName/FooTest.DoesBlah/0` for `"cat"` 1243 * `AnotherInstantiationName/FooTest.DoesBlah/1` for `"dog"` 1244 * `AnotherInstantiationName/FooTest.HasBlahBlah/0` for `"cat"` 1245 * `AnotherInstantiationName/FooTest.HasBlahBlah/1` for `"dog"` 1246 1247Please note that `INSTANTIATE_TEST_CASE_P` will instantiate _all_ 1248tests in the given test case, whether their definitions come before or 1249_after_ the `INSTANTIATE_TEST_CASE_P` statement. 1250 1251You can see 1252[these](../samples/sample7_unittest.cc) 1253[files](../samples/sample8_unittest.cc) for more examples. 1254 1255_Availability_: Linux, Windows (requires MSVC 8.0 or above), Mac; since version 1.2.0. 1256 1257## Creating Value-Parameterized Abstract Tests ## 1258 1259In the above, we define and instantiate `FooTest` in the same source 1260file. Sometimes you may want to define value-parameterized tests in a 1261library and let other people instantiate them later. This pattern is 1262known as <i>abstract tests</i>. As an example of its application, when you 1263are designing an interface you can write a standard suite of abstract 1264tests (perhaps using a factory function as the test parameter) that 1265all implementations of the interface are expected to pass. When 1266someone implements the interface, he can instantiate your suite to get 1267all the interface-conformance tests for free. 1268 1269To define abstract tests, you should organize your code like this: 1270 1271 1. Put the definition of the parameterized test fixture class (e.g. `FooTest`) in a header file, say `foo_param_test.h`. Think of this as _declaring_ your abstract tests. 1272 1. Put the `TEST_P` definitions in `foo_param_test.cc`, which includes `foo_param_test.h`. Think of this as _implementing_ your abstract tests. 1273 1274Once they are defined, you can instantiate them by including 1275`foo_param_test.h`, invoking `INSTANTIATE_TEST_CASE_P()`, and linking 1276with `foo_param_test.cc`. You can instantiate the same abstract test 1277case multiple times, possibly in different source files. 1278 1279# Typed Tests # 1280 1281Suppose you have multiple implementations of the same interface and 1282want to make sure that all of them satisfy some common requirements. 1283Or, you may have defined several types that are supposed to conform to 1284the same "concept" and you want to verify it. In both cases, you want 1285the same test logic repeated for different types. 1286 1287While you can write one `TEST` or `TEST_F` for each type you want to 1288test (and you may even factor the test logic into a function template 1289that you invoke from the `TEST`), it's tedious and doesn't scale: 1290if you want _m_ tests over _n_ types, you'll end up writing _m\*n_ 1291`TEST`s. 1292 1293_Typed tests_ allow you to repeat the same test logic over a list of 1294types. You only need to write the test logic once, although you must 1295know the type list when writing typed tests. Here's how you do it: 1296 1297First, define a fixture class template. It should be parameterized 1298by a type. Remember to derive it from `::testing::Test`: 1299 1300``` 1301template <typename T> 1302class FooTest : public ::testing::Test { 1303 public: 1304 ... 1305 typedef std::list<T> List; 1306 static T shared_; 1307 T value_; 1308}; 1309``` 1310 1311Next, associate a list of types with the test case, which will be 1312repeated for each type in the list: 1313 1314``` 1315typedef ::testing::Types<char, int, unsigned int> MyTypes; 1316TYPED_TEST_CASE(FooTest, MyTypes); 1317``` 1318 1319The `typedef` is necessary for the `TYPED_TEST_CASE` macro to parse 1320correctly. Otherwise the compiler will think that each comma in the 1321type list introduces a new macro argument. 1322 1323Then, use `TYPED_TEST()` instead of `TEST_F()` to define a typed test 1324for this test case. You can repeat this as many times as you want: 1325 1326``` 1327TYPED_TEST(FooTest, DoesBlah) { 1328 // Inside a test, refer to the special name TypeParam to get the type 1329 // parameter. Since we are inside a derived class template, C++ requires 1330 // us to visit the members of FooTest via 'this'. 1331 TypeParam n = this->value_; 1332 1333 // To visit static members of the fixture, add the 'TestFixture::' 1334 // prefix. 1335 n += TestFixture::shared_; 1336 1337 // To refer to typedefs in the fixture, add the 'typename TestFixture::' 1338 // prefix. The 'typename' is required to satisfy the compiler. 1339 typename TestFixture::List values; 1340 values.push_back(n); 1341 ... 1342} 1343 1344TYPED_TEST(FooTest, HasPropertyA) { ... } 1345``` 1346 1347You can see `samples/sample6_unittest.cc` for a complete example. 1348 1349_Availability:_ Linux, Windows (requires MSVC 8.0 or above), Mac; 1350since version 1.1.0. 1351 1352# Type-Parameterized Tests # 1353 1354_Type-parameterized tests_ are like typed tests, except that they 1355don't require you to know the list of types ahead of time. Instead, 1356you can define the test logic first and instantiate it with different 1357type lists later. You can even instantiate it more than once in the 1358same program. 1359 1360If you are designing an interface or concept, you can define a suite 1361of type-parameterized tests to verify properties that any valid 1362implementation of the interface/concept should have. Then, the author 1363of each implementation can just instantiate the test suite with his 1364type to verify that it conforms to the requirements, without having to 1365write similar tests repeatedly. Here's an example: 1366 1367First, define a fixture class template, as we did with typed tests: 1368 1369``` 1370template <typename T> 1371class FooTest : public ::testing::Test { 1372 ... 1373}; 1374``` 1375 1376Next, declare that you will define a type-parameterized test case: 1377 1378``` 1379TYPED_TEST_CASE_P(FooTest); 1380``` 1381 1382The `_P` suffix is for "parameterized" or "pattern", whichever you 1383prefer to think. 1384 1385Then, use `TYPED_TEST_P()` to define a type-parameterized test. You 1386can repeat this as many times as you want: 1387 1388``` 1389TYPED_TEST_P(FooTest, DoesBlah) { 1390 // Inside a test, refer to TypeParam to get the type parameter. 1391 TypeParam n = 0; 1392 ... 1393} 1394 1395TYPED_TEST_P(FooTest, HasPropertyA) { ... } 1396``` 1397 1398Now the tricky part: you need to register all test patterns using the 1399`REGISTER_TYPED_TEST_CASE_P` macro before you can instantiate them. 1400The first argument of the macro is the test case name; the rest are 1401the names of the tests in this test case: 1402 1403``` 1404REGISTER_TYPED_TEST_CASE_P(FooTest, 1405 DoesBlah, HasPropertyA); 1406``` 1407 1408Finally, you are free to instantiate the pattern with the types you 1409want. If you put the above code in a header file, you can `#include` 1410it in multiple C++ source files and instantiate it multiple times. 1411 1412``` 1413typedef ::testing::Types<char, int, unsigned int> MyTypes; 1414INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, MyTypes); 1415``` 1416 1417To distinguish different instances of the pattern, the first argument 1418to the `INSTANTIATE_TYPED_TEST_CASE_P` macro is a prefix that will be 1419added to the actual test case name. Remember to pick unique prefixes 1420for different instances. 1421 1422In the special case where the type list contains only one type, you 1423can write that type directly without `::testing::Types<...>`, like this: 1424 1425``` 1426INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, int); 1427``` 1428 1429You can see `samples/sample6_unittest.cc` for a complete example. 1430 1431_Availability:_ Linux, Windows (requires MSVC 8.0 or above), Mac; 1432since version 1.1.0. 1433 1434# Testing Private Code # 1435 1436If you change your software's internal implementation, your tests should not 1437break as long as the change is not observable by users. Therefore, per the 1438_black-box testing principle_, most of the time you should test your code 1439through its public interfaces. 1440 1441If you still find yourself needing to test internal implementation code, 1442consider if there's a better design that wouldn't require you to do so. If you 1443absolutely have to test non-public interface code though, you can. There are 1444two cases to consider: 1445 1446 * Static functions (_not_ the same as static member functions!) or unnamed namespaces, and 1447 * Private or protected class members 1448 1449## Static Functions ## 1450 1451Both static functions and definitions/declarations in an unnamed namespace are 1452only visible within the same translation unit. To test them, you can `#include` 1453the entire `.cc` file being tested in your `*_test.cc` file. (`#include`ing `.cc` 1454files is not a good way to reuse code - you should not do this in production 1455code!) 1456 1457However, a better approach is to move the private code into the 1458`foo::internal` namespace, where `foo` is the namespace your project normally 1459uses, and put the private declarations in a `*-internal.h` file. Your 1460production `.cc` files and your tests are allowed to include this internal 1461header, but your clients are not. This way, you can fully test your internal 1462implementation without leaking it to your clients. 1463 1464## Private Class Members ## 1465 1466Private class members are only accessible from within the class or by friends. 1467To access a class' private members, you can declare your test fixture as a 1468friend to the class and define accessors in your fixture. Tests using the 1469fixture can then access the private members of your production class via the 1470accessors in the fixture. Note that even though your fixture is a friend to 1471your production class, your tests are not automatically friends to it, as they 1472are technically defined in sub-classes of the fixture. 1473 1474Another way to test private members is to refactor them into an implementation 1475class, which is then declared in a `*-internal.h` file. Your clients aren't 1476allowed to include this header but your tests can. Such is called the Pimpl 1477(Private Implementation) idiom. 1478 1479Or, you can declare an individual test as a friend of your class by adding this 1480line in the class body: 1481 1482``` 1483FRIEND_TEST(TestCaseName, TestName); 1484``` 1485 1486For example, 1487``` 1488// foo.h 1489#include "gtest/gtest_prod.h" 1490 1491// Defines FRIEND_TEST. 1492class Foo { 1493 ... 1494 private: 1495 FRIEND_TEST(FooTest, BarReturnsZeroOnNull); 1496 int Bar(void* x); 1497}; 1498 1499// foo_test.cc 1500... 1501TEST(FooTest, BarReturnsZeroOnNull) { 1502 Foo foo; 1503 EXPECT_EQ(0, foo.Bar(NULL)); 1504 // Uses Foo's private member Bar(). 1505} 1506``` 1507 1508Pay special attention when your class is defined in a namespace, as you should 1509define your test fixtures and tests in the same namespace if you want them to 1510be friends of your class. For example, if the code to be tested looks like: 1511 1512``` 1513namespace my_namespace { 1514 1515class Foo { 1516 friend class FooTest; 1517 FRIEND_TEST(FooTest, Bar); 1518 FRIEND_TEST(FooTest, Baz); 1519 ... 1520 definition of the class Foo 1521 ... 1522}; 1523 1524} // namespace my_namespace 1525``` 1526 1527Your test code should be something like: 1528 1529``` 1530namespace my_namespace { 1531class FooTest : public ::testing::Test { 1532 protected: 1533 ... 1534}; 1535 1536TEST_F(FooTest, Bar) { ... } 1537TEST_F(FooTest, Baz) { ... } 1538 1539} // namespace my_namespace 1540``` 1541 1542# Catching Failures # 1543 1544If you are building a testing utility on top of Google Test, you'll 1545want to test your utility. What framework would you use to test it? 1546Google Test, of course. 1547 1548The challenge is to verify that your testing utility reports failures 1549correctly. In frameworks that report a failure by throwing an 1550exception, you could catch the exception and assert on it. But Google 1551Test doesn't use exceptions, so how do we test that a piece of code 1552generates an expected failure? 1553 1554`"gtest/gtest-spi.h"` contains some constructs to do this. After 1555`#include`ing this header, you can use 1556 1557| `EXPECT_FATAL_FAILURE(`_statement, substring_`);` | 1558|:--------------------------------------------------| 1559 1560to assert that _statement_ generates a fatal (e.g. `ASSERT_*`) failure 1561whose message contains the given _substring_, or use 1562 1563| `EXPECT_NONFATAL_FAILURE(`_statement, substring_`);` | 1564|:-----------------------------------------------------| 1565 1566if you are expecting a non-fatal (e.g. `EXPECT_*`) failure. 1567 1568For technical reasons, there are some caveats: 1569 1570 1. You cannot stream a failure message to either macro. 1571 1. _statement_ in `EXPECT_FATAL_FAILURE()` cannot reference local non-static variables or non-static members of `this` object. 1572 1. _statement_ in `EXPECT_FATAL_FAILURE()` cannot return a value. 1573 1574_Note:_ Google Test is designed with threads in mind. Once the 1575synchronization primitives in `"gtest/internal/gtest-port.h"` have 1576been implemented, Google Test will become thread-safe, meaning that 1577you can then use assertions in multiple threads concurrently. Before 1578that, however, Google Test only supports single-threaded usage. Once 1579thread-safe, `EXPECT_FATAL_FAILURE()` and `EXPECT_NONFATAL_FAILURE()` 1580will capture failures in the current thread only. If _statement_ 1581creates new threads, failures in these threads will be ignored. If 1582you want to capture failures from all threads instead, you should use 1583the following macros: 1584 1585| `EXPECT_FATAL_FAILURE_ON_ALL_THREADS(`_statement, substring_`);` | 1586|:-----------------------------------------------------------------| 1587| `EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(`_statement, substring_`);` | 1588 1589# Getting the Current Test's Name # 1590 1591Sometimes a function may need to know the name of the currently running test. 1592For example, you may be using the `SetUp()` method of your test fixture to set 1593the golden file name based on which test is running. The `::testing::TestInfo` 1594class has this information: 1595 1596``` 1597namespace testing { 1598 1599class TestInfo { 1600 public: 1601 // Returns the test case name and the test name, respectively. 1602 // 1603 // Do NOT delete or free the return value - it's managed by the 1604 // TestInfo class. 1605 const char* test_case_name() const; 1606 const char* name() const; 1607}; 1608 1609} // namespace testing 1610``` 1611 1612 1613> To obtain a `TestInfo` object for the currently running test, call 1614`current_test_info()` on the `UnitTest` singleton object: 1615 1616``` 1617// Gets information about the currently running test. 1618// Do NOT delete the returned object - it's managed by the UnitTest class. 1619const ::testing::TestInfo* const test_info = 1620 ::testing::UnitTest::GetInstance()->current_test_info(); 1621printf("We are in test %s of test case %s.\n", 1622 test_info->name(), test_info->test_case_name()); 1623``` 1624 1625`current_test_info()` returns a null pointer if no test is running. In 1626particular, you cannot find the test case name in `TestCaseSetUp()`, 1627`TestCaseTearDown()` (where you know the test case name implicitly), or 1628functions called from them. 1629 1630_Availability:_ Linux, Windows, Mac. 1631 1632# Extending Google Test by Handling Test Events # 1633 1634Google Test provides an <b>event listener API</b> to let you receive 1635notifications about the progress of a test program and test 1636failures. The events you can listen to include the start and end of 1637the test program, a test case, or a test method, among others. You may 1638use this API to augment or replace the standard console output, 1639replace the XML output, or provide a completely different form of 1640output, such as a GUI or a database. You can also use test events as 1641checkpoints to implement a resource leak checker, for example. 1642 1643_Availability:_ Linux, Windows, Mac; since v1.4.0. 1644 1645## Defining Event Listeners ## 1646 1647To define a event listener, you subclass either 1648[testing::TestEventListener](../include/gtest/gtest.h#L991) 1649or [testing::EmptyTestEventListener](../include/gtest/gtest.h#L1044). 1650The former is an (abstract) interface, where <i>each pure virtual method<br> 1651can be overridden to handle a test event</i> (For example, when a test 1652starts, the `OnTestStart()` method will be called.). The latter provides 1653an empty implementation of all methods in the interface, such that a 1654subclass only needs to override the methods it cares about. 1655 1656When an event is fired, its context is passed to the handler function 1657as an argument. The following argument types are used: 1658 * [UnitTest](../include/gtest/gtest.h#L1151) reflects the state of the entire test program, 1659 * [TestCase](../include/gtest/gtest.h#L778) has information about a test case, which can contain one or more tests, 1660 * [TestInfo](../include/gtest/gtest.h#L644) contains the state of a test, and 1661 * [TestPartResult](../include/gtest/gtest-test-part.h#L47) represents the result of a test assertion. 1662 1663An event handler function can examine the argument it receives to find 1664out interesting information about the event and the test program's 1665state. Here's an example: 1666 1667``` 1668 class MinimalistPrinter : public ::testing::EmptyTestEventListener { 1669 // Called before a test starts. 1670 virtual void OnTestStart(const ::testing::TestInfo& test_info) { 1671 printf("*** Test %s.%s starting.\n", 1672 test_info.test_case_name(), test_info.name()); 1673 } 1674 1675 // Called after a failed assertion or a SUCCEED() invocation. 1676 virtual void OnTestPartResult( 1677 const ::testing::TestPartResult& test_part_result) { 1678 printf("%s in %s:%d\n%s\n", 1679 test_part_result.failed() ? "*** Failure" : "Success", 1680 test_part_result.file_name(), 1681 test_part_result.line_number(), 1682 test_part_result.summary()); 1683 } 1684 1685 // Called after a test ends. 1686 virtual void OnTestEnd(const ::testing::TestInfo& test_info) { 1687 printf("*** Test %s.%s ending.\n", 1688 test_info.test_case_name(), test_info.name()); 1689 } 1690 }; 1691``` 1692 1693## Using Event Listeners ## 1694 1695To use the event listener you have defined, add an instance of it to 1696the Google Test event listener list (represented by class 1697[TestEventListeners](../include/gtest/gtest.h#L1064) 1698- note the "s" at the end of the name) in your 1699`main()` function, before calling `RUN_ALL_TESTS()`: 1700``` 1701int main(int argc, char** argv) { 1702 ::testing::InitGoogleTest(&argc, argv); 1703 // Gets hold of the event listener list. 1704 ::testing::TestEventListeners& listeners = 1705 ::testing::UnitTest::GetInstance()->listeners(); 1706 // Adds a listener to the end. Google Test takes the ownership. 1707 listeners.Append(new MinimalistPrinter); 1708 return RUN_ALL_TESTS(); 1709} 1710``` 1711 1712There's only one problem: the default test result printer is still in 1713effect, so its output will mingle with the output from your minimalist 1714printer. To suppress the default printer, just release it from the 1715event listener list and delete it. You can do so by adding one line: 1716``` 1717 ... 1718 delete listeners.Release(listeners.default_result_printer()); 1719 listeners.Append(new MinimalistPrinter); 1720 return RUN_ALL_TESTS(); 1721``` 1722 1723Now, sit back and enjoy a completely different output from your 1724tests. For more details, you can read this 1725[sample](../samples/sample9_unittest.cc). 1726 1727You may append more than one listener to the list. When an `On*Start()` 1728or `OnTestPartResult()` event is fired, the listeners will receive it in 1729the order they appear in the list (since new listeners are added to 1730the end of the list, the default text printer and the default XML 1731generator will receive the event first). An `On*End()` event will be 1732received by the listeners in the _reverse_ order. This allows output by 1733listeners added later to be framed by output from listeners added 1734earlier. 1735 1736## Generating Failures in Listeners ## 1737 1738You may use failure-raising macros (`EXPECT_*()`, `ASSERT_*()`, 1739`FAIL()`, etc) when processing an event. There are some restrictions: 1740 1741 1. You cannot generate any failure in `OnTestPartResult()` (otherwise it will cause `OnTestPartResult()` to be called recursively). 1742 1. A listener that handles `OnTestPartResult()` is not allowed to generate any failure. 1743 1744When you add listeners to the listener list, you should put listeners 1745that handle `OnTestPartResult()` _before_ listeners that can generate 1746failures. This ensures that failures generated by the latter are 1747attributed to the right test by the former. 1748 1749We have a sample of failure-raising listener 1750[here](../samples/sample10_unittest.cc). 1751 1752# Running Test Programs: Advanced Options # 1753 1754Google Test test programs are ordinary executables. Once built, you can run 1755them directly and affect their behavior via the following environment variables 1756and/or command line flags. For the flags to work, your programs must call 1757`::testing::InitGoogleTest()` before calling `RUN_ALL_TESTS()`. 1758 1759To see a list of supported flags and their usage, please run your test 1760program with the `--help` flag. You can also use `-h`, `-?`, or `/?` 1761for short. This feature is added in version 1.3.0. 1762 1763If an option is specified both by an environment variable and by a 1764flag, the latter takes precedence. Most of the options can also be 1765set/read in code: to access the value of command line flag 1766`--gtest_foo`, write `::testing::GTEST_FLAG(foo)`. A common pattern is 1767to set the value of a flag before calling `::testing::InitGoogleTest()` 1768to change the default value of the flag: 1769``` 1770int main(int argc, char** argv) { 1771 // Disables elapsed time by default. 1772 ::testing::GTEST_FLAG(print_time) = false; 1773 1774 // This allows the user to override the flag on the command line. 1775 ::testing::InitGoogleTest(&argc, argv); 1776 1777 return RUN_ALL_TESTS(); 1778} 1779``` 1780 1781## Selecting Tests ## 1782 1783This section shows various options for choosing which tests to run. 1784 1785### Listing Test Names ### 1786 1787Sometimes it is necessary to list the available tests in a program before 1788running them so that a filter may be applied if needed. Including the flag 1789`--gtest_list_tests` overrides all other flags and lists tests in the following 1790format: 1791``` 1792TestCase1. 1793 TestName1 1794 TestName2 1795TestCase2. 1796 TestName 1797``` 1798 1799None of the tests listed are actually run if the flag is provided. There is no 1800corresponding environment variable for this flag. 1801 1802_Availability:_ Linux, Windows, Mac. 1803 1804### Running a Subset of the Tests ### 1805 1806By default, a Google Test program runs all tests the user has defined. 1807Sometimes, you want to run only a subset of the tests (e.g. for debugging or 1808quickly verifying a change). If you set the `GTEST_FILTER` environment variable 1809or the `--gtest_filter` flag to a filter string, Google Test will only run the 1810tests whose full names (in the form of `TestCaseName.TestName`) match the 1811filter. 1812 1813The format of a filter is a '`:`'-separated list of wildcard patterns (called 1814the positive patterns) optionally followed by a '`-`' and another 1815'`:`'-separated pattern list (called the negative patterns). A test matches the 1816filter if and only if it matches any of the positive patterns but does not 1817match any of the negative patterns. 1818 1819A pattern may contain `'*'` (matches any string) or `'?'` (matches any single 1820character). For convenience, the filter `'*-NegativePatterns'` can be also 1821written as `'-NegativePatterns'`. 1822 1823For example: 1824 1825 * `./foo_test` Has no flag, and thus runs all its tests. 1826 * `./foo_test --gtest_filter=*` Also runs everything, due to the single match-everything `*` value. 1827 * `./foo_test --gtest_filter=FooTest.*` Runs everything in test case `FooTest`. 1828 * `./foo_test --gtest_filter=*Null*:*Constructor*` Runs any test whose full name contains either `"Null"` or `"Constructor"`. 1829 * `./foo_test --gtest_filter=-*DeathTest.*` Runs all non-death tests. 1830 * `./foo_test --gtest_filter=FooTest.*-FooTest.Bar` Runs everything in test case `FooTest` except `FooTest.Bar`. 1831 1832_Availability:_ Linux, Windows, Mac. 1833 1834### Temporarily Disabling Tests ### 1835 1836If you have a broken test that you cannot fix right away, you can add the 1837`DISABLED_` prefix to its name. This will exclude it from execution. This is 1838better than commenting out the code or using `#if 0`, as disabled tests are 1839still compiled (and thus won't rot). 1840 1841If you need to disable all tests in a test case, you can either add `DISABLED_` 1842to the front of the name of each test, or alternatively add it to the front of 1843the test case name. 1844 1845For example, the following tests won't be run by Google Test, even though they 1846will still be compiled: 1847 1848``` 1849// Tests that Foo does Abc. 1850TEST(FooTest, DISABLED_DoesAbc) { ... } 1851 1852class DISABLED_BarTest : public ::testing::Test { ... }; 1853 1854// Tests that Bar does Xyz. 1855TEST_F(DISABLED_BarTest, DoesXyz) { ... } 1856``` 1857 1858_Note:_ This feature should only be used for temporary pain-relief. You still 1859have to fix the disabled tests at a later date. As a reminder, Google Test will 1860print a banner warning you if a test program contains any disabled tests. 1861 1862_Tip:_ You can easily count the number of disabled tests you have 1863using `grep`. This number can be used as a metric for improving your 1864test quality. 1865 1866_Availability:_ Linux, Windows, Mac. 1867 1868### Temporarily Enabling Disabled Tests ### 1869 1870To include [disabled tests](#temporarily-disabling-tests) in test 1871execution, just invoke the test program with the 1872`--gtest_also_run_disabled_tests` flag or set the 1873`GTEST_ALSO_RUN_DISABLED_TESTS` environment variable to a value other 1874than `0`. You can combine this with the 1875[--gtest\_filter](#running-a-subset-of-the-tests) flag to further select 1876which disabled tests to run. 1877 1878_Availability:_ Linux, Windows, Mac; since version 1.3.0. 1879 1880## Repeating the Tests ## 1881 1882Once in a while you'll run into a test whose result is hit-or-miss. Perhaps it 1883will fail only 1% of the time, making it rather hard to reproduce the bug under 1884a debugger. This can be a major source of frustration. 1885 1886The `--gtest_repeat` flag allows you to repeat all (or selected) test methods 1887in a program many times. Hopefully, a flaky test will eventually fail and give 1888you a chance to debug. Here's how to use it: 1889 1890| `$ foo_test --gtest_repeat=1000` | Repeat foo\_test 1000 times and don't stop at failures. | 1891|:---------------------------------|:--------------------------------------------------------| 1892| `$ foo_test --gtest_repeat=-1` | A negative count means repeating forever. | 1893| `$ foo_test --gtest_repeat=1000 --gtest_break_on_failure` | Repeat foo\_test 1000 times, stopping at the first failure. This is especially useful when running under a debugger: when the testfails, it will drop into the debugger and you can then inspect variables and stacks. | 1894| `$ foo_test --gtest_repeat=1000 --gtest_filter=FooBar` | Repeat the tests whose name matches the filter 1000 times. | 1895 1896If your test program contains global set-up/tear-down code registered 1897using `AddGlobalTestEnvironment()`, it will be repeated in each 1898iteration as well, as the flakiness may be in it. You can also specify 1899the repeat count by setting the `GTEST_REPEAT` environment variable. 1900 1901_Availability:_ Linux, Windows, Mac. 1902 1903## Shuffling the Tests ## 1904 1905You can specify the `--gtest_shuffle` flag (or set the `GTEST_SHUFFLE` 1906environment variable to `1`) to run the tests in a program in a random 1907order. This helps to reveal bad dependencies between tests. 1908 1909By default, Google Test uses a random seed calculated from the current 1910time. Therefore you'll get a different order every time. The console 1911output includes the random seed value, such that you can reproduce an 1912order-related test failure later. To specify the random seed 1913explicitly, use the `--gtest_random_seed=SEED` flag (or set the 1914`GTEST_RANDOM_SEED` environment variable), where `SEED` is an integer 1915between 0 and 99999. The seed value 0 is special: it tells Google Test 1916to do the default behavior of calculating the seed from the current 1917time. 1918 1919If you combine this with `--gtest_repeat=N`, Google Test will pick a 1920different random seed and re-shuffle the tests in each iteration. 1921 1922_Availability:_ Linux, Windows, Mac; since v1.4.0. 1923 1924## Controlling Test Output ## 1925 1926This section teaches how to tweak the way test results are reported. 1927 1928### Colored Terminal Output ### 1929 1930Google Test can use colors in its terminal output to make it easier to spot 1931the separation between tests, and whether tests passed. 1932 1933You can set the GTEST\_COLOR environment variable or set the `--gtest_color` 1934command line flag to `yes`, `no`, or `auto` (the default) to enable colors, 1935disable colors, or let Google Test decide. When the value is `auto`, Google 1936Test will use colors if and only if the output goes to a terminal and (on 1937non-Windows platforms) the `TERM` environment variable is set to `xterm` or 1938`xterm-color`. 1939 1940_Availability:_ Linux, Windows, Mac. 1941 1942### Suppressing the Elapsed Time ### 1943 1944By default, Google Test prints the time it takes to run each test. To 1945suppress that, run the test program with the `--gtest_print_time=0` 1946command line flag. Setting the `GTEST_PRINT_TIME` environment 1947variable to `0` has the same effect. 1948 1949_Availability:_ Linux, Windows, Mac. (In Google Test 1.3.0 and lower, 1950the default behavior is that the elapsed time is **not** printed.) 1951 1952### Generating an XML Report ### 1953 1954Google Test can emit a detailed XML report to a file in addition to its normal 1955textual output. The report contains the duration of each test, and thus can 1956help you identify slow tests. 1957 1958To generate the XML report, set the `GTEST_OUTPUT` environment variable or the 1959`--gtest_output` flag to the string `"xml:_path_to_output_file_"`, which will 1960create the file at the given location. You can also just use the string 1961`"xml"`, in which case the output can be found in the `test_detail.xml` file in 1962the current directory. 1963 1964If you specify a directory (for example, `"xml:output/directory/"` on Linux or 1965`"xml:output\directory\"` on Windows), Google Test will create the XML file in 1966that directory, named after the test executable (e.g. `foo_test.xml` for test 1967program `foo_test` or `foo_test.exe`). If the file already exists (perhaps left 1968over from a previous run), Google Test will pick a different name (e.g. 1969`foo_test_1.xml`) to avoid overwriting it. 1970 1971The report uses the format described here. It is based on the 1972`junitreport` Ant task and can be parsed by popular continuous build 1973systems like [Hudson](https://hudson.dev.java.net/). Since that format 1974was originally intended for Java, a little interpretation is required 1975to make it apply to Google Test tests, as shown here: 1976 1977``` 1978<testsuites name="AllTests" ...> 1979 <testsuite name="test_case_name" ...> 1980 <testcase name="test_name" ...> 1981 <failure message="..."/> 1982 <failure message="..."/> 1983 <failure message="..."/> 1984 </testcase> 1985 </testsuite> 1986</testsuites> 1987``` 1988 1989 * The root `<testsuites>` element corresponds to the entire test program. 1990 * `<testsuite>` elements correspond to Google Test test cases. 1991 * `<testcase>` elements correspond to Google Test test functions. 1992 1993For instance, the following program 1994 1995``` 1996TEST(MathTest, Addition) { ... } 1997TEST(MathTest, Subtraction) { ... } 1998TEST(LogicTest, NonContradiction) { ... } 1999``` 2000 2001could generate this report: 2002 2003``` 2004<?xml version="1.0" encoding="UTF-8"?> 2005<testsuites tests="3" failures="1" errors="0" time="35" name="AllTests"> 2006 <testsuite name="MathTest" tests="2" failures="1" errors="0" time="15"> 2007 <testcase name="Addition" status="run" time="7" classname=""> 2008 <failure message="Value of: add(1, 1)
 Actual: 3
Expected: 2" type=""/> 2009 <failure message="Value of: add(1, -1)
 Actual: 1
Expected: 0" type=""/> 2010 </testcase> 2011 <testcase name="Subtraction" status="run" time="5" classname=""> 2012 </testcase> 2013 </testsuite> 2014 <testsuite name="LogicTest" tests="1" failures="0" errors="0" time="5"> 2015 <testcase name="NonContradiction" status="run" time="5" classname=""> 2016 </testcase> 2017 </testsuite> 2018</testsuites> 2019``` 2020 2021Things to note: 2022 2023 * The `tests` attribute of a `<testsuites>` or `<testsuite>` element tells how many test functions the Google Test program or test case contains, while the `failures` attribute tells how many of them failed. 2024 * The `time` attribute expresses the duration of the test, test case, or entire test program in milliseconds. 2025 * Each `<failure>` element corresponds to a single failed Google Test assertion. 2026 * Some JUnit concepts don't apply to Google Test, yet we have to conform to the DTD. Therefore you'll see some dummy elements and attributes in the report. You can safely ignore these parts. 2027 2028_Availability:_ Linux, Windows, Mac. 2029 2030## Controlling How Failures Are Reported ## 2031 2032### Turning Assertion Failures into Break-Points ### 2033 2034When running test programs under a debugger, it's very convenient if the 2035debugger can catch an assertion failure and automatically drop into interactive 2036mode. Google Test's _break-on-failure_ mode supports this behavior. 2037 2038To enable it, set the `GTEST_BREAK_ON_FAILURE` environment variable to a value 2039other than `0` . Alternatively, you can use the `--gtest_break_on_failure` 2040command line flag. 2041 2042_Availability:_ Linux, Windows, Mac. 2043 2044### Disabling Catching Test-Thrown Exceptions ### 2045 2046Google Test can be used either with or without exceptions enabled. If 2047a test throws a C++ exception or (on Windows) a structured exception 2048(SEH), by default Google Test catches it, reports it as a test 2049failure, and continues with the next test method. This maximizes the 2050coverage of a test run. Also, on Windows an uncaught exception will 2051cause a pop-up window, so catching the exceptions allows you to run 2052the tests automatically. 2053 2054When debugging the test failures, however, you may instead want the 2055exceptions to be handled by the debugger, such that you can examine 2056the call stack when an exception is thrown. To achieve that, set the 2057`GTEST_CATCH_EXCEPTIONS` environment variable to `0`, or use the 2058`--gtest_catch_exceptions=0` flag when running the tests. 2059 2060**Availability**: Linux, Windows, Mac. 2061 2062### Letting Another Testing Framework Drive ### 2063 2064If you work on a project that has already been using another testing 2065framework and is not ready to completely switch to Google Test yet, 2066you can get much of Google Test's benefit by using its assertions in 2067your existing tests. Just change your `main()` function to look 2068like: 2069 2070``` 2071#include "gtest/gtest.h" 2072 2073int main(int argc, char** argv) { 2074 ::testing::GTEST_FLAG(throw_on_failure) = true; 2075 // Important: Google Test must be initialized. 2076 ::testing::InitGoogleTest(&argc, argv); 2077 2078 ... whatever your existing testing framework requires ... 2079} 2080``` 2081 2082With that, you can use Google Test assertions in addition to the 2083native assertions your testing framework provides, for example: 2084 2085``` 2086void TestFooDoesBar() { 2087 Foo foo; 2088 EXPECT_LE(foo.Bar(1), 100); // A Google Test assertion. 2089 CPPUNIT_ASSERT(foo.IsEmpty()); // A native assertion. 2090} 2091``` 2092 2093If a Google Test assertion fails, it will print an error message and 2094throw an exception, which will be treated as a failure by your host 2095testing framework. If you compile your code with exceptions disabled, 2096a failed Google Test assertion will instead exit your program with a 2097non-zero code, which will also signal a test failure to your test 2098runner. 2099 2100If you don't write `::testing::GTEST_FLAG(throw_on_failure) = true;` in 2101your `main()`, you can alternatively enable this feature by specifying 2102the `--gtest_throw_on_failure` flag on the command-line or setting the 2103`GTEST_THROW_ON_FAILURE` environment variable to a non-zero value. 2104 2105Death tests are _not_ supported when other test framework is used to organize tests. 2106 2107_Availability:_ Linux, Windows, Mac; since v1.3.0. 2108 2109## Distributing Test Functions to Multiple Machines ## 2110 2111If you have more than one machine you can use to run a test program, 2112you might want to run the test functions in parallel and get the 2113result faster. We call this technique _sharding_, where each machine 2114is called a _shard_. 2115 2116Google Test is compatible with test sharding. To take advantage of 2117this feature, your test runner (not part of Google Test) needs to do 2118the following: 2119 2120 1. Allocate a number of machines (shards) to run the tests. 2121 1. On each shard, set the `GTEST_TOTAL_SHARDS` environment variable to the total number of shards. It must be the same for all shards. 2122 1. On each shard, set the `GTEST_SHARD_INDEX` environment variable to the index of the shard. Different shards must be assigned different indices, which must be in the range `[0, GTEST_TOTAL_SHARDS - 1]`. 2123 1. Run the same test program on all shards. When Google Test sees the above two environment variables, it will select a subset of the test functions to run. Across all shards, each test function in the program will be run exactly once. 2124 1. Wait for all shards to finish, then collect and report the results. 2125 2126Your project may have tests that were written without Google Test and 2127thus don't understand this protocol. In order for your test runner to 2128figure out which test supports sharding, it can set the environment 2129variable `GTEST_SHARD_STATUS_FILE` to a non-existent file path. If a 2130test program supports sharding, it will create this file to 2131acknowledge the fact (the actual contents of the file are not 2132important at this time; although we may stick some useful information 2133in it in the future.); otherwise it will not create it. 2134 2135Here's an example to make it clear. Suppose you have a test program 2136`foo_test` that contains the following 5 test functions: 2137``` 2138TEST(A, V) 2139TEST(A, W) 2140TEST(B, X) 2141TEST(B, Y) 2142TEST(B, Z) 2143``` 2144and you have 3 machines at your disposal. To run the test functions in 2145parallel, you would set `GTEST_TOTAL_SHARDS` to 3 on all machines, and 2146set `GTEST_SHARD_INDEX` to 0, 1, and 2 on the machines respectively. 2147Then you would run the same `foo_test` on each machine. 2148 2149Google Test reserves the right to change how the work is distributed 2150across the shards, but here's one possible scenario: 2151 2152 * Machine #0 runs `A.V` and `B.X`. 2153 * Machine #1 runs `A.W` and `B.Y`. 2154 * Machine #2 runs `B.Z`. 2155 2156_Availability:_ Linux, Windows, Mac; since version 1.3.0. 2157 2158# Fusing Google Test Source Files # 2159 2160Google Test's implementation consists of ~30 files (excluding its own 2161tests). Sometimes you may want them to be packaged up in two files (a 2162`.h` and a `.cc`) instead, such that you can easily copy them to a new 2163machine and start hacking there. For this we provide an experimental 2164Python script `fuse_gtest_files.py` in the `scripts/` directory (since release 1.3.0). 2165Assuming you have Python 2.4 or above installed on your machine, just 2166go to that directory and run 2167``` 2168python fuse_gtest_files.py OUTPUT_DIR 2169``` 2170 2171and you should see an `OUTPUT_DIR` directory being created with files 2172`gtest/gtest.h` and `gtest/gtest-all.cc` in it. These files contain 2173everything you need to use Google Test. Just copy them to anywhere 2174you want and you are ready to write tests. You can use the 2175[scripts/test/Makefile](../scripts/test/Makefile) 2176file as an example on how to compile your tests against them. 2177 2178# Where to Go from Here # 2179 2180Congratulations! You've now learned more advanced Google Test tools and are 2181ready to tackle more complex testing tasks. If you want to dive even deeper, you 2182can read the [Frequently-Asked Questions](FAQ.md). 2183