...
Run Format

Source file src/testing/testing.go

Documentation: testing

  // Copyright 2009 The Go Authors. All rights reserved.
  // Use of this source code is governed by a BSD-style
  // license that can be found in the LICENSE file.
  
  // Package testing provides support for automated testing of Go packages.
  // It is intended to be used in concert with the ``go test'' command, which automates
  // execution of any function of the form
  //     func TestXxx(*testing.T)
  // where Xxx can be any alphanumeric string (but the first letter must not be in
  // [a-z]) and serves to identify the test routine.
  //
  // Within these functions, use the Error, Fail or related methods to signal failure.
  //
  // To write a new test suite, create a file whose name ends _test.go that
  // contains the TestXxx functions as described here. Put the file in the same
  // package as the one being tested. The file will be excluded from regular
  // package builds but will be included when the ``go test'' command is run.
  // For more detail, run ``go help test'' and ``go help testflag''.
  //
  // Tests and benchmarks may be skipped if not applicable with a call to
  // the Skip method of *T and *B:
  //     func TestTimeConsuming(t *testing.T) {
  //         if testing.Short() {
  //             t.Skip("skipping test in short mode.")
  //         }
  //         ...
  //     }
  //
  // Benchmarks
  //
  // Functions of the form
  //     func BenchmarkXxx(*testing.B)
  // are considered benchmarks, and are executed by the "go test" command when
  // its -bench flag is provided. Benchmarks are run sequentially.
  //
  // For a description of the testing flags, see
  // https://golang.org/cmd/go/#hdr-Description_of_testing_flags.
  //
  // A sample benchmark function looks like this:
  //     func BenchmarkHello(b *testing.B) {
  //         for i := 0; i < b.N; i++ {
  //             fmt.Sprintf("hello")
  //         }
  //     }
  //
  // The benchmark function must run the target code b.N times.
  // During benchmark execution, b.N is adjusted until the benchmark function lasts
  // long enough to be timed reliably. The output
  //     BenchmarkHello    10000000    282 ns/op
  // means that the loop ran 10000000 times at a speed of 282 ns per loop.
  //
  // If a benchmark needs some expensive setup before running, the timer
  // may be reset:
  //
  //     func BenchmarkBigLen(b *testing.B) {
  //         big := NewBig()
  //         b.ResetTimer()
  //         for i := 0; i < b.N; i++ {
  //             big.Len()
  //         }
  //     }
  //
  // If a benchmark needs to test performance in a parallel setting, it may use
  // the RunParallel helper function; such benchmarks are intended to be used with
  // the go test -cpu flag:
  //
  //     func BenchmarkTemplateParallel(b *testing.B) {
  //         templ := template.Must(template.New("test").Parse("Hello, {{.}}!"))
  //         b.RunParallel(func(pb *testing.PB) {
  //             var buf bytes.Buffer
  //             for pb.Next() {
  //                 buf.Reset()
  //                 templ.Execute(&buf, "World")
  //             }
  //         })
  //     }
  //
  // Examples
  //
  // The package also runs and verifies example code. Example functions may
  // include a concluding line comment that begins with "Output:" and is compared with
  // the standard output of the function when the tests are run. (The comparison
  // ignores leading and trailing space.) These are examples of an example:
  //
  //     func ExampleHello() {
  //         fmt.Println("hello")
  //         // Output: hello
  //     }
  //
  //     func ExampleSalutations() {
  //         fmt.Println("hello, and")
  //         fmt.Println("goodbye")
  //         // Output:
  //         // hello, and
  //         // goodbye
  //     }
  //
  // The comment prefix "Unordered output:" is like "Output:", but matches any
  // line order:
  //
  //     func ExamplePerm() {
  //         for _, value := range Perm(4) {
  //             fmt.Println(value)
  //         }
  //         // Unordered output: 4
  //         // 2
  //         // 1
  //         // 3
  //         // 0
  //     }
  //
  // Example functions without output comments are compiled but not executed.
  //
  // The naming convention to declare examples for the package, a function F, a type T and
  // method M on type T are:
  //
  //     func Example() { ... }
  //     func ExampleF() { ... }
  //     func ExampleT() { ... }
  //     func ExampleT_M() { ... }
  //
  // Multiple example functions for a package/type/function/method may be provided by
  // appending a distinct suffix to the name. The suffix must start with a
  // lower-case letter.
  //
  //     func Example_suffix() { ... }
  //     func ExampleF_suffix() { ... }
  //     func ExampleT_suffix() { ... }
  //     func ExampleT_M_suffix() { ... }
  //
  // The entire test file is presented as the example when it contains a single
  // example function, at least one other function, type, variable, or constant
  // declaration, and no test or benchmark functions.
  //
  // Subtests and Sub-benchmarks
  //
  // The Run methods of T and B allow defining subtests and sub-benchmarks,
  // without having to define separate functions for each. This enables uses
  // like table-driven benchmarks and creating hierarchical tests.
  // It also provides a way to share common setup and tear-down code:
  //
  //     func TestFoo(t *testing.T) {
  //         // <setup code>
  //         t.Run("A=1", func(t *testing.T) { ... })
  //         t.Run("A=2", func(t *testing.T) { ... })
  //         t.Run("B=1", func(t *testing.T) { ... })
  //         // <tear-down code>
  //     }
  //
  // Each subtest and sub-benchmark has a unique name: the combination of the name
  // of the top-level test and the sequence of names passed to Run, separated by
  // slashes, with an optional trailing sequence number for disambiguation.
  //
  // The argument to the -run and -bench command-line flags is an unanchored regular
  // expression that matches the test's name. For tests with multiple slash-separated
  // elements, such as subtests, the argument is itself slash-separated, with
  // expressions matching each name element in turn. Because it is unanchored, an
  // empty expression matches any string.
  // For example, using "matching" to mean "whose name contains":
  //
  //     go test -run ''      # Run all tests.
  //     go test -run Foo     # Run top-level tests matching "Foo", such as "TestFooBar".
  //     go test -run Foo/A=  # For top-level tests matching "Foo", run subtests matching "A=".
  //     go test -run /A=1    # For all top-level tests, run subtests matching "A=1".
  //
  // Subtests can also be used to control parallelism. A parent test will only
  // complete once all of its subtests complete. In this example, all tests are
  // run in parallel with each other, and only with each other, regardless of
  // other top-level tests that may be defined:
  //
  //     func TestGroupedParallel(t *testing.T) {
  //         for _, tc := range tests {
  //             tc := tc // capture range variable
  //             t.Run(tc.Name, func(t *testing.T) {
  //                 t.Parallel()
  //                 ...
  //             })
  //         }
  //     }
  //
  // Run does not return until parallel subtests have completed, providing a way
  // to clean up after a group of parallel tests:
  //
  //     func TestTeardownParallel(t *testing.T) {
  //         // This Run will not return until the parallel tests finish.
  //         t.Run("group", func(t *testing.T) {
  //             t.Run("Test1", parallelTest1)
  //             t.Run("Test2", parallelTest2)
  //             t.Run("Test3", parallelTest3)
  //         })
  //         // <tear-down code>
  //     }
  //
  // Main
  //
  // It is sometimes necessary for a test program to do extra setup or teardown
  // before or after testing. It is also sometimes necessary for a test to control
  // which code runs on the main thread. To support these and other cases,
  // if a test file contains a function:
  //
  //	func TestMain(m *testing.M)
  //
  // then the generated test will call TestMain(m) instead of running the tests
  // directly. TestMain runs in the main goroutine and can do whatever setup
  // and teardown is necessary around a call to m.Run. It should then call
  // os.Exit with the result of m.Run. When TestMain is called, flag.Parse has
  // not been run. If TestMain depends on command-line flags, including those
  // of the testing package, it should call flag.Parse explicitly.
  //
  // A simple implementation of TestMain is:
  //
  //	func TestMain(m *testing.M) {
  //		// call flag.Parse() here if TestMain uses flags
  //		os.Exit(m.Run())
  //	}
  //
  package testing
  
  import (
  	"bytes"
  	"errors"
  	"flag"
  	"fmt"
  	"internal/race"
  	"io"
  	"os"
  	"runtime"
  	"runtime/debug"
  	"runtime/trace"
  	"strconv"
  	"strings"
  	"sync"
  	"sync/atomic"
  	"time"
  )
  
  var (
  	// The short flag requests that tests run more quickly, but its functionality
  	// is provided by test writers themselves. The testing package is just its
  	// home. The all.bash installation script sets it to make installation more
  	// efficient, but by default the flag is off so a plain "go test" will do a
  	// full test of the package.
  	short = flag.Bool("test.short", false, "run smaller test suite to save time")
  
  	// The directory in which to create profile files and the like. When run from
  	// "go test", the binary always runs in the source directory for the package;
  	// this flag lets "go test" tell the binary to write the files in the directory where
  	// the "go test" command is run.
  	outputDir = flag.String("test.outputdir", "", "write profiles to `dir`")
  
  	// Report as tests are run; default is silent for success.
  	chatty               = flag.Bool("test.v", false, "verbose: print additional output")
  	count                = flag.Uint("test.count", 1, "run tests and benchmarks `n` times")
  	coverProfile         = flag.String("test.coverprofile", "", "write a coverage profile to `file`")
  	matchList            = flag.String("test.list", "", "list tests, examples, and benchmarch maching `regexp` then exit")
  	match                = flag.String("test.run", "", "run only tests and examples matching `regexp`")
  	memProfile           = flag.String("test.memprofile", "", "write a memory profile to `file`")
  	memProfileRate       = flag.Int("test.memprofilerate", 0, "set memory profiling `rate` (see runtime.MemProfileRate)")
  	cpuProfile           = flag.String("test.cpuprofile", "", "write a cpu profile to `file`")
  	blockProfile         = flag.String("test.blockprofile", "", "write a goroutine blocking profile to `file`")
  	blockProfileRate     = flag.Int("test.blockprofilerate", 1, "set blocking profile `rate` (see runtime.SetBlockProfileRate)")
  	mutexProfile         = flag.String("test.mutexprofile", "", "write a mutex contention profile to the named file after execution")
  	mutexProfileFraction = flag.Int("test.mutexprofilefraction", 1, "if >= 0, calls runtime.SetMutexProfileFraction()")
  	traceFile            = flag.String("test.trace", "", "write an execution trace to `file`")
  	timeout              = flag.Duration("test.timeout", 0, "panic test binary after duration `d` (0 means unlimited)")
  	cpuListStr           = flag.String("test.cpu", "", "comma-separated `list` of cpu counts to run each test with")
  	parallel             = flag.Int("test.parallel", runtime.GOMAXPROCS(0), "run at most `n` tests in parallel")
  
  	haveExamples bool // are there examples?
  
  	cpuList []int
  )
  
  // common holds the elements common between T and B and
  // captures common methods such as Errorf.
  type common struct {
  	mu      sync.RWMutex        // guards this group of fields
  	output  []byte              // Output generated by test or benchmark.
  	w       io.Writer           // For flushToParent.
  	ran     bool                // Test or benchmark (or one of its subtests) was executed.
  	failed  bool                // Test or benchmark has failed.
  	skipped bool                // Test of benchmark has been skipped.
  	done    bool                // Test is finished and all subtests have completed.
  	helpers map[string]struct{} // functions to be skipped when writing file/line info
  
  	chatty     bool   // A copy of the chatty flag.
  	finished   bool   // Test function has completed.
  	hasSub     int32  // written atomically
  	raceErrors int    // number of races detected during test
  	runner     string // function name of tRunner running the test
  
  	parent   *common
  	level    int       // Nesting depth of test or benchmark.
  	name     string    // Name of test or benchmark.
  	start    time.Time // Time test or benchmark started
  	duration time.Duration
  	barrier  chan bool // To signal parallel subtests they may start.
  	signal   chan bool // To signal a test is done.
  	sub      []*T      // Queue of subtests to be run in parallel.
  }
  
  // Short reports whether the -test.short flag is set.
  func Short() bool {
  	return *short
  }
  
  // CoverMode reports what the test coverage mode is set to. The
  // values are "set", "count", or "atomic". The return value will be
  // empty if test coverage is not enabled.
  func CoverMode() string {
  	return cover.Mode
  }
  
  // Verbose reports whether the -test.v flag is set.
  func Verbose() bool {
  	return *chatty
  }
  
  // frameSkip searches, starting after skip frames, for the first caller frame
  // in a function not marked as a helper and returns the frames to skip
  // to reach that site. The search stops if it finds a tRunner function that
  // was the entry point into the test.
  // This function must be called with c.mu held.
  func (c *common) frameSkip(skip int) int {
  	if c.helpers == nil {
  		return skip
  	}
  	var pc [50]uintptr
  	// Skip two extra frames to account for this function
  	// and runtime.Callers itself.
  	n := runtime.Callers(skip+2, pc[:])
  	if n == 0 {
  		panic("testing: zero callers found")
  	}
  	frames := runtime.CallersFrames(pc[:n])
  	var frame runtime.Frame
  	more := true
  	for i := 0; more; i++ {
  		frame, more = frames.Next()
  		if frame.Function == c.runner {
  			// We've gone up all the way to the tRunner calling
  			// the test function (so the user must have
  			// called tb.Helper from inside that test function).
  			// Only skip up to the test function itself.
  			return skip + i - 1
  		}
  		if _, ok := c.helpers[frame.Function]; !ok {
  			// Found a frame that wasn't inside a helper function.
  			return skip + i
  		}
  	}
  	return skip
  }
  
  // decorate prefixes the string with the file and line of the call site
  // and inserts the final newline if needed and indentation tabs for formatting.
  // This function must be called with c.mu held.
  func (c *common) decorate(s string) string {
  	skip := c.frameSkip(3) // decorate + log + public function.
  	_, file, line, ok := runtime.Caller(skip)
  	if ok {
  		// Truncate file name at last file name separator.
  		if index := strings.LastIndex(file, "/"); index >= 0 {
  			file = file[index+1:]
  		} else if index = strings.LastIndex(file, "\\"); index >= 0 {
  			file = file[index+1:]
  		}
  	} else {
  		file = "???"
  		line = 1
  	}
  	buf := new(bytes.Buffer)
  	// Every line is indented at least one tab.
  	buf.WriteByte('\t')
  	fmt.Fprintf(buf, "%s:%d: ", file, line)
  	lines := strings.Split(s, "\n")
  	if l := len(lines); l > 1 && lines[l-1] == "" {
  		lines = lines[:l-1]
  	}
  	for i, line := range lines {
  		if i > 0 {
  			// Second and subsequent lines are indented an extra tab.
  			buf.WriteString("\n\t\t")
  		}
  		buf.WriteString(line)
  	}
  	buf.WriteByte('\n')
  	return buf.String()
  }
  
  // flushToParent writes c.output to the parent after first writing the header
  // with the given format and arguments.
  func (c *common) flushToParent(format string, args ...interface{}) {
  	p := c.parent
  	p.mu.Lock()
  	defer p.mu.Unlock()
  
  	fmt.Fprintf(p.w, format, args...)
  
  	c.mu.Lock()
  	defer c.mu.Unlock()
  	io.Copy(p.w, bytes.NewReader(c.output))
  	c.output = c.output[:0]
  }
  
  type indenter struct {
  	c *common
  }
  
  func (w indenter) Write(b []byte) (n int, err error) {
  	n = len(b)
  	for len(b) > 0 {
  		end := bytes.IndexByte(b, '\n')
  		if end == -1 {
  			end = len(b)
  		} else {
  			end++
  		}
  		// An indent of 4 spaces will neatly align the dashes with the status
  		// indicator of the parent.
  		const indent = "    "
  		w.c.output = append(w.c.output, indent...)
  		w.c.output = append(w.c.output, b[:end]...)
  		b = b[end:]
  	}
  	return
  }
  
  // fmtDuration returns a string representing d in the form "87.00s".
  func fmtDuration(d time.Duration) string {
  	return fmt.Sprintf("%.2fs", d.Seconds())
  }
  
  // TB is the interface common to T and B.
  type TB interface {
  	Error(args ...interface{})
  	Errorf(format string, args ...interface{})
  	Fail()
  	FailNow()
  	Failed() bool
  	Fatal(args ...interface{})
  	Fatalf(format string, args ...interface{})
  	Log(args ...interface{})
  	Logf(format string, args ...interface{})
  	Name() string
  	Skip(args ...interface{})
  	SkipNow()
  	Skipf(format string, args ...interface{})
  	Skipped() bool
  	Helper()
  
  	// A private method to prevent users implementing the
  	// interface and so future additions to it will not
  	// violate Go 1 compatibility.
  	private()
  }
  
  var _ TB = (*T)(nil)
  var _ TB = (*B)(nil)
  
  // T is a type passed to Test functions to manage test state and support formatted test logs.
  // Logs are accumulated during execution and dumped to standard output when done.
  //
  // A test ends when its Test function returns or calls any of the methods
  // FailNow, Fatal, Fatalf, SkipNow, Skip, or Skipf. Those methods, as well as
  // the Parallel method, must be called only from the goroutine running the
  // Test function.
  //
  // The other reporting methods, such as the variations of Log and Error,
  // may be called simultaneously from multiple goroutines.
  type T struct {
  	common
  	isParallel bool
  	context    *testContext // For running tests and subtests.
  }
  
  func (c *common) private() {}
  
  // Name returns the name of the running test or benchmark.
  func (c *common) Name() string {
  	return c.name
  }
  
  func (c *common) setRan() {
  	if c.parent != nil {
  		c.parent.setRan()
  	}
  	c.mu.Lock()
  	defer c.mu.Unlock()
  	c.ran = true
  }
  
  // Fail marks the function as having failed but continues execution.
  func (c *common) Fail() {
  	if c.parent != nil {
  		c.parent.Fail()
  	}
  	c.mu.Lock()
  	defer c.mu.Unlock()
  	// c.done needs to be locked to synchronize checks to c.done in parent tests.
  	if c.done {
  		panic("Fail in goroutine after " + c.name + " has completed")
  	}
  	c.failed = true
  }
  
  // Failed reports whether the function has failed.
  func (c *common) Failed() bool {
  	c.mu.RLock()
  	failed := c.failed
  	c.mu.RUnlock()
  	return failed || c.raceErrors+race.Errors() > 0
  }
  
  // FailNow marks the function as having failed and stops its execution.
  // Execution will continue at the next test or benchmark.
  // FailNow must be called from the goroutine running the
  // test or benchmark function, not from other goroutines
  // created during the test. Calling FailNow does not stop
  // those other goroutines.
  func (c *common) FailNow() {
  	c.Fail()
  
  	// Calling runtime.Goexit will exit the goroutine, which
  	// will run the deferred functions in this goroutine,
  	// which will eventually run the deferred lines in tRunner,
  	// which will signal to the test loop that this test is done.
  	//
  	// A previous version of this code said:
  	//
  	//	c.duration = ...
  	//	c.signal <- c.self
  	//	runtime.Goexit()
  	//
  	// This previous version duplicated code (those lines are in
  	// tRunner no matter what), but worse the goroutine teardown
  	// implicit in runtime.Goexit was not guaranteed to complete
  	// before the test exited. If a test deferred an important cleanup
  	// function (like removing temporary files), there was no guarantee
  	// it would run on a test failure. Because we send on c.signal during
  	// a top-of-stack deferred function now, we know that the send
  	// only happens after any other stacked defers have completed.
  	c.finished = true
  	runtime.Goexit()
  }
  
  // log generates the output. It's always at the same stack depth.
  func (c *common) log(s string) {
  	c.mu.Lock()
  	defer c.mu.Unlock()
  	c.output = append(c.output, c.decorate(s)...)
  }
  
  // Log formats its arguments using default formatting, analogous to Println,
  // and records the text in the error log. For tests, the text will be printed only if
  // the test fails or the -test.v flag is set. For benchmarks, the text is always
  // printed to avoid having performance depend on the value of the -test.v flag.
  func (c *common) Log(args ...interface{}) { c.log(fmt.Sprintln(args...)) }
  
  // Logf formats its arguments according to the format, analogous to Printf, and
  // records the text in the error log. A final newline is added if not provided. For
  // tests, the text will be printed only if the test fails or the -test.v flag is
  // set. For benchmarks, the text is always printed to avoid having performance
  // depend on the value of the -test.v flag.
  func (c *common) Logf(format string, args ...interface{}) { c.log(fmt.Sprintf(format, args...)) }
  
  // Error is equivalent to Log followed by Fail.
  func (c *common) Error(args ...interface{}) {
  	c.log(fmt.Sprintln(args...))
  	c.Fail()
  }
  
  // Errorf is equivalent to Logf followed by Fail.
  func (c *common) Errorf(format string, args ...interface{}) {
  	c.log(fmt.Sprintf(format, args...))
  	c.Fail()
  }
  
  // Fatal is equivalent to Log followed by FailNow.
  func (c *common) Fatal(args ...interface{}) {
  	c.log(fmt.Sprintln(args...))
  	c.FailNow()
  }
  
  // Fatalf is equivalent to Logf followed by FailNow.
  func (c *common) Fatalf(format string, args ...interface{}) {
  	c.log(fmt.Sprintf(format, args...))
  	c.FailNow()
  }
  
  // Skip is equivalent to Log followed by SkipNow.
  func (c *common) Skip(args ...interface{}) {
  	c.log(fmt.Sprintln(args...))
  	c.SkipNow()
  }
  
  // Skipf is equivalent to Logf followed by SkipNow.
  func (c *common) Skipf(format string, args ...interface{}) {
  	c.log(fmt.Sprintf(format, args...))
  	c.SkipNow()
  }
  
  // SkipNow marks the test as having been skipped and stops its execution.
  // If a test fails (see Error, Errorf, Fail) and is then skipped,
  // it is still considered to have failed.
  // Execution will continue at the next test or benchmark. See also FailNow.
  // SkipNow must be called from the goroutine running the test, not from
  // other goroutines created during the test. Calling SkipNow does not stop
  // those other goroutines.
  func (c *common) SkipNow() {
  	c.skip()
  	c.finished = true
  	runtime.Goexit()
  }
  
  func (c *common) skip() {
  	c.mu.Lock()
  	defer c.mu.Unlock()
  	c.skipped = true
  }
  
  // Skipped reports whether the test was skipped.
  func (c *common) Skipped() bool {
  	c.mu.RLock()
  	defer c.mu.RUnlock()
  	return c.skipped
  }
  
  // Helper marks the calling function as a test helper function.
  // When printing file and line information, that function will be skipped.
  // Helper may be called simultaneously from multiple goroutines.
  // Helper has no effect if it is called directly from a TestXxx/BenchmarkXxx
  // function or a subtest/sub-benchmark function.
  func (c *common) Helper() {
  	c.mu.Lock()
  	defer c.mu.Unlock()
  	if c.helpers == nil {
  		c.helpers = make(map[string]struct{})
  	}
  	c.helpers[callerName(1)] = struct{}{}
  }
  
  // callerName gives the function name (qualified with a package path)
  // for the caller after skip frames (where 0 means the current function).
  func callerName(skip int) string {
  	// Make room for the skip PC.
  	var pc [2]uintptr
  	n := runtime.Callers(skip+2, pc[:]) // skip + runtime.Callers + callerName
  	if n == 0 {
  		panic("testing: zero callers found")
  	}
  	frames := runtime.CallersFrames(pc[:n])
  	frame, _ := frames.Next()
  	return frame.Function
  }
  
  // Parallel signals that this test is to be run in parallel with (and only with)
  // other parallel tests. When a test is run multiple times due to use of
  // -test.count or -test.cpu, multiple instances of a single test never run in
  // parallel with each other.
  func (t *T) Parallel() {
  	if t.isParallel {
  		panic("testing: t.Parallel called multiple times")
  	}
  	t.isParallel = true
  
  	// We don't want to include the time we spend waiting for serial tests
  	// in the test duration. Record the elapsed time thus far and reset the
  	// timer afterwards.
  	t.duration += time.Since(t.start)
  
  	// Add to the list of tests to be released by the parent.
  	t.parent.sub = append(t.parent.sub, t)
  	t.raceErrors += race.Errors()
  
  	t.signal <- true   // Release calling test.
  	<-t.parent.barrier // Wait for the parent test to complete.
  	t.context.waitParallel()
  	t.start = time.Now()
  	t.raceErrors += -race.Errors()
  }
  
  // An internal type but exported because it is cross-package; part of the implementation
  // of the "go test" command.
  type InternalTest struct {
  	Name string
  	F    func(*T)
  }
  
  func tRunner(t *T, fn func(t *T)) {
  	t.runner = callerName(0)
  
  	// When this goroutine is done, either because fn(t)
  	// returned normally or because a test failure triggered
  	// a call to runtime.Goexit, record the duration and send
  	// a signal saying that the test is done.
  	defer func() {
  		if t.raceErrors+race.Errors() > 0 {
  			t.Errorf("race detected during execution of test")
  		}
  
  		t.duration += time.Now().Sub(t.start)
  		// If the test panicked, print any test output before dying.
  		err := recover()
  		if !t.finished && err == nil {
  			err = fmt.Errorf("test executed panic(nil) or runtime.Goexit")
  		}
  		if err != nil {
  			t.Fail()
  			t.report()
  			panic(err)
  		}
  
  		if len(t.sub) > 0 {
  			// Run parallel subtests.
  			// Decrease the running count for this test.
  			t.context.release()
  			// Release the parallel subtests.
  			close(t.barrier)
  			// Wait for subtests to complete.
  			for _, sub := range t.sub {
  				<-sub.signal
  			}
  			if !t.isParallel {
  				// Reacquire the count for sequential tests. See comment in Run.
  				t.context.waitParallel()
  			}
  		} else if t.isParallel {
  			// Only release the count for this test if it was run as a parallel
  			// test. See comment in Run method.
  			t.context.release()
  		}
  		t.report() // Report after all subtests have finished.
  
  		// Do not lock t.done to allow race detector to detect race in case
  		// the user does not appropriately synchronizes a goroutine.
  		t.done = true
  		if t.parent != nil && atomic.LoadInt32(&t.hasSub) == 0 {
  			t.setRan()
  		}
  		t.signal <- true
  	}()
  
  	t.start = time.Now()
  	t.raceErrors = -race.Errors()
  	fn(t)
  	t.finished = true
  }
  
  // Run runs f as a subtest of t called name. It reports whether f succeeded. Run
  // runs f in a separate goroutine and will block until all its parallel subtests
  // have completed.
  //
  // Run may be called simultaneously from multiple goroutines, but all such calls
  // must return before the outer test function for t returns.
  func (t *T) Run(name string, f func(t *T)) bool {
  	atomic.StoreInt32(&t.hasSub, 1)
  	testName, ok, _ := t.context.match.fullName(&t.common, name)
  	if !ok {
  		return true
  	}
  	t = &T{
  		common: common{
  			barrier: make(chan bool),
  			signal:  make(chan bool),
  			name:    testName,
  			parent:  &t.common,
  			level:   t.level + 1,
  			chatty:  t.chatty,
  		},
  		context: t.context,
  	}
  	t.w = indenter{&t.common}
  
  	if t.chatty {
  		// Print directly to root's io.Writer so there is no delay.
  		root := t.parent
  		for ; root.parent != nil; root = root.parent {
  		}
  		root.mu.Lock()
  		fmt.Fprintf(root.w, "=== RUN   %s\n", t.name)
  		root.mu.Unlock()
  	}
  	// Instead of reducing the running count of this test before calling the
  	// tRunner and increasing it afterwards, we rely on tRunner keeping the
  	// count correct. This ensures that a sequence of sequential tests runs
  	// without being preempted, even when their parent is a parallel test. This
  	// may especially reduce surprises if *parallel == 1.
  	go tRunner(t, f)
  	<-t.signal
  	return !t.failed
  }
  
  // testContext holds all fields that are common to all tests. This includes
  // synchronization primitives to run at most *parallel tests.
  type testContext struct {
  	match *matcher
  
  	mu sync.Mutex
  
  	// Channel used to signal tests that are ready to be run in parallel.
  	startParallel chan bool
  
  	// running is the number of tests currently running in parallel.
  	// This does not include tests that are waiting for subtests to complete.
  	running int
  
  	// numWaiting is the number tests waiting to be run in parallel.
  	numWaiting int
  
  	// maxParallel is a copy of the parallel flag.
  	maxParallel int
  }
  
  func newTestContext(maxParallel int, m *matcher) *testContext {
  	return &testContext{
  		match:         m,
  		startParallel: make(chan bool),
  		maxParallel:   maxParallel,
  		running:       1, // Set the count to 1 for the main (sequential) test.
  	}
  }
  
  func (c *testContext) waitParallel() {
  	c.mu.Lock()
  	if c.running < c.maxParallel {
  		c.running++
  		c.mu.Unlock()
  		return
  	}
  	c.numWaiting++
  	c.mu.Unlock()
  	<-c.startParallel
  }
  
  func (c *testContext) release() {
  	c.mu.Lock()
  	if c.numWaiting == 0 {
  		c.running--
  		c.mu.Unlock()
  		return
  	}
  	c.numWaiting--
  	c.mu.Unlock()
  	c.startParallel <- true // Pick a waiting test to be run.
  }
  
  // No one should be using func Main anymore.
  // See the doc comment on func Main and use MainStart instead.
  var errMain = errors.New("testing: unexpected use of func Main")
  
  type matchStringOnly func(pat, str string) (bool, error)
  
  func (f matchStringOnly) MatchString(pat, str string) (bool, error)   { return f(pat, str) }
  func (f matchStringOnly) StartCPUProfile(w io.Writer) error           { return errMain }
  func (f matchStringOnly) StopCPUProfile()                             {}
  func (f matchStringOnly) WriteHeapProfile(w io.Writer) error          { return errMain }
  func (f matchStringOnly) WriteProfileTo(string, io.Writer, int) error { return errMain }
  func (f matchStringOnly) ImportPath() string                          { return "" }
  
  // Main is an internal function, part of the implementation of the "go test" command.
  // It was exported because it is cross-package and predates "internal" packages.
  // It is no longer used by "go test" but preserved, as much as possible, for other
  // systems that simulate "go test" using Main, but Main sometimes cannot be updated as
  // new functionality is added to the testing package.
  // Systems simulating "go test" should be updated to use MainStart.
  func Main(matchString func(pat, str string) (bool, error), tests []InternalTest, benchmarks []InternalBenchmark, examples []InternalExample) {
  	os.Exit(MainStart(matchStringOnly(matchString), tests, benchmarks, examples).Run())
  }
  
  // M is a type passed to a TestMain function to run the actual tests.
  type M struct {
  	deps       testDeps
  	tests      []InternalTest
  	benchmarks []InternalBenchmark
  	examples   []InternalExample
  }
  
  // testDeps is an internal interface of functionality that is
  // passed into this package by a test's generated main package.
  // The canonical implementation of this interface is
  // testing/internal/testdeps's TestDeps.
  type testDeps interface {
  	MatchString(pat, str string) (bool, error)
  	StartCPUProfile(io.Writer) error
  	StopCPUProfile()
  	WriteHeapProfile(io.Writer) error
  	WriteProfileTo(string, io.Writer, int) error
  	ImportPath() string
  }
  
  // MainStart is meant for use by tests generated by 'go test'.
  // It is not meant to be called directly and is not subject to the Go 1 compatibility document.
  // It may change signature from release to release.
  func MainStart(deps testDeps, tests []InternalTest, benchmarks []InternalBenchmark, examples []InternalExample) *M {
  	return &M{
  		deps:       deps,
  		tests:      tests,
  		benchmarks: benchmarks,
  		examples:   examples,
  	}
  }
  
  // Run runs the tests. It returns an exit code to pass to os.Exit.
  func (m *M) Run() int {
  	// TestMain may have already called flag.Parse.
  	if !flag.Parsed() {
  		flag.Parse()
  	}
  
  	if len(*matchList) != 0 {
  		listTests(m.deps.MatchString, m.tests, m.benchmarks, m.examples)
  		return 0
  	}
  
  	parseCpuList()
  
  	m.before()
  	startAlarm()
  	haveExamples = len(m.examples) > 0
  	testRan, testOk := runTests(m.deps.MatchString, m.tests)
  	exampleRan, exampleOk := runExamples(m.deps.MatchString, m.examples)
  	stopAlarm()
  	if !testRan && !exampleRan && *matchBenchmarks == "" {
  		fmt.Fprintln(os.Stderr, "testing: warning: no tests to run")
  	}
  	if !testOk || !exampleOk || !runBenchmarks(m.deps.ImportPath(), m.deps.MatchString, m.benchmarks) || race.Errors() > 0 {
  		fmt.Println("FAIL")
  		m.after()
  		return 1
  	}
  
  	fmt.Println("PASS")
  	m.after()
  	return 0
  }
  
  func (t *T) report() {
  	if t.parent == nil {
  		return
  	}
  	dstr := fmtDuration(t.duration)
  	format := "--- %s: %s (%s)\n"
  	if t.Failed() {
  		t.flushToParent(format, "FAIL", t.name, dstr)
  	} else if t.chatty {
  		if t.Skipped() {
  			t.flushToParent(format, "SKIP", t.name, dstr)
  		} else {
  			t.flushToParent(format, "PASS", t.name, dstr)
  		}
  	}
  }
  
  func listTests(matchString func(pat, str string) (bool, error), tests []InternalTest, benchmarks []InternalBenchmark, examples []InternalExample) {
  	if _, err := matchString(*matchList, "non-empty"); err != nil {
  		fmt.Fprintf(os.Stderr, "testing: invalid regexp in -test.list (%q): %s\n", *matchList, err)
  		os.Exit(1)
  	}
  
  	for _, test := range tests {
  		if ok, _ := matchString(*matchList, test.Name); ok {
  			fmt.Println(test.Name)
  		}
  	}
  	for _, bench := range benchmarks {
  		if ok, _ := matchString(*matchList, bench.Name); ok {
  			fmt.Println(bench.Name)
  		}
  	}
  	for _, example := range examples {
  		if ok, _ := matchString(*matchList, example.Name); ok {
  			fmt.Println(example.Name)
  		}
  	}
  }
  
  // An internal function but exported because it is cross-package; part of the implementation
  // of the "go test" command.
  func RunTests(matchString func(pat, str string) (bool, error), tests []InternalTest) (ok bool) {
  	ran, ok := runTests(matchString, tests)
  	if !ran && !haveExamples {
  		fmt.Fprintln(os.Stderr, "testing: warning: no tests to run")
  	}
  	return ok
  }
  
  func runTests(matchString func(pat, str string) (bool, error), tests []InternalTest) (ran, ok bool) {
  	ok = true
  	for _, procs := range cpuList {
  		runtime.GOMAXPROCS(procs)
  		ctx := newTestContext(*parallel, newMatcher(matchString, *match, "-test.run"))
  		t := &T{
  			common: common{
  				signal:  make(chan bool),
  				barrier: make(chan bool),
  				w:       os.Stdout,
  				chatty:  *chatty,
  			},
  			context: ctx,
  		}
  		tRunner(t, func(t *T) {
  			for _, test := range tests {
  				t.Run(test.Name, test.F)
  			}
  			// Run catching the signal rather than the tRunner as a separate
  			// goroutine to avoid adding a goroutine during the sequential
  			// phase as this pollutes the stacktrace output when aborting.
  			go func() { <-t.signal }()
  		})
  		ok = ok && !t.Failed()
  		ran = ran || t.ran
  	}
  	return ran, ok
  }
  
  // before runs before all testing.
  func (m *M) before() {
  	if *memProfileRate > 0 {
  		runtime.MemProfileRate = *memProfileRate
  	}
  	if *cpuProfile != "" {
  		f, err := os.Create(toOutputDir(*cpuProfile))
  		if err != nil {
  			fmt.Fprintf(os.Stderr, "testing: %s\n", err)
  			return
  		}
  		if err := m.deps.StartCPUProfile(f); err != nil {
  			fmt.Fprintf(os.Stderr, "testing: can't start cpu profile: %s\n", err)
  			f.Close()
  			return
  		}
  		// Could save f so after can call f.Close; not worth the effort.
  	}
  	if *traceFile != "" {
  		f, err := os.Create(toOutputDir(*traceFile))
  		if err != nil {
  			fmt.Fprintf(os.Stderr, "testing: %s\n", err)
  			return
  		}
  		if err := trace.Start(f); err != nil {
  			fmt.Fprintf(os.Stderr, "testing: can't start tracing: %s\n", err)
  			f.Close()
  			return
  		}
  		// Could save f so after can call f.Close; not worth the effort.
  	}
  	if *blockProfile != "" && *blockProfileRate >= 0 {
  		runtime.SetBlockProfileRate(*blockProfileRate)
  	}
  	if *mutexProfile != "" && *mutexProfileFraction >= 0 {
  		runtime.SetMutexProfileFraction(*mutexProfileFraction)
  	}
  	if *coverProfile != "" && cover.Mode == "" {
  		fmt.Fprintf(os.Stderr, "testing: cannot use -test.coverprofile because test binary was not built with coverage enabled\n")
  		os.Exit(2)
  	}
  }
  
  // after runs after all testing.
  func (m *M) after() {
  	if *cpuProfile != "" {
  		m.deps.StopCPUProfile() // flushes profile to disk
  	}
  	if *traceFile != "" {
  		trace.Stop() // flushes trace to disk
  	}
  	if *memProfile != "" {
  		f, err := os.Create(toOutputDir(*memProfile))
  		if err != nil {
  			fmt.Fprintf(os.Stderr, "testing: %s\n", err)
  			os.Exit(2)
  		}
  		runtime.GC() // materialize all statistics
  		if err = m.deps.WriteHeapProfile(f); err != nil {
  			fmt.Fprintf(os.Stderr, "testing: can't write %s: %s\n", *memProfile, err)
  			os.Exit(2)
  		}
  		f.Close()
  	}
  	if *blockProfile != "" && *blockProfileRate >= 0 {
  		f, err := os.Create(toOutputDir(*blockProfile))
  		if err != nil {
  			fmt.Fprintf(os.Stderr, "testing: %s\n", err)
  			os.Exit(2)
  		}
  		if err = m.deps.WriteProfileTo("block", f, 0); err != nil {
  			fmt.Fprintf(os.Stderr, "testing: can't write %s: %s\n", *blockProfile, err)
  			os.Exit(2)
  		}
  		f.Close()
  	}
  	if *mutexProfile != "" && *mutexProfileFraction >= 0 {
  		f, err := os.Create(toOutputDir(*mutexProfile))
  		if err != nil {
  			fmt.Fprintf(os.Stderr, "testing: %s\n", err)
  			os.Exit(2)
  		}
  		if err = m.deps.WriteProfileTo("mutex", f, 0); err != nil {
  			fmt.Fprintf(os.Stderr, "testing: can't write %s: %s\n", *blockProfile, err)
  			os.Exit(2)
  		}
  		f.Close()
  	}
  	if cover.Mode != "" {
  		coverReport()
  	}
  }
  
  // toOutputDir returns the file name relocated, if required, to outputDir.
  // Simple implementation to avoid pulling in path/filepath.
  func toOutputDir(path string) string {
  	if *outputDir == "" || path == "" {
  		return path
  	}
  	if runtime.GOOS == "windows" {
  		// On Windows, it's clumsy, but we can be almost always correct
  		// by just looking for a drive letter and a colon.
  		// Absolute paths always have a drive letter (ignoring UNC).
  		// Problem: if path == "C:A" and outputdir == "C:\Go" it's unclear
  		// what to do, but even then path/filepath doesn't help.
  		// TODO: Worth doing better? Probably not, because we're here only
  		// under the management of go test.
  		if len(path) >= 2 {
  			letter, colon := path[0], path[1]
  			if ('a' <= letter && letter <= 'z' || 'A' <= letter && letter <= 'Z') && colon == ':' {
  				// If path starts with a drive letter we're stuck with it regardless.
  				return path
  			}
  		}
  	}
  	if os.IsPathSeparator(path[0]) {
  		return path
  	}
  	return fmt.Sprintf("%s%c%s", *outputDir, os.PathSeparator, path)
  }
  
  var timer *time.Timer
  
  // startAlarm starts an alarm if requested.
  func startAlarm() {
  	if *timeout > 0 {
  		timer = time.AfterFunc(*timeout, func() {
  			debug.SetTraceback("all")
  			panic(fmt.Sprintf("test timed out after %v", *timeout))
  		})
  	}
  }
  
  // stopAlarm turns off the alarm.
  func stopAlarm() {
  	if *timeout > 0 {
  		timer.Stop()
  	}
  }
  
  func parseCpuList() {
  	for _, val := range strings.Split(*cpuListStr, ",") {
  		val = strings.TrimSpace(val)
  		if val == "" {
  			continue
  		}
  		cpu, err := strconv.Atoi(val)
  		if err != nil || cpu <= 0 {
  			fmt.Fprintf(os.Stderr, "testing: invalid value %q for -test.cpu\n", val)
  			os.Exit(1)
  		}
  		for i := uint(0); i < *count; i++ {
  			cpuList = append(cpuList, cpu)
  		}
  	}
  	if cpuList == nil {
  		for i := uint(0); i < *count; i++ {
  			cpuList = append(cpuList, runtime.GOMAXPROCS(-1))
  		}
  	}
  }
  

View as plain text