codehost.go

     1  // Copyright 2018 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  // Package codehost defines the interface implemented by a code hosting source,
     6  // along with support code for use by implementations.
     7  package codehost
     8  
     9  import (
    10  	"bytes"
    11  	"context"
    12  	"crypto/sha256"
    13  	"fmt"
    14  	"io"
    15  	"io/fs"
    16  	"os"
    17  	"os/exec"
    18  	"path/filepath"
    19  	"strings"
    20  	"sync"
    21  	"time"
    22  
    23  	"cmd/go/internal/cfg"
    24  	"cmd/go/internal/lockedfile"
    25  	"cmd/go/internal/str"
    26  
    27  	"golang.org/x/mod/module"
    28  	"golang.org/x/mod/semver"
    29  )
    30  
    31  // Downloaded size limits.
    32  const (
    33  	MaxGoMod   = 16 << 20  // maximum size of go.mod file
    34  	MaxLICENSE = 16 << 20  // maximum size of LICENSE file
    35  	MaxZipFile = 500 << 20 // maximum size of downloaded zip file
    36  )
    37  
    38  // A Repo represents a code hosting source.
    39  // Typical implementations include local version control repositories,
    40  // remote version control servers, and code hosting sites.
    41  //
    42  // A Repo must be safe for simultaneous use by multiple goroutines,
    43  // and callers must not modify returned values, which may be cached and shared.
    44  type Repo interface {
    45  	// CheckReuse checks whether the old origin information
    46  	// remains up to date. If so, whatever cached object it was
    47  	// taken from can be reused.
    48  	// The subdir gives subdirectory name where the module root is expected to be found,
    49  	// "" for the root or "sub/dir" for a subdirectory (no trailing slash).
    50  	CheckReuse(ctx context.Context, old *Origin, subdir string) error
    51  
    52  	// Tags lists all tags with the given prefix.
    53  	Tags(ctx context.Context, prefix string) (*Tags, error)
    54  
    55  	// Stat returns information about the revision rev.
    56  	// A revision can be any identifier known to the underlying service:
    57  	// commit hash, branch, tag, and so on.
    58  	Stat(ctx context.Context, rev string) (*RevInfo, error)
    59  
    60  	// Latest returns the latest revision on the default branch,
    61  	// whatever that means in the underlying implementation.
    62  	Latest(ctx context.Context) (*RevInfo, error)
    63  
    64  	// ReadFile reads the given file in the file tree corresponding to revision rev.
    65  	// It should refuse to read more than maxSize bytes.
    66  	//
    67  	// If the requested file does not exist it should return an error for which
    68  	// os.IsNotExist(err) returns true.
    69  	ReadFile(ctx context.Context, rev, file string, maxSize int64) (data []byte, err error)
    70  
    71  	// ReadZip downloads a zip file for the subdir subdirectory
    72  	// of the given revision to a new file in a given temporary directory.
    73  	// It should refuse to read more than maxSize bytes.
    74  	// It returns a ReadCloser for a streamed copy of the zip file.
    75  	// All files in the zip file are expected to be
    76  	// nested in a single top-level directory, whose name is not specified.
    77  	ReadZip(ctx context.Context, rev, subdir string, maxSize int64) (zip io.ReadCloser, err error)
    78  
    79  	// RecentTag returns the most recent tag on rev or one of its predecessors
    80  	// with the given prefix. allowed may be used to filter out unwanted versions.
    81  	RecentTag(ctx context.Context, rev, prefix string, allowed func(tag string) bool) (tag string, err error)
    82  
    83  	// DescendsFrom reports whether rev or any of its ancestors has the given tag.
    84  	//
    85  	// DescendsFrom must return true for any tag returned by RecentTag for the
    86  	// same revision.
    87  	DescendsFrom(ctx context.Context, rev, tag string) (bool, error)
    88  }
    89  
    90  // An Origin describes the provenance of a given repo method result.
    91  // It can be passed to CheckReuse (usually in a different go command invocation)
    92  // to see whether the result remains up-to-date.
    93  type Origin struct {
    94  	VCS    string `json:",omitempty"` // "git" etc
    95  	URL    string `json:",omitempty"` // URL of repository
    96  	Subdir string `json:",omitempty"` // subdirectory in repo
    97  
    98  	Hash string `json:",omitempty"` // commit hash or ID
    99  
   100  	// If TagSum is non-empty, then the resolution of this module version
   101  	// depends on the set of tags present in the repo, specifically the tags
   102  	// of the form TagPrefix + a valid semver version.
   103  	// If the matching repo tags and their commit hashes still hash to TagSum,
   104  	// the Origin is still valid (at least as far as the tags are concerned).
   105  	// The exact checksum is up to the Repo implementation; see (*gitRepo).Tags.
   106  	TagPrefix string `json:",omitempty"`
   107  	TagSum    string `json:",omitempty"`
   108  
   109  	// If Ref is non-empty, then the resolution of this module version
   110  	// depends on Ref resolving to the revision identified by Hash.
   111  	// If Ref still resolves to Hash, the Origin is still valid (at least as far as Ref is concerned).
   112  	// For Git, the Ref is a full ref like "refs/heads/main" or "refs/tags/v1.2.3",
   113  	// and the Hash is the Git object hash the ref maps to.
   114  	// Other VCS might choose differently, but the idea is that Ref is the name
   115  	// with a mutable meaning while Hash is a name with an immutable meaning.
   116  	Ref string `json:",omitempty"`
   117  
   118  	// If RepoSum is non-empty, then the resolution of this module version
   119  	// failed due to the repo being available but the version not being present.
   120  	// This depends on the entire state of the repo, which RepoSum summarizes.
   121  	// For Git, this is a hash of all the refs and their hashes.
   122  	RepoSum string `json:",omitempty"`
   123  }
   124  
   125  // A Tags describes the available tags in a code repository.
   126  type Tags struct {
   127  	Origin *Origin
   128  	List   []Tag
   129  }
   130  
   131  // A Tag describes a single tag in a code repository.
   132  type Tag struct {
   133  	Name string
   134  	Hash string // content hash identifying tag's content, if available
   135  }
   136  
   137  // isOriginTag reports whether tag should be preserved
   138  // in the Tags method's Origin calculation.
   139  // We can safely ignore tags that are not look like pseudo-versions,
   140  // because ../coderepo.go's (*codeRepo).Versions ignores them too.
   141  // We can also ignore non-semver tags, but we have to include semver
   142  // tags with extra suffixes, because the pseudo-version base finder uses them.
   143  func isOriginTag(tag string) bool {
   144  	// modfetch.(*codeRepo).Versions uses Canonical == tag,
   145  	// but pseudo-version calculation has a weaker condition that
   146  	// the canonical is a prefix of the tag.
   147  	// Include those too, so that if any new one appears, we'll invalidate the cache entry.
   148  	// This will lead to spurious invalidation of version list results,
   149  	// but tags of this form being created should be fairly rare
   150  	// (and invalidate pseudo-version results anyway).
   151  	c := semver.Canonical(tag)
   152  	return c != "" && strings.HasPrefix(tag, c) && !module.IsPseudoVersion(tag)
   153  }
   154  
   155  // A RevInfo describes a single revision in a source code repository.
   156  type RevInfo struct {
   157  	Origin  *Origin
   158  	Name    string    // complete ID in underlying repository
   159  	Short   string    // shortened ID, for use in pseudo-version
   160  	Version string    // version used in lookup
   161  	Time    time.Time // commit time
   162  	Tags    []string  // known tags for commit
   163  }
   164  
   165  // UnknownRevisionError is an error equivalent to fs.ErrNotExist, but for a
   166  // revision rather than a file.
   167  type UnknownRevisionError struct {
   168  	Rev string
   169  }
   170  
   171  func (e *UnknownRevisionError) Error() string {
   172  	return "unknown revision " + e.Rev
   173  }
   174  func (UnknownRevisionError) Is(err error) bool {
   175  	return err == fs.ErrNotExist
   176  }
   177  
   178  // ErrNoCommits is an error equivalent to fs.ErrNotExist indicating that a given
   179  // repository or module contains no commits.
   180  var ErrNoCommits error = noCommitsError{}
   181  
   182  type noCommitsError struct{}
   183  
   184  func (noCommitsError) Error() string {
   185  	return "no commits"
   186  }
   187  func (noCommitsError) Is(err error) bool {
   188  	return err == fs.ErrNotExist
   189  }
   190  
   191  // AllHex reports whether the revision rev is entirely lower-case hexadecimal digits.
   192  func AllHex(rev string) bool {
   193  	for i := 0; i < len(rev); i++ {
   194  		c := rev[i]
   195  		if '0' <= c && c <= '9' || 'a' <= c && c <= 'f' {
   196  			continue
   197  		}
   198  		return false
   199  	}
   200  	return true
   201  }
   202  
   203  // ShortenSHA1 shortens a SHA1 hash (40 hex digits) to the canonical length
   204  // used in pseudo-versions (12 hex digits).
   205  func ShortenSHA1(rev string) string {
   206  	if AllHex(rev) && len(rev) == 40 {
   207  		return rev[:12]
   208  	}
   209  	return rev
   210  }
   211  
   212  // WorkDir returns the name of the cached work directory to use for the
   213  // given repository type and name.
   214  func WorkDir(ctx context.Context, typ, name string) (dir, lockfile string, err error) {
   215  	if cfg.GOMODCACHE == "" {
   216  		return "", "", fmt.Errorf("neither GOPATH nor GOMODCACHE are set")
   217  	}
   218  
   219  	// We name the work directory for the SHA256 hash of the type and name.
   220  	// We intentionally avoid the actual name both because of possible
   221  	// conflicts with valid file system paths and because we want to ensure
   222  	// that one checkout is never nested inside another. That nesting has
   223  	// led to security problems in the past.
   224  	if strings.Contains(typ, ":") {
   225  		return "", "", fmt.Errorf("codehost.WorkDir: type cannot contain colon")
   226  	}
   227  	key := typ + ":" + name
   228  	dir = filepath.Join(cfg.GOMODCACHE, "cache/vcs", fmt.Sprintf("%x", sha256.Sum256([]byte(key))))
   229  
   230  	xLog, buildX := cfg.BuildXWriter(ctx)
   231  	if buildX {
   232  		fmt.Fprintf(xLog, "mkdir -p %s # %s %s\n", filepath.Dir(dir), typ, name)
   233  	}
   234  	if err := os.MkdirAll(filepath.Dir(dir), 0777); err != nil {
   235  		return "", "", err
   236  	}
   237  
   238  	lockfile = dir + ".lock"
   239  	if buildX {
   240  		fmt.Fprintf(xLog, "# lock %s\n", lockfile)
   241  	}
   242  
   243  	unlock, err := lockedfile.MutexAt(lockfile).Lock()
   244  	if err != nil {
   245  		return "", "", fmt.Errorf("codehost.WorkDir: can't find or create lock file: %v", err)
   246  	}
   247  	defer unlock()
   248  
   249  	data, err := os.ReadFile(dir + ".info")
   250  	info, err2 := os.Stat(dir)
   251  	if err == nil && err2 == nil && info.IsDir() {
   252  		// Info file and directory both already exist: reuse.
   253  		have := strings.TrimSuffix(string(data), "\n")
   254  		if have != key {
   255  			return "", "", fmt.Errorf("%s exists with wrong content (have %q want %q)", dir+".info", have, key)
   256  		}
   257  		if buildX {
   258  			fmt.Fprintf(xLog, "# %s for %s %s\n", dir, typ, name)
   259  		}
   260  		return dir, lockfile, nil
   261  	}
   262  
   263  	// Info file or directory missing. Start from scratch.
   264  	if xLog != nil {
   265  		fmt.Fprintf(xLog, "mkdir -p %s # %s %s\n", dir, typ, name)
   266  	}
   267  	os.RemoveAll(dir)
   268  	if err := os.MkdirAll(dir, 0777); err != nil {
   269  		return "", "", err
   270  	}
   271  	if err := os.WriteFile(dir+".info", []byte(key), 0666); err != nil {
   272  		os.RemoveAll(dir)
   273  		return "", "", err
   274  	}
   275  	return dir, lockfile, nil
   276  }
   277  
   278  type RunError struct {
   279  	Cmd      string
   280  	Err      error
   281  	Stderr   []byte
   282  	HelpText string
   283  }
   284  
   285  func (e *RunError) Error() string {
   286  	text := e.Cmd + ": " + e.Err.Error()
   287  	stderr := bytes.TrimRight(e.Stderr, "\n")
   288  	if len(stderr) > 0 {
   289  		text += ":\n\t" + strings.ReplaceAll(string(stderr), "\n", "\n\t")
   290  	}
   291  	if len(e.HelpText) > 0 {
   292  		text += "\n" + e.HelpText
   293  	}
   294  	return text
   295  }
   296  
   297  var dirLock sync.Map
   298  
   299  // Run runs the command line in the given directory
   300  // (an empty dir means the current directory).
   301  // It returns the standard output and, for a non-zero exit,
   302  // a *RunError indicating the command, exit status, and standard error.
   303  // Standard error is unavailable for commands that exit successfully.
   304  func Run(ctx context.Context, dir string, cmdline ...any) ([]byte, error) {
   305  	return RunWithStdin(ctx, dir, nil, cmdline...)
   306  }
   307  
   308  // bashQuoter escapes characters that have special meaning in double-quoted strings in the bash shell.
   309  // See https://www.gnu.org/software/bash/manual/html_node/Double-Quotes.html.
   310  var bashQuoter = strings.NewReplacer(`"`, `\"`, `$`, `\$`, "`", "\\`", `\`, `\\`)
   311  
   312  func RunWithStdin(ctx context.Context, dir string, stdin io.Reader, cmdline ...any) ([]byte, error) {
   313  	if dir != "" {
   314  		muIface, ok := dirLock.Load(dir)
   315  		if !ok {
   316  			muIface, _ = dirLock.LoadOrStore(dir, new(sync.Mutex))
   317  		}
   318  		mu := muIface.(*sync.Mutex)
   319  		mu.Lock()
   320  		defer mu.Unlock()
   321  	}
   322  
   323  	cmd := str.StringList(cmdline...)
   324  	if os.Getenv("TESTGOVCS") == "panic" {
   325  		panic(fmt.Sprintf("use of vcs: %v", cmd))
   326  	}
   327  	if xLog, ok := cfg.BuildXWriter(ctx); ok {
   328  		text := new(strings.Builder)
   329  		if dir != "" {
   330  			text.WriteString("cd ")
   331  			text.WriteString(dir)
   332  			text.WriteString("; ")
   333  		}
   334  		for i, arg := range cmd {
   335  			if i > 0 {
   336  				text.WriteByte(' ')
   337  			}
   338  			switch {
   339  			case strings.ContainsAny(arg, "'"):
   340  				// Quote args that could be mistaken for quoted args.
   341  				text.WriteByte('"')
   342  				text.WriteString(bashQuoter.Replace(arg))
   343  				text.WriteByte('"')
   344  			case strings.ContainsAny(arg, "$`\\*?[\"\t\n\v\f\r \u0085\u00a0"):
   345  				// Quote args that contain special characters, glob patterns, or spaces.
   346  				text.WriteByte('\'')
   347  				text.WriteString(arg)
   348  				text.WriteByte('\'')
   349  			default:
   350  				text.WriteString(arg)
   351  			}
   352  		}
   353  		fmt.Fprintf(xLog, "%s\n", text)
   354  		start := time.Now()
   355  		defer func() {
   356  			fmt.Fprintf(xLog, "%.3fs # %s\n", time.Since(start).Seconds(), text)
   357  		}()
   358  	}
   359  	// TODO: Impose limits on command output size.
   360  	// TODO: Set environment to get English error messages.
   361  	var stderr bytes.Buffer
   362  	var stdout bytes.Buffer
   363  	c := exec.CommandContext(ctx, cmd[0], cmd[1:]...)
   364  	c.Cancel = func() error { return c.Process.Signal(os.Interrupt) }
   365  	c.Dir = dir
   366  	c.Stdin = stdin
   367  	c.Stderr = &stderr
   368  	c.Stdout = &stdout
   369  	// For Git commands, manually supply GIT_DIR so Git works with safe.bareRepository=explicit set. Noop for other commands.
   370  	c.Env = append(c.Environ(), "GIT_DIR="+dir)
   371  	err := c.Run()
   372  	if err != nil {
   373  		err = &RunError{Cmd: strings.Join(cmd, " ") + " in " + dir, Stderr: stderr.Bytes(), Err: err}
   374  	}
   375  	return stdout.Bytes(), err
   376  }
   377
View as plain text