vcs.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
     6  
     7  import (
     8  	"context"
     9  	"errors"
    10  	"fmt"
    11  	"internal/lazyregexp"
    12  	"io"
    13  	"io/fs"
    14  	"os"
    15  	"path/filepath"
    16  	"sort"
    17  	"strconv"
    18  	"strings"
    19  	"sync"
    20  	"sync/atomic"
    21  	"time"
    22  
    23  	"cmd/go/internal/base"
    24  	"cmd/go/internal/cfg"
    25  	"cmd/go/internal/lockedfile"
    26  	"cmd/go/internal/str"
    27  	"cmd/internal/par"
    28  
    29  	"golang.org/x/mod/semver"
    30  )
    31  
    32  // A VCSError indicates an error using a version control system.
    33  // The implication of a VCSError is that we know definitively where
    34  // to get the code, but we can't access it due to the error.
    35  // The caller should report this error instead of continuing to probe
    36  // other possible module paths.
    37  //
    38  // TODO(golang.org/issue/31730): See if we can invert this. (Return a
    39  // distinguished error for “repo not found” and treat everything else
    40  // as terminal.)
    41  type VCSError struct {
    42  	Err error
    43  }
    44  
    45  func (e *VCSError) Error() string { return e.Err.Error() }
    46  
    47  func (e *VCSError) Unwrap() error { return e.Err }
    48  
    49  func vcsErrorf(format string, a ...any) error {
    50  	return &VCSError{Err: fmt.Errorf(format, a...)}
    51  }
    52  
    53  type vcsCacheKey struct {
    54  	vcs    string
    55  	remote string
    56  	local  bool
    57  }
    58  
    59  func NewRepo(ctx context.Context, vcs, remote string, local bool) (Repo, error) {
    60  	return vcsRepoCache.Do(vcsCacheKey{vcs, remote, local}, func() (Repo, error) {
    61  		repo, err := newVCSRepo(ctx, vcs, remote, local)
    62  		if err != nil {
    63  			return nil, &VCSError{err}
    64  		}
    65  		return repo, nil
    66  	})
    67  }
    68  
    69  var vcsRepoCache par.ErrCache[vcsCacheKey, Repo]
    70  
    71  type vcsRepo struct {
    72  	mu lockedfile.Mutex // protects all commands, so we don't have to decide which are safe on a per-VCS basis
    73  
    74  	remote string
    75  	cmd    *vcsCmd
    76  	dir    string
    77  	local  bool
    78  
    79  	tagsOnce sync.Once
    80  	tags     map[string]bool
    81  
    82  	branchesOnce sync.Once
    83  	branches     map[string]bool
    84  
    85  	fetchOnce sync.Once
    86  	fetchErr  error
    87  	fetched   atomic.Bool
    88  
    89  	repoSumOnce sync.Once
    90  	repoSum     string
    91  }
    92  
    93  func newVCSRepo(ctx context.Context, vcs, remote string, local bool) (Repo, error) {
    94  	if vcs == "git" {
    95  		return newGitRepo(ctx, remote, local)
    96  	}
    97  	r := &vcsRepo{remote: remote, local: local}
    98  	cmd := vcsCmds[vcs]
    99  	if cmd == nil {
   100  		return nil, fmt.Errorf("unknown vcs: %s %s", vcs, remote)
   101  	}
   102  	r.cmd = cmd
   103  	if local {
   104  		info, err := os.Stat(remote)
   105  		if err != nil {
   106  			return nil, err
   107  		}
   108  		if !info.IsDir() {
   109  			return nil, fmt.Errorf("%s exists but is not a directory", remote)
   110  		}
   111  		r.dir = remote
   112  		r.mu.Path = r.dir + ".lock"
   113  		return r, nil
   114  	}
   115  	if !strings.Contains(remote, "://") {
   116  		return nil, fmt.Errorf("invalid vcs remote: %s %s", vcs, remote)
   117  	}
   118  	var err error
   119  	r.dir, r.mu.Path, err = WorkDir(ctx, vcsWorkDirType+vcs, r.remote)
   120  	if err != nil {
   121  		return nil, err
   122  	}
   123  
   124  	if cmd.init == nil {
   125  		return r, nil
   126  	}
   127  
   128  	unlock, err := r.mu.Lock()
   129  	if err != nil {
   130  		return nil, err
   131  	}
   132  	defer unlock()
   133  
   134  	if _, err := os.Stat(filepath.Join(r.dir, "."+vcs)); err != nil {
   135  		release, err := base.AcquireNet()
   136  		if err != nil {
   137  			return nil, err
   138  		}
   139  		_, err = Run(ctx, r.dir, cmd.init(r.remote))
   140  		if err == nil && cmd.postInit != nil {
   141  			err = cmd.postInit(ctx, r)
   142  		}
   143  		release()
   144  
   145  		if err != nil {
   146  			os.RemoveAll(r.dir)
   147  			return nil, err
   148  		}
   149  	}
   150  	return r, nil
   151  }
   152  
   153  const vcsWorkDirType = "vcs1."
   154  
   155  type vcsCmd struct {
   156  	vcs                string                                            // vcs name "hg"
   157  	init               func(remote string) []string                      // cmd to init repo to track remote
   158  	postInit           func(context.Context, *vcsRepo) error             // func to init repo after .init runs
   159  	repoSum            func(remote string) []string                      // cmd to calculate reposum of remote repo
   160  	lookupRef          func(remote, ref string) []string                 // cmd to look up ref in remote repo
   161  	tags               func(remote string) []string                      // cmd to list local tags
   162  	tagsNeedsFetch     bool                                              // run fetch before tags
   163  	tagRE              *lazyregexp.Regexp                                // regexp to extract tag names from output of tags cmd
   164  	branches           func(remote string) []string                      // cmd to list local branches
   165  	branchesNeedsFetch bool                                              // run branches before tags
   166  	branchRE           *lazyregexp.Regexp                                // regexp to extract branch names from output of tags cmd
   167  	badLocalRevRE      *lazyregexp.Regexp                                // regexp of names that must not be served out of local cache without doing fetch first
   168  	statLocal          func(rev, remote string) []string                 // cmd to stat local rev
   169  	parseStat          func(rev, out string) (*RevInfo, error)           // func to parse output of statLocal
   170  	fetch              []string                                          // cmd to fetch everything from remote
   171  	latest             string                                            // name of latest commit on remote (tip, HEAD, etc)
   172  	descendsFrom       func(rev, tag string) []string                    // cmd to check whether rev descends from tag
   173  	recentTags         func(rev string) []string                         // cmd to print tag ancestors of rev
   174  	readFile           func(rev, file, remote string) []string           // cmd to read rev's file
   175  	readZip            func(rev, subdir, remote, target string) []string // cmd to read rev's subdir as zip file
   176  
   177  	// arbitrary function to read rev's subdir as zip file
   178  	doReadZip func(ctx context.Context, dst io.Writer, workDir, rev, subdir, remote string) error
   179  }
   180  
   181  var re = lazyregexp.New
   182  
   183  var vcsCmds = map[string]*vcsCmd{
   184  	"hg": {
   185  		vcs: "hg",
   186  		repoSum: func(remote string) []string {
   187  			return []string{
   188  				"hg",
   189  				"--config=extensions.goreposum=" + filepath.Join(cfg.GOROOT, "lib/hg/goreposum.py"),
   190  				"goreposum",
   191  				remote,
   192  			}
   193  		},
   194  		lookupRef: func(remote, ref string) []string {
   195  			return []string{
   196  				"hg",
   197  				"--config=extensions.goreposum=" + filepath.Join(cfg.GOROOT, "lib/hg/goreposum.py"),
   198  				"golookup",
   199  				remote,
   200  				ref,
   201  			}
   202  		},
   203  		init: func(remote string) []string {
   204  			return []string{"hg", "init", "."}
   205  		},
   206  		postInit: hgAddRemote,
   207  		tags: func(remote string) []string {
   208  			return []string{"hg", "tags", "-q"}
   209  		},
   210  		tagsNeedsFetch: true,
   211  		tagRE:          re(`(?m)^[^\n]+$`),
   212  		branches: func(remote string) []string {
   213  			return []string{"hg", "branches", "-c", "-q"}
   214  		},
   215  		branchesNeedsFetch: true,
   216  		branchRE:           re(`(?m)^[^\n]+$`),
   217  		badLocalRevRE:      re(`(?m)^(tip)$`),
   218  		statLocal: func(rev, remote string) []string {
   219  			return []string{"hg", "log", "-l1", "-r", rev, "--template", "{node} {date|hgdate} {tags}"}
   220  		},
   221  		parseStat: hgParseStat,
   222  		fetch:     []string{"hg", "pull", "-f"},
   223  		latest:    "tip",
   224  		descendsFrom: func(rev, tag string) []string {
   225  			return []string{"hg", "log", "-r", "ancestors(" + rev + ") and " + tag}
   226  		},
   227  		recentTags: func(rev string) []string {
   228  			return []string{"hg", "log", "-r", "ancestors(" + rev + ") and tag()", "--template", "{tags}\n"}
   229  		},
   230  		readFile: func(rev, file, remote string) []string {
   231  			return []string{"hg", "cat", "-r", rev, file}
   232  		},
   233  		readZip: func(rev, subdir, remote, target string) []string {
   234  			pattern := []string{}
   235  			if subdir != "" {
   236  				pattern = []string{"-I", subdir + "/**"}
   237  			}
   238  			return str.StringList("hg", "archive", "-t", "zip", "--no-decode", "-r", rev, "--prefix=prefix/", pattern, "--", target)
   239  		},
   240  	},
   241  
   242  	"svn": {
   243  		vcs:  "svn",
   244  		init: nil, // no local checkout
   245  		tags: func(remote string) []string {
   246  			return []string{"svn", "list", "--", strings.TrimSuffix(remote, "/trunk") + "/tags"}
   247  		},
   248  		tagRE: re(`(?m)^(.*?)/?$`),
   249  		statLocal: func(rev, remote string) []string {
   250  			suffix := "@" + rev
   251  			if rev == "latest" {
   252  				suffix = ""
   253  			}
   254  			return []string{"svn", "log", "-l1", "--xml", "--", remote + suffix}
   255  		},
   256  		parseStat: svnParseStat,
   257  		latest:    "latest",
   258  		readFile: func(rev, file, remote string) []string {
   259  			return []string{"svn", "cat", "--", remote + "/" + file + "@" + rev}
   260  		},
   261  		doReadZip: svnReadZip,
   262  	},
   263  
   264  	"bzr": {
   265  		vcs: "bzr",
   266  		init: func(remote string) []string {
   267  			return []string{"bzr", "branch", "--use-existing-dir", "--", remote, "."}
   268  		},
   269  		fetch: []string{
   270  			"bzr", "pull", "--overwrite-tags",
   271  		},
   272  		tags: func(remote string) []string {
   273  			return []string{"bzr", "tags"}
   274  		},
   275  		tagRE:         re(`(?m)^\S+`),
   276  		badLocalRevRE: re(`^revno:-`),
   277  		statLocal: func(rev, remote string) []string {
   278  			return []string{"bzr", "log", "-l1", "--long", "--show-ids", "-r", rev}
   279  		},
   280  		parseStat: bzrParseStat,
   281  		latest:    "revno:-1",
   282  		readFile: func(rev, file, remote string) []string {
   283  			return []string{"bzr", "cat", "-r", rev, file}
   284  		},
   285  		readZip: func(rev, subdir, remote, target string) []string {
   286  			extra := []string{}
   287  			if subdir != "" {
   288  				extra = []string{"./" + subdir}
   289  			}
   290  			return str.StringList("bzr", "export", "--format=zip", "-r", rev, "--root=prefix/", "--", target, extra)
   291  		},
   292  	},
   293  
   294  	"fossil": {
   295  		vcs: "fossil",
   296  		init: func(remote string) []string {
   297  			return []string{"fossil", "clone", "--", remote, ".fossil"}
   298  		},
   299  		fetch: []string{"fossil", "pull", "-R", ".fossil"},
   300  		tags: func(remote string) []string {
   301  			return []string{"fossil", "tag", "-R", ".fossil", "list"}
   302  		},
   303  		tagRE: re(`XXXTODO`),
   304  		statLocal: func(rev, remote string) []string {
   305  			return []string{"fossil", "info", "-R", ".fossil", rev}
   306  		},
   307  		parseStat: fossilParseStat,
   308  		latest:    "trunk",
   309  		readFile: func(rev, file, remote string) []string {
   310  			return []string{"fossil", "cat", "-R", ".fossil", "-r", rev, file}
   311  		},
   312  		readZip: func(rev, subdir, remote, target string) []string {
   313  			extra := []string{}
   314  			if subdir != "" && !strings.ContainsAny(subdir, "*?[],") {
   315  				extra = []string{"--include", subdir}
   316  			}
   317  			// Note that vcsRepo.ReadZip below rewrites this command
   318  			// to run in a different directory, to work around a fossil bug.
   319  			return str.StringList("fossil", "zip", "-R", ".fossil", "--name", "prefix", extra, "--", rev, target)
   320  		},
   321  	},
   322  }
   323  
   324  func (r *vcsRepo) loadTags(ctx context.Context) {
   325  	if r.cmd.tagsNeedsFetch {
   326  		r.fetchOnce.Do(func() { r.fetch(ctx) })
   327  	}
   328  
   329  	out, err := Run(ctx, r.dir, r.cmd.tags(r.remote))
   330  	if err != nil {
   331  		return
   332  	}
   333  
   334  	// Run tag-listing command and extract tags.
   335  	r.tags = make(map[string]bool)
   336  	for _, tag := range r.cmd.tagRE.FindAllString(string(out), -1) {
   337  		if r.cmd.badLocalRevRE != nil && r.cmd.badLocalRevRE.MatchString(tag) {
   338  			continue
   339  		}
   340  		r.tags[tag] = true
   341  	}
   342  }
   343  
   344  func (r *vcsRepo) loadBranches(ctx context.Context) {
   345  	if r.cmd.branches == nil {
   346  		return
   347  	}
   348  
   349  	if r.cmd.branchesNeedsFetch {
   350  		r.fetchOnce.Do(func() { r.fetch(ctx) })
   351  	}
   352  
   353  	out, err := Run(ctx, r.dir, r.cmd.branches(r.remote))
   354  	if err != nil {
   355  		return
   356  	}
   357  
   358  	r.branches = make(map[string]bool)
   359  	for _, branch := range r.cmd.branchRE.FindAllString(string(out), -1) {
   360  		if r.cmd.badLocalRevRE != nil && r.cmd.badLocalRevRE.MatchString(branch) {
   361  			continue
   362  		}
   363  		r.branches[branch] = true
   364  	}
   365  }
   366  
   367  func (r *vcsRepo) loadRepoSum(ctx context.Context) {
   368  	if r.cmd.repoSum == nil {
   369  		return
   370  	}
   371  	where := r.remote
   372  	if r.fetched.Load() {
   373  		where = "." // use local repo
   374  	}
   375  	out, err := Run(ctx, r.dir, r.cmd.repoSum(where))
   376  	if err != nil {
   377  		return
   378  	}
   379  	r.repoSum = strings.TrimSpace(string(out))
   380  }
   381  
   382  func (r *vcsRepo) lookupRef(ctx context.Context, ref string) (string, error) {
   383  	if r.cmd.lookupRef == nil {
   384  		return "", fmt.Errorf("no lookupRef")
   385  	}
   386  	out, err := Run(ctx, r.dir, r.cmd.lookupRef(r.remote, ref))
   387  	if err != nil {
   388  		return "", err
   389  	}
   390  	return strings.TrimSpace(string(out)), nil
   391  }
   392  
   393  // repoSumOrigin returns an Origin containing a RepoSum.
   394  func (r *vcsRepo) repoSumOrigin(ctx context.Context) *Origin {
   395  	origin := &Origin{
   396  		VCS:     r.cmd.vcs,
   397  		URL:     r.remote,
   398  		RepoSum: r.repoSum,
   399  	}
   400  	r.repoSumOnce.Do(func() { r.loadRepoSum(ctx) })
   401  	origin.RepoSum = r.repoSum
   402  	return origin
   403  }
   404  
   405  func (r *vcsRepo) CheckReuse(ctx context.Context, old *Origin, subdir string) error {
   406  	if old == nil {
   407  		return fmt.Errorf("missing origin")
   408  	}
   409  	if old.VCS != r.cmd.vcs || old.URL != r.remote {
   410  		return fmt.Errorf("origin moved from %v %q to %v %q", old.VCS, old.URL, r.cmd.vcs, r.remote)
   411  	}
   412  	if old.Subdir != subdir {
   413  		return fmt.Errorf("origin moved from %v %q %q to %v %q %q", old.VCS, old.URL, old.Subdir, r.cmd.vcs, r.remote, subdir)
   414  	}
   415  
   416  	if old.Ref == "" && old.RepoSum == "" && old.Hash != "" {
   417  		// Hash has to remain in repo.
   418  		hash, err := r.lookupRef(ctx, old.Hash)
   419  		if err == nil && hash == old.Hash {
   420  			return nil
   421  		}
   422  		if err != nil {
   423  			return fmt.Errorf("looking up hash: %v", err)
   424  		}
   425  		return fmt.Errorf("hash changed") // weird but maybe they made a tag
   426  	}
   427  
   428  	if old.Ref != "" && old.RepoSum == "" {
   429  		hash, err := r.lookupRef(ctx, old.Ref)
   430  		if err == nil && hash != "" && hash == old.Hash {
   431  			return nil
   432  		}
   433  	}
   434  
   435  	r.repoSumOnce.Do(func() { r.loadRepoSum(ctx) })
   436  	if r.repoSum != "" {
   437  		if old.RepoSum == "" {
   438  			return fmt.Errorf("non-specific origin")
   439  		}
   440  		if old.RepoSum != r.repoSum {
   441  			return fmt.Errorf("repo changed")
   442  		}
   443  		return nil
   444  	}
   445  	return fmt.Errorf("vcs %s: CheckReuse: %w", r.cmd.vcs, errors.ErrUnsupported)
   446  }
   447  
   448  func (r *vcsRepo) Tags(ctx context.Context, prefix string) (*Tags, error) {
   449  	unlock, err := r.mu.Lock()
   450  	if err != nil {
   451  		return nil, err
   452  	}
   453  	defer unlock()
   454  
   455  	r.tagsOnce.Do(func() { r.loadTags(ctx) })
   456  	tags := &Tags{
   457  		Origin: r.repoSumOrigin(ctx),
   458  		List:   []Tag{},
   459  	}
   460  	for tag := range r.tags {
   461  		if strings.HasPrefix(tag, prefix) {
   462  			tags.List = append(tags.List, Tag{tag, ""})
   463  		}
   464  	}
   465  	sort.Slice(tags.List, func(i, j int) bool {
   466  		return tags.List[i].Name < tags.List[j].Name
   467  	})
   468  	return tags, nil
   469  }
   470  
   471  func (r *vcsRepo) Stat(ctx context.Context, rev string) (*RevInfo, error) {
   472  	unlock, err := r.mu.Lock()
   473  	if err != nil {
   474  		return nil, err
   475  	}
   476  	defer unlock()
   477  
   478  	if rev == "latest" {
   479  		rev = r.cmd.latest
   480  	}
   481  	r.branchesOnce.Do(func() { r.loadBranches(ctx) })
   482  	if r.local {
   483  		// Ignore the badLocalRevRE precondition in local only mode.
   484  		// We cannot fetch latest upstream changes so only serve what's in the local cache.
   485  		return r.statLocal(ctx, rev)
   486  	}
   487  	revOK := (r.cmd.badLocalRevRE == nil || !r.cmd.badLocalRevRE.MatchString(rev)) && !r.branches[rev]
   488  	if revOK {
   489  		if info, err := r.statLocal(ctx, rev); err == nil {
   490  			return info, nil
   491  		}
   492  	}
   493  
   494  	r.fetchOnce.Do(func() { r.fetch(ctx) })
   495  	if r.fetchErr != nil {
   496  		return nil, r.fetchErr
   497  	}
   498  	info, err := r.statLocal(ctx, rev)
   499  	if err != nil {
   500  		return info, err
   501  	}
   502  	if !revOK {
   503  		info.Version = info.Name
   504  	}
   505  	return info, nil
   506  }
   507  
   508  func (r *vcsRepo) fetch(ctx context.Context) {
   509  	if len(r.cmd.fetch) > 0 {
   510  		release, err := base.AcquireNet()
   511  		if err != nil {
   512  			r.fetchErr = err
   513  			return
   514  		}
   515  		_, r.fetchErr = Run(ctx, r.dir, r.cmd.fetch)
   516  		release()
   517  		r.fetched.Store(true)
   518  	}
   519  }
   520  
   521  func (r *vcsRepo) statLocal(ctx context.Context, rev string) (*RevInfo, error) {
   522  	out, err := Run(ctx, r.dir, r.cmd.statLocal(rev, r.remote))
   523  	if err != nil {
   524  		info := &RevInfo{Origin: r.repoSumOrigin(ctx)}
   525  		return info, &UnknownRevisionError{Rev: rev}
   526  	}
   527  	info, err := r.cmd.parseStat(rev, string(out))
   528  	if err != nil {
   529  		return nil, err
   530  	}
   531  	if info.Origin == nil {
   532  		info.Origin = new(Origin)
   533  	}
   534  	info.Origin.VCS = r.cmd.vcs
   535  	info.Origin.URL = r.remote
   536  	info.Origin.Ref = rev
   537  	if strings.HasPrefix(info.Name, rev) && len(rev) >= 12 {
   538  		info.Origin.Ref = "" // duplicates Hash
   539  	}
   540  	return info, nil
   541  }
   542  
   543  func (r *vcsRepo) Latest(ctx context.Context) (*RevInfo, error) {
   544  	return r.Stat(ctx, "latest")
   545  }
   546  
   547  func (r *vcsRepo) ReadFile(ctx context.Context, rev, file string, maxSize int64) ([]byte, error) {
   548  	if rev == "latest" {
   549  		rev = r.cmd.latest
   550  	}
   551  	_, err := r.Stat(ctx, rev) // download rev into local repo
   552  	if err != nil {
   553  		return nil, err
   554  	}
   555  
   556  	// r.Stat acquires r.mu, so lock after that.
   557  	unlock, err := r.mu.Lock()
   558  	if err != nil {
   559  		return nil, err
   560  	}
   561  	defer unlock()
   562  
   563  	out, err := Run(ctx, r.dir, r.cmd.readFile(rev, file, r.remote))
   564  	if err != nil {
   565  		return nil, fs.ErrNotExist
   566  	}
   567  	return out, nil
   568  }
   569  
   570  func (r *vcsRepo) RecentTag(ctx context.Context, rev, prefix string, allowed func(string) bool) (tag string, err error) {
   571  	// Only lock for the subprocess execution, not for the tag scan.
   572  	// allowed may call other methods that acquire the lock.
   573  	unlock, err := r.mu.Lock()
   574  	if err != nil {
   575  		return "", err
   576  	}
   577  
   578  	if r.cmd.recentTags == nil {
   579  		unlock()
   580  		return "", vcsErrorf("vcs %s: RecentTag: %w", r.cmd.vcs, errors.ErrUnsupported)
   581  	}
   582  	out, err := Run(ctx, r.dir, r.cmd.recentTags(rev))
   583  	unlock()
   584  	if err != nil {
   585  		return "", err
   586  	}
   587  
   588  	highest := ""
   589  	for _, tag := range strings.Fields(string(out)) {
   590  		if !strings.HasPrefix(tag, prefix) || !allowed(tag) {
   591  			continue
   592  		}
   593  		semtag := tag[len(prefix):]
   594  		if semver.Compare(semtag, highest) > 0 {
   595  			highest = semtag
   596  		}
   597  	}
   598  	if highest != "" {
   599  		return prefix + highest, nil
   600  	}
   601  	return "", nil
   602  }
   603  
   604  func (r *vcsRepo) DescendsFrom(ctx context.Context, rev, tag string) (bool, error) {
   605  	unlock, err := r.mu.Lock()
   606  	if err != nil {
   607  		return false, err
   608  	}
   609  	defer unlock()
   610  
   611  	if r.cmd.descendsFrom == nil {
   612  		return false, vcsErrorf("vcs %s: DescendsFrom: %w", r.cmd.vcs, errors.ErrUnsupported)
   613  	}
   614  
   615  	out, err := Run(ctx, r.dir, r.cmd.descendsFrom(rev, tag))
   616  	if err != nil {
   617  		return false, err
   618  	}
   619  	return strings.TrimSpace(string(out)) != "", nil
   620  }
   621  
   622  func (r *vcsRepo) ReadZip(ctx context.Context, rev, subdir string, maxSize int64) (zip io.ReadCloser, err error) {
   623  	if r.cmd.readZip == nil && r.cmd.doReadZip == nil {
   624  		return nil, vcsErrorf("vcs %s: ReadZip: %w", r.cmd.vcs, errors.ErrUnsupported)
   625  	}
   626  
   627  	if rev == "latest" {
   628  		rev = r.cmd.latest
   629  	}
   630  	_, err = r.Stat(ctx, rev) // download rev into local repo
   631  	if err != nil {
   632  		return nil, err
   633  	}
   634  
   635  	unlock, err := r.mu.Lock()
   636  	if err != nil {
   637  		return nil, err
   638  	}
   639  	defer unlock()
   640  
   641  	f, err := os.CreateTemp("", "go-readzip-*.zip")
   642  	if err != nil {
   643  		return nil, err
   644  	}
   645  	if r.cmd.doReadZip != nil {
   646  		lw := &limitedWriter{
   647  			W:               f,
   648  			N:               maxSize,
   649  			ErrLimitReached: errors.New("ReadZip: encoded file exceeds allowed size"),
   650  		}
   651  		err = r.cmd.doReadZip(ctx, lw, r.dir, rev, subdir, r.remote)
   652  		if err == nil {
   653  			_, err = f.Seek(0, io.SeekStart)
   654  		}
   655  	} else if r.cmd.vcs == "fossil" {
   656  		// If you run
   657  		//	fossil zip -R .fossil --name prefix trunk /tmp/x.zip
   658  		// fossil fails with "unable to create directory /tmp" [sic].
   659  		// Change the command to run in /tmp instead,
   660  		// replacing the -R argument with an absolute path.
   661  		args := r.cmd.readZip(rev, subdir, r.remote, filepath.Base(f.Name()))
   662  		for i := range args {
   663  			if args[i] == ".fossil" {
   664  				args[i] = filepath.Join(r.dir, ".fossil")
   665  			}
   666  		}
   667  		_, err = Run(ctx, filepath.Dir(f.Name()), args)
   668  	} else {
   669  		_, err = Run(ctx, r.dir, r.cmd.readZip(rev, subdir, r.remote, f.Name()))
   670  	}
   671  	if err != nil {
   672  		f.Close()
   673  		os.Remove(f.Name())
   674  		return nil, err
   675  	}
   676  	return &deleteCloser{f}, nil
   677  }
   678  
   679  // deleteCloser is a file that gets deleted on Close.
   680  type deleteCloser struct {
   681  	*os.File
   682  }
   683  
   684  func (d *deleteCloser) Close() error {
   685  	defer os.Remove(d.File.Name())
   686  	return d.File.Close()
   687  }
   688  
   689  func hgAddRemote(ctx context.Context, r *vcsRepo) error {
   690  	// Write .hg/hgrc with remote URL in it.
   691  	return os.WriteFile(filepath.Join(r.dir, ".hg/hgrc"), []byte(fmt.Sprintf("[paths]\ndefault = %s\n", r.remote)), 0666)
   692  }
   693  
   694  func hgParseStat(rev, out string) (*RevInfo, error) {
   695  	f := strings.Fields(out)
   696  	if len(f) < 3 {
   697  		return nil, vcsErrorf("unexpected response from hg log: %q", out)
   698  	}
   699  	hash := f[0]
   700  	version := rev
   701  	if strings.HasPrefix(hash, version) {
   702  		version = hash // extend to full hash
   703  	}
   704  	t, err := strconv.ParseInt(f[1], 10, 64)
   705  	if err != nil {
   706  		return nil, vcsErrorf("invalid time from hg log: %q", out)
   707  	}
   708  
   709  	var tags []string
   710  	for _, tag := range f[3:] {
   711  		if tag != "tip" {
   712  			tags = append(tags, tag)
   713  		}
   714  	}
   715  	sort.Strings(tags)
   716  
   717  	info := &RevInfo{
   718  		Origin:  &Origin{Hash: hash},
   719  		Name:    hash,
   720  		Short:   ShortenSHA1(hash),
   721  		Time:    time.Unix(t, 0).UTC(),
   722  		Version: version,
   723  		Tags:    tags,
   724  	}
   725  	return info, nil
   726  }
   727  
   728  func bzrParseStat(rev, out string) (*RevInfo, error) {
   729  	var revno int64
   730  	var tm time.Time
   731  	var tags []string
   732  	for line := range strings.SplitSeq(out, "\n") {
   733  		if line == "" || line[0] == ' ' || line[0] == '\t' {
   734  			// End of header, start of commit message.
   735  			break
   736  		}
   737  		if line[0] == '-' {
   738  			continue
   739  		}
   740  		before, after, found := strings.Cut(line, ":")
   741  		if !found {
   742  			// End of header, start of commit message.
   743  			break
   744  		}
   745  		key, val := before, strings.TrimSpace(after)
   746  		switch key {
   747  		case "revno":
   748  			if j := strings.Index(val, " "); j >= 0 {
   749  				val = val[:j]
   750  			}
   751  			i, err := strconv.ParseInt(val, 10, 64)
   752  			if err != nil {
   753  				return nil, vcsErrorf("unexpected revno from bzr log: %q", line)
   754  			}
   755  			revno = i
   756  		case "timestamp":
   757  			j := strings.Index(val, " ")
   758  			if j < 0 {
   759  				return nil, vcsErrorf("unexpected timestamp from bzr log: %q", line)
   760  			}
   761  			t, err := time.Parse("2006-01-02 15:04:05 -0700", val[j+1:])
   762  			if err != nil {
   763  				return nil, vcsErrorf("unexpected timestamp from bzr log: %q", line)
   764  			}
   765  			tm = t.UTC()
   766  		case "tags":
   767  			tags = strings.Split(val, ", ")
   768  		}
   769  	}
   770  	if revno == 0 || tm.IsZero() {
   771  		return nil, vcsErrorf("unexpected response from bzr log: %q", out)
   772  	}
   773  
   774  	info := &RevInfo{
   775  		Name:    strconv.FormatInt(revno, 10),
   776  		Short:   fmt.Sprintf("%012d", revno),
   777  		Time:    tm,
   778  		Version: rev,
   779  		Tags:    tags,
   780  	}
   781  	return info, nil
   782  }
   783  
   784  func fossilParseStat(rev, out string) (*RevInfo, error) {
   785  	for line := range strings.SplitSeq(out, "\n") {
   786  		if strings.HasPrefix(line, "uuid:") || strings.HasPrefix(line, "hash:") {
   787  			f := strings.Fields(line)
   788  			if len(f) != 5 || len(f[1]) != 40 || f[4] != "UTC" {
   789  				return nil, vcsErrorf("unexpected response from fossil info: %q", line)
   790  			}
   791  			t, err := time.Parse(time.DateTime, f[2]+" "+f[3])
   792  			if err != nil {
   793  				return nil, vcsErrorf("unexpected response from fossil info: %q", line)
   794  			}
   795  			hash := f[1]
   796  			version := rev
   797  			if strings.HasPrefix(hash, version) {
   798  				version = hash // extend to full hash
   799  			}
   800  			info := &RevInfo{
   801  				Origin:  &Origin{Hash: hash},
   802  				Name:    hash,
   803  				Short:   ShortenSHA1(hash),
   804  				Time:    t,
   805  				Version: version,
   806  			}
   807  			return info, nil
   808  		}
   809  	}
   810  	return nil, vcsErrorf("unexpected response from fossil info: %q", out)
   811  }
   812  
   813  type limitedWriter struct {
   814  	W               io.Writer
   815  	N               int64
   816  	ErrLimitReached error
   817  }
   818  
   819  func (l *limitedWriter) Write(p []byte) (n int, err error) {
   820  	if l.N > 0 {
   821  		max := len(p)
   822  		if l.N < int64(max) {
   823  			max = int(l.N)
   824  		}
   825  		n, err = l.W.Write(p[:max])
   826  		l.N -= int64(n)
   827  		if err != nil || n >= len(p) {
   828  			return n, err
   829  		}
   830  	}
   831  
   832  	return n, l.ErrLimitReached
   833  }
   834
View as plain text