git.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  	"bytes"
     9  	"context"
    10  	"crypto/sha256"
    11  	"encoding/base64"
    12  	"errors"
    13  	"fmt"
    14  	"io"
    15  	"io/fs"
    16  	"net/url"
    17  	"os"
    18  	"os/exec"
    19  	"path/filepath"
    20  	"runtime"
    21  	"slices"
    22  	"sort"
    23  	"strconv"
    24  	"strings"
    25  	"sync"
    26  	"time"
    27  
    28  	"cmd/go/internal/base"
    29  	"cmd/go/internal/lockedfile"
    30  	"cmd/go/internal/web"
    31  	"cmd/internal/par"
    32  
    33  	"golang.org/x/mod/semver"
    34  )
    35  
    36  // A notExistError wraps another error to retain its original text
    37  // but makes it opaquely equivalent to fs.ErrNotExist.
    38  type notExistError struct {
    39  	err error
    40  }
    41  
    42  func (e notExistError) Error() string   { return e.err.Error() }
    43  func (notExistError) Is(err error) bool { return err == fs.ErrNotExist }
    44  
    45  const gitWorkDirType = "git3"
    46  
    47  func newGitRepo(ctx context.Context, remote string, local bool) (Repo, error) {
    48  	r := &gitRepo{remote: remote, local: local}
    49  	if local {
    50  		if strings.Contains(remote, "://") { // Local flag, but URL provided
    51  			return nil, fmt.Errorf("git remote (%s) lookup disabled", remote)
    52  		}
    53  		info, err := os.Stat(remote)
    54  		if err != nil {
    55  			return nil, err
    56  		}
    57  		if !info.IsDir() {
    58  			return nil, fmt.Errorf("%s exists but is not a directory", remote)
    59  		}
    60  		r.dir = remote
    61  		r.mu.Path = r.dir + ".lock"
    62  		r.sha256Hashes = r.checkConfigSHA256(ctx)
    63  		return r, nil
    64  	}
    65  	// This is a remote path lookup.
    66  	if !strings.Contains(remote, "://") { // No URL scheme, could be host:path
    67  		if strings.Contains(remote, ":") {
    68  			return nil, fmt.Errorf("git remote (%s) must not be local directory (use URL syntax not host:path syntax)", remote)
    69  		}
    70  		return nil, fmt.Errorf("git remote (%s) must not be local directory", remote)
    71  	}
    72  	var err error
    73  	r.dir, r.mu.Path, err = WorkDir(ctx, gitWorkDirType, r.remote)
    74  	if err != nil {
    75  		return nil, err
    76  	}
    77  
    78  	unlock, err := r.mu.Lock()
    79  	if err != nil {
    80  		return nil, err
    81  	}
    82  	defer unlock()
    83  
    84  	if _, err := os.Stat(filepath.Join(r.dir, "objects")); err != nil {
    85  		repoSha256Hash := false
    86  		if refs, lrErr := r.loadRefs(ctx); lrErr == nil {
    87  			// Check any ref's hash, it doesn't matter which; they won't be mixed
    88  			// between sha1 and sha256 for the moment.
    89  			for _, refHash := range refs {
    90  				repoSha256Hash = len(refHash) == (256 / 4)
    91  				break
    92  			}
    93  		}
    94  		objFormatFlag := []string{}
    95  		if repoSha256Hash {
    96  			objFormatFlag = []string{"--object-format=sha256"}
    97  		}
    98  		if _, err := Run(ctx, r.dir, "git", "init", "--bare", objFormatFlag); err != nil {
    99  			os.RemoveAll(r.dir)
   100  			return nil, err
   101  		}
   102  		// We could just say git fetch https://whatever later,
   103  		// but this lets us say git fetch origin instead, which
   104  		// is a little nicer. More importantly, using a named remote
   105  		// avoids a problem with Git LFS. See golang.org/issue/25605.
   106  		if _, err := r.runGit(ctx, "git", "remote", "add", "origin", "--", r.remote); err != nil {
   107  			os.RemoveAll(r.dir)
   108  			return nil, err
   109  		}
   110  		if runtime.GOOS == "windows" {
   111  			// Git for Windows by default does not support paths longer than
   112  			// MAX_PATH (260 characters) because that may interfere with navigation
   113  			// in some Windows programs. However, cmd/go should be able to handle
   114  			// long paths just fine, and we expect people to use 'go clean' to
   115  			// manipulate the module cache, so it should be harmless to set here,
   116  			// and in some cases may be necessary in order to download modules with
   117  			// long branch names.
   118  			//
   119  			// See https://github.com/git-for-windows/git/wiki/Git-cannot-create-a-file-or-directory-with-a-long-path.
   120  			if _, err := r.runGit(ctx, "git", "config", "core.longpaths", "true"); err != nil {
   121  				os.RemoveAll(r.dir)
   122  				return nil, err
   123  			}
   124  		}
   125  	}
   126  	r.sha256Hashes = r.checkConfigSHA256(ctx)
   127  	r.remoteURL = r.remote
   128  	r.remote = "origin"
   129  	return r, nil
   130  }
   131  
   132  type gitRepo struct {
   133  	ctx context.Context
   134  
   135  	remote, remoteURL string
   136  	local             bool // local only lookups; no remote fetches
   137  	dir               string
   138  
   139  	// Repo uses the SHA256 for hashes, so expect the hashes to be 256/4 == 64-bytes in hex.
   140  	sha256Hashes bool
   141  
   142  	mu lockedfile.Mutex // protects fetchLevel and git repo state
   143  
   144  	fetchLevel int
   145  
   146  	statCache par.ErrCache[string, *RevInfo]
   147  
   148  	refsOnce sync.Once
   149  	// refs maps branch and tag refs (e.g., "HEAD", "refs/heads/master")
   150  	// to commits (e.g., "37ffd2e798afde829a34e8955b716ab730b2a6d6")
   151  	refs    map[string]string
   152  	refsErr error
   153  
   154  	localTagsOnce sync.Once
   155  	localTags     sync.Map // map[string]bool
   156  }
   157  
   158  const (
   159  	// How much have we fetched into the git repo (in this process)?
   160  	fetchNone = iota // nothing yet
   161  	fetchSome        // shallow fetches of individual hashes
   162  	fetchAll         // "fetch -t origin": get all remote branches and tags
   163  )
   164  
   165  // loadLocalTags loads tag references from the local git cache
   166  // into the map r.localTags.
   167  func (r *gitRepo) loadLocalTags(ctx context.Context) {
   168  	// The git protocol sends all known refs and ls-remote filters them on the client side,
   169  	// so we might as well record both heads and tags in one shot.
   170  	// Most of the time we only care about tags but sometimes we care about heads too.
   171  	out, err := r.runGit(ctx, "git", "tag", "-l")
   172  	if err != nil {
   173  		return
   174  	}
   175  
   176  	for _, line := range strings.Split(string(out), "\n") {
   177  		if line != "" {
   178  			r.localTags.Store(line, true)
   179  		}
   180  	}
   181  }
   182  
   183  func (r *gitRepo) CheckReuse(ctx context.Context, old *Origin, subdir string) error {
   184  	if old == nil {
   185  		return fmt.Errorf("missing origin")
   186  	}
   187  	if old.VCS != "git" || old.URL != r.remoteURL {
   188  		return fmt.Errorf("origin moved from %v %q to %v %q", old.VCS, old.URL, "git", r.remoteURL)
   189  	}
   190  	if old.Subdir != subdir {
   191  		return fmt.Errorf("origin moved from %v %q %q to %v %q %q", old.VCS, old.URL, old.Subdir, "git", r.remoteURL, subdir)
   192  	}
   193  
   194  	// Note: Can have Hash with no Ref and no TagSum and no RepoSum,
   195  	// meaning the Hash simply has to remain in the repo.
   196  	// In that case we assume it does in the absence of any real way to check.
   197  	// But if neither Hash nor TagSum is present, we have nothing to check,
   198  	// which we take to mean we didn't record enough information to be sure.
   199  	if old.Hash == "" && old.TagSum == "" && old.RepoSum == "" {
   200  		return fmt.Errorf("non-specific origin")
   201  	}
   202  
   203  	r.loadRefs(ctx)
   204  	if r.refsErr != nil {
   205  		return r.refsErr
   206  	}
   207  
   208  	if old.Ref != "" {
   209  		hash, ok := r.refs[old.Ref]
   210  		if !ok {
   211  			return fmt.Errorf("ref %q deleted", old.Ref)
   212  		}
   213  		if hash != old.Hash {
   214  			return fmt.Errorf("ref %q moved from %s to %s", old.Ref, old.Hash, hash)
   215  		}
   216  	}
   217  	if old.TagSum != "" {
   218  		tags, err := r.Tags(ctx, old.TagPrefix)
   219  		if err != nil {
   220  			return err
   221  		}
   222  		if tags.Origin.TagSum != old.TagSum {
   223  			return fmt.Errorf("tags changed")
   224  		}
   225  	}
   226  	if old.RepoSum != "" {
   227  		if r.repoSum(r.refs) != old.RepoSum {
   228  			return fmt.Errorf("refs changed")
   229  		}
   230  	}
   231  	return nil
   232  }
   233  
   234  // loadRefs loads heads and tags references from the remote into the map r.refs.
   235  // The result is cached in memory.
   236  func (r *gitRepo) loadRefs(ctx context.Context) (map[string]string, error) {
   237  	if r.local { // Return results from the cache if local only.
   238  		// In the future, we could consider loading r.refs using local git commands
   239  		// if desired.
   240  		return nil, nil
   241  	}
   242  	r.refsOnce.Do(func() {
   243  		// The git protocol sends all known refs and ls-remote filters them on the client side,
   244  		// so we might as well record both heads and tags in one shot.
   245  		// Most of the time we only care about tags but sometimes we care about heads too.
   246  		release, err := base.AcquireNet()
   247  		if err != nil {
   248  			r.refsErr = err
   249  			return
   250  		}
   251  		out, gitErr := r.runGit(ctx, "git", "ls-remote", "-q", r.remote)
   252  		release()
   253  
   254  		if gitErr != nil {
   255  			if rerr, ok := gitErr.(*RunError); ok {
   256  				if bytes.Contains(rerr.Stderr, []byte("fatal: could not read Username")) {
   257  					rerr.HelpText = "Confirm the import path was entered correctly.\nIf this is a private repository, see https://golang.org/doc/faq#git_https for additional information."
   258  				}
   259  			}
   260  
   261  			// If the remote URL doesn't exist at all, ideally we should treat the whole
   262  			// repository as nonexistent by wrapping the error in a notExistError.
   263  			// For HTTP and HTTPS, that's easy to detect: we'll try to fetch the URL
   264  			// ourselves and see what code it serves.
   265  			if u, err := url.Parse(r.remoteURL); err == nil && (u.Scheme == "http" || u.Scheme == "https") {
   266  				if _, err := web.GetBytes(u); errors.Is(err, fs.ErrNotExist) {
   267  					gitErr = notExistError{gitErr}
   268  				}
   269  			}
   270  
   271  			r.refsErr = gitErr
   272  			return
   273  		}
   274  
   275  		refs := make(map[string]string)
   276  		for _, line := range strings.Split(string(out), "\n") {
   277  			f := strings.Fields(line)
   278  			if len(f) != 2 {
   279  				continue
   280  			}
   281  			if f[1] == "HEAD" || strings.HasPrefix(f[1], "refs/heads/") || strings.HasPrefix(f[1], "refs/tags/") {
   282  				refs[f[1]] = f[0]
   283  			}
   284  		}
   285  		for ref, hash := range refs {
   286  			if k, found := strings.CutSuffix(ref, "^{}"); found { // record unwrapped annotated tag as value of tag
   287  				refs[k] = hash
   288  				delete(refs, ref)
   289  			}
   290  		}
   291  		r.refs = refs
   292  	})
   293  	return r.refs, r.refsErr
   294  }
   295  
   296  func (r *gitRepo) Tags(ctx context.Context, prefix string) (*Tags, error) {
   297  	refs, err := r.loadRefs(ctx)
   298  	if err != nil {
   299  		return nil, err
   300  	}
   301  
   302  	tags := &Tags{
   303  		Origin: &Origin{
   304  			VCS:       "git",
   305  			URL:       r.remoteURL,
   306  			TagPrefix: prefix,
   307  		},
   308  		List: []Tag{},
   309  	}
   310  	for ref, hash := range refs {
   311  		if !strings.HasPrefix(ref, "refs/tags/") {
   312  			continue
   313  		}
   314  		tag := ref[len("refs/tags/"):]
   315  		if !strings.HasPrefix(tag, prefix) {
   316  			continue
   317  		}
   318  		tags.List = append(tags.List, Tag{tag, hash})
   319  	}
   320  	sort.Slice(tags.List, func(i, j int) bool {
   321  		return tags.List[i].Name < tags.List[j].Name
   322  	})
   323  
   324  	dir := prefix[:strings.LastIndex(prefix, "/")+1]
   325  	h := sha256.New()
   326  	for _, tag := range tags.List {
   327  		if isOriginTag(strings.TrimPrefix(tag.Name, dir)) {
   328  			fmt.Fprintf(h, "%q %s\n", tag.Name, tag.Hash)
   329  		}
   330  	}
   331  	tags.Origin.TagSum = "t1:" + base64.StdEncoding.EncodeToString(h.Sum(nil))
   332  	return tags, nil
   333  }
   334  
   335  // repoSum returns a checksum of the entire repo state,
   336  // which can be checked (as Origin.RepoSum) to cache
   337  // the absence of a specific module version.
   338  // The caller must supply refs, the result of a successful r.loadRefs.
   339  func (r *gitRepo) repoSum(refs map[string]string) string {
   340  	list := make([]string, 0, len(refs))
   341  	for ref := range refs {
   342  		list = append(list, ref)
   343  	}
   344  	sort.Strings(list)
   345  	h := sha256.New()
   346  	for _, ref := range list {
   347  		fmt.Fprintf(h, "%q %s\n", ref, refs[ref])
   348  	}
   349  	return "r1:" + base64.StdEncoding.EncodeToString(h.Sum(nil))
   350  }
   351  
   352  // unknownRevisionInfo returns a RevInfo containing an Origin containing a RepoSum of refs,
   353  // for use when returning an UnknownRevisionError.
   354  func (r *gitRepo) unknownRevisionInfo(refs map[string]string) *RevInfo {
   355  	return &RevInfo{
   356  		Origin: &Origin{
   357  			VCS:     "git",
   358  			URL:     r.remoteURL,
   359  			RepoSum: r.repoSum(refs),
   360  		},
   361  	}
   362  }
   363  
   364  func (r *gitRepo) Latest(ctx context.Context) (*RevInfo, error) {
   365  	refs, err := r.loadRefs(ctx)
   366  	if err != nil {
   367  		return nil, err
   368  	}
   369  	if refs["HEAD"] == "" {
   370  		return nil, ErrNoCommits
   371  	}
   372  	statInfo, err := r.Stat(ctx, refs["HEAD"])
   373  	if err != nil {
   374  		return nil, err
   375  	}
   376  
   377  	// Stat may return cached info, so make a copy to modify here.
   378  	info := new(RevInfo)
   379  	*info = *statInfo
   380  	info.Origin = new(Origin)
   381  	if statInfo.Origin != nil {
   382  		*info.Origin = *statInfo.Origin
   383  	}
   384  	info.Origin.Ref = "HEAD"
   385  	info.Origin.Hash = refs["HEAD"]
   386  
   387  	return info, nil
   388  }
   389  
   390  // findRef finds some ref name for the given hash,
   391  // for use when the server requires giving a ref instead of a hash.
   392  // There may be multiple ref names for a given hash,
   393  // in which case this returns some name - it doesn't matter which.
   394  func (r *gitRepo) findRef(ctx context.Context, hash string) (ref string, ok bool) {
   395  	refs, err := r.loadRefs(ctx)
   396  	if err != nil {
   397  		return "", false
   398  	}
   399  	for ref, h := range refs {
   400  		if h == hash {
   401  			return ref, true
   402  		}
   403  	}
   404  	return "", false
   405  }
   406  
   407  func (r *gitRepo) checkConfigSHA256(ctx context.Context) bool {
   408  	if hashType, sha256CfgErr := r.runGit(ctx, "git", "config", "extensions.objectformat"); sha256CfgErr == nil {
   409  		return "sha256" == strings.TrimSpace(string(hashType))
   410  	}
   411  	return false
   412  }
   413  
   414  func (r *gitRepo) hexHashLen() int {
   415  	if !r.sha256Hashes {
   416  		return 160 / 4
   417  	}
   418  	return 256 / 4
   419  }
   420  
   421  // shortenObjectHash shortens a SHA1 or SHA256 hash (40 or 64 hex digits) to
   422  // the canonical length used in pseudo-versions (12 hex digits).
   423  func (r *gitRepo) shortenObjectHash(rev string) string {
   424  	if !r.sha256Hashes {
   425  		return ShortenSHA1(rev)
   426  	}
   427  	if AllHex(rev) && len(rev) == 256/4 {
   428  		return rev[:12]
   429  	}
   430  	return rev
   431  }
   432  
   433  // minHashDigits is the minimum number of digits to require
   434  // before accepting a hex digit sequence as potentially identifying
   435  // a specific commit in a git repo. (Of course, users can always
   436  // specify more digits, and many will paste in all 40 digits,
   437  // but many of git's commands default to printing short hashes
   438  // as 7 digits.)
   439  const minHashDigits = 7
   440  
   441  // stat stats the given rev in the local repository,
   442  // or else it fetches more info from the remote repository and tries again.
   443  func (r *gitRepo) stat(ctx context.Context, rev string) (info *RevInfo, err error) {
   444  	// Fast path: maybe rev is a hash we already have locally.
   445  	didStatLocal := false
   446  	if len(rev) >= minHashDigits && len(rev) <= r.hexHashLen() && AllHex(rev) {
   447  		if info, err := r.statLocal(ctx, rev, rev); err == nil {
   448  			return info, nil
   449  		}
   450  		didStatLocal = true
   451  	}
   452  
   453  	// Maybe rev is a tag we already have locally.
   454  	// (Note that we're excluding branches, which can be stale.)
   455  	r.localTagsOnce.Do(func() { r.loadLocalTags(ctx) })
   456  	if _, ok := r.localTags.Load(rev); ok {
   457  		return r.statLocal(ctx, rev, "refs/tags/"+rev)
   458  	}
   459  
   460  	// Maybe rev is the name of a tag or branch on the remote server.
   461  	// Or maybe it's the prefix of a hash of a named ref.
   462  	// Try to resolve to both a ref (git name) and full (40-hex-digit for
   463  	// sha1 64 for sha256) commit hash.
   464  	refs, err := r.loadRefs(ctx)
   465  	if err != nil {
   466  		return nil, err
   467  	}
   468  	// loadRefs may return an error if git fails, for example segfaults, or
   469  	// could not load a private repo, but defer checking to the else block
   470  	// below, in case we already have the rev in question in the local cache.
   471  	var ref, hash string
   472  	if refs["refs/tags/"+rev] != "" {
   473  		ref = "refs/tags/" + rev
   474  		hash = refs[ref]
   475  		// Keep rev as is: tags are assumed not to change meaning.
   476  	} else if refs["refs/heads/"+rev] != "" {
   477  		ref = "refs/heads/" + rev
   478  		hash = refs[ref]
   479  		rev = hash // Replace rev, because meaning of refs/heads/foo can change.
   480  	} else if rev == "HEAD" && refs["HEAD"] != "" {
   481  		ref = "HEAD"
   482  		hash = refs[ref]
   483  		rev = hash // Replace rev, because meaning of HEAD can change.
   484  	} else if len(rev) >= minHashDigits && len(rev) <= r.hexHashLen() && AllHex(rev) {
   485  		// At the least, we have a hash prefix we can look up after the fetch below.
   486  		// Maybe we can map it to a full hash using the known refs.
   487  		prefix := rev
   488  		// Check whether rev is prefix of known ref hash.
   489  		for k, h := range refs {
   490  			if strings.HasPrefix(h, prefix) {
   491  				if hash != "" && hash != h {
   492  					// Hash is an ambiguous hash prefix.
   493  					// More information will not change that.
   494  					return nil, fmt.Errorf("ambiguous revision %s", rev)
   495  				}
   496  				if ref == "" || ref > k { // Break ties deterministically when multiple refs point at same hash.
   497  					ref = k
   498  				}
   499  				rev = h
   500  				hash = h
   501  			}
   502  		}
   503  		if hash == "" && len(rev) == r.hexHashLen() { // Didn't find a ref, but rev is a full hash.
   504  			hash = rev
   505  		}
   506  	} else {
   507  		return r.unknownRevisionInfo(refs), &UnknownRevisionError{Rev: rev}
   508  	}
   509  
   510  	defer func() {
   511  		if info != nil {
   512  			info.Origin.Hash = info.Name
   513  			// There's a ref = hash below; don't write that hash down as Origin.Ref.
   514  			if ref != info.Origin.Hash {
   515  				info.Origin.Ref = ref
   516  			}
   517  		}
   518  	}()
   519  
   520  	// Protect r.fetchLevel and the "fetch more and more" sequence.
   521  	unlock, err := r.mu.Lock()
   522  	if err != nil {
   523  		return nil, err
   524  	}
   525  	defer unlock()
   526  
   527  	// Perhaps r.localTags did not have the ref when we loaded local tags,
   528  	// but we've since done fetches that pulled down the hash we need
   529  	// (or already have the hash we need, just without its tag).
   530  	// Either way, try a local stat before falling back to network I/O.
   531  	if !didStatLocal {
   532  		if info, err := r.statLocal(ctx, rev, hash); err == nil {
   533  			tag, fromTag := strings.CutPrefix(ref, "refs/tags/")
   534  			if fromTag && !slices.Contains(info.Tags, tag) {
   535  				// The local repo includes the commit hash we want, but it is missing
   536  				// the corresponding tag. Add that tag and try again.
   537  				_, err := r.runGit(ctx, "git", "tag", tag, hash)
   538  				if err != nil {
   539  					return nil, err
   540  				}
   541  				r.localTags.Store(tag, true)
   542  				return r.statLocal(ctx, rev, ref)
   543  			}
   544  			return info, err
   545  		}
   546  	}
   547  
   548  	if r.local { // at this point, we have determined that we need to fetch rev, fail early if local only mode.
   549  		return nil, fmt.Errorf("revision does not exist locally: %s", rev)
   550  	}
   551  
   552  	// If we know a specific commit we need and its ref, fetch it.
   553  	// We do NOT fetch arbitrary hashes (when we don't know the ref)
   554  	// because we want to avoid ever importing a commit that isn't
   555  	// reachable from refs/tags/* or refs/heads/* or HEAD.
   556  	// Both Gerrit and GitHub expose every CL/PR as a named ref,
   557  	// and we don't want those commits masquerading as being real
   558  	// pseudo-versions in the main repo.
   559  	if r.fetchLevel <= fetchSome && ref != "" && hash != "" {
   560  		r.fetchLevel = fetchSome
   561  		var refspec string
   562  		if ref == "HEAD" {
   563  			// Fetch the hash but give it a local name (refs/dummy),
   564  			// because that triggers the fetch behavior of creating any
   565  			// other known remote tags for the hash. We never use
   566  			// refs/dummy (it's not refs/tags/dummy) and it will be
   567  			// overwritten in the next command, and that's fine.
   568  			ref = hash
   569  			refspec = hash + ":refs/dummy"
   570  		} else {
   571  			// If we do know the ref name, save the mapping locally
   572  			// so that (if it is a tag) it can show up in localTags
   573  			// on a future call. Also, some servers refuse to allow
   574  			// full hashes in ref specs, so prefer a ref name if known.
   575  			refspec = ref + ":" + ref
   576  		}
   577  
   578  		release, err := base.AcquireNet()
   579  		if err != nil {
   580  			return nil, err
   581  		}
   582  		// We explicitly set protocol.version=2 for this command to work around
   583  		// an apparent Git bug introduced in Git 2.21 (commit 61c771),
   584  		// which causes the handler for protocol version 1 to sometimes miss
   585  		// tags that point to the requested commit (see https://go.dev/issue/56881).
   586  		_, err = r.runGit(ctx, "git", "-c", "protocol.version=2", "fetch", "-f", "--depth=1", r.remote, refspec)
   587  		release()
   588  
   589  		if err == nil {
   590  			return r.statLocal(ctx, rev, ref)
   591  		}
   592  		// Don't try to be smart about parsing the error.
   593  		// It's too complex and varies too much by git version.
   594  		// No matter what went wrong, fall back to a complete fetch.
   595  	}
   596  
   597  	// Last resort.
   598  	// Fetch all heads and tags and hope the hash we want is in the history.
   599  	if err := r.fetchRefsLocked(ctx); err != nil {
   600  		return nil, err
   601  	}
   602  
   603  	return r.statLocal(ctx, rev, rev)
   604  }
   605  
   606  // fetchRefsLocked fetches all heads and tags from the origin, along with the
   607  // ancestors of those commits.
   608  //
   609  // We only fetch heads and tags, not arbitrary other commits: we don't want to
   610  // pull in off-branch commits (such as rejected GitHub pull requests) that the
   611  // server may be willing to provide. (See the comments within the stat method
   612  // for more detail.)
   613  //
   614  // fetchRefsLocked requires that r.mu remain locked for the duration of the call.
   615  func (r *gitRepo) fetchRefsLocked(ctx context.Context) error {
   616  	if r.local {
   617  		panic("go: fetchRefsLocked called in local only mode.")
   618  	}
   619  	if r.fetchLevel < fetchAll {
   620  		// NOTE: To work around a bug affecting Git clients up to at least 2.23.0
   621  		// (2019-08-16), we must first expand the set of local refs, and only then
   622  		// unshallow the repository as a separate fetch operation. (See
   623  		// golang.org/issue/34266 and
   624  		// https://github.com/git/git/blob/4c86140027f4a0d2caaa3ab4bd8bfc5ce3c11c8a/transport.c#L1303-L1309.)
   625  
   626  		release, err := base.AcquireNet()
   627  		if err != nil {
   628  			return err
   629  		}
   630  		defer release()
   631  
   632  		if _, err := r.runGit(ctx, "git", "fetch", "-f", r.remote, "refs/heads/*:refs/heads/*", "refs/tags/*:refs/tags/*"); err != nil {
   633  			return err
   634  		}
   635  
   636  		if _, err := os.Stat(filepath.Join(r.dir, "shallow")); err == nil {
   637  			if _, err := r.runGit(ctx, "git", "fetch", "--unshallow", "-f", r.remote); err != nil {
   638  				return err
   639  			}
   640  		}
   641  
   642  		r.fetchLevel = fetchAll
   643  	}
   644  	return nil
   645  }
   646  
   647  // statLocal returns a new RevInfo describing rev in the local git repository.
   648  // It uses version as info.Version.
   649  func (r *gitRepo) statLocal(ctx context.Context, version, rev string) (*RevInfo, error) {
   650  	out, err := r.runGit(ctx, "git", "-c", "log.showsignature=false", "log", "--no-decorate", "-n1", "--format=format:%H %ct %D", rev, "--")
   651  	if err != nil {
   652  		// Return info with Origin.RepoSum if possible to allow caching of negative lookup.
   653  		var info *RevInfo
   654  		if refs, err := r.loadRefs(ctx); err == nil {
   655  			info = r.unknownRevisionInfo(refs)
   656  		}
   657  		return info, &UnknownRevisionError{Rev: rev}
   658  	}
   659  	f := strings.Fields(string(out))
   660  	if len(f) < 2 {
   661  		return nil, fmt.Errorf("unexpected response from git log: %q", out)
   662  	}
   663  	hash := f[0]
   664  	if strings.HasPrefix(hash, version) {
   665  		version = hash // extend to full hash
   666  	}
   667  	t, err := strconv.ParseInt(f[1], 10, 64)
   668  	if err != nil {
   669  		return nil, fmt.Errorf("invalid time from git log: %q", out)
   670  	}
   671  
   672  	info := &RevInfo{
   673  		Origin: &Origin{
   674  			VCS:  "git",
   675  			URL:  r.remoteURL,
   676  			Hash: hash,
   677  		},
   678  		Name:    hash,
   679  		Short:   r.shortenObjectHash(hash),
   680  		Time:    time.Unix(t, 0).UTC(),
   681  		Version: hash,
   682  	}
   683  	if !strings.HasPrefix(hash, rev) {
   684  		info.Origin.Ref = rev
   685  	}
   686  
   687  	// Add tags. Output looks like:
   688  	//	ede458df7cd0fdca520df19a33158086a8a68e81 1523994202 HEAD -> master, tag: v1.2.4-annotated, tag: v1.2.3, origin/master, origin/HEAD
   689  	for i := 2; i < len(f); i++ {
   690  		if f[i] == "tag:" {
   691  			i++
   692  			if i < len(f) {
   693  				info.Tags = append(info.Tags, strings.TrimSuffix(f[i], ","))
   694  			}
   695  		}
   696  	}
   697  
   698  	// Git 2.47.1 does not send the tags during shallow clone anymore
   699  	// (perhaps the exact version that changed behavior is an earlier one),
   700  	// so we have to also add tags from the refs list we fetched with ls-remote.
   701  	if refs, err := r.loadRefs(ctx); err == nil {
   702  		for ref, h := range refs {
   703  			if h == hash {
   704  				if tag, found := strings.CutPrefix(ref, "refs/tags/"); found {
   705  					info.Tags = append(info.Tags, tag)
   706  				}
   707  			}
   708  		}
   709  	}
   710  	slices.Sort(info.Tags)
   711  	info.Tags = slices.Compact(info.Tags)
   712  
   713  	// Used hash as info.Version above.
   714  	// Use caller's suggested version if it appears in the tag list
   715  	// (filters out branch names, HEAD).
   716  	for _, tag := range info.Tags {
   717  		if version == tag {
   718  			info.Version = version
   719  		}
   720  	}
   721  
   722  	return info, nil
   723  }
   724  
   725  func (r *gitRepo) Stat(ctx context.Context, rev string) (*RevInfo, error) {
   726  	if rev == "latest" {
   727  		return r.Latest(ctx)
   728  	}
   729  	return r.statCache.Do(rev, func() (*RevInfo, error) {
   730  		return r.stat(ctx, rev)
   731  	})
   732  }
   733  
   734  func (r *gitRepo) ReadFile(ctx context.Context, rev, file string, maxSize int64) ([]byte, error) {
   735  	// TODO: Could use git cat-file --batch.
   736  	info, err := r.Stat(ctx, rev) // download rev into local git repo
   737  	if err != nil {
   738  		return nil, err
   739  	}
   740  	out, err := r.runGit(ctx, "git", "cat-file", "blob", info.Name+":"+file)
   741  	if err != nil {
   742  		return nil, fs.ErrNotExist
   743  	}
   744  	return out, nil
   745  }
   746  
   747  func (r *gitRepo) RecentTag(ctx context.Context, rev, prefix string, allowed func(tag string) bool) (tag string, err error) {
   748  	info, err := r.Stat(ctx, rev)
   749  	if err != nil {
   750  		return "", err
   751  	}
   752  	rev = info.Name // expand hash prefixes
   753  
   754  	// describe sets tag and err using 'git for-each-ref' and reports whether the
   755  	// result is definitive.
   756  	describe := func() (definitive bool) {
   757  		var out []byte
   758  		out, err = r.runGit(ctx, "git", "for-each-ref", "--format", "%(refname)", "refs/tags", "--merged", rev)
   759  		if err != nil {
   760  			return true
   761  		}
   762  
   763  		// prefixed tags aren't valid semver tags so compare without prefix, but only tags with correct prefix
   764  		var highest string
   765  		for _, line := range strings.Split(string(out), "\n") {
   766  			line = strings.TrimSpace(line)
   767  			// git do support lstrip in for-each-ref format, but it was added in v2.13.0. Stripping here
   768  			// instead gives support for git v2.7.0.
   769  			if !strings.HasPrefix(line, "refs/tags/") {
   770  				continue
   771  			}
   772  			line = line[len("refs/tags/"):]
   773  
   774  			if !strings.HasPrefix(line, prefix) {
   775  				continue
   776  			}
   777  			if !allowed(line) {
   778  				continue
   779  			}
   780  
   781  			semtag := line[len(prefix):]
   782  			if semver.Compare(semtag, highest) > 0 {
   783  				highest = semtag
   784  			}
   785  		}
   786  
   787  		if highest != "" {
   788  			tag = prefix + highest
   789  		}
   790  
   791  		return tag != "" && !AllHex(tag)
   792  	}
   793  
   794  	if describe() {
   795  		return tag, err
   796  	}
   797  
   798  	// Git didn't find a version tag preceding the requested rev.
   799  	// See whether any plausible tag exists.
   800  	tags, err := r.Tags(ctx, prefix+"v")
   801  	if err != nil {
   802  		return "", err
   803  	}
   804  	if len(tags.List) == 0 {
   805  		return "", nil
   806  	}
   807  
   808  	if r.local { // at this point, we have determined that we need to fetch rev, fail early if local only mode.
   809  		return "", fmt.Errorf("revision does not exist locally: %s", rev)
   810  	}
   811  	// There are plausible tags, but we don't know if rev is a descendent of any of them.
   812  	// Fetch the history to find out.
   813  
   814  	unlock, err := r.mu.Lock()
   815  	if err != nil {
   816  		return "", err
   817  	}
   818  	defer unlock()
   819  
   820  	if err := r.fetchRefsLocked(ctx); err != nil {
   821  		return "", err
   822  	}
   823  
   824  	// If we've reached this point, we have all of the commits that are reachable
   825  	// from all heads and tags.
   826  	//
   827  	// The only refs we should be missing are those that are no longer reachable
   828  	// (or never were reachable) from any branch or tag, including the master
   829  	// branch, and we don't want to resolve them anyway (they're probably
   830  	// unreachable for a reason).
   831  	//
   832  	// Try one last time in case some other goroutine fetched rev while we were
   833  	// waiting on the lock.
   834  	describe()
   835  	return tag, err
   836  }
   837  
   838  func (r *gitRepo) DescendsFrom(ctx context.Context, rev, tag string) (bool, error) {
   839  	// The "--is-ancestor" flag was added to "git merge-base" in version 1.8.0, so
   840  	// this won't work with Git 1.7.1. According to golang.org/issue/28550, cmd/go
   841  	// already doesn't work with Git 1.7.1, so at least it's not a regression.
   842  	//
   843  	// git merge-base --is-ancestor exits with status 0 if rev is an ancestor, or
   844  	// 1 if not.
   845  	_, err := r.runGit(ctx, "git", "merge-base", "--is-ancestor", "--", tag, rev)
   846  
   847  	// Git reports "is an ancestor" with exit code 0 and "not an ancestor" with
   848  	// exit code 1.
   849  	// Unfortunately, if we've already fetched rev with a shallow history, git
   850  	// merge-base has been observed to report a false-negative, so don't stop yet
   851  	// even if the exit code is 1!
   852  	if err == nil {
   853  		return true, nil
   854  	}
   855  
   856  	// See whether the tag and rev even exist.
   857  	tags, err := r.Tags(ctx, tag)
   858  	if err != nil {
   859  		return false, err
   860  	}
   861  	if len(tags.List) == 0 {
   862  		return false, nil
   863  	}
   864  
   865  	// NOTE: r.stat is very careful not to fetch commits that we shouldn't know
   866  	// about, like rejected GitHub pull requests, so don't try to short-circuit
   867  	// that here.
   868  	if _, err = r.stat(ctx, rev); err != nil {
   869  		return false, err
   870  	}
   871  
   872  	if r.local { // at this point, we have determined that we need to fetch rev, fail early if local only mode.
   873  		return false, fmt.Errorf("revision does not exist locally: %s", rev)
   874  	}
   875  
   876  	// Now fetch history so that git can search for a path.
   877  	unlock, err := r.mu.Lock()
   878  	if err != nil {
   879  		return false, err
   880  	}
   881  	defer unlock()
   882  
   883  	if r.fetchLevel < fetchAll {
   884  		// Fetch the complete history for all refs and heads. It would be more
   885  		// efficient to only fetch the history from rev to tag, but that's much more
   886  		// complicated, and any kind of shallow fetch is fairly likely to trigger
   887  		// bugs in JGit servers and/or the go command anyway.
   888  		if err := r.fetchRefsLocked(ctx); err != nil {
   889  			return false, err
   890  		}
   891  	}
   892  
   893  	_, err = r.runGit(ctx, "git", "merge-base", "--is-ancestor", "--", tag, rev)
   894  	if err == nil {
   895  		return true, nil
   896  	}
   897  	if ee, ok := err.(*RunError).Err.(*exec.ExitError); ok && ee.ExitCode() == 1 {
   898  		return false, nil
   899  	}
   900  	return false, err
   901  }
   902  
   903  func (r *gitRepo) ReadZip(ctx context.Context, rev, subdir string, maxSize int64) (zip io.ReadCloser, err error) {
   904  	// TODO: Use maxSize or drop it.
   905  	args := []string{}
   906  	if subdir != "" {
   907  		args = append(args, "--", subdir)
   908  	}
   909  	info, err := r.Stat(ctx, rev) // download rev into local git repo
   910  	if err != nil {
   911  		return nil, err
   912  	}
   913  
   914  	unlock, err := r.mu.Lock()
   915  	if err != nil {
   916  		return nil, err
   917  	}
   918  	defer unlock()
   919  
   920  	if err := ensureGitAttributes(r.dir); err != nil {
   921  		return nil, err
   922  	}
   923  
   924  	// Incredibly, git produces different archives depending on whether
   925  	// it is running on a Windows system or not, in an attempt to normalize
   926  	// text file line endings. Setting -c core.autocrlf=input means only
   927  	// translate files on the way into the repo, not on the way out (archive).
   928  	// The -c core.eol=lf should be unnecessary but set it anyway.
   929  	archive, err := r.runGit(ctx, "git", "-c", "core.autocrlf=input", "-c", "core.eol=lf", "archive", "--format=zip", "--prefix=prefix/", info.Name, args)
   930  	if err != nil {
   931  		if bytes.Contains(err.(*RunError).Stderr, []byte("did not match any files")) {
   932  			return nil, fs.ErrNotExist
   933  		}
   934  		return nil, err
   935  	}
   936  
   937  	return io.NopCloser(bytes.NewReader(archive)), nil
   938  }
   939  
   940  // ensureGitAttributes makes sure export-subst and export-ignore features are
   941  // disabled for this repo. This is intended to be run prior to running git
   942  // archive so that zip files are generated that produce consistent ziphashes
   943  // for a given revision, independent of variables such as git version and the
   944  // size of the repo.
   945  //
   946  // See: https://github.com/golang/go/issues/27153
   947  func ensureGitAttributes(repoDir string) (err error) {
   948  	const attr = "\n* -export-subst -export-ignore\n"
   949  
   950  	d := repoDir + "/info"
   951  	p := d + "/attributes"
   952  
   953  	if err := os.MkdirAll(d, 0755); err != nil {
   954  		return err
   955  	}
   956  
   957  	f, err := os.OpenFile(p, os.O_CREATE|os.O_APPEND|os.O_RDWR, 0666)
   958  	if err != nil {
   959  		return err
   960  	}
   961  	defer func() {
   962  		closeErr := f.Close()
   963  		if closeErr != nil {
   964  			err = closeErr
   965  		}
   966  	}()
   967  
   968  	b, err := io.ReadAll(f)
   969  	if err != nil {
   970  		return err
   971  	}
   972  	if !bytes.HasSuffix(b, []byte(attr)) {
   973  		_, err := f.WriteString(attr)
   974  		return err
   975  	}
   976  
   977  	return nil
   978  }
   979  
   980  func (r *gitRepo) runGit(ctx context.Context, cmdline ...any) ([]byte, error) {
   981  	args := RunArgs{cmdline: cmdline, dir: r.dir, local: r.local}
   982  	if !r.local {
   983  		// Manually supply GIT_DIR so Git works with safe.bareRepository=explicit set.
   984  		// This is necessary only for remote repositories as they are initialized with git init --bare.
   985  		args.env = []string{"GIT_DIR=" + r.dir}
   986  	}
   987  	return RunWithArgs(ctx, args)
   988  }
   989
View as plain text