internal/gitutil/gitutil.go
326 lines · 12337 bytes
1// Package gitutil wraps the system git binary. All repository access goes
2// through git subprocesses; there is no in-process git implementation.
3package gitutil
4
5import (
6 "context"
7 "fmt"
8 "io"
9 "io/fs"
10 "os"
11 "os/exec"
12 "path/filepath"
13 "strings"
14
15 "gitbay.org/gitbay/internal/toolpath"
16)
17
18// InitBare creates a bare repository with the shared hooks directory wired
19// via core.hooksPath.
20func InitBare(path, defaultBranch, hooksPath string) error {
21 if err := os.MkdirAll(filepath.Dir(path), 0o750); err != nil {
22 return err
23 }
24 cmd := exec.Command(toolpath.Look("git"), "init", "--bare", "--initial-branch="+defaultBranch, path)
25 if out, err := cmd.CombinedOutput(); err != nil {
26 return fmt.Errorf("git init: %v\n%s", err, out)
27 }
28 cmd = exec.Command(toolpath.Look("git"), "-C", path, "config", "core.hooksPath", hooksPath)
29 if out, err := cmd.CombinedOutput(); err != nil {
30 return fmt.Errorf("git config core.hooksPath: %v\n%s", err, out)
31 }
32 return nil
33}
34
35// Transport streams one git transport service (upload-pack, receive-pack,
36// upload-archive). extraEnv entries are appended to the process environment;
37// hooks read the GITBAY_* variables from it. maxPack caps incoming pack
38// bytes on receive-pack (0 = unlimited). Closing cancel kills the service
39// and everything it started; a push killed before its pre-receive hook
40// answers updates no refs. A nil cancel never fires.
41func Transport(service, repoPath string, stdin io.Reader, stdout, errW io.Writer, extraEnv []string, maxPack int64, cancel <-chan struct{}) error {
42 var args []string
43 switch service {
44 case "git-upload-pack", "git-receive-pack", "git-upload-archive":
45 if service == "git-receive-pack" && maxPack > 0 {
46 args = []string{"-c", fmt.Sprintf("receive.maxInputSize=%d", maxPack)}
47 }
48 args = append(args, strings.TrimPrefix(service, "git-"), repoPath)
49 default:
50 return fmt.Errorf("unknown service %q", service)
51 }
52 cmd := exec.Command(toolpath.Look("git"), args...)
53 cmd.Env = append(os.Environ(), extraEnv...)
54 cmd.Stdin = stdin
55 cmd.Stdout = stdout
56 cmd.Stderr = errW
57 ownProcessGroup(cmd)
58 if err := cmd.Start(); err != nil {
59 return err
60 }
61 finished := make(chan struct{})
62 go func() {
63 select {
64 case <-cancel:
65 killTree(cmd)
66 case <-finished:
67 }
68 }()
69 err := cmd.Wait()
70 close(finished)
71 return err
72}
73
74// IsAncestor reports whether old is an ancestor of new in the repository at
75// dir. It must run with the caller's environment intact so that quarantined
76// objects during pre-receive remain visible.
77func IsAncestor(dir, old, new string) (bool, error) {
78 cmd := exec.Command(toolpath.Look("git"), "-C", dir, "merge-base", "--is-ancestor", old, new)
79 err := cmd.Run()
80 if err == nil {
81 return true, nil
82 }
83 if ee, ok := err.(*exec.ExitError); ok && ee.ExitCode() == 1 {
84 return false, nil
85 }
86 return false, err
87}
88
89// Reachable reports whether sha is reachable from some ref in the
90// repository at dir — the condition a clone must find true to have any
91// chance of checking it out. That includes refs outside refs/heads and
92// refs/tags: a merge request head fetched from a fork lives at
93// refs/merge-requests/<n>/head, and is exactly as fetchable as a branch
94// tip, so it must count as reachable too — a first pass of this function
95// restricted the check to branches and tags and read every fork MR's
96// queued build as unreachable, cancelling it. false comes from two
97// shapes, and Reachable does not need to tell them apart: the object is
98// already gone (pruned), or it is still in the object store but nothing
99// points at it any more (a force-push moved the branch, gc has not run
100// yet). Either way the answer is the same: no ref reaches it.
101//
102// A non-nil error means the check itself did not run to a clean answer —
103// missing repository, git failing for its own reasons — and false is
104// meaningless in that case. The caller must not read err as "unreachable":
105// that would cancel a build the check never actually looked at.
106func Reachable(dir, sha string) (bool, error) {
107 if _, err := os.Stat(dir); err != nil {
108 return false, fmt.Errorf("reachable %s: %w", sha, err)
109 }
110 // cat-file -e <object>, unpeeled, is git's own existence predicate with
111 // a documented exit code: 1 means the object is not there, the same
112 // contract IsAncestor above already trusts from merge-base
113 // --is-ancestor. Peeling to ^{commit} breaks that contract: a missing
114 // object then exits 128, indistinguishable from "not a git repository"
115 // or a corrupted one — the ambiguous case that must never read as
116 // "unreachable" and cancel a build the check never actually looked at.
117 err := exec.Command(toolpath.Look("git"), "-C", dir, "cat-file", "-e", "--end-of-options", sha).Run()
118 if err != nil {
119 if ee, ok := err.(*exec.ExitError); ok && ee.ExitCode() == 1 {
120 return false, nil
121 }
122 return false, fmt.Errorf("cat-file -e %s: %w", sha, err)
123 }
124 // The object exists; --contains lists every ref whose history includes
125 // it, with no namespace restriction — a branch, a tag, or a
126 // refs/merge-requests/<n>/head are equally "a ref reaches this", and
127 // that is the actual question, not whether it happens to be a branch
128 // or a tag. Empty output with no error is the force-push case: present
129 // in the object store, reachable from nothing.
130 out, err := exec.Command(toolpath.Look("git"), "-C", dir, "for-each-ref",
131 "--count=1", "--format=x", "--contains="+sha).Output()
132 if err != nil {
133 return false, fmt.Errorf("for-each-ref --contains %s: %w", sha, err)
134 }
135 return len(out) > 0, nil
136}
137
138// ZeroSHA reports whether s is an all-zero object id (SHA-1 or SHA-256).
139func ZeroSHA(s string) bool {
140 if len(s) != 40 && len(s) != 64 {
141 return false
142 }
143 for i := 0; i < len(s); i++ {
144 if s[i] != '0' {
145 return false
146 }
147 }
148 return true
149}
150
151// RevList returns up to limit commit SHAs reachable from ref, newest first.
152func RevList(dir, ref string, limit int) ([]string, error) {
153 cmd := exec.Command(toolpath.Look("git"), "-C", dir, "rev-list", fmt.Sprintf("--max-count=%d", limit), "--end-of-options", ref)
154 out, err := cmd.Output()
155 if err != nil {
156 return nil, fmt.Errorf("rev-list %s: %w", ref, err)
157 }
158 var shas []string
159 for _, l := range strings.Split(strings.TrimSpace(string(out)), "\n") {
160 if l != "" {
161 shas = append(shas, l)
162 }
163 }
164 return shas, nil
165}
166
167// RevListPath returns up to limit commit SHAs reachable from ref that
168// touch filePath, newest first. The "--" keeps the path from ever being
169// read as an option or ref.
170func RevListPath(dir, ref, filePath string, limit int) ([]string, error) {
171 cmd := exec.Command(toolpath.Look("git"), "-C", dir, "rev-list",
172 fmt.Sprintf("--max-count=%d", limit), "--end-of-options", ref, "--", filePath)
173 out, err := cmd.Output()
174 if err != nil {
175 return nil, fmt.Errorf("rev-list %s -- %s: %w", ref, filePath, err)
176 }
177 var shas []string
178 for _, l := range strings.Split(strings.TrimSpace(string(out)), "\n") {
179 if l != "" {
180 shas = append(shas, l)
181 }
182 }
183 return shas, nil
184}
185
186// PeelToCommit resolves a ref or object to its commit — annotated tags
187// peel to the commit they point at.
188func PeelToCommit(dir, ref string) (string, error) {
189 out, err := exec.Command(toolpath.Look("git"), "-C", dir, "rev-parse", "--verify", "--end-of-options", ref+"^{commit}").Output()
190 if err != nil {
191 return "", fmt.Errorf("rev-parse %s^{commit}: %w", ref, err)
192 }
193 return strings.TrimSpace(string(out)), nil
194}
195
196// ReadCommit returns the raw commit object bytes.
197func ReadCommit(dir, sha string) ([]byte, error) {
198 cmd := exec.Command(toolpath.Look("git"), "-C", dir, "cat-file", "commit", "--end-of-options", sha)
199 out, err := cmd.Output()
200 if err != nil {
201 return nil, fmt.Errorf("cat-file commit %s: %w", sha, err)
202 }
203 return out, nil
204}
205
206// FetchMirror pulls all branches, tags, and notes from a foreign URL into
207// the bare repository at dir, forcing updates. Progress streams to errW so
208// an interactive caller can watch. extraEnv carries credentials via
209// GIT_ASKPASS; the URL itself must never contain them.
210func FetchMirror(ctx context.Context, dir, url string, errW io.Writer, extraEnv []string) error {
211 cmd := exec.CommandContext(ctx, toolpath.Look("git"), "-C", dir, "fetch", "--progress", "--no-write-fetch-head", url,
212 "+refs/heads/*:refs/heads/*",
213 "+refs/tags/*:refs/tags/*",
214 "+refs/notes/*:refs/notes/*")
215 cmd.Env = append(os.Environ(), extraEnv...)
216 cmd.Stderr = errW
217 if err := cmd.Run(); err != nil {
218 return fmt.Errorf("fetch from %s: %w", url, err)
219 }
220 return nil
221}
222
223// FetchPullHeads pulls a GitHub repository's pull-request heads into
224// refs/gh-pull/*, so an imported pull request has something to diff.
225// GitHub publishes every PR head at refs/pull/<n>/head on the git remote,
226// but a mirror made with the default refspecs does not carry them, which
227// is why an import used to produce merge requests with no head at all.
228//
229// One fetch for every pull request rather than one each: the ref count is
230// the repository's history, and asking a hundred times is a hundred
231// handshakes. extraEnv carries credentials via GIT_ASKPASS; the URL must
232// never contain them.
233func FetchPullHeads(ctx context.Context, dir, url string, errW io.Writer, extraEnv []string) error {
234 cmd := exec.CommandContext(ctx, toolpath.Look("git"), "-C", dir, "fetch", "--no-write-fetch-head", "--no-tags",
235 url, "+refs/pull/*/head:refs/gh-pull/*")
236 cmd.Env = append(os.Environ(), extraEnv...)
237 cmd.Stderr = errW
238 if err := cmd.Run(); err != nil {
239 return fmt.Errorf("fetch pull heads from %s: %w", url, err)
240 }
241 return nil
242}
243
244// RemoteDefaultBranch asks the remote which branch HEAD points at.
245func RemoteDefaultBranch(ctx context.Context, url string, extraEnv []string) (string, error) {
246 cmd := exec.CommandContext(ctx, toolpath.Look("git"), "ls-remote", "--symref", url, "HEAD")
247 cmd.Env = append(os.Environ(), extraEnv...)
248 out, err := cmd.Output()
249 if err != nil {
250 return "", fmt.Errorf("ls-remote %s: %w", url, err)
251 }
252 // "ref: refs/heads/<branch>\tHEAD"
253 for _, line := range strings.Split(string(out), "\n") {
254 if rest, ok := strings.CutPrefix(line, "ref: refs/heads/"); ok {
255 if branch, _, ok := strings.Cut(rest, "\t"); ok {
256 return branch, nil
257 }
258 }
259 }
260 return "", fmt.Errorf("remote %s did not advertise a default branch", url)
261}
262
263// SetHead points the bare repo's HEAD at a branch.
264func SetHead(dir, branch string) error {
265 cmd := exec.Command(toolpath.Look("git"), "-C", dir, "symbolic-ref", "HEAD", "refs/heads/"+branch)
266 if out, err := cmd.CombinedOutput(); err != nil {
267 return fmt.Errorf("symbolic-ref: %v\n%s", err, out)
268 }
269 return nil
270}
271
272// gitDefaultDescription is the placeholder git init writes; treated as no
273// description at all.
274const gitDefaultDescription = "Unnamed repository; edit this file 'description' to name the repository."
275
276// ReadDescription returns the repo's description from the classic
277// <repo>.git/description file, empty for the git-init placeholder.
278func ReadDescription(dir string) string {
279 raw, err := os.ReadFile(filepath.Join(dir, "description"))
280 if err != nil {
281 return ""
282 }
283 desc := strings.TrimSpace(string(raw))
284 if desc == gitDefaultDescription {
285 return ""
286 }
287 return desc
288}
289
290// WriteDescription sets the description file: first line only, capped.
291func WriteDescription(dir, desc string) error {
292 desc, _, _ = strings.Cut(strings.TrimSpace(desc), "\n")
293 if len(desc) > 256 {
294 desc = desc[:256]
295 }
296 return os.WriteFile(filepath.Join(dir, "description"), []byte(desc+"\n"), 0o644)
297}
298
299// DirSize sums file sizes under dir; unreadable entries count as zero.
300func DirSize(dir string) int64 {
301 var total int64
302 filepath.WalkDir(dir, func(_ string, d fs.DirEntry, err error) error {
303 if err != nil || d.IsDir() {
304 return nil
305 }
306 if fi, err := d.Info(); err == nil {
307 total += fi.Size()
308 }
309 return nil
310 })
311 return total
312}
313
314// Every git this package runs prints paths as they are, not quoted with
315// octal escapes the way core.quotepath does by default, so a file called
316// übersicht.txt lists, greps, blames and diffs under its own name.
317// GIT_CONFIG_PARAMETERS reaches every subprocess, hooks included,
318// without touching each call site (#129).
319func init() {
320 const q = "'core.quotepath=off'"
321 if cur := os.Getenv("GIT_CONFIG_PARAMETERS"); cur != "" {
322 os.Setenv("GIT_CONFIG_PARAMETERS", cur+" "+q)
323 } else {
324 os.Setenv("GIT_CONFIG_PARAMETERS", q)
325 }
326}