Serge Bazanski | 4915425 | 2020-10-24 20:22:08 +0200 | [diff] [blame] | 1 | package main |
| 2 | |
| 3 | // tarify implements a minimal, self-contained, hermetic tarball builder. |
| 4 | // It is currently used with gostatic to take a non-hermetic directory and |
| 5 | // turn it into a hermetic tarball via a glob. |
| 6 | // |
| 7 | // For more information about tree artifacts and hermeticity, see: |
| 8 | // https://jmmv.dev/2019/12/bazel-dynamic-execution-tree-artifacts.html |
| 9 | |
| 10 | import ( |
| 11 | "archive/tar" |
| 12 | "flag" |
| 13 | "fmt" |
| 14 | "io" |
| 15 | "os" |
| 16 | "path/filepath" |
| 17 | "sort" |
| 18 | "strings" |
| 19 | |
| 20 | "github.com/golang/glog" |
| 21 | ) |
| 22 | |
| 23 | var ( |
| 24 | flagSite string |
| 25 | flagTarball string |
| 26 | ) |
| 27 | |
| 28 | func init() { |
| 29 | flag.Set("logtostderr", "true") |
| 30 | } |
| 31 | |
| 32 | func main() { |
| 33 | flag.StringVar(&flagSite, "site", "", "Site sources") |
| 34 | flag.StringVar(&flagTarball, "tarball", "", "Output tarball") |
| 35 | flag.Parse() |
| 36 | |
| 37 | if flagSite == "" { |
| 38 | glog.Exitf("-site must be set") |
| 39 | } |
| 40 | if flagTarball == "" { |
| 41 | glog.Exitf("-tarball must be set") |
| 42 | } |
| 43 | |
| 44 | f, err := os.Create(flagTarball) |
| 45 | if err != nil { |
| 46 | glog.Exitf("Create(%q): %v", flagTarball, err) |
| 47 | } |
| 48 | defer f.Close() |
| 49 | w := tar.NewWriter(f) |
| 50 | defer w.Close() |
| 51 | |
| 52 | flagSite = strings.TrimSuffix(flagSite, "/") |
| 53 | |
| 54 | // First retrieve all files and sort. This is required for idempotency. |
| 55 | elems := []struct { |
| 56 | path string |
| 57 | info os.FileInfo |
| 58 | }{} |
| 59 | err = filepath.Walk(flagSite, func(inPath string, _ os.FileInfo, err error) error { |
| 60 | // We don't use the given fileinfo, as we want to deref symlinks. |
| 61 | info, err := os.Stat(inPath) |
| 62 | if err != nil { |
| 63 | return fmt.Errorf("Stat: %w", err) |
| 64 | } |
| 65 | elems = append(elems, struct { |
| 66 | path string |
| 67 | info os.FileInfo |
| 68 | }{inPath, info}) |
| 69 | return nil |
| 70 | }) |
| 71 | if err != nil { |
| 72 | glog.Exitf("Walk(%q, _): %v", flagSite, err) |
| 73 | } |
| 74 | sort.Slice(elems, func(i, j int) bool { return elems[i].path < elems[j].path }) |
| 75 | |
| 76 | // Now that we have a sorted list, tar 'em up. |
| 77 | for _, elem := range elems { |
| 78 | inPath := elem.path |
| 79 | info := elem.info |
| 80 | |
| 81 | outPath := strings.TrimPrefix(strings.TrimPrefix(inPath, flagSite), "/") |
| 82 | if outPath == "" { |
| 83 | continue |
| 84 | } |
| 85 | if info.IsDir() { |
| 86 | glog.Infof("D %s", outPath) |
| 87 | if err := w.WriteHeader(&tar.Header{ |
| 88 | Typeflag: tar.TypeDir, |
| 89 | Name: outPath, |
| 90 | Mode: 0755, |
| 91 | }); err != nil { |
| 92 | glog.Exitf("Writing directory header for %q failed: %v", inPath, err) |
| 93 | } |
| 94 | } else { |
| 95 | glog.Infof("F %s", outPath) |
| 96 | if err := w.WriteHeader(&tar.Header{ |
| 97 | Typeflag: tar.TypeReg, |
| 98 | Name: outPath, |
| 99 | Mode: 0644, |
| 100 | // TODO(q3k): this can race (TOCTOU Stat/Open, resulting in "archive/tar: write Too long") |
| 101 | // No idea, how to handle this better though without reading the entire file into memory, |
| 102 | // or trying to do filesystem locks? Besides, in practical use with Bazel this will never |
| 103 | // happen. |
| 104 | Size: info.Size(), |
| 105 | }); err != nil { |
| 106 | glog.Exitf("Writing file header for %q failed: %v", inPath, err) |
| 107 | } |
| 108 | r, err := os.Open(inPath) |
| 109 | if err != nil { |
| 110 | glog.Exitf("Open(%q): %v", inPath, err) |
| 111 | } |
| 112 | defer r.Close() |
| 113 | if _, err := io.Copy(w, r); err != nil { |
| 114 | glog.Exitf("Copy(%q): %v", inPath, err) |
| 115 | } |
| 116 | |
| 117 | } |
| 118 | } |
| 119 | } |