| 1 | package main |
| 2 | |
| 3 | import ( |
| 4 | "crypto/sha256" |
| 5 | "encoding/hex" |
| 6 | "errors" |
| 7 | "fmt" |
| 8 | "io" |
| 9 | "os" |
| 10 | "path/filepath" |
| 11 | "sort" |
| 12 | "strings" |
| 13 | |
| 14 | "reasonix/internal/store" |
| 15 | ) |
| 16 | |
| 17 | // topicMigrationMarker records a completed legacy→topic migration for the |
| 18 | // directory's current session signature. New CLI sessions and same-size |
| 19 | // rewrites change the signature and force re-migration without relying only on |
| 20 | // coarse directory mtimes. |
| 21 | // v2 also re-evaluates recovery-named sessions that v1 skipped by filename. |
| 22 | const topicMigrationMarker = ".topics-migrated-v2" |
| 23 | const topicIndexRepairMarker = ".topic-indexes-repaired-v2" |
| 24 | |
| 25 | const ( |
| 26 | migrationFingerprintWindow = int64(2 << 10) |
| 27 | migrationFullFingerprintLimit = int64(256 << 10) |
| 28 | ) |
| 29 | |
| 30 | func invalidateTopicDirMarkers(dir string) error { |
| 31 | dir = strings.TrimSpace(dir) |
| 32 | if dir == "" { |
| 33 | return nil |
| 34 | } |
| 35 | var errs []error |
| 36 | for _, marker := range []string{topicMigrationMarker, topicIndexRepairMarker} { |
| 37 | if err := os.Remove(filepath.Join(dir, marker)); err != nil && !os.IsNotExist(err) { |
| 38 | errs = append(errs, err) |
| 39 | } |
| 40 | } |
| 41 | return errors.Join(errs...) |
| 42 | } |
| 43 | |
| 44 | func topicDirMarkerDone(dir, marker string) bool { |
| 45 | dir = strings.TrimSpace(dir) |
| 46 | marker = strings.TrimSpace(marker) |
| 47 | if dir == "" || marker == "" { |
| 48 | return false |
| 49 | } |
| 50 | data, err := os.ReadFile(filepath.Join(dir, marker)) |
| 51 | if err != nil { |
| 52 | return false |
| 53 | } |
| 54 | sig, err := sessionDirMigrationSignature(dir) |
| 55 | if err != nil { |
| 56 | // Transient directory read failure: treat as not done so the next |
| 57 | // reconcile retries rather than permanently skipping migration. |
| 58 | return false |
| 59 | } |
| 60 | // Accept both signature content and legacy empty markers that still match |
| 61 | // only when the directory has no session files (empty sig of empty dir). |
| 62 | got := strings.TrimSpace(string(data)) |
| 63 | if got == "" { |
| 64 | // Legacy empty marker: valid only when the dir currently has no |
| 65 | // migratable session/meta files. Any new transcript must re-run. |
| 66 | return sig == emptySessionDirSignature |
| 67 | } |
| 68 | return got == sig |
| 69 | } |
| 70 | |
| 71 | const emptySessionDirSignature = "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" |
| 72 | |
| 73 | func sessionDirMigrationSignature(dir string) (string, error) { |
| 74 | entries, err := os.ReadDir(dir) |
| 75 | if err != nil { |
| 76 | return "", err |
| 77 | } |
| 78 | lines := make([]string, 0, len(entries)) |
| 79 | for _, entry := range entries { |
| 80 | if entry.IsDir() { |
| 81 | continue |
| 82 | } |
| 83 | name := entry.Name() |
| 84 | if !migrationSignatureArtifact(name) { |
| 85 | continue |
| 86 | } |
| 87 | record, err := migrationArtifactSignature(filepath.Join(dir, name), name) |
| 88 | if err != nil { |
| 89 | return "", err |
| 90 | } |
| 91 | lines = append(lines, record) |
| 92 | } |
| 93 | sort.Strings(lines) |
| 94 | sum := sha256.Sum256([]byte(strings.Join(lines, "\n"))) |
| 95 | return hex.EncodeToString(sum[:]), nil |
| 96 | } |
| 97 | |
| 98 | func migrationSignatureArtifact(name string) bool { |
| 99 | return store.IsSessionTranscriptName(name) || |
| 100 | strings.HasSuffix(name, ".events.jsonl") || |
| 101 | strings.HasSuffix(name, ".jsonl.meta") |
| 102 | } |
| 103 | |
| 104 | // migrationArtifactSignature hashes bounded transcript windows so a large |
| 105 | // history is never fully read. Metadata gets a full digest under the size cap; |
| 106 | // mtime, size, prefix, and tail cover ordinary and restored-time rewrites. |
| 107 | func migrationArtifactSignature(path, name string) (string, error) { |
| 108 | f, err := os.Open(path) |
| 109 | if err != nil { |
| 110 | return "", err |
| 111 | } |
| 112 | defer f.Close() |
| 113 | |
| 114 | before, err := f.Stat() |
| 115 | if err != nil { |
| 116 | return "", err |
| 117 | } |
| 118 | if before.IsDir() { |
| 119 | return "", fmt.Errorf("migration signature artifact %q is a directory", path) |
| 120 | } |
| 121 | digest, err := migrationArtifactContentDigest(f, before.Size(), strings.HasSuffix(name, ".jsonl.meta")) |
| 122 | if err != nil { |
| 123 | return "", err |
| 124 | } |
| 125 | after, err := f.Stat() |
| 126 | if err != nil { |
| 127 | return "", err |
| 128 | } |
| 129 | if before.Size() != after.Size() || !before.ModTime().Equal(after.ModTime()) { |
| 130 | return "", fmt.Errorf("migration signature artifact changed while reading: %q", path) |
| 131 | } |
| 132 | return fmt.Sprintf("%q\t%d\t%d\t%s", name, before.Size(), before.ModTime().UnixNano(), digest), nil |
| 133 | } |
| 134 | |
| 135 | func migrationArtifactContentDigest(f *os.File, size int64, preferFull bool) (string, error) { |
| 136 | h := sha256.New() |
| 137 | if size <= migrationFingerprintWindow*2 || (preferFull && size <= migrationFullFingerprintLimit) { |
| 138 | if _, err := f.Seek(0, io.SeekStart); err != nil { |
| 139 | return "", err |
| 140 | } |
| 141 | if _, err := io.CopyN(h, f, size); err != nil { |
| 142 | return "", err |
| 143 | } |
| 144 | return hex.EncodeToString(h.Sum(nil)), nil |
| 145 | } |
| 146 | for _, sample := range []struct { |
| 147 | offset int64 |
| 148 | length int64 |
| 149 | }{ |
| 150 | {offset: 0, length: migrationFingerprintWindow}, |
| 151 | {offset: size - migrationFingerprintWindow, length: migrationFingerprintWindow}, |
| 152 | } { |
| 153 | if _, err := fmt.Fprintf(h, "@%d:%d\n", sample.offset, sample.length); err != nil { |
| 154 | return "", err |
| 155 | } |
| 156 | if _, err := f.Seek(sample.offset, io.SeekStart); err != nil { |
| 157 | return "", err |
| 158 | } |
| 159 | if _, err := io.CopyN(h, f, sample.length); err != nil { |
| 160 | return "", err |
| 161 | } |
| 162 | } |
| 163 | return hex.EncodeToString(h.Sum(nil)), nil |
| 164 | } |
| 165 | |
| 166 | func topicMigrationDone(dir string) bool { |
| 167 | return topicDirMarkerDone(dir, topicMigrationMarker) |
| 168 | } |
| 169 | |
| 170 | func topicIndexRepairDone(dir string) bool { |
| 171 | return topicDirMarkerDone(dir, topicIndexRepairMarker) |
| 172 | } |
| 173 | |
| 174 | func markTopicDirMarkerDone(dir, marker string) { |
| 175 | dir = strings.TrimSpace(dir) |
| 176 | marker = strings.TrimSpace(marker) |
| 177 | if dir == "" || marker == "" { |
| 178 | return |
| 179 | } |
| 180 | if err := os.MkdirAll(dir, 0o755); err != nil { |
| 181 | return |
| 182 | } |
| 183 | sig, err := sessionDirMigrationSignature(dir) |
| 184 | if err != nil { |
| 185 | return |
| 186 | } |
| 187 | _ = os.WriteFile(filepath.Join(dir, marker), []byte(sig+"\n"), 0o644) |
| 188 | } |
| 189 | |
| 190 | func markTopicMigrationDone(dir string) { |
| 191 | markTopicDirMarkerDone(dir, topicMigrationMarker) |
| 192 | } |
| 193 | |
| 194 | func markTopicIndexRepairDone(dir string) { |
| 195 | markTopicDirMarkerDone(dir, topicIndexRepairMarker) |
| 196 | } |
| 197 |