| 1 | package agent |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "crypto/sha256" |
| 6 | "encoding/json" |
| 7 | "errors" |
| 8 | "fmt" |
| 9 | "io" |
| 10 | "os" |
| 11 | |
| 12 | "reasonix/internal/provider" |
| 13 | "reasonix/internal/store" |
| 14 | ) |
| 15 | |
| 16 | // MigrationMessageStream is the bounded legacy migration result. Messages is |
| 17 | // the final normalized count after replace records and persisted-safe repairs. |
| 18 | type MigrationMessageStream struct { |
| 19 | Messages int |
| 20 | FromEvents bool |
| 21 | } |
| 22 | |
| 23 | // StreamSessionMessagesForMigration emits the authoritative frozen transcript |
| 24 | // without retaining cumulative schema-1/checkpoint history. reset is called |
| 25 | // before the initial stream and whenever a schema-1 replace record supersedes |
| 26 | // its prefix. Schema-2 DAGs still use their graph-specific compatibility |
| 27 | // reader; their disk-index migration is a separate adapter. |
| 28 | func StreamSessionMessagesForMigration(ctx context.Context, path, headID string, reset func() error, emit func(provider.Message) error) (MigrationMessageStream, error) { |
| 29 | if ctx == nil { |
| 30 | ctx = context.Background() |
| 31 | } |
| 32 | if reset == nil || emit == nil { |
| 33 | return MigrationMessageStream{}, errors.New("session migration stream requires reset and emit callbacks") |
| 34 | } |
| 35 | probe, err := probeSessionEventLogWithLimits(path, migrationSessionReplayLimits()) |
| 36 | if err != nil { |
| 37 | return MigrationMessageStream{}, err |
| 38 | } |
| 39 | if probe.futureSchema { |
| 40 | return MigrationMessageStream{}, fmt.Errorf("session event log for %s uses unsupported schema %d", path, probe.schemaVersion) |
| 41 | } |
| 42 | if probe.dag || headID != "" { |
| 43 | var loaded *Session |
| 44 | if headID == "" { |
| 45 | loaded, err = LoadSessionForMigration(ctx, path) |
| 46 | } else { |
| 47 | loaded, err = LoadSessionHeadForMigration(ctx, path, headID) |
| 48 | } |
| 49 | if err != nil { |
| 50 | return MigrationMessageStream{}, err |
| 51 | } |
| 52 | if err := reset(); err != nil { |
| 53 | return MigrationMessageStream{}, err |
| 54 | } |
| 55 | messages := loaded.Snapshot() |
| 56 | for _, message := range messages { |
| 57 | if err := ctx.Err(); err != nil { |
| 58 | return MigrationMessageStream{}, err |
| 59 | } |
| 60 | if err := emit(message); err != nil { |
| 61 | return MigrationMessageStream{}, err |
| 62 | } |
| 63 | } |
| 64 | return MigrationMessageStream{Messages: len(messages), FromEvents: probe.dag}, nil |
| 65 | } |
| 66 | emitter := newMigrationTurnEmitter(path, reset, emit) |
| 67 | if err := emitter.resetAll(); err != nil { |
| 68 | return MigrationMessageStream{}, err |
| 69 | } |
| 70 | if probe.native && probe.size > 0 { |
| 71 | records, err := streamSchemaOneMigration(ctx, store.SessionEventLog(path), emitter) |
| 72 | if err != nil { |
| 73 | return MigrationMessageStream{FromEvents: true}, err |
| 74 | } |
| 75 | if records > 0 { |
| 76 | if err := emitter.finish(); err != nil { |
| 77 | return MigrationMessageStream{FromEvents: true}, err |
| 78 | } |
| 79 | return MigrationMessageStream{Messages: emitter.count, FromEvents: true}, nil |
| 80 | } |
| 81 | } |
| 82 | if err := emitter.resetAll(); err != nil { |
| 83 | return MigrationMessageStream{}, err |
| 84 | } |
| 85 | file, err := os.Open(path) |
| 86 | if err != nil { |
| 87 | return MigrationMessageStream{}, err |
| 88 | } |
| 89 | defer file.Close() |
| 90 | decoder := json.NewDecoder(&contextReader{ctx: ctx, reader: file}) |
| 91 | for { |
| 92 | var message provider.Message |
| 93 | if err := decoder.Decode(&message); err != nil { |
| 94 | if errors.Is(err, io.EOF) { |
| 95 | break |
| 96 | } |
| 97 | return MigrationMessageStream{}, fmt.Errorf("decode %s: %w", path, err) |
| 98 | } |
| 99 | if err := emitter.add(message); err != nil { |
| 100 | return MigrationMessageStream{}, err |
| 101 | } |
| 102 | } |
| 103 | if err := emitter.finish(); err != nil { |
| 104 | return MigrationMessageStream{}, err |
| 105 | } |
| 106 | return MigrationMessageStream{Messages: emitter.count}, nil |
| 107 | } |
| 108 | |
| 109 | type migrationTurnEmitter struct { |
| 110 | path string |
| 111 | reset func() error |
| 112 | emit func(provider.Message) error |
| 113 | pending []provider.Message |
| 114 | runningHash hashWriter |
| 115 | count int |
| 116 | sourceCount int |
| 117 | } |
| 118 | |
| 119 | type hashWriter interface { |
| 120 | Write([]byte) (int, error) |
| 121 | Sum([]byte) []byte |
| 122 | Reset() |
| 123 | } |
| 124 | |
| 125 | func newMigrationTurnEmitter(path string, reset func() error, emit func(provider.Message) error) *migrationTurnEmitter { |
| 126 | return &migrationTurnEmitter{path: path, reset: reset, emit: emit, runningHash: sha256.New()} |
| 127 | } |
| 128 | |
| 129 | func (e *migrationTurnEmitter) resetAll() error { |
| 130 | e.pending = nil |
| 131 | e.runningHash.Reset() |
| 132 | e.count, e.sourceCount = 0, 0 |
| 133 | return e.reset() |
| 134 | } |
| 135 | |
| 136 | func (e *migrationTurnEmitter) add(message provider.Message) error { |
| 137 | // Persisted-safe normalization never pairs work across a real user turn. |
| 138 | // Buffering one turn preserves receipt/tool repair semantics without |
| 139 | // retaining the cumulative conversation. |
| 140 | if message.Role == provider.RoleUser && !message.LocalOnly && len(e.pending) > 0 { |
| 141 | if err := e.flushTurn(); err != nil { |
| 142 | return err |
| 143 | } |
| 144 | } |
| 145 | e.pending = append(e.pending, message) |
| 146 | e.sourceCount++ |
| 147 | return nil |
| 148 | } |
| 149 | |
| 150 | func (e *migrationTurnEmitter) finish() error { return e.flushTurn() } |
| 151 | |
| 152 | func (e *migrationTurnEmitter) flushTurn() error { |
| 153 | if len(e.pending) == 0 { |
| 154 | return nil |
| 155 | } |
| 156 | messages := migrateLegacyProviderContent(NormalizeSession(e.pending)) |
| 157 | branchID := BranchID(e.path) |
| 158 | for _, message := range messages { |
| 159 | identity, err := json.Marshal(messageForSessionIdentity(message)) |
| 160 | if err != nil { |
| 161 | return err |
| 162 | } |
| 163 | _, _ = e.runningHash.Write(identity) |
| 164 | _, _ = e.runningHash.Write([]byte{'\n'}) |
| 165 | if message.ID == "" { |
| 166 | message.ID = legacyMessageID(branchID, e.count, e.runningHash.Sum(nil)) |
| 167 | } |
| 168 | if err := e.emit(message); err != nil { |
| 169 | return err |
| 170 | } |
| 171 | e.count++ |
| 172 | } |
| 173 | e.pending = nil |
| 174 | return nil |
| 175 | } |
| 176 | |
| 177 | func streamSchemaOneMigration(ctx context.Context, path string, emitter *migrationTurnEmitter) (int, error) { |
| 178 | file, err := os.Open(path) |
| 179 | if err != nil { |
| 180 | return 0, err |
| 181 | } |
| 182 | defer file.Close() |
| 183 | decoder := json.NewDecoder(&contextReader{ctx: ctx, reader: file}) |
| 184 | records := 0 |
| 185 | for { |
| 186 | if err := ctx.Err(); err != nil { |
| 187 | return records, err |
| 188 | } |
| 189 | token, err := decoder.Token() |
| 190 | if errors.Is(err, io.EOF) { |
| 191 | return records, nil |
| 192 | } |
| 193 | if err != nil { |
| 194 | return records, fmt.Errorf("%w: decode schema-1 record: %w", ErrSessionHistoryDamaged, err) |
| 195 | } |
| 196 | if delimiter, ok := token.(json.Delim); !ok || delimiter != '{' { |
| 197 | return records, fmt.Errorf("%w: schema-1 record is not an object", ErrSessionHistoryDamaged) |
| 198 | } |
| 199 | var schema int |
| 200 | var kind string |
| 201 | var messageIndex int |
| 202 | sawMessages := false |
| 203 | for decoder.More() { |
| 204 | nameToken, err := decoder.Token() |
| 205 | if err != nil { |
| 206 | return records, fmt.Errorf("%w: decode schema-1 field: %w", ErrSessionHistoryDamaged, err) |
| 207 | } |
| 208 | name, ok := nameToken.(string) |
| 209 | if !ok { |
| 210 | return records, fmt.Errorf("%w: invalid schema-1 field name", ErrSessionHistoryDamaged) |
| 211 | } |
| 212 | switch name { |
| 213 | case "schema_version": |
| 214 | err = decoder.Decode(&schema) |
| 215 | case "type": |
| 216 | err = decoder.Decode(&kind) |
| 217 | case "message_index": |
| 218 | err = decoder.Decode(&messageIndex) |
| 219 | case "messages": |
| 220 | if schema != sessionEventSchemaVersion || (kind != sessionEventTypeReplace && kind != sessionEventTypeAppend) { |
| 221 | return records, fmt.Errorf("%w: schema/type must precede messages", ErrSessionHistoryDamaged) |
| 222 | } |
| 223 | if kind == sessionEventTypeReplace { |
| 224 | if err := emitter.resetAll(); err != nil { |
| 225 | return records, err |
| 226 | } |
| 227 | } else if messageIndex != emitter.sourceCount { |
| 228 | return records, fmt.Errorf("%w: append index %d does not match %d", ErrSessionHistoryDamaged, messageIndex, emitter.sourceCount) |
| 229 | } |
| 230 | err = decodeMigrationMessageArray(ctx, decoder, emitter) |
| 231 | sawMessages = true |
| 232 | default: |
| 233 | var discard json.RawMessage |
| 234 | err = decoder.Decode(&discard) |
| 235 | } |
| 236 | if err != nil { |
| 237 | return records, fmt.Errorf("%w: decode schema-1 %s: %w", ErrSessionHistoryDamaged, name, err) |
| 238 | } |
| 239 | } |
| 240 | if _, err := decoder.Token(); err != nil { |
| 241 | return records, fmt.Errorf("%w: close schema-1 record: %w", ErrSessionHistoryDamaged, err) |
| 242 | } |
| 243 | if schema != sessionEventSchemaVersion || !sawMessages { |
| 244 | return records, fmt.Errorf("%w: incomplete schema-1 record", ErrSessionHistoryDamaged) |
| 245 | } |
| 246 | records++ |
| 247 | } |
| 248 | } |
| 249 | |
| 250 | func decodeMigrationMessageArray(ctx context.Context, decoder *json.Decoder, emitter *migrationTurnEmitter) error { |
| 251 | token, err := decoder.Token() |
| 252 | if err != nil { |
| 253 | return err |
| 254 | } |
| 255 | if delimiter, ok := token.(json.Delim); !ok || delimiter != '[' { |
| 256 | return errors.New("messages must be an array") |
| 257 | } |
| 258 | for decoder.More() { |
| 259 | if err := ctx.Err(); err != nil { |
| 260 | return err |
| 261 | } |
| 262 | var message provider.Message |
| 263 | if err := decoder.Decode(&message); err != nil { |
| 264 | return err |
| 265 | } |
| 266 | if err := emitter.add(message); err != nil { |
| 267 | return err |
| 268 | } |
| 269 | } |
| 270 | _, err = decoder.Token() |
| 271 | return err |
| 272 | } |
| 273 |