| 1 | package agent |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "database/sql" |
| 6 | "encoding/json" |
| 7 | "errors" |
| 8 | "io" |
| 9 | "os" |
| 10 | "strings" |
| 11 | |
| 12 | "reasonix/internal/historywork" |
| 13 | "reasonix/internal/provider" |
| 14 | ) |
| 15 | |
| 16 | // Each committed point is either a complete record or a complete element of |
| 17 | // its messages array. The staged rows and parser state commit together, even |
| 18 | // when a legacy writer puts type/schema fields after the messages array. |
| 19 | type eventPagerScanProgress struct { |
| 20 | Version int `json:"version"` |
| 21 | SourceKey string `json:"sourceKey"` |
| 22 | Offset int64 `json:"offset"` |
| 23 | Count int `json:"count"` |
| 24 | Pending int `json:"pending"` |
| 25 | InMessages bool `json:"inMessages"` |
| 26 | Done bool `json:"done"` |
| 27 | Record sessionEventRecord `json:"record"` |
| 28 | } |
| 29 | |
| 30 | type eventPagerScanner struct { |
| 31 | ctx context.Context |
| 32 | db *sql.DB |
| 33 | tx *sql.Tx |
| 34 | decoder *json.Decoder |
| 35 | base int64 |
| 36 | progress eventPagerScanProgress |
| 37 | resuming bool |
| 38 | observed func(string, int) |
| 39 | records int |
| 40 | } |
| 41 | |
| 42 | func (s *eventPagerScanner) commit(inMessages, done bool) error { |
| 43 | s.progress.Offset = s.base + s.decoder.InputOffset() |
| 44 | s.progress.InMessages, s.progress.Done = inMessages, done |
| 45 | body, err := json.Marshal(s.progress) |
| 46 | if err == nil { |
| 47 | _, err = s.tx.ExecContext(s.ctx, `INSERT OR REPLACE INTO metadata VALUES('event_scan_progress',?)`, string(body)) |
| 48 | } |
| 49 | if err != nil { |
| 50 | return err |
| 51 | } |
| 52 | if err := s.tx.Commit(); err != nil { |
| 53 | return err |
| 54 | } |
| 55 | s.tx = nil |
| 56 | s.records = 0 |
| 57 | if s.observed != nil { |
| 58 | if inMessages { |
| 59 | s.observed("messages", s.progress.Pending) |
| 60 | } else { |
| 61 | s.observed("records", s.progress.Count) |
| 62 | } |
| 63 | } |
| 64 | return s.ctx.Err() |
| 65 | } |
| 66 | |
| 67 | func (s *eventPagerScanner) transaction() error { |
| 68 | if s.tx != nil { |
| 69 | return nil |
| 70 | } |
| 71 | var err error |
| 72 | s.tx, err = s.db.BeginTx(s.ctx, nil) |
| 73 | return err |
| 74 | } |
| 75 | |
| 76 | func (s *eventPagerScanner) messages() error { |
| 77 | if !s.resuming { |
| 78 | s.progress.Pending = 0 |
| 79 | if _, err := s.tx.ExecContext(s.ctx, `DELETE FROM event_pending`); err != nil { |
| 80 | return err |
| 81 | } |
| 82 | } |
| 83 | token, err := s.decoder.Token() |
| 84 | if err != nil || token == nil { |
| 85 | return err |
| 86 | } |
| 87 | if token != json.Delim('[') { |
| 88 | return ErrSessionDisplayReadModelDamaged |
| 89 | } |
| 90 | if s.resuming { |
| 91 | // Consume the synthetic prior element; the following separator and |
| 92 | // remainder are validated by encoding/json with their original grammar. |
| 93 | var previous any |
| 94 | if err := s.decoder.Decode(&previous); err != nil { |
| 95 | return err |
| 96 | } |
| 97 | s.resuming = false |
| 98 | } |
| 99 | for s.decoder.More() { |
| 100 | if err := s.transaction(); err != nil { |
| 101 | return err |
| 102 | } |
| 103 | start := s.base + s.decoder.InputOffset() |
| 104 | var message provider.Message |
| 105 | if err := s.decoder.Decode(&message); err != nil { |
| 106 | return err |
| 107 | } |
| 108 | length := s.base + s.decoder.InputOffset() - start |
| 109 | if _, err := s.tx.ExecContext(s.ctx, `INSERT INTO event_pending VALUES(?,?,?)`, s.progress.Pending, start, length); err != nil { |
| 110 | return err |
| 111 | } |
| 112 | s.progress.Pending++ |
| 113 | if s.progress.Pending%historywork.BatchEntries == 0 { |
| 114 | if err := s.commit(true, false); err != nil { |
| 115 | return err |
| 116 | } |
| 117 | } |
| 118 | } |
| 119 | if _, err := s.decoder.Token(); err != nil { |
| 120 | return err |
| 121 | } |
| 122 | return s.transaction() |
| 123 | } |
| 124 | |
| 125 | func (s *eventPagerScanner) fields() error { |
| 126 | for s.decoder.More() { |
| 127 | if err := s.ctx.Err(); err != nil { |
| 128 | return err |
| 129 | } |
| 130 | field, err := s.decoder.Token() |
| 131 | if err != nil { |
| 132 | return err |
| 133 | } |
| 134 | switch field { |
| 135 | case "schema_version": |
| 136 | err = s.decoder.Decode(&s.progress.Record.SchemaVersion) |
| 137 | case "type": |
| 138 | err = s.decoder.Decode(&s.progress.Record.Type) |
| 139 | case "message_index": |
| 140 | err = s.decoder.Decode(&s.progress.Record.MessageIndex) |
| 141 | case "created_at": |
| 142 | err = s.decoder.Decode(&s.progress.Record.CreatedAt) |
| 143 | case "messages": |
| 144 | err = s.messages() |
| 145 | default: |
| 146 | err = skipDisplayJSONValue(s.ctx, s.decoder) |
| 147 | } |
| 148 | if err != nil { |
| 149 | return err |
| 150 | } |
| 151 | } |
| 152 | _, err := s.decoder.Token() |
| 153 | return err |
| 154 | } |
| 155 | |
| 156 | func (s *eventPagerScanner) applyRecord() error { |
| 157 | record := s.progress.Record |
| 158 | if record.SchemaVersion != sessionEventSchemaVersion { |
| 159 | return ErrSessionDisplayReadModelDamaged |
| 160 | } |
| 161 | at := int64(0) |
| 162 | switch record.Type { |
| 163 | case sessionEventTypeReplace: |
| 164 | if _, err := s.tx.ExecContext(s.ctx, `DELETE FROM event_locations`); err != nil { |
| 165 | return err |
| 166 | } |
| 167 | s.progress.Count = 0 |
| 168 | case sessionEventTypeAppend: |
| 169 | if record.MessageIndex != s.progress.Count { |
| 170 | return ErrSessionDisplayReadModelDamaged |
| 171 | } |
| 172 | if !record.CreatedAt.IsZero() { |
| 173 | at = record.CreatedAt.UnixMilli() |
| 174 | } |
| 175 | default: |
| 176 | return ErrSessionDisplayReadModelDamaged |
| 177 | } |
| 178 | if _, err := s.tx.ExecContext(s.ctx, `INSERT INTO event_locations SELECT position+?,offset,length,? FROM event_pending`, s.progress.Count, at); err != nil { |
| 179 | return err |
| 180 | } |
| 181 | s.progress.Count += s.progress.Pending |
| 182 | s.progress.Pending, s.progress.Record = 0, sessionEventRecord{} |
| 183 | _, err := s.tx.ExecContext(s.ctx, `DELETE FROM event_pending`) |
| 184 | return err |
| 185 | } |
| 186 | |
| 187 | func (s *eventPagerScanner) scan(file *os.File) (result error) { |
| 188 | if s.progress.Done { |
| 189 | return nil |
| 190 | } |
| 191 | if _, err := file.Seek(s.progress.Offset, io.SeekStart); err != nil { |
| 192 | return err |
| 193 | } |
| 194 | var reader io.Reader = &historywork.Reader{Context: s.ctx, Source: file} |
| 195 | s.base = s.progress.Offset |
| 196 | s.resuming = s.progress.InMessages |
| 197 | if s.resuming { |
| 198 | const framing = `{"messages":[null` |
| 199 | reader = io.MultiReader(strings.NewReader(framing), reader) |
| 200 | s.base -= int64(len(framing)) |
| 201 | } |
| 202 | s.decoder = json.NewDecoder(reader) |
| 203 | defer func() { |
| 204 | if s.tx != nil { |
| 205 | _ = s.tx.Rollback() |
| 206 | } |
| 207 | }() |
| 208 | for { |
| 209 | if err := s.transaction(); err != nil { |
| 210 | return err |
| 211 | } |
| 212 | token, err := s.decoder.Token() |
| 213 | if errors.Is(err, io.EOF) { |
| 214 | return s.commit(false, true) |
| 215 | } |
| 216 | if err != nil || token != json.Delim('{') { |
| 217 | return errors.Join(ErrSessionDisplayReadModelDamaged, err) |
| 218 | } |
| 219 | if err := s.fields(); err != nil { |
| 220 | return errors.Join(ErrSessionDisplayReadModelDamaged, err) |
| 221 | } |
| 222 | if err := s.applyRecord(); err != nil { |
| 223 | return err |
| 224 | } |
| 225 | s.records++ |
| 226 | if s.records >= historywork.BatchEntries { |
| 227 | if err := s.commit(false, false); err != nil { |
| 228 | return err |
| 229 | } |
| 230 | } |
| 231 | } |
| 232 | } |
| 233 |