| 1 | package session |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "context" |
| 6 | "database/sql" |
| 7 | "encoding/json" |
| 8 | "errors" |
| 9 | "os" |
| 10 | "path/filepath" |
| 11 | "strings" |
| 12 | "unicode/utf8" |
| 13 | |
| 14 | "reasonix/internal/projectiondb" |
| 15 | "reasonix/internal/sessioncontent" |
| 16 | ) |
| 17 | |
| 18 | // Per-field message body reading (history-window-v1 companion): a client |
| 19 | // requests one field of one message by stable identity — never a file path — |
| 20 | // and receives a bounded, UTF-8 aligned fragment with the full length and the |
| 21 | // next range offset. Fragments never split inside a JSON string escape |
| 22 | // sequence, so concatenated ranges re-parse as the original value. |
| 23 | const ( |
| 24 | // MessageFieldMaxChunk bounds one response fragment: the plan's detail |
| 25 | // reading budget of at most 256 KiB per read. |
| 26 | MessageFieldMaxChunk = 256 << 10 |
| 27 | // messageFieldMaxBody bounds the source message body a field read will |
| 28 | // decode; bodies above it must go through the streaming export path. |
| 29 | messageFieldMaxBody = sessioncontent.MaxReadRange |
| 30 | ) |
| 31 | |
| 32 | // MessageFieldPage is one bounded fragment of one message field. |
| 33 | type MessageFieldPage struct { |
| 34 | Status string `json:"status"` // ready | not_found | preparing | unsupported |
| 35 | MessageID string `json:"messageId"` |
| 36 | Version int `json:"version"` |
| 37 | Field string `json:"field"` |
| 38 | TotalBytes int64 `json:"totalBytes"` |
| 39 | Offset int64 `json:"offset"` |
| 40 | Data []byte `json:"data,omitempty"` |
| 41 | NextOffset int64 `json:"nextOffset,omitempty"` |
| 42 | Encoding string `json:"encoding"` |
| 43 | } |
| 44 | |
| 45 | // ReadMessageField returns [offset, offset+length) of one top-level field of |
| 46 | // the canonical message JSON. The field "canonicalMessage" addresses the |
| 47 | // whole stored body. Version 0 selects the message's latest version inside |
| 48 | // the current durable snapshot. Range credentials come from a prior window |
| 49 | // or page that displayed the message; a read without one is rejected. |
| 50 | func (q *Query) ReadMessageField(ctx context.Context, ref SessionRef, messageID string, version int, field string, offset, length int64) (MessageFieldPage, error) { |
| 51 | if q == nil { |
| 52 | return MessageFieldPage{}, errors.New("session: nil query") |
| 53 | } |
| 54 | if err := ref.validate(q.hostID); err != nil { |
| 55 | return MessageFieldPage{}, err |
| 56 | } |
| 57 | messageID = strings.TrimSpace(messageID) |
| 58 | if messageID == "" || field == "" { |
| 59 | return MessageFieldPage{}, errors.New("session: message id and field are required") |
| 60 | } |
| 61 | if offset < 0 { |
| 62 | return MessageFieldPage{}, errors.New("session: invalid field offset") |
| 63 | } |
| 64 | if length <= 0 { |
| 65 | length = MessageFieldMaxChunk |
| 66 | } |
| 67 | length = min(length, MessageFieldMaxChunk) |
| 68 | |
| 69 | filesystem, ok := q.persistence.(*FilesystemPersistence) |
| 70 | if !ok { |
| 71 | return MessageFieldPage{}, errors.New("session: field reads require filesystem persistence") |
| 72 | } |
| 73 | path := historyIndexPath(filesystem.Root, ref.SessionID) |
| 74 | if _, err := os.Stat(path); errors.Is(err, os.ErrNotExist) { |
| 75 | preparation := q.prepareHistoryLocator(filesystem, ref.SessionID, path) |
| 76 | select { |
| 77 | case <-preparation.done: |
| 78 | if preparation.err != nil { |
| 79 | return MessageFieldPage{Status: "not_found"}, preparation.err |
| 80 | } |
| 81 | default: |
| 82 | return MessageFieldPage{Status: "preparing", MessageID: messageID, Field: field}, nil |
| 83 | } |
| 84 | } |
| 85 | lock := q.projectionLock("history", ref.SessionID) |
| 86 | lock.Lock() |
| 87 | err := ensureHistoryIndex(ctx, filesystem, ref.SessionID, path) |
| 88 | lock.Unlock() |
| 89 | if err != nil { |
| 90 | return MessageFieldPage{}, err |
| 91 | } |
| 92 | handle, err := projectiondb.Open(ctx, projectiondb.OpenOptions{Path: path, Migrations: historyMigrations, RequireDisk: true, MaxOpenConns: 1}) |
| 93 | if err != nil { |
| 94 | return MessageFieldPage{}, err |
| 95 | } |
| 96 | defer handle.DB.Close() |
| 97 | metadata, err := readHistoryIndexMetadata(ctx, handle.DB) |
| 98 | if err != nil { |
| 99 | return MessageFieldPage{}, err |
| 100 | } |
| 101 | snapshot := metadata.durableSequence |
| 102 | |
| 103 | page := MessageFieldPage{Status: "ready", MessageID: messageID, Version: version, Field: field, Offset: offset, Encoding: "utf-8"} |
| 104 | query := `SELECT inline,content_digest,content_bytes,content_index_digest FROM messages WHERE message_id=? AND event_sequence<=? AND (valid_to=0 OR valid_to>?)` |
| 105 | args := []any{messageID, snapshot, snapshot} |
| 106 | if version > 0 { |
| 107 | query += ` AND version=?` |
| 108 | args = append(args, version) |
| 109 | } |
| 110 | query += ` ORDER BY version DESC LIMIT 1` |
| 111 | var inline []byte |
| 112 | var digest, indexDigest string |
| 113 | var contentBytes int64 |
| 114 | err = handle.DB.QueryRowContext(ctx, query, args...).Scan(&inline, &digest, &contentBytes, &indexDigest) |
| 115 | if errors.Is(err, sql.ErrNoRows) { |
| 116 | page.Status = "not_found" |
| 117 | return page, nil |
| 118 | } |
| 119 | if err != nil { |
| 120 | return MessageFieldPage{}, err |
| 121 | } |
| 122 | var body []byte |
| 123 | if digest != "" { |
| 124 | contentRef := sessioncontent.Ref{Digest: digest, Bytes: contentBytes, IndexDigest: indexDigest, IntegrityBlock: sessioncontent.IntegrityBlockBytes, MediaType: "application/json"} |
| 125 | if !q.contentAuthorized(ref.SessionID, digest, contentBytes, indexDigest) { |
| 126 | return MessageFieldPage{}, errors.New("session: content reference is not authorized for this session") |
| 127 | } |
| 128 | if contentBytes > messageFieldMaxBody { |
| 129 | return MessageFieldPage{}, errors.New("session: message body exceeds the field-read budget; use the streaming export path") |
| 130 | } |
| 131 | body, err = contentStoreForSessionDir(filepath.Join(filesystem.Root, ref.SessionID)).ReadRange(ctx, contentRef, 0, contentBytes) |
| 132 | if err != nil { |
| 133 | return MessageFieldPage{}, err |
| 134 | } |
| 135 | } else { |
| 136 | body = inline |
| 137 | } |
| 138 | |
| 139 | fieldBytes, err := extractMessageField(body, field) |
| 140 | if errors.Is(err, errMessageFieldAbsent) { |
| 141 | // A message that simply does not carry the field is a finished empty |
| 142 | // read, not a failure: optional fields like reasoning are routinely |
| 143 | // absent, and a hard error would surface as a broken body. |
| 144 | return page, nil |
| 145 | } |
| 146 | if err != nil { |
| 147 | return MessageFieldPage{}, err |
| 148 | } |
| 149 | page.TotalBytes = int64(len(fieldBytes)) |
| 150 | if offset >= page.TotalBytes { |
| 151 | return page, nil |
| 152 | } |
| 153 | end := min(offset+length, page.TotalBytes) |
| 154 | // Never split a UTF-8 rune: align the fragment end back to a rune start. |
| 155 | for end > offset && end < page.TotalBytes && !utf8.RuneStart(fieldBytes[end]) { |
| 156 | end-- |
| 157 | } |
| 158 | // Never split a JSON escape sequence: a cut between a backslash and the |
| 159 | // character it escapes would corrupt concatenation. A cut between two |
| 160 | // complete escapes inside a string is fine. |
| 161 | if endsInsideEscape(fieldBytes, offset, end) { |
| 162 | end-- |
| 163 | } |
| 164 | page.Data = append([]byte(nil), fieldBytes[offset:end]...) |
| 165 | page.Offset = offset |
| 166 | if end < page.TotalBytes { |
| 167 | page.NextOffset = end |
| 168 | } |
| 169 | return page, nil |
| 170 | } |
| 171 | |
| 172 | // extractMessageField streams the canonical message JSON and captures the raw |
| 173 | // bytes of one top-level field value. Only the field itself is materialized — |
| 174 | // never the whole message. |
| 175 | func extractMessageField(body []byte, field string) ([]byte, error) { |
| 176 | if field == "canonicalMessage" { |
| 177 | return body, nil |
| 178 | } |
| 179 | decoder := json.NewDecoder(bytes.NewReader(body)) |
| 180 | open, err := decoder.Token() |
| 181 | if err != nil { |
| 182 | return nil, err |
| 183 | } |
| 184 | if delim, ok := open.(json.Delim); !ok || delim != '{' { |
| 185 | return nil, errors.New("session: canonical message is not a JSON object") |
| 186 | } |
| 187 | for decoder.More() { |
| 188 | keyToken, err := decoder.Token() |
| 189 | if err != nil { |
| 190 | return nil, err |
| 191 | } |
| 192 | key, ok := keyToken.(string) |
| 193 | if !ok { |
| 194 | return nil, errors.New("session: canonical message has a non-string key") |
| 195 | } |
| 196 | var raw json.RawMessage |
| 197 | if err := decoder.Decode(&raw); err != nil { |
| 198 | return nil, err |
| 199 | } |
| 200 | if key == field { |
| 201 | return raw, nil |
| 202 | } |
| 203 | } |
| 204 | return nil, errMessageFieldAbsent |
| 205 | } |
| 206 | |
| 207 | // errMessageFieldAbsent separates "this message has no such field" from a |
| 208 | // genuinely unreadable body: the caller answers the former with an empty |
| 209 | // finished read and the latter with an error. |
| 210 | var errMessageFieldAbsent = errors.New("session: message has no such field") |
| 211 | |
| 212 | // endsInsideEscape reports whether a fragment ending at end would split a |
| 213 | // JSON escape sequence: an odd run of backslashes immediately before the cut |
| 214 | // means the last one dangles. begin is the fragment start, where a |
| 215 | // well-formed previous fragment guaranteed no dangling escape. |
| 216 | func endsInsideEscape(data []byte, begin, end int64) bool { |
| 217 | if end <= begin { |
| 218 | return false |
| 219 | } |
| 220 | backslashes := int64(0) |
| 221 | for i := end - 1; i >= begin; i-- { |
| 222 | if data[i] != '\\' { |
| 223 | break |
| 224 | } |
| 225 | backslashes++ |
| 226 | } |
| 227 | return backslashes%2 == 1 |
| 228 | } |
| 229 |