| 1 | // Message search over the local store (#586) — the web twin of the native |
| 2 | // MessageStore.search (Mac header / iOS thread list). |
| 3 | // |
| 4 | // The store holds CIPHERTEXT rows (store.js), so searching means decrypting — |
| 5 | // done lazily into an in-MEMORY index (never written anywhere): each thread |
| 6 | // decrypts once per session, then queries are instant scans. At this |
| 7 | // product's scale a substring scan is plenty — no FTS engine in the browser, |
| 8 | // matching the issue's call. Decrypt is injected, so the index is Node-testable. |
| 9 | // |
| 10 | // Memory: the index holds each ensured thread's decrypted text for the |
| 11 | // session — roughly the size of the conversation itself (tens of MB only at |
| 12 | // the far tail of huge, fully-backfilled spaces). Deliberately uncapped: an |
| 13 | // eviction scheme would silently narrow search coverage; the tab's lifetime |
| 14 | // bounds it. Revisit if a real space ever makes this hurt — the agreed |
| 15 | // upgrade (a persisted, encrypted-at-rest term index; triggers, shape, and |
| 16 | // invariants) is docs/sealed-search-index.md. |
| 17 | |
| 18 | // Pure: does `text` match the query? Case-insensitive; every whitespace- |
| 19 | // separated term must appear (AND) — the familiar chat-search behaviour. |
| 20 | export function matchText(text, query) { |
| 21 | if (!text || !query) return false; |
| 22 | const hay = text.toLowerCase(); |
| 23 | const terms = query.toLowerCase().split(/\s+/).filter(Boolean); |
| 24 | return terms.length > 0 && terms.every((t) => hay.includes(t)); |
| 25 | } |
| 26 | |
| 27 | // Pure: a readable window around the first hit, ellipsised at cut edges. |
| 28 | export function snippetFor(text, query, len = 90) { |
| 29 | const t = (text || "").replace(/\s+/g, " ").trim(); |
| 30 | if (t.length <= len) return t; |
| 31 | const first = (query || "").toLowerCase().split(/\s+/).filter(Boolean)[0] || ""; |
| 32 | const at = first ? t.toLowerCase().indexOf(first) : 0; |
| 33 | const start = Math.max(0, Math.min(at - Math.floor(len / 3), t.length - len)); |
| 34 | const cut = t.slice(start, start + len); |
| 35 | return (start > 0 ? "…" : "") + cut + (start + len < t.length ? "…" : ""); |
| 36 | } |
| 37 | |
| 38 | // A session-scoped decrypted-text index. Rows arrive as raw store rows |
| 39 | // ({id, created_at, blob, tombstone_of}); decryptRow(row, tid) resolves |
| 40 | // { text, name, ts } or null (undecryptable / typed app event / reaction — |
| 41 | // not searchable prose). Incremental per thread by a seen-id SET, not a |
| 42 | // high-water mark: rows arrive out of ID order by design — live arrivals are |
| 43 | // newer, but a backfill walk (#457) delivers ever-OLDER pages, and both must |
| 44 | // index exactly once. |
| 45 | export class SearchIndex { |
| 46 | constructor(decryptRow) { |
| 47 | this.decryptRow = decryptRow; |
| 48 | this.threads = new Map(); // tid → { seen: Set<id>, entries: [{id, ts, text, name}] } |
| 49 | } |
| 50 | |
| 51 | async ensureThread(tid, rows) { |
| 52 | let t = this.threads.get(tid); |
| 53 | if (!t) { |
| 54 | t = { seen: new Set(), entries: [] }; |
| 55 | this.threads.set(tid, t); |
| 56 | } |
| 57 | for (const row of rows) { |
| 58 | if (!row || !row.id || t.seen.has(row.id)) continue; |
| 59 | t.seen.add(row.id); |
| 60 | if (row.tombstone_of) { |
| 61 | // The unsent message must stop matching — drop it from the index too. |
| 62 | const gone = row.tombstone_of; |
| 63 | t.entries = t.entries.filter((e) => e.id !== gone); |
| 64 | continue; |
| 65 | } |
| 66 | const d = await this.decryptRow(row, tid); |
| 67 | if (d && d.text) t.entries.push({ id: row.id, ts: d.ts || row.created_at * 1000 || 0, text: d.text, name: d.name || "" }); |
| 68 | } |
| 69 | return t; |
| 70 | } |
| 71 | |
| 72 | // Forget one message (a live delete frame) or a whole thread (left it). |
| 73 | dropMessage(tid, id) { |
| 74 | const t = this.threads.get(tid); |
| 75 | if (t) t.entries = t.entries.filter((e) => e.id !== id); |
| 76 | } |
| 77 | dropThread(tid) { |
| 78 | this.threads.delete(tid); |
| 79 | } |
| 80 | |
| 81 | // Newest-first matches across every ensured thread — native parity |
| 82 | // (MessageStore.search: instance-wide, limit 50). |
| 83 | query(q, limit = 50) { |
| 84 | const out = []; |
| 85 | for (const [tid, t] of this.threads) { |
| 86 | for (const e of t.entries) { |
| 87 | if (matchText(e.text, q)) out.push({ tid, ...e }); |
| 88 | } |
| 89 | } |
| 90 | out.sort((a, b) => b.id - a.id); |
| 91 | return out.slice(0, limit); |
| 92 | } |
| 93 | } |
| 94 | |
amadan