agent_coder
01637fa86b
S1: fetchDetails now builds rank/highlight/matchedFields from the same [...positive, ...required] set matchedTerms uses, so a required-only query (`+кофейня`) no longer reports matchedFields:[] / rank:null. W1: documented that CANDIDATE_CAP bounds pagination reachability (JS-side), not query compute — a SQL LIMIT would corrupt the exact permission-filtered total. S2: one-line note that literal-`%`/`_` search was intentionally dropped (total:0). W3: restored the trgm-index EXPLAIN guard (search-lookup-explain.int-spec) that asserts the coalesce-free substring predicates use idx_pages_*_trgm, not Seq Scan. W4: added an integration ordering test proving title-exact > title-substring > text-only tier dominance under RRF. S3: added a test that an exact-title hit survives a tiny CANDIDATE_CAP window. S5: server pretest now also builds @docmost/mcp so the ToolWriteClass import resolves from clean CI (mirrors editor-ext/prosemirror-markdown). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
764 lines
28 KiB
TypeScript
764 lines
28 KiB
TypeScript
import { Injectable } from '@nestjs/common';
|
|
import { SearchDTO, SearchSuggestionDTO } from './dto/search.dto';
|
|
import {
|
|
SearchResponseDto,
|
|
SearchResultDto,
|
|
} from './dto/search-response.dto';
|
|
import { InjectKysely } from 'nestjs-kysely';
|
|
import { KyselyDB } from '@docmost/db/types/kysely.types';
|
|
import { RawBuilder, sql } from 'kysely';
|
|
import { PageRepo } from '@docmost/db/repos/page/page.repo';
|
|
import { SpaceMemberRepo } from '@docmost/db/repos/space/space-member.repo';
|
|
import { ShareRepo } from '@docmost/db/repos/share/share.repo';
|
|
import { PagePermissionRepo } from '@docmost/db/repos/page/page-permission.repo';
|
|
import {
|
|
ParsedQuery,
|
|
ParsedTerm,
|
|
parseSearchQuery,
|
|
SearchBooleanMode,
|
|
SearchMatchMode,
|
|
hasPositiveRecall,
|
|
} from './search-query-parser';
|
|
|
|
// eslint-disable-next-line @typescript-eslint/no-require-imports
|
|
const tsquery = require('pg-tsquery')();
|
|
|
|
// The FTS text-search configuration used on BOTH the stored side (pages.tsv via
|
|
// its trigger, see migration 20260707T120000) and the query side here. #529
|
|
// acceptance #13 invariant: column config and query config always change as a
|
|
// pair — flip this only alongside the migration.
|
|
const TS_CONFIG = 'ru_en';
|
|
|
|
// RRF constant (Cormack et al. 2009; the Elasticsearch/OpenSearch default). RRF
|
|
// fuses RANKS (not raw scores) so the incomparable ts_rank_cd (lexical) and the
|
|
// substring tier scales never need normalizing — that is the whole point.
|
|
const RRF_K = 60;
|
|
|
|
// Legacy prefix-tsquery builder (kept for the /suggest path + back-compat unit
|
|
// tests). The #529 engine below no longer uses it — it parses the query into an
|
|
// AST instead — but `buildTsQuery` remains exported and behaviour-identical.
|
|
//
|
|
// Strips everything that is not a letter/number/space BEFORE handing text to
|
|
// pg-tsquery so adversarial to_tsquery operators degrade to a neutral query
|
|
// instead of a 500.
|
|
export function buildTsQuery(raw: string): string {
|
|
const cleaned = (raw ?? '')
|
|
.normalize('NFC')
|
|
.replace(/[^\p{L}\p{N}\s]+/gu, ' ')
|
|
.replace(/\s+/g, ' ')
|
|
.trim();
|
|
|
|
if (!cleaned) return '';
|
|
return tsquery(cleaned + '*');
|
|
}
|
|
|
|
// Escape the LIKE metacharacters (`%`, `_`, `\`) so every character is matched
|
|
// LITERALLY by a `col LIKE '%' || q || '%'` predicate. Without this a query of
|
|
// `%` or `_` would match every row.
|
|
//
|
|
// S2 (acceptance #10): escaping is intentionally SYMMETRIC on `%` and `_` — the
|
|
// ability to literal-search a bare `%` (or `_`) as a real character was
|
|
// deliberately dropped. Such a query parses to no positive recall and returns
|
|
// total:0 rather than matching everything. That trade-off is accepted.
|
|
export function escapeLikePattern(raw: string): string {
|
|
return (raw ?? '')
|
|
.replace(/\\/g, '\\\\')
|
|
.replace(/%/g, '\\%')
|
|
.replace(/_/g, '\\_');
|
|
}
|
|
|
|
// Substring tier (highest first), used to rank the substring branch before RRF.
|
|
export enum SearchLookupTier {
|
|
TITLE_EXACT = 3,
|
|
TITLE_SUBSTRING = 2,
|
|
TEXT = 1,
|
|
}
|
|
|
|
// Env-tunable fusion window: the top-N candidates (by RRF) that pagination can
|
|
// reach. The tail beyond it is unreachable (truncatedAtCap:true). Default 500.
|
|
function getCandidateCap(): number {
|
|
const raw = Number(process.env.SEARCH_CANDIDATE_CAP);
|
|
return Number.isFinite(raw) && raw > 0 ? Math.floor(raw) : 500;
|
|
}
|
|
|
|
// Global AND/OR toggle: SEARCH_MODE overrides the request `mode` default. Both
|
|
// are valid on the ru_en tsv — this is a parser toggle, NOT a config rollback.
|
|
function defaultBooleanMode(): SearchBooleanMode {
|
|
return process.env.SEARCH_MODE === 'and' ? 'and' : 'or';
|
|
}
|
|
|
|
@Injectable()
|
|
export class SearchService {
|
|
constructor(
|
|
@InjectKysely() private readonly db: KyselyDB,
|
|
private pageRepo: PageRepo,
|
|
private shareRepo: ShareRepo,
|
|
private spaceMemberRepo: SpaceMemberRepo,
|
|
private pagePermissionRepo: PagePermissionRepo,
|
|
) {}
|
|
|
|
// === #529 SQL fragment builders (parameterized AST, never string-concat) =====
|
|
|
|
// to_tsquery for a single FTS term. The term's own (already metachar-stripped)
|
|
// words are joined with `&` and given a `:*` suffix for the prefix branch. The
|
|
// whole lexeme string is a BOUND parameter to to_tsquery — no user operator can
|
|
// reach the parser.
|
|
private ftsTermTsq(term: ParsedTerm): RawBuilder<unknown> {
|
|
const words = term.text.split(/\s+/).filter(Boolean);
|
|
const suffix = term.branch === 'ftsPrefix' ? ':*' : '';
|
|
const lexeme = words.map((w) => w + suffix).join(' & ');
|
|
return sql`to_tsquery(${TS_CONFIG}, f_unaccent(${lexeme}))`;
|
|
}
|
|
|
|
private phraseTermTsq(term: ParsedTerm): RawBuilder<unknown> {
|
|
return sql`phraseto_tsquery(${TS_CONFIG}, f_unaccent(${term.text}))`;
|
|
}
|
|
|
|
private termTsq(term: ParsedTerm): RawBuilder<unknown> {
|
|
return term.branch === 'phrase'
|
|
? this.phraseTermTsq(term)
|
|
: this.ftsTermTsq(term);
|
|
}
|
|
|
|
// Fold FTS/phrase term tsqueries into ONE tsquery via the SQL `||` operator
|
|
// (AND via `&&` when mode is 'and'). Returns null when there are no such terms.
|
|
private combineTsq(
|
|
terms: ParsedTerm[],
|
|
mode: SearchBooleanMode,
|
|
): RawBuilder<unknown> | null {
|
|
const ftsTerms = terms.filter((t) => t.branch !== 'substring');
|
|
if (ftsTerms.length === 0) return null;
|
|
const op = mode === 'and' ? sql`&&` : sql`||`;
|
|
let acc: RawBuilder<unknown> = sql`(${this.termTsq(ftsTerms[0])})`;
|
|
for (let i = 1; i < ftsTerms.length; i++) {
|
|
acc = sql`(${acc}) ${op} (${this.termTsq(ftsTerms[i])})`;
|
|
}
|
|
return acc;
|
|
}
|
|
|
|
// A LIKE '%needle%' pattern in the LOWER(f_unaccent(...)) space. Coalesce-free
|
|
// so it can use the trigram GIN indexes (a coalesce wrapper would force a seq
|
|
// scan). NULL title/text simply doesn't match, exactly as '' wouldn't.
|
|
private likePattern(term: ParsedTerm): RawBuilder<unknown> {
|
|
const body = '%' + escapeLikePattern(term.text) + '%';
|
|
return sql`LOWER(f_unaccent(${body}))`;
|
|
}
|
|
|
|
private substringPred(
|
|
term: ParsedTerm,
|
|
titleOnly: boolean,
|
|
): RawBuilder<unknown> {
|
|
const pat = this.likePattern(term);
|
|
const title = sql`LOWER(f_unaccent(pages.title)) LIKE ${pat} ESCAPE '\\'`;
|
|
if (titleOnly) return sql`(${title})`;
|
|
const text = sql`LOWER(f_unaccent(pages.text_content)) LIKE ${pat} ESCAPE '\\'`;
|
|
return sql`(${title} OR ${text})`;
|
|
}
|
|
|
|
// Whole-candidate match predicate for a single term (required / excluded
|
|
// predicates, and per-hit matchedTerms).
|
|
private termMatchPred(
|
|
term: ParsedTerm,
|
|
titleOnly: boolean,
|
|
): RawBuilder<unknown> {
|
|
if (term.branch === 'substring') return this.substringPred(term, titleOnly);
|
|
return sql`(pages.tsv @@ ${this.termTsq(term)})`;
|
|
}
|
|
|
|
// The positive RECALL predicate: OR (or AND, per mode) of every positive term's
|
|
// match — the FTS terms via one combined tsquery, the substring terms via LIKE.
|
|
private positiveRecall(
|
|
parsed: ParsedQuery,
|
|
titleOnly: boolean,
|
|
): RawBuilder<unknown> {
|
|
const parts: RawBuilder<unknown>[] = [];
|
|
const tsq = this.combineTsq(parsed.positive, parsed.mode);
|
|
if (tsq) parts.push(sql`(pages.tsv @@ (${tsq}))`);
|
|
for (const t of parsed.positive.filter((x) => x.branch === 'substring')) {
|
|
parts.push(this.substringPred(t, titleOnly));
|
|
}
|
|
if (parts.length === 0) return sql`false`;
|
|
const op = parsed.mode === 'and' ? sql` AND ` : sql` OR `;
|
|
return sql`(${sql.join(parts, op)})`;
|
|
}
|
|
|
|
// The full candidate WHERE: positive recall AND every required AND none of the
|
|
// excluded. Required terms with no positive term ARE the recall (A2).
|
|
private candidatePredicate(
|
|
parsed: ParsedQuery,
|
|
titleOnly: boolean,
|
|
): RawBuilder<unknown> {
|
|
const preds: RawBuilder<unknown>[] = [];
|
|
if (parsed.positive.length > 0) {
|
|
preds.push(this.positiveRecall(parsed, titleOnly));
|
|
}
|
|
for (const t of parsed.required) {
|
|
preds.push(this.termMatchPred(t, titleOnly));
|
|
}
|
|
for (const t of parsed.excluded) {
|
|
preds.push(sql`NOT ${this.termMatchPred(t, titleOnly)}`);
|
|
}
|
|
if (preds.length === 0) return sql`false`;
|
|
return sql`(${sql.join(preds, sql` AND `)})`;
|
|
}
|
|
|
|
// ts_rank_cd over the positive FTS tsquery. NULL when the query has no FTS
|
|
// positive term (pure-substring query) — the row is then absent from the FTS
|
|
// RRF branch.
|
|
private ftsScoreExpr(parsed: ParsedQuery): RawBuilder<unknown> {
|
|
const tsq = this.combineTsq(parsed.positive, parsed.mode);
|
|
if (!tsq) return sql`NULL::float`;
|
|
return sql`CASE WHEN pages.tsv @@ (${tsq}) THEN ts_rank_cd(pages.tsv, (${tsq})) ELSE NULL END`;
|
|
}
|
|
|
|
// Substring tier (3 title-exact / 2 title-substring / 1 text) or NULL when the
|
|
// row matched no positive substring term. Drives the substring RRF branch.
|
|
private subTierExpr(
|
|
parsed: ParsedQuery,
|
|
titleOnly: boolean,
|
|
): RawBuilder<unknown> {
|
|
const subTerms = parsed.positive.filter((t) => t.branch === 'substring');
|
|
if (subTerms.length === 0) return sql`NULL::int`;
|
|
const exact: RawBuilder<unknown>[] = [];
|
|
const titleSub: RawBuilder<unknown>[] = [];
|
|
const textSub: RawBuilder<unknown>[] = [];
|
|
for (const t of subTerms) {
|
|
const needle = sql`LOWER(f_unaccent(${t.text}))`;
|
|
const pat = this.likePattern(t);
|
|
exact.push(sql`LOWER(f_unaccent(pages.title)) = ${needle}`);
|
|
titleSub.push(sql`LOWER(f_unaccent(pages.title)) LIKE ${pat} ESCAPE '\\'`);
|
|
textSub.push(
|
|
sql`LOWER(f_unaccent(pages.text_content)) LIKE ${pat} ESCAPE '\\'`,
|
|
);
|
|
}
|
|
const exactOr = sql`(${sql.join(exact, sql` OR `)})`;
|
|
const titleOr = sql`(${sql.join(titleSub, sql` OR `)})`;
|
|
const textOr = titleOnly
|
|
? sql`false`
|
|
: sql`(${sql.join(textSub, sql` OR `)})`;
|
|
return sql`CASE WHEN ${exactOr} THEN 3 WHEN ${titleOr} THEN 2 WHEN ${textOr} THEN 1 ELSE NULL END`;
|
|
}
|
|
|
|
// === Main entry (A6 unified path) ==========================================
|
|
|
|
async searchPage(
|
|
searchParams: SearchDTO,
|
|
opts: { userId?: string; workspaceId: string },
|
|
): Promise<SearchResponseDto> {
|
|
const rawQuery = (searchParams.query ?? '').trim();
|
|
const mode = defaultBooleanMode();
|
|
const match = (searchParams.match as SearchMatchMode) ?? 'auto';
|
|
const parsed = parseSearchQuery(rawQuery, { match, mode });
|
|
const titleOnly = Boolean(searchParams.titleOnly);
|
|
const limit = Math.min(Math.max(searchParams.limit || 25, 1), 100);
|
|
const offset = Math.max(searchParams.offset || 0, 0);
|
|
const cap = getCandidateCap();
|
|
|
|
const queryMeta = {
|
|
raw: parsed.raw,
|
|
parsed: {
|
|
positive: parsed.positive.map((t) => t.text),
|
|
required: parsed.required.map((t) => t.text),
|
|
excluded: parsed.excluded.map((t) => t.text),
|
|
reason: parsed.reason,
|
|
},
|
|
mode: parsed.mode,
|
|
match,
|
|
};
|
|
|
|
const empty = (reason?: string): SearchResponseDto => ({
|
|
items: [],
|
|
total: 0,
|
|
hasMore: false,
|
|
truncatedAtCap: false,
|
|
offset,
|
|
query: reason
|
|
? { ...queryMeta, parsed: { ...queryMeta.parsed, reason } }
|
|
: queryMeta,
|
|
});
|
|
|
|
// Only-negation / empty query: never run an expensive NOT-only scan (A2).
|
|
if (!hasPositiveRecall(parsed)) {
|
|
return empty(parsed.reason);
|
|
}
|
|
|
|
// --- Resolve scope (A6). --------------------------------------------------
|
|
const scope = await this.resolveScope(searchParams, opts);
|
|
if (scope.kind === 'none') return empty();
|
|
|
|
// --- Optional parentPageId subtree scope. ---------------------------------
|
|
let descendantIds: string[] | null = null;
|
|
if (searchParams.parentPageId) {
|
|
const descendants = await this.pageRepo.getPageAndDescendants(
|
|
searchParams.parentPageId,
|
|
{ includeContent: false },
|
|
);
|
|
descendantIds = descendants.map((p: any) => p.id);
|
|
if (descendantIds.length === 0) return empty();
|
|
}
|
|
|
|
// --- Build the shared candidate WHERE fragment. ---------------------------
|
|
const scopePreds: RawBuilder<unknown>[] = [sql`pages.deleted_at IS NULL`];
|
|
if (scope.kind === 'spaces') {
|
|
scopePreds.push(
|
|
sql`pages.space_id = ANY(${scope.spaceIds}::uuid[])`,
|
|
sql`pages.workspace_id = ${scope.workspaceId}`,
|
|
);
|
|
} else {
|
|
// share: an explicit (already permission-filtered) id set.
|
|
scopePreds.push(
|
|
sql`pages.id = ANY(${scope.pageIds}::uuid[])`,
|
|
sql`pages.workspace_id = ${scope.workspaceId}`,
|
|
);
|
|
}
|
|
if (searchParams.creatorId) {
|
|
scopePreds.push(sql`pages.creator_id = ${searchParams.creatorId}`);
|
|
}
|
|
if (descendantIds) {
|
|
scopePreds.push(sql`pages.id = ANY(${descendantIds}::uuid[])`);
|
|
}
|
|
const scopeSql = sql`(${sql.join(scopePreds, sql` AND `)})`;
|
|
const candidateSql = sql`(${scopeSql} AND ${this.candidatePredicate(parsed, titleOnly)})`;
|
|
|
|
// --- Ranked candidate ids (ALL matches, RRF order). -----------------------
|
|
// Two independent branches ranked SEPARATELY (FTS by ts_rank_cd, substring by
|
|
// tier) then fused with RRF: score = Σ 1/(k + rank_branch). Deterministic
|
|
// ORDER BY rrf DESC, id so pagination never dupes/skips (A4, acceptance #8).
|
|
const rankedRows = await sql<{ id: string }>`
|
|
WITH candidates AS (
|
|
SELECT pages.id AS id,
|
|
${this.ftsScoreExpr(parsed)} AS fts_score,
|
|
${this.subTierExpr(parsed, titleOnly)} AS sub_tier
|
|
FROM pages
|
|
WHERE ${candidateSql}
|
|
),
|
|
ranked AS (
|
|
SELECT id, fts_score, sub_tier,
|
|
row_number() OVER (ORDER BY fts_score DESC NULLS LAST, id) AS rn_fts,
|
|
row_number() OVER (ORDER BY sub_tier DESC NULLS LAST, id) AS rn_sub
|
|
FROM candidates
|
|
)
|
|
SELECT id
|
|
FROM ranked
|
|
ORDER BY
|
|
(CASE WHEN fts_score IS NOT NULL THEN 1.0/(${RRF_K} + rn_fts) ELSE 0 END)
|
|
+ (CASE WHEN sub_tier IS NOT NULL THEN 1.0/(${RRF_K} + rn_sub) ELSE 0 END) DESC,
|
|
id ASC
|
|
`.execute(this.db);
|
|
|
|
let orderedIds = rankedRows.rows.map((r) => r.id);
|
|
|
|
// --- Permission filter (fail-closed, exact total). ------------------------
|
|
// filterAccessiblePageIds runs the #348 hasRestricted pre-check internally:
|
|
// no restricted pages in scope → returns the ids untouched (fast path, total
|
|
// stays trivially exact); otherwise a SQL anti-join drops hidden pages. An
|
|
// error PROPAGATES (500) — it never degrades to an empty lexical result, so
|
|
// hidden-page cardinality never leaks into `total`. The share path is already
|
|
// permission-filtered by getPageAndDescendantsExcludingRestricted.
|
|
if (opts.userId && scope.kind === 'spaces' && orderedIds.length > 0) {
|
|
const accessible = await this.pagePermissionRepo.filterAccessiblePageIds({
|
|
pageIds: orderedIds,
|
|
userId: opts.userId,
|
|
spaceId: searchParams.spaceId,
|
|
workspaceId: opts.workspaceId,
|
|
});
|
|
const accessibleSet = new Set(accessible);
|
|
orderedIds = orderedIds.filter((id) => accessibleSet.has(id));
|
|
}
|
|
|
|
// W1 — CANDIDATE_CAP bounds pagination REACHABILITY (JS-side), NOT query
|
|
// compute. The `ranked` CTE deliberately has no SQL LIMIT: an exact `total`
|
|
// requires counting EVERY match, and the permission anti-join must run over
|
|
// the full match set, so a SQL LIMIT would corrupt `total`. The consequence
|
|
// is that a broad common term in a tenant with restricted pages sorts +
|
|
// anti-joins the whole match set on every request; the cap only truncates the
|
|
// reachable window afterwards (rows past it are unreachable → truncatedAtCap).
|
|
// Accepted for the small-tenant fork target — do NOT convert `cap` into a SQL
|
|
// LIMIT thinking it bounds compute; it does not, and doing so breaks the exact
|
|
// total contract (acceptance #8/#11).
|
|
const total = orderedIds.length;
|
|
const truncatedAtCap = total > cap;
|
|
const window = orderedIds.slice(0, cap);
|
|
const pageIds = window.slice(offset, offset + limit);
|
|
const hasMore = offset + pageIds.length < window.length;
|
|
|
|
if (pageIds.length === 0) {
|
|
return { items: [], total, hasMore, truncatedAtCap, offset, query: queryMeta };
|
|
}
|
|
|
|
// --- Detail fetch for the page slice only (ts_headline/snippet are costly, so
|
|
// compute them ONLY for the returned rows), preserving RRF order. -------
|
|
const items = await this.fetchDetails(pageIds, parsed, titleOnly);
|
|
|
|
return { items, total, hasMore, truncatedAtCap, offset, query: queryMeta };
|
|
}
|
|
|
|
// Resolve the search scope: explicit space, the authed user's member spaces, or
|
|
// a public share's descendant id set (permission-filtered). Mirrors the legacy
|
|
// branch logic (A6) but returns a small tagged union.
|
|
private async resolveScope(
|
|
searchParams: SearchDTO,
|
|
opts: { userId?: string; workspaceId: string },
|
|
): Promise<
|
|
| { kind: 'none' }
|
|
| { kind: 'spaces'; spaceIds: string[]; workspaceId: string }
|
|
| { kind: 'ids'; pageIds: string[]; workspaceId: string }
|
|
> {
|
|
if (searchParams.spaceId) {
|
|
return {
|
|
kind: 'spaces',
|
|
spaceIds: [searchParams.spaceId],
|
|
workspaceId: opts.workspaceId,
|
|
};
|
|
}
|
|
|
|
if (opts.userId && !searchParams.shareId) {
|
|
const spaceIds = await this.spaceMemberRepo.getUserSpaceIds(opts.userId);
|
|
if (spaceIds.length === 0) return { kind: 'none' };
|
|
return { kind: 'spaces', spaceIds, workspaceId: opts.workspaceId };
|
|
}
|
|
|
|
if (searchParams.shareId && !opts.userId) {
|
|
const share = await this.shareRepo.findById(searchParams.shareId);
|
|
if (!share || share.workspaceId !== opts.workspaceId) {
|
|
return { kind: 'none' };
|
|
}
|
|
const isRestricted = await this.pagePermissionRepo.hasRestrictedAncestor(
|
|
share.pageId,
|
|
);
|
|
if (isRestricted) return { kind: 'none' };
|
|
|
|
const pageIds: string[] = [];
|
|
if (share.includeSubPages) {
|
|
// A6: exclude restricted descendants + honour a restricted ancestor.
|
|
const pageList =
|
|
await this.pageRepo.getPageAndDescendantsExcludingRestricted(
|
|
share.pageId,
|
|
{ includeContent: false },
|
|
);
|
|
pageIds.push(...pageList.map((p) => p.id));
|
|
} else {
|
|
pageIds.push(share.pageId);
|
|
}
|
|
if (pageIds.length === 0) return { kind: 'none' };
|
|
return { kind: 'ids', pageIds, workspaceId: opts.workspaceId };
|
|
}
|
|
|
|
return { kind: 'none' };
|
|
}
|
|
|
|
// Fetch the display superset (A7) for the given page ids, in the given order.
|
|
private async fetchDetails(
|
|
pageIds: string[],
|
|
parsed: ParsedQuery,
|
|
titleOnly: boolean,
|
|
): Promise<SearchResultDto[]> {
|
|
// S1: rank/highlight/matchedFields must reflect required terms too, not just
|
|
// bare positives — a required-only query like `+кофейня` (no bare positive)
|
|
// really matches, so it must not report matchedFields:[] / rank:null. Build
|
|
// every detail expression from the SAME [...positive, ...required] set the
|
|
// matchedTerms path uses (`detailTerms` below). Excluded terms stay out (they
|
|
// are anti-matches). The candidate WHERE already enforced required-AND, so
|
|
// OR-combining them here is purely for display scoring/highlighting.
|
|
const detailTerms = [...parsed.positive, ...parsed.required];
|
|
const tsq = this.combineTsq(detailTerms, parsed.mode);
|
|
// rank/highlight are FTS-only (null for substring-only hits — the web already
|
|
// falls back). ts_headline runs over the ORIGINAL text.
|
|
const rankExpr = tsq
|
|
? sql<number>`CASE WHEN pages.tsv @@ (${tsq}) THEN ts_rank_cd(pages.tsv, (${tsq})) ELSE NULL END`
|
|
: sql<number>`NULL::float`;
|
|
const highlightExpr = tsq
|
|
? sql<string>`CASE WHEN pages.tsv @@ (${tsq}) THEN ts_headline(${TS_CONFIG}, coalesce(pages.text_content, ''), (${tsq}), 'MinWords=9, MaxWords=10, MaxFragments=3') ELSE NULL END`
|
|
: sql<string>`NULL::text`;
|
|
|
|
// Windowed plain snippet: leading window, else the FTS headline (plain).
|
|
const snippetExpr = titleOnly
|
|
? sql<string>`''`
|
|
: sql<string>`coalesce(
|
|
case
|
|
when coalesce(pages.text_content, '') <> ''
|
|
then substring(LOWER(f_unaccent(pages.text_content)) from 1 for 300)
|
|
${
|
|
tsq
|
|
? sql`else ts_headline(${TS_CONFIG}, coalesce(pages.text_content, ''), (${tsq}), 'MinWords=25, MaxWords=40, MaxFragments=3')`
|
|
: sql``
|
|
}
|
|
end, '')`;
|
|
|
|
// matchedFields: title / text.
|
|
const titleMatchParts: RawBuilder<unknown>[] = [];
|
|
const textMatchParts: RawBuilder<unknown>[] = [];
|
|
if (tsq) {
|
|
titleMatchParts.push(
|
|
sql`to_tsvector(${TS_CONFIG}, f_unaccent(coalesce(pages.title, ''))) @@ (${tsq})`,
|
|
);
|
|
if (!titleOnly) {
|
|
textMatchParts.push(
|
|
sql`(pages.tsv @@ (${tsq}) AND NOT to_tsvector(${TS_CONFIG}, f_unaccent(coalesce(pages.title, ''))) @@ (${tsq}))`,
|
|
);
|
|
}
|
|
}
|
|
for (const t of detailTerms.filter((x) => x.branch === 'substring')) {
|
|
const pat = this.likePattern(t);
|
|
titleMatchParts.push(
|
|
sql`LOWER(f_unaccent(pages.title)) LIKE ${pat} ESCAPE '\\'`,
|
|
);
|
|
if (!titleOnly) {
|
|
textMatchParts.push(
|
|
sql`LOWER(f_unaccent(pages.text_content)) LIKE ${pat} ESCAPE '\\'`,
|
|
);
|
|
}
|
|
}
|
|
const titleMatchExpr = titleMatchParts.length
|
|
? sql<boolean>`(${sql.join(titleMatchParts, sql` OR `)})`
|
|
: sql<boolean>`false`;
|
|
const textMatchExpr = textMatchParts.length
|
|
? sql<boolean>`(${sql.join(textMatchParts, sql` OR `)})`
|
|
: sql<boolean>`false`;
|
|
|
|
// Per-term matched flags (positive + required), assembled into matchedTerms.
|
|
// Same set as `detailTerms` above. NB: aliases must be UNDERSCORE-FREE — the
|
|
// app's Kysely runs the CamelCasePlugin, which would rewrite a `t_0` result
|
|
// key to `t0`. `tmatch0` survives the round-trip unchanged.
|
|
const allTerms = detailTerms;
|
|
const termCols = allTerms.map(
|
|
(t, i) =>
|
|
sql`(${this.termMatchPred(t, titleOnly)}) AS "tmatch${sql.raw(String(i))}"`,
|
|
);
|
|
|
|
let q = this.db
|
|
.selectFrom('pages')
|
|
.select([
|
|
'pages.id as id',
|
|
'pages.slugId as slugId',
|
|
'pages.title as title',
|
|
'pages.icon as icon',
|
|
'pages.parentPageId as parentPageId',
|
|
'pages.creatorId as creatorId',
|
|
'pages.spaceId as spaceId',
|
|
'pages.createdAt as createdAt',
|
|
'pages.updatedAt as updatedAt',
|
|
rankExpr.as('rank'),
|
|
highlightExpr.as('highlight'),
|
|
snippetExpr.as('snippet'),
|
|
titleMatchExpr.as('titleMatch'),
|
|
textMatchExpr.as('textMatch'),
|
|
])
|
|
.select((eb) => this.pageRepo.withSpace(eb))
|
|
.where(sql<boolean>`pages.id = ANY(${pageIds}::uuid[])` as any);
|
|
|
|
if (termCols.length > 0) {
|
|
q = q.select(termCols as any);
|
|
}
|
|
|
|
const rows: any[] = await q.execute();
|
|
|
|
const byId = new Map<string, any>(rows.map((r) => [r.id, r]));
|
|
const pathById = await this.buildAncestorPaths(pageIds);
|
|
|
|
const items: SearchResultDto[] = [];
|
|
for (const id of pageIds) {
|
|
const r = byId.get(id);
|
|
if (!r) continue;
|
|
const matchedFields: string[] = [];
|
|
if (r.titleMatch) matchedFields.push('title');
|
|
if (r.textMatch) matchedFields.push('text');
|
|
const matchedTerms: string[] = [];
|
|
allTerms.forEach((t, i) => {
|
|
if (r[`tmatch${i}`]) matchedTerms.push(t.text);
|
|
});
|
|
const rank = r.rank == null ? null : Number(r.rank);
|
|
const highlight = r.highlight
|
|
? String(r.highlight)
|
|
.replace(/\r\n|\r|\n/g, ' ')
|
|
.replace(/\s+/g, ' ')
|
|
: null;
|
|
const snippet = (r.snippet ?? '')
|
|
.replace(/\r\n|\r|\n/g, ' ')
|
|
.replace(/\s+/g, ' ')
|
|
.trim();
|
|
items.push({
|
|
id: r.id,
|
|
pageId: r.id,
|
|
slugId: r.slugId,
|
|
icon: r.icon,
|
|
title: r.title,
|
|
space: r.space,
|
|
creatorId: r.creatorId,
|
|
createdAt: r.createdAt,
|
|
updatedAt: r.updatedAt,
|
|
rank,
|
|
highlight,
|
|
snippet,
|
|
path: pathById.get(id) ?? [],
|
|
// A per-response ordering proxy for legacy consumers; falls back to rank.
|
|
score: rank ?? 0,
|
|
matchedFields,
|
|
matchedTerms: Array.from(new Set(matchedTerms)),
|
|
});
|
|
}
|
|
return items;
|
|
}
|
|
|
|
/**
|
|
* Batch ancestor-titles helper: ONE recursive CTE seeded with ALL hit ids,
|
|
* walking UP parentPageId. Returns hitId → ancestor titles ordered root →
|
|
* direct parent (the hit's own title excluded).
|
|
*
|
|
* A8 fix: the recursive walk now skips soft-deleted ancestors (`deleted_at IS
|
|
* NULL`) and stays within the hit's own space, so a deleted or cross-space
|
|
* ancestor's title can no longer leak into `path`.
|
|
*/
|
|
private async buildAncestorPaths(
|
|
hitIds: string[],
|
|
): Promise<Map<string, string[]>> {
|
|
const result = new Map<string, string[]>();
|
|
if (hitIds.length === 0) return result;
|
|
|
|
const rows = await this.db
|
|
.withRecursive('ancestry', (db) =>
|
|
db
|
|
.selectFrom('pages')
|
|
.select([
|
|
'pages.id as hitId',
|
|
'pages.id as pageId',
|
|
'pages.title as title',
|
|
'pages.parentPageId as parentPageId',
|
|
'pages.spaceId as spaceId',
|
|
sql<number>`0`.as('depth'),
|
|
])
|
|
.where('pages.id', 'in', hitIds)
|
|
.where('pages.deletedAt', 'is', null)
|
|
.unionAll((exp) =>
|
|
exp
|
|
.selectFrom('pages as p')
|
|
.innerJoin('ancestry as a', 'p.id', 'a.parentPageId')
|
|
// A8: don't cross into a deleted ancestor or another space.
|
|
.where('p.deletedAt', 'is', null)
|
|
.whereRef('p.spaceId', '=', 'a.spaceId')
|
|
.select([
|
|
'a.hitId as hitId',
|
|
'p.id as pageId',
|
|
'p.title as title',
|
|
'p.parentPageId as parentPageId',
|
|
'a.spaceId as spaceId',
|
|
sql<number>`a.depth + 1`.as('depth'),
|
|
]),
|
|
),
|
|
)
|
|
.selectFrom('ancestry')
|
|
.select(['hitId', 'title', 'depth'])
|
|
.where('depth', '>', 0)
|
|
.orderBy('hitId')
|
|
.orderBy('depth', 'desc')
|
|
.execute();
|
|
|
|
for (const r of rows as any[]) {
|
|
const list = result.get(r.hitId) ?? [];
|
|
list.push(r.title);
|
|
result.set(r.hitId, list);
|
|
}
|
|
return result;
|
|
}
|
|
|
|
async searchSuggestions(
|
|
suggestion: SearchSuggestionDTO,
|
|
userId: string,
|
|
workspaceId: string,
|
|
) {
|
|
let users = [];
|
|
let groups = [];
|
|
let pages = [];
|
|
|
|
const limit = suggestion?.limit || 10;
|
|
const query = suggestion.query.toLowerCase().trim();
|
|
|
|
if (suggestion.includeUsers) {
|
|
const userQuery = this.db
|
|
.selectFrom('users')
|
|
.select(['id', 'name', 'email', 'avatarUrl'])
|
|
.where('workspaceId', '=', workspaceId)
|
|
.where('deletedAt', 'is', null)
|
|
.where((eb) =>
|
|
eb.or([
|
|
eb(
|
|
sql`LOWER(f_unaccent(users.name))`,
|
|
'like',
|
|
sql`LOWER(f_unaccent(${`%${query}%`}))`,
|
|
),
|
|
eb(sql`users.email`, 'ilike', sql`f_unaccent(${`%${query}%`})`),
|
|
]),
|
|
)
|
|
.limit(limit);
|
|
|
|
users = await userQuery.execute();
|
|
}
|
|
|
|
if (suggestion.includeGroups) {
|
|
groups = await this.db
|
|
.selectFrom('groups')
|
|
.select(['id', 'name', 'description'])
|
|
.where((eb) =>
|
|
eb(
|
|
sql`LOWER(f_unaccent(groups.name))`,
|
|
'like',
|
|
sql`LOWER(f_unaccent(${`%${query}%`}))`,
|
|
),
|
|
)
|
|
.where('workspaceId', '=', workspaceId)
|
|
.limit(limit)
|
|
.execute();
|
|
}
|
|
|
|
if (suggestion.includePages) {
|
|
let pageSearch = this.db
|
|
.selectFrom('pages')
|
|
.select(['id', 'slugId', 'title', 'icon', 'spaceId'])
|
|
.select((eb) => this.pageRepo.withSpace(eb))
|
|
.where((eb) =>
|
|
eb(
|
|
sql`LOWER(f_unaccent(pages.title))`,
|
|
'like',
|
|
sql`LOWER(f_unaccent(${`%${query}%`}))`,
|
|
),
|
|
)
|
|
.where('deletedAt', 'is', null)
|
|
.where('workspaceId', '=', workspaceId)
|
|
.limit(limit);
|
|
|
|
if (suggestion.onlyTemplates) {
|
|
pageSearch = pageSearch.where('isTemplate', '=', true);
|
|
}
|
|
|
|
const userSpaceIds = await this.spaceMemberRepo.getUserSpaceIds(userId);
|
|
|
|
if (userSpaceIds?.length > 0) {
|
|
pageSearch = pageSearch.where('spaceId', 'in', userSpaceIds);
|
|
|
|
if (suggestion?.spaceId) {
|
|
pageSearch = pageSearch.orderBy(
|
|
sql`CASE WHEN pages."space_id" = ${suggestion.spaceId} THEN 0 ELSE 1 END`,
|
|
'asc',
|
|
);
|
|
}
|
|
|
|
pages = await pageSearch.execute();
|
|
}
|
|
|
|
if (pages.length > 0) {
|
|
const pageIds = pages.map((p) => p.id);
|
|
const accessibleIds =
|
|
await this.pagePermissionRepo.filterAccessiblePageIds({
|
|
pageIds,
|
|
userId,
|
|
workspaceId,
|
|
});
|
|
const accessibleSet = new Set(accessibleIds);
|
|
pages = pages.filter((p) => accessibleSet.has(p.id));
|
|
}
|
|
}
|
|
|
|
return { users, groups, pages };
|
|
}
|
|
}
|