app/lib/search/query.mjs

Functions

Imports

It imports no other module of this repo. Imported by app/lib/search/search.server.ts, app/routes/search.tsx.

Source

1// Rules apply IN ORDER and each consumes its tokens: tag:2019 stays a tag filter only because the
2// operator rule runs before the year rule.
3
4export { RRF_K, fuse, toMatchExpression } from "@drdustinedwards/site-helpers";
5
6/**
7 * @typedef {object} ParsedQuery
8 * @property {string} raw exactly what was typed, for echoing back
9 * @property {string[]} terms bare words, ANDed
10 * @property {string[]} phrases quoted runs, matched as phrases
11 * @property {string[]} tags from `tag:` operators and the ?tag= parameter
12 * @property {string[]} types from `type:` operators and the ?type= parameter
13 * @property {string[]} bases from `base:` operators and the ?base= parameter: a knowledge base's id (app/kb/search-bases.mjs)
14 * @property {number | null} year a bare four-digit year inside the corpus range
15 * @property {boolean} isEmpty true when there is nothing to match on
16 */
17
18/**
19 * @typedef {object} ParseOptions
20 * @property {number} [minYear] earliest year the corpus contains
21 * @property {number} [maxYear] latest year the corpus contains
22 */
23
24/** Narrow enough that 1024 or 8080 stays a search term rather than silently filtering everything. */
25const DEFAULT_MIN_YEAR = 2000;
26
27/**
28 * @param {string} input
29 * @param {ParseOptions} [options]
30 * @returns {ParsedQuery}
31 */
32export function parseQuery(input, options = {}) {
33  const raw = input ?? "";
34  const minYear = options.minYear ?? DEFAULT_MIN_YEAR;
35  // Not computed at module scope: that would freeze the ceiling at deploy time
36  // and stop accepting next year's posts after New Year.
37  const maxYear = options.maxYear ?? new Date().getUTCFullYear() + 1;
38
39  let rest = raw;
40
41  // 1. Quoted phrases first, so an operator inside quotes stays literal text.
42  /** @type {string[]} */
43  const phrases = [];
44  rest = rest.replace(/"([^"]+)"/g, (/** @type {string} */ _m, /** @type {string} */ phrase) => {
45    const trimmed = phrase.trim();
46    if (trimmed) phrases.push(trimmed);
47    return " ";
48  });
49
50  /** @type {string[]} */
51  const tags = [];
52  /** @type {string[]} */
53  const types = [];
54  /** @type {string[]} */
55  const bases = [];
56  rest = rest.replace(
57    /\b(tag|type|base):([^\s]+)/gi,
58    (/** @type {string} */ _m, /** @type {string} */ field, /** @type {string} */ value) => {
59      const clean = value.trim().toLowerCase();
60      if (clean) {
61        const f = field.toLowerCase();
62        if (f === "tag") tags.push(clean);
63        else if (f === "base") bases.push(clean);
64        else types.push(clean);
65      }
66      return " ";
67    },
68  );
69
70  // 3. A bare year becomes a date filter: prose rarely spells out its own date, so as text it matches nothing.
71  /** @type {number | null} */
72  let year = null;
73  const tokens = rest.split(/\s+/).filter(Boolean);
74  /** @type {string[]} */
75  const terms = [];
76  for (const token of tokens) {
77    if (year === null && /^\d{4}$/.test(token)) {
78      const candidate = Number(token);
79      if (candidate >= minYear && candidate <= maxYear) {
80        year = candidate;
81        continue;
82      }
83    }
84    terms.push(token);
85  }
86
87  return {
88    raw,
89    terms,
90    phrases,
91    tags,
92    types,
93    bases,
94    year,
95    isEmpty: terms.length === 0 && phrases.length === 0,
96  };
97}
98
99/**
100 * Filter-only queries (tag:x, a bare year) leave no text for fts5, so they take the browse path.
101 *
102 * @param {ParsedQuery} parsed
103 * @returns {boolean}
104 */
105export function hasFilters(parsed) {
106  return parsed.tags.length > 0 || parsed.types.length > 0 || parsed.bases.length > 0 || parsed.year !== null;
107}