study-bible-project/src/utils/personalSearch.js

// Search across everything the reader has written or saved on this device: study notes, project
// titles and passages, bookmarks and highlights. (Bible text itself is searched in the Reader.)
import { bookOptions, formatChunkReference } from './bibleUtils.jsx';

const NOTE_FIELDS = [
  ['observation', 'Observation'], ['interpretation', 'Interpretation'], ['application', 'Application'],
  ['generalNotes', 'Notes'], ['finalScript', 'Script'],
];

const snippetAround = (raw, at, len) => {
  const start = Math.max(0, at - 40);
  const end = at + len + 60;
  return (start > 0 ? '…' : '') + raw.slice(start, end).replace(/\s+/g, ' ') + (end < raw.length ? '…' : '');
};

const DAY = 86400000;

// "grace in Romans last year" -> { text: 'grace', book: 'Romans', since, until }. Words it understands:
// "in <book>", "last|this week|month|year" (last = the past 7 / 30 / 365 days; this year = since 1 January) and a
// four-digit year. Whatever is left is the text to look for.
export function parseSearchQuery(query, now = Date.now()) {
  let text = ` ${query} `;
  let book = null;
  let since = null;
  let until = null;
  for (const b of [...bookOptions].sort((x, y) => y.name.length - x.name.length)) {
    const re = new RegExp(`\\s(?:in|from)\\s+${b.name.toLowerCase().replace(/\s+/g, '\\s+')}(?=\\s)`, 'i');
    if (re.test(text)) { book = b.name; text = text.replace(re, ' '); break; }
  }
  const rel = text.match(/\s(last|this)\s+(week|month|year)(?=\s)/i);
  if (rel) {
    const [, which, unit] = rel;
    const span = { week: 7, month: 30, year: 365 }[unit.toLowerCase()];
    since = which.toLowerCase() === 'last' ? now - span * DAY : (unit.toLowerCase() === 'year' ? new Date(new Date(now).getFullYear(), 0, 1).getTime() : now - span * DAY);
    text = text.replace(rel[0], ' ');
  }
  const year = text.match(/\s(20\d\d)(?=\s)/);
  if (year) {
    since = new Date(Number(year[1]), 0, 1).getTime();
    until = new Date(Number(year[1]) + 1, 0, 1).getTime();
    text = text.replace(year[0], ' ');
  }
  return { text: text.replace(/\s+/g, ' ').trim(), book, since, until };
}

// projects: full project objects. -> [{ projectId, projectTitle, chunkId, ref, field, snippet }], at most
// `limit`, one hit per chunk (the first field that matches).
export function searchNotes(projects, query, { limit = 20, now = Date.now() } = {}) {
  const parsed = parseSearchQuery(query, now);
  const q = parsed.text.toLowerCase();
  const filtered = !!(parsed.book || parsed.since);
  if (q.length < 3 && !(filtered && q.length === 0)) return [];
  const hits = [];
  for (const project of projects) {
    for (const [chapterIndex, chapter] of (project.chapters ?? []).entries()) {
      if (parsed.book && chapter.book !== parsed.book) continue;
      for (const chunk of chapter.chunks ?? []) {
        const when = chunk.editedAt ?? project.lastEdited ?? 0;
        if (parsed.since && when < parsed.since) continue;
        if (parsed.until && when >= parsed.until) continue;
        for (const [field, label] of NOTE_FIELDS) {
          const raw = chunk[field] ?? '';
          if (!q && !raw.trim()) continue; // a passage/date search lists chunks that have something written
          const at = q ? raw.toLowerCase().indexOf(q) : 0;
          if (at === -1) continue;
          hits.push({
            projectId: project.id, projectTitle: project.title, chunkId: chunk.id,
            ref: formatChunkReference(project, chapterIndex, chunk, '–'), field: label, snippet: snippetAround(raw, at, q.length),
          });
          break;
        }
        if (hits.length >= limit) return hits;
      }
    }
  }
  return hits;
}

// Projects whose title, passages or tags contain the text.
export function searchProjects(index, query, { limit = 6 } = {}) {
  const q = query.trim().toLowerCase();
  if (!q) return [];
  return index.filter((e) => [e.title, e.chapterSummary, ...(e.tags ?? [])].some((t) => (t ?? '').toLowerCase().includes(q))).slice(0, limit);
}

const bookName = (abbrev) => bookOptions.find((b) => b.abbrev === abbrev)?.name ?? abbrev;

// Highlights (by their words or what the colour is named) and bookmarks (by reference).
// -> [{ kind: 'highlight' | 'bookmark', book, bookName, chapter, verse, color, text?, label? }]
export function searchMarks({ bookmarks = {}, highlights = [], labels = {} }, query, { limit = 10 } = {}) {
  const q = query.trim().toLowerCase();
  if (q.length < 2) return [];
  const out = [];
  for (const h of highlights) {
    const label = labels[h.color] ?? '';
    const ref = `${bookName(h.book)} ${h.chapter}:${h.verse}`.toLowerCase();
    if ((h.text ?? '').toLowerCase().includes(q) || label.toLowerCase().includes(q) || ref.includes(q)) {
      out.push({ kind: 'highlight', book: h.book, bookName: bookName(h.book), chapter: h.chapter, verse: h.verse, color: h.color, text: h.text, label });
    }
  }
  for (const [key, color] of Object.entries(bookmarks)) {
    const [book, chapter, verse] = key.split('-');
    const ref = `${bookName(book)} ${chapter}:${verse}`;
    if (ref.toLowerCase().includes(q)) out.push({ kind: 'bookmark', book, bookName: bookName(book), chapter: Number(chapter), verse: Number(verse), color });
  }
  return out.slice(0, limit);
}