docorbit 0.1.0 → 0.1.2

package/dist/packages/storage/src/repositories/chunk-repository.js

@@ -0,0 +1,250 @@
+export class ChunkRepository {
+    db;
+    constructor(db) {
+        this.db = db;
+    }
+    /**
+     * Saves chunks, code snippets, symbols, and relationships in a single atomic transaction.
+     * Transactionally synchronizes the FTS5 chunks_fts table with chunks.
+     */
+    saveChunks(chunks, relationships = [], codeSnippets = [], symbols = []) {
+        if (chunks.length === 0)
+            return;
+        const raw = this.db.getRawDb();
+        const createdAt = new Date().toISOString();
+        // Group symbols by chunkId for fast indexing into chunks_fts
+        const symbolsByChunk = new Map();
+        for (const s of symbols) {
+            const list = symbolsByChunk.get(s.chunkId) || [];
+            list.push(s.name);
+            symbolsByChunk.set(s.chunkId, list);
+        }
+        raw.exec('BEGIN IMMEDIATE TRANSACTION;');
+        try {
+            const chunkStmt = raw.prepare(`
+        INSERT INTO chunks (
+          id, page_id, snapshot_id, title, section_path_json, content,
+          chunk_type, language, token_estimate, ordinal, content_hash,
+          doc_version, provenance_json, created_at
+        ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+        ON CONFLICT(id) DO UPDATE SET
+          title = excluded.title,
+          section_path_json = excluded.section_path_json,
+          content = excluded.content,
+          chunk_type = excluded.chunk_type,
+          language = excluded.language,
+          token_estimate = excluded.token_estimate,
+          ordinal = excluded.ordinal,
+          content_hash = excluded.content_hash,
+          doc_version = excluded.doc_version,
+          provenance_json = excluded.provenance_json
+      `);
+            const ftsDelStmt = this.db.isFtsAvailable()
+                ? raw.prepare('DELETE FROM chunks_fts WHERE chunk_id = ?')
+                : null;
+            const ftsInsStmt = this.db.isFtsAvailable()
+                ? raw.prepare(`
+            INSERT INTO chunks_fts (chunk_id, title, section_path, content, symbols)
+            VALUES (?, ?, ?, ?, ?)
+          `)
+                : null;
+            for (const chunk of chunks) {
+                const sectionPath = Array.isArray(chunk.sectionPath)
+                    ? chunk.sectionPath
+                    : (Array.isArray(chunk.headingPath) ? chunk.headingPath : []);
+                const chunkType = chunk.chunkType || chunk.type || 'prose';
+                const tokenEstimate = chunk.tokenEstimate ?? chunk.tokenCount ?? 0;
+                const ordinal = chunk.ordinal ?? chunk.orderIndex ?? 0;
+                const provenance = chunk.provenance ? JSON.stringify(chunk.provenance) : null;
+                chunkStmt.run(chunk.id, chunk.pageId || null, chunk.snapshotId || 'default', chunk.title || null, JSON.stringify(sectionPath), chunk.content, chunkType, chunk.language || null, tokenEstimate, ordinal, chunk.contentHash, chunk.docVersion || null, provenance, createdAt);
+                if (ftsDelStmt && ftsInsStmt) {
+                    ftsDelStmt.run(chunk.id);
+                    const chunkSymbols = symbolsByChunk.get(chunk.id)?.join(' ') || '';
+                    const sectionPathStr = sectionPath.join(' > ');
+                    ftsInsStmt.run(chunk.id, chunk.title || '', sectionPathStr, chunk.content, chunkSymbols);
+                }
+            }
+            // Save code snippets
+            if (codeSnippets.length > 0) {
+                const codeStmt = raw.prepare(`
+          INSERT OR REPLACE INTO chunk_code (id, chunk_id, language, code)
+          VALUES (?, ?, ?, ?)
+        `);
+                for (const cs of codeSnippets) {
+                    codeStmt.run(cs.id, cs.chunkId, cs.language || null, cs.code);
+                }
+            }
+            // Save symbols
+            if (symbols.length > 0) {
+                const symStmt = raw.prepare(`
+          INSERT OR REPLACE INTO symbol_references (id, chunk_id, name, kind)
+          VALUES (?, ?, ?, ?)
+        `);
+                for (const s of symbols) {
+                    symStmt.run(s.id, s.chunkId, s.name, s.kind);
+                }
+            }
+            // Save relationships
+            if (relationships.length > 0) {
+                const relStmt = raw.prepare(`
+          INSERT OR REPLACE INTO chunk_relationships (source_chunk_id, target_chunk_id, type)
+          VALUES (?, ?, ?)
+        `);
+                for (const r of relationships) {
+                    relStmt.run(r.sourceChunkId, r.targetChunkId, r.type);
+                }
+            }
+            raw.exec('COMMIT;');
+        }
+        catch (err) {
+            raw.exec('ROLLBACK;');
+            throw err;
+        }
+    }
+    getChunk(id) {
+        const raw = this.db.getRawDb();
+        const row = raw.prepare('SELECT * FROM chunks WHERE id = ?').get(id);
+        if (!row)
+            return null;
+        return this.hydrateChunk(row);
+    }
+    getChunksByPage(pageId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT * FROM chunks WHERE page_id = ? ORDER BY ordinal ASC').all(pageId);
+        return rows.map(r => this.hydrateChunk(r));
+    }
+    getChunksBySnapshot(snapshotId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT * FROM chunks WHERE snapshot_id = ? ORDER BY page_id ASC, ordinal ASC').all(snapshotId);
+        return rows.map(r => this.hydrateChunk(r));
+    }
+    getChunksByVersion(docVersion) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT * FROM chunks WHERE doc_version = ? ORDER BY page_id ASC, ordinal ASC').all(docVersion);
+        return rows.map(r => this.hydrateChunk(r));
+    }
+    getChunkCode(chunkId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT id, chunk_id, language, code FROM chunk_code WHERE chunk_id = ?').all(chunkId);
+        return rows.map(r => ({
+            id: String(r.id),
+            chunkId: String(r.chunk_id),
+            language: r.language ? String(r.language) : undefined,
+            code: String(r.code),
+        }));
+    }
+    getChunkSymbols(chunkId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT id, chunk_id, name, kind FROM symbol_references WHERE chunk_id = ?').all(chunkId);
+        return rows.map(r => ({
+            id: String(r.id),
+            chunkId: String(r.chunk_id),
+            name: String(r.name),
+            kind: r.kind,
+        }));
+    }
+    getChunkRelationships(chunkId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare(`
+      SELECT source_chunk_id, target_chunk_id, type
+      FROM chunk_relationships
+      WHERE source_chunk_id = ? OR target_chunk_id = ?
+    `).all(chunkId, chunkId);
+        return rows.map(r => ({
+            sourceChunkId: String(r.source_chunk_id),
+            targetChunkId: String(r.target_chunk_id),
+            type: r.type,
+        }));
+    }
+    countChunks(snapshotId) {
+        const raw = this.db.getRawDb();
+        if (snapshotId) {
+            const row = raw.prepare('SELECT COUNT(*) as count FROM chunks WHERE snapshot_id = ?').get(snapshotId);
+            return row?.count || 0;
+        }
+        const row = raw.prepare('SELECT COUNT(*) as count FROM chunks').get();
+        return row?.count || 0;
+    }
+    searchChunksFts(query, options = {}) {
+        if (!this.db.isFtsAvailable()) {
+            return [];
+        }
+        const raw = this.db.getRawDb();
+        const limit = options.limit || 20;
+        // Sanitize query for FTS5: strip dangerous punctuation, quote bare words or phrases
+        const cleanTokens = query
+            .replace(/[^\w\s-]/g, ' ')
+            .trim()
+            .split(/\s+/)
+            .filter(t => t.length > 0);
+        if (cleanTokens.length === 0) {
+            return [];
+        }
+        // Use OR-joined quoted tokens or phrase search
+        const sanitizedFtsQuery = cleanTokens.map(t => `"${t.replace(/"/g, '""')}"`).join(' OR ');
+        try {
+            let sql = `
+        SELECT c.*, fts.rank as fts_rank
+        FROM chunks c
+        JOIN chunks_fts fts ON c.id = fts.chunk_id
+        WHERE chunks_fts MATCH ?
+      `;
+            const params = [sanitizedFtsQuery];
+            if (options.snapshotId) {
+                sql += ' AND c.snapshot_id = ?';
+                params.push(options.snapshotId);
+            }
+            if (options.chunkType) {
+                sql += ' AND c.chunk_type = ?';
+                params.push(options.chunkType);
+            }
+            if (options.docVersion) {
+                sql += ' AND c.doc_version = ?';
+                params.push(options.docVersion);
+            }
+            sql += ' ORDER BY fts.rank LIMIT ?';
+            params.push(limit);
+            const rows = raw.prepare(sql).all(...params);
+            return rows.map(r => {
+                const chunk = this.hydrateChunk(r);
+                return {
+                    chunk,
+                    ftsRank: Number(r.fts_rank),
+                    symbols: this.getChunkSymbols(chunk.id),
+                    codeSnippets: this.getChunkCode(chunk.id),
+                };
+            });
+        }
+        catch {
+            return [];
+        }
+    }
+    hydrateChunk(row) {
+        const sectionPath = row.section_path_json ? JSON.parse(String(row.section_path_json)) : [];
+        const provenance = row.provenance_json
+            ? JSON.parse(String(row.provenance_json))
+            : {
+                sourceUrl: '',
+                targetUrl: '',
+                fetchedAt: String(row.created_at || ''),
+                discoveredBy: 'direct',
+                contentHash: String(row.content_hash || ''),
+                snapshotId: String(row.snapshot_id || ''),
+            };
+        return {
+            id: String(row.id),
+            pageId: String(row.page_id),
+            snapshotId: String(row.snapshot_id),
+            title: row.title ? String(row.title) : undefined,
+            sectionPath,
+            content: String(row.content),
+            chunkType: row.chunk_type,
+            language: row.language ? String(row.language) : undefined,
+            tokenEstimate: Number(row.token_estimate),
+            ordinal: Number(row.ordinal),
+            contentHash: String(row.content_hash),
+            docVersion: row.doc_version ? String(row.doc_version) : undefined,
+            provenance,
+        };
+    }
+}

package/dist/packages/storage/src/repositories/example-repository.d.ts

@@ -0,0 +1,16 @@
+import type { IndexedExample } from '../../../shared/src/index.ts';
+import type { DocOrbitDb } from '../db.ts';
+import type { IExampleRepository } from '../interfaces.ts';
+export declare class ExampleRepository implements IExampleRepository {
+    private db;
+    constructor(db: DocOrbitDb);
+    saveIndexedExamples(examples: IndexedExample[]): void;
+    getIndexedExamplesBySnapshot(snapshotId: string): IndexedExample[];
+    searchIndexedExamples(query: string, options?: {
+        language?: string;
+        framework?: string;
+        docVersion?: string;
+        limit?: number;
+    }): IndexedExample[];
+    private hydrateIndexedExample;
+}

package/dist/packages/storage/src/repositories/example-repository.js

@@ -0,0 +1,158 @@
+import { extractSearchTokens } from "../search-tokens.js";
+export class ExampleRepository {
+    db;
+    constructor(db) {
+        this.db = db;
+    }
+    saveIndexedExamples(examples) {
+        if (examples.length === 0)
+            return;
+        const raw = this.db.getRawDb();
+        const insertEx = raw.prepare(`
+      INSERT INTO indexed_examples (
+        id, chunk_id, page_id, snapshot_id, language, framework,
+        task, code, source_url, source_authority, related_api,
+        related_symbol, doc_version, provenance_json, created_at
+      ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+      ON CONFLICT(id) DO UPDATE SET
+        chunk_id = excluded.chunk_id,
+        language = excluded.language,
+        framework = excluded.framework,
+        task = excluded.task,
+        code = excluded.code,
+        source_url = excluded.source_url,
+        source_authority = excluded.source_authority,
+        related_api = excluded.related_api,
+        related_symbol = excluded.related_symbol,
+        doc_version = excluded.doc_version,
+        provenance_json = excluded.provenance_json
+    `);
+        const insertFts = this.db.isFtsAvailable()
+            ? raw.prepare(`
+          INSERT INTO indexed_examples_fts (example_id, task, language, framework, code, related_api, related_symbol)
+          VALUES (?, ?, ?, ?, ?, ?, ?)
+        `)
+            : null;
+        const deleteFts = this.db.isFtsAvailable()
+            ? raw.prepare('DELETE FROM indexed_examples_fts WHERE example_id = ?')
+            : null;
+        raw.exec('BEGIN TRANSACTION;');
+        try {
+            for (const ex of examples) {
+                insertEx.run(ex.id, ex.chunkId || null, ex.pageId, ex.snapshotId || 'snap_default', ex.language, ex.framework || null, ex.task, ex.code, ex.sourceUrl, ex.sourceAuthority, ex.relatedApi || null, ex.relatedSymbol || null, ex.docVersion || null, ex.provenance ? JSON.stringify(ex.provenance) : null, ex.createdAt || new Date().toISOString());
+                if (insertFts && deleteFts) {
+                    deleteFts.run(ex.id);
+                    insertFts.run(ex.id, ex.task, ex.language, ex.framework || '', ex.code, ex.relatedApi || '', ex.relatedSymbol || '');
+                }
+            }
+            raw.exec('COMMIT;');
+        }
+        catch (err) {
+            raw.exec('ROLLBACK;');
+            throw err;
+        }
+    }
+    getIndexedExamplesBySnapshot(snapshotId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT * FROM indexed_examples WHERE snapshot_id = ?').all(snapshotId);
+        return rows.map(r => this.hydrateIndexedExample(r));
+    }
+    searchIndexedExamples(query, options = {}) {
+        const raw = this.db.getRawDb();
+        const limit = options.limit || 10;
+        const term = query.trim();
+        let results = [];
+        // FTS attempt
+        if (this.db.isFtsAvailable() && term.length > 0) {
+            const cleanTokens = extractSearchTokens(term);
+            if (cleanTokens.length > 0) {
+                const sanitizedFts = cleanTokens
+                    .map(t => t.length >= 4 ? `"${t.replace(/"/g, '""')}"*` : `"${t.replace(/"/g, '""')}"`)
+                    .join(' OR ');
+                try {
+                    let sql = `
+            SELECT ex.*, fts.rank as fts_rank
+            FROM indexed_examples ex
+            JOIN indexed_examples_fts fts ON ex.id = fts.example_id
+            WHERE indexed_examples_fts MATCH ?
+          `;
+                    const params = [sanitizedFts];
+                    if (options.language) {
+                        sql += ' AND lower(ex.language) = ?';
+                        params.push(options.language.toLowerCase());
+                    }
+                    if (options.framework) {
+                        sql += ' AND lower(ex.framework) = ?';
+                        params.push(options.framework.toLowerCase());
+                    }
+                    if (options.docVersion) {
+                        sql += ' AND (ex.doc_version = ? OR ex.doc_version IS NULL)';
+                        params.push(options.docVersion);
+                    }
+                    sql += " ORDER BY CASE WHEN ex.source_authority = 'official' THEN 0 ELSE 1 END, fts.rank LIMIT ?";
+                    params.push(limit);
+                    const rows = raw.prepare(sql).all(...params);
+                    results = rows.map(r => this.hydrateIndexedExample(r));
+                }
+                catch {
+                    // Fall through to LIKE
+                }
+            }
+        }
+        if (results.length === 0) {
+            // LIKE fallback
+            let sql = 'SELECT * FROM indexed_examples WHERE 1=1';
+            const params = [];
+            if (options.language) {
+                sql += ' AND lower(language) = ?';
+                params.push(options.language.toLowerCase());
+            }
+            if (options.framework) {
+                sql += ' AND lower(framework) = ?';
+                params.push(options.framework.toLowerCase());
+            }
+            if (options.docVersion) {
+                sql += ' AND (doc_version = ? OR doc_version IS NULL)';
+                params.push(options.docVersion);
+            }
+            if (term.length > 0) {
+                const tokens = extractSearchTokens(term);
+                if (tokens.length > 0) {
+                    const tokenClauses = tokens.map(() => '(task LIKE ? OR code LIKE ? OR related_api LIKE ? OR related_symbol LIKE ?)');
+                    sql += ` AND (${tokenClauses.join(' OR ')})`;
+                    for (const t of tokens) {
+                        const lp = `%${t}%`;
+                        params.push(lp, lp, lp, lp);
+                    }
+                }
+                else {
+                    return [];
+                }
+            }
+            sql += " ORDER BY CASE WHEN source_authority = 'official' THEN 0 ELSE 1 END LIMIT ?";
+            params.push(limit);
+            const rows = raw.prepare(sql).all(...params);
+            results = rows.map(r => this.hydrateIndexedExample(r));
+        }
+        return results;
+    }
+    hydrateIndexedExample(row) {
+        return {
+            id: String(row.id),
+            chunkId: row.chunk_id ? String(row.chunk_id) : undefined,
+            pageId: String(row.page_id),
+            snapshotId: String(row.snapshot_id),
+            language: String(row.language),
+            framework: row.framework ? String(row.framework) : undefined,
+            task: String(row.task),
+            code: String(row.code),
+            sourceUrl: String(row.source_url),
+            sourceAuthority: String(row.source_authority),
+            relatedApi: row.related_api ? String(row.related_api) : undefined,
+            relatedSymbol: row.related_symbol ? String(row.related_symbol) : undefined,
+            docVersion: row.doc_version ? String(row.doc_version) : undefined,
+            provenance: row.provenance_json ? JSON.parse(String(row.provenance_json)) : undefined,
+            createdAt: String(row.created_at),
+        };
+    }
+}

package/dist/packages/storage/src/repositories/page-repository.d.ts

@@ -0,0 +1,16 @@
+import type { NormalizedPage } from '../../../shared/src/index.ts';
+import type { DocOrbitDb } from '../db.ts';
+export declare class PageRepository {
+    private db;
+    constructor(db: DocOrbitDb);
+    savePage(page: NormalizedPage): void;
+    getPage(id: string): NormalizedPage | null;
+    getPageByUrl(url: string): NormalizedPage | null;
+    getPagesBySource(sourceId: string): NormalizedPage[];
+    listPages(limit?: number, offset?: number): NormalizedPage[];
+    countPages(sourceId?: string): number;
+    searchPagesFts(query: string, limit?: number): NormalizedPage[];
+    getPageLinks(pageId: string): string[];
+    savePageLinks(pageId: string, targetUrls: string[]): void;
+    hydratePage(row: Record<string, unknown>): NormalizedPage;
+}

package/dist/packages/storage/src/repositories/page-repository.js

@@ -0,0 +1,161 @@
+export class PageRepository {
+    db;
+    constructor(db) {
+        this.db = db;
+    }
+    savePage(page) {
+        const raw = this.db.getRawDb();
+        const pageStmt = raw.prepare(`
+      INSERT INTO pages (
+        id, source_id, title, url, content, content_hash,
+        fetched_at, raw_bytes, estimated_tokens,
+        headings_json, security_annotations_json, provenance_json
+      ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+      ON CONFLICT(url) DO UPDATE SET
+        title = excluded.title,
+        content = excluded.content,
+        content_hash = excluded.content_hash,
+        fetched_at = excluded.fetched_at,
+        raw_bytes = excluded.raw_bytes,
+        estimated_tokens = excluded.estimated_tokens,
+        headings_json = excluded.headings_json,
+        security_annotations_json = excluded.security_annotations_json,
+        provenance_json = excluded.provenance_json
+    `);
+        pageStmt.run(page.id, page.sourceId || null, page.title, page.url, page.content, page.contentHash, page.fetchedAt, page.rawBytes, page.estimatedTokens, JSON.stringify(page.headings), JSON.stringify(page.securityAnnotations), page.provenance ? JSON.stringify(page.provenance) : null);
+        raw.prepare('DELETE FROM links WHERE page_id = ?').run(page.id);
+        const linkStmt = raw.prepare(`
+      INSERT INTO links (id, page_id, text, url, is_external)
+      VALUES (?, ?, ?, ?, ?)
+    `);
+        for (let i = 0; i < page.links.length; i++) {
+            const link = page.links[i];
+            const linkId = `${page.id}_link_${i}`;
+            linkStmt.run(linkId, page.id, link.text, link.url, link.isExternal ? 1 : 0);
+        }
+        raw.prepare('DELETE FROM code_examples WHERE page_id = ?').run(page.id);
+        const codeStmt = raw.prepare(`
+      INSERT INTO code_examples (id, page_id, language, code, caption)
+      VALUES (?, ?, ?, ?, ?)
+    `);
+        for (let i = 0; i < page.codeExamples.length; i++) {
+            const example = page.codeExamples[i];
+            const exampleId = `${page.id}_code_${i}`;
+            codeStmt.run(exampleId, page.id, example.language, example.code, example.caption || null);
+        }
+        if (this.db.isFtsAvailable()) {
+            try {
+                raw.prepare('DELETE FROM pages_fts WHERE page_id = ?').run(page.id);
+                raw.prepare(`
+          INSERT INTO pages_fts (page_id, title, content)
+          VALUES (?, ?, ?)
+        `).run(page.id, page.title, page.content);
+            }
+            catch {
+                // Continue
+            }
+        }
+    }
+    getPage(id) {
+        const raw = this.db.getRawDb();
+        const row = raw.prepare('SELECT * FROM pages WHERE id = ?').get(id);
+        if (!row)
+            return null;
+        return this.hydratePage(row);
+    }
+    getPageByUrl(url) {
+        const raw = this.db.getRawDb();
+        const row = raw.prepare('SELECT * FROM pages WHERE url = ?').get(url);
+        if (!row)
+            return null;
+        return this.hydratePage(row);
+    }
+    getPagesBySource(sourceId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT * FROM pages WHERE source_id = ? ORDER BY fetched_at ASC').all(sourceId);
+        return rows.map(r => this.hydratePage(r));
+    }
+    listPages(limit = 50, offset = 0) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT * FROM pages ORDER BY fetched_at DESC LIMIT ? OFFSET ?').all(limit, offset);
+        return rows.map(r => this.hydratePage(r));
+    }
+    countPages(sourceId) {
+        const raw = this.db.getRawDb();
+        if (sourceId) {
+            const row = raw.prepare('SELECT COUNT(*) as count FROM pages WHERE source_id = ?').get(sourceId);
+            return row?.count || 0;
+        }
+        const row = raw.prepare('SELECT COUNT(*) as count FROM pages').get();
+        return row?.count || 0;
+    }
+    searchPagesFts(query, limit = 10) {
+        if (!this.db.isFtsAvailable()) {
+            return [];
+        }
+        const raw = this.db.getRawDb();
+        const sanitizedQuery = `"${query.replace(/"/g, '""')}"`;
+        try {
+            const rows = raw.prepare(`
+        SELECT p.* FROM pages p
+        JOIN pages_fts fts ON p.id = fts.page_id
+        WHERE pages_fts MATCH ?
+        ORDER BY rank
+        LIMIT ?
+      `).all(sanitizedQuery, limit);
+            return rows.map(r => this.hydratePage(r));
+        }
+        catch {
+            return [];
+        }
+    }
+    getPageLinks(pageId) {
+        const raw = this.db.getRawDb();
+        const rows = raw.prepare('SELECT url FROM links WHERE page_id = ?').all(pageId);
+        return rows.map(r => String(r.url));
+    }
+    savePageLinks(pageId, targetUrls) {
+        const raw = this.db.getRawDb();
+        raw.prepare('DELETE FROM links WHERE page_id = ?').run(pageId);
+        const stmt = raw.prepare('INSERT OR IGNORE INTO links (id, page_id, text, url, is_external) VALUES (?, ?, ?, ?, ?)');
+        for (let i = 0; i < targetUrls.length; i++) {
+            stmt.run(`${pageId}_link_${i}`, pageId, '', targetUrls[i], 0);
+        }
+    }
+    hydratePage(row) {
+        const raw = this.db.getRawDb();
+        const pageId = String(row.id);
+        const linkRows = raw.prepare('SELECT text, url, is_external FROM links WHERE page_id = ?').all(pageId);
+        const links = linkRows.map(lr => ({
+            text: String(lr.text),
+            url: String(lr.url),
+            isExternal: Number(lr.is_external) === 1,
+        }));
+        const codeRows = raw.prepare('SELECT id, language, code, caption FROM code_examples WHERE page_id = ?').all(pageId);
+        const codeExamples = codeRows.map(cr => ({
+            id: String(cr.id),
+            language: String(cr.language),
+            code: String(cr.code),
+            caption: cr.caption ? String(cr.caption) : undefined,
+        }));
+        const headings = row.headings_json ? JSON.parse(String(row.headings_json)) : [];
+        const securityAnnotations = row.security_annotations_json ? JSON.parse(String(row.security_annotations_json)) : [];
+        const provenance = row.provenance_json ? JSON.parse(String(row.provenance_json)) : undefined;
+        return {
+            id: pageId,
+            sourceId: String(row.source_id),
+            title: String(row.title),
+            url: String(row.url),
+            content: String(row.content),
+            headings,
+            links,
+            codeExamples,
+            contentHash: String(row.content_hash),
+            fetchedAt: String(row.fetched_at),
+            rawBytes: Number(row.raw_bytes),
+            estimatedTokens: Number(row.estimated_tokens),
+            securityAnnotations,
+            provenance,
+        };
+    }
+}

package/dist/packages/storage/src/repositories/pitfall-repository.d.ts

@@ -0,0 +1,15 @@
+import type { Pitfall, PitfallKind } from '../../../shared/src/index.ts';
+import type { DocOrbitDb } from '../db.ts';
+import type { IPitfallRepository } from '../interfaces.ts';
+export declare class PitfallRepository implements IPitfallRepository {
+    private db;
+    constructor(db: DocOrbitDb);
+    savePitfalls(pitfalls: Pitfall[]): void;
+    getPitfallsBySnapshot(snapshotId: string): Pitfall[];
+    searchPitfalls(query: string, options?: {
+        kind?: PitfallKind;
+        docVersion?: string;
+        limit?: number;
+    }): Pitfall[];
+    private hydratePitfall;
+}