require('module-alias/register') const Schema = require('@helpers/Schema') const db = require('@config/database') afterAll(() => { db.end() }) afterEach(() => { Schema.__reset() }) describe('verifyScrapeSchema against the real dev schema', () => { test('reports ok once the migration has been applied', async () => { const result = await Schema.verifyScrapeSchema() if (!result.ok) { throw new Error( `dev scrape schema not applied - run ./migrateDatabase.sh. ` + `missing objects: ${result.missingObjects.join(', ') || 'none'}; ` + `missing privileges: ${result.missingPrivileges.join(', ') || 'none'}` ) } expect(result.missingObjects).toEqual([]) expect(result.missingPrivileges).toEqual([]) expect(result.remedy).toBe('') }) test('isScrapeSchemaReady reflects the last verification', async () => { expect(Schema.isScrapeSchemaReady()).toBe(false) await Schema.verifyScrapeSchema() expect(Schema.isScrapeSchemaReady()).toBe(true) Schema.__reset() expect(Schema.isScrapeSchemaReady()).toBe(false) }) test('__setReady is honoured, so request paths can be driven in tests', () => { Schema.__setReady(true) expect(Schema.isScrapeSchemaReady()).toBe(true) Schema.__setReady(false) expect(Schema.isScrapeSchemaReady()).toBe(false) }) }) // // The FULLTEXT index is the one object whose loss is silent: MATCH..AGAINST // against a BTREE index returns nothing without erroring. Guard it by type. // describe('FULLTEXT index guard', () => { test('attachment_text_index exists and is actually FULLTEXT', async () => { await expect(Schema.__fulltextIndexExists('attachment', 'attachment_text_index')) .resolves.toBe(true) }) test('a BTREE index of the same shape would NOT satisfy the check', async () => { // attachment_scrape_state is a real BTREE index, so this proves the // check discriminates on type rather than just on name existing. await expect(Schema.__indexExists('attachment', 'attachment_scrape_state')) .resolves.toBe(true) await expect(Schema.__fulltextIndexExists('attachment', 'attachment_scrape_state')) .resolves.toBe(false) }) test('verifyScrapeSchema includes it among the verified objects', async () => { const result = await Schema.verifyScrapeSchema() expect(result.ok).toBe(true) expect(result.missingObjects).toEqual([]) }) test('MATCH..AGAINST actually works, not just that an index exists', async () => { // The functional end of the guard. An index of the right type that // somehow could not be searched would still be a broken search. const [rows] = await db.promise().query( `SELECT COUNT(*) AS n FROM attachment WHERE MATCH(text) AGAINST(? IN NATURAL LANGUAGE MODE)`, ['the'] ) // Any result (including 0 matches) proves the query planner accepted // MATCH..AGAINST against this column, which a BTREE index would reject. expect(typeof rows[0].n).toBe('number') }) }) describe('object existence checks', () => { test('finds the real table, columns and index', async () => { await expect(Schema.__tableExists('scrape_cache')).resolves.toBe(true) await expect(Schema.__columnExists('attachment', 'scrape_state')).resolves.toBe(true) await expect(Schema.__columnExists('attachment', 'scrape_attempts')).resolves.toBe(true) await expect(Schema.__columnExists('attachment', 'enqueued_at')).resolves.toBe(true) await expect(Schema.__indexExists('attachment', 'attachment_scrape_state')).resolves.toBe(true) }) test('reports absence rather than throwing for things that do not exist', async () => { await expect(Schema.__tableExists('no_such_table_xyz')).resolves.toBe(false) await expect(Schema.__columnExists('attachment', 'no_such_column_xyz')).resolves.toBe(false) await expect(Schema.__indexExists('attachment', 'no_such_index_xyz')).resolves.toBe(false) }) }) // // The privilege check is what turns a future narrow-grants problem into one // clear boot message instead of a swallowed runtime failure. // describe('parseGrants', () => { const DB = 'application' const TABLES = ['attachment', 'scrape_cache'] const parse = (lines) => Schema.parseGrants(lines, DB, TABLES) test('ALL PRIVILEGES ON *.* grants everything', () => { const granted = parse(['GRANT ALL PRIVILEGES ON *.* TO `root`@`localhost` WITH GRANT OPTION']) expect(granted.has('SELECT')).toBe(true) expect(granted.has('INSERT')).toBe(true) expect(granted.has('UPDATE')).toBe(true) expect(granted.has('DELETE')).toBe(true) }) test('ALL PRIVILEGES scoped to the database grants everything', () => { const granted = parse(['GRANT ALL PRIVILEGES ON `application`.* TO `app`@`localhost`']) expect(granted.has('DELETE')).toBe(true) }) test('reads an explicit privilege list', () => { const granted = parse([ 'GRANT USAGE ON *.* TO `app`@`localhost`', 'GRANT SELECT, INSERT, UPDATE, DELETE ON `application`.* TO `app`@`localhost`', ]) expect([...granted].sort()).toEqual(['DELETE', 'INSERT', 'SELECT', 'UPDATE']) }) // The case that caused this whole change class. test('spots a missing DELETE', () => { const granted = parse(['GRANT SELECT, INSERT, UPDATE ON `application`.* TO `app`@`localhost`']) expect(granted.has('UPDATE')).toBe(true) expect(granted.has('DELETE')).toBe(false) }) test('accepts table-scoped grants for our tables', () => { const granted = parse([ 'GRANT SELECT, DELETE ON `application`.`attachment` TO `app`@`localhost`', 'GRANT INSERT, UPDATE ON `application`.`scrape_cache` TO `app`@`localhost`', ]) expect([...granted].sort()).toEqual(['DELETE', 'INSERT', 'SELECT', 'UPDATE']) }) test('ignores grants on an unrelated database', () => { const granted = parse(['GRANT ALL PRIVILEGES ON `somethingelse`.* TO `app`@`localhost`']) expect(granted.size).toBe(0) }) test('ignores USAGE, which conveys no privileges', () => { const granted = parse(['GRANT USAGE ON *.* TO `app`@`localhost`']) expect(granted.size).toBe(0) }) test('handles grants written without backticks', () => { const granted = parse(['GRANT SELECT, DELETE ON application.* TO app@localhost']) expect(granted.has('DELETE')).toBe(true) }) test('strips column-level qualifiers', () => { const granted = parse(['GRANT SELECT (id, url), UPDATE (text) ON `application`.`attachment` TO `app`@`localhost`']) expect(granted.has('SELECT')).toBe(true) expect(granted.has('UPDATE')).toBe(true) }) test('ignores lines that are not grant statements', () => { expect(parse(['', 'not a grant line', null]).size).toBe(0) }) }) // // The regression that matters most. // // Schema.js used to issue CREATE TABLE / ALTER TABLE through the application's // own connection pool, which failed on production's least-privilege user with // "CREATE command denied to user 'app'@'localhost'". Nothing in this module may // ever issue DDL again. // describe('issues no DDL', () => { test('verifyScrapeSchema runs only read-only statements', async () => { const statements = [] // db.promise() hands back a NEW wrapper on every call, so spying on one // instance would miss the one Schema.js obtains internally. Intercept // the factory and instrument each wrapper it returns. const realPromise = db.promise.bind(db) const spy = jest.spyOn(db, 'promise').mockImplementation(() => { const wrapper = realPromise() const originalQuery = wrapper.query.bind(wrapper) wrapper.query = (sql, params) => { statements.push(String(sql)) return originalQuery(sql, params) } return wrapper }) await Schema.verifyScrapeSchema() spy.mockRestore() expect(statements.length).toBeGreaterThan(0) const forbidden = /\b(CREATE|ALTER|DROP|TRUNCATE|RENAME)\b/i const offenders = statements.filter(statement => forbidden.test(statement)) expect(offenders).toEqual([]) }) test('the module source contains no DDL outside comments', () => { const fs = require('fs') const path = require('path') const source = fs.readFileSync( path.join(__dirname, '..', '..', 'helpers', 'Schema.js'), 'utf8' ) const code = source .split('\n') .filter(line => !line.trim().startsWith('//')) .join('\n') expect(code).not.toMatch(/\bCREATE\s+TABLE\b/i) expect(code).not.toMatch(/\bALTER\s+TABLE\b/i) expect(code).not.toMatch(/\bCREATE\s+INDEX\b/i) expect(code).not.toMatch(/\bDROP\b/i) }) })