bare-operating-system/packages/bare-os-coreutils/test/awk-sed-posix-smoke.test.mjs

/**
 * CI smoke checks for awk/sed sources (POSIX-oriented coverage is partial by design).
 */
import { readFile } from 'node:fs/promises'
import { dirname, join } from 'node:path'
import { fileURLToPath } from 'node:url'
import test from 'brittle'
import { bareAwkRun } from '../lib/awk-engine.js'

const __dirname = dirname(fileURLToPath(import.meta.url))
const srcDir = join(__dirname, '../src')
const fixtureDir = join(__dirname, 'fixtures')

test('sed.js exists and documents line-oriented operation', async (t) => {
  const s = await readFile(join(srcDir, 'sed.js'), 'utf8')
  t.ok(s.includes('async function run'))
  t.ok(s.length > 200)
})

test('awk.js exists and wires field parsing', async (t) => {
  const s = await readFile(join(srcDir, 'awk.js'), 'utf8')
  t.ok(s.includes('async function run'))
  t.ok(s.includes('field-separator') || s.includes('split'))
})

test('awk-engine supports delete arr[key] statement', async (t) => {
  const s = await readFile(join(__dirname, '../lib/awk-engine.js'), 'utf8')
  t.ok(s.includes("id === 'delete'"))
  t.ok(s.includes("case 'delete':"))
})

test('printf FORMAT octal escape helper is defined', async (t) => {
  const s = await readFile(join(srcDir, 'printf.js'), 'utf8')
  t.ok(s.includes('barePrintfUnescapeFormat'))
})

test('awk nextfile skips remainder of first file', async (t) => {
  /** @type {string[]} */
  const lines = []
  const enc = new TextEncoder()
  const io = {
    stdinLines: [],
    bytesToString(buf) {
      return buf ? new TextDecoder().decode(buf) : ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    readFile(p) {
      if (p === 'a') return enc.encode('A1\nA2\nA3\n')
      if (p === 'b') return enc.encode('B1\nB2\n')
      return null
    }
  }
  await bareAwkRun(
    'FNR==2 && FILENAME == "a" { nextfile } { print $0 }',
    { fs: ' ', argc: 3, argv: ['awk', 'a', 'b'], environ: {} },
    io
  )
  t.is(lines.join('\n'), 'A1\nB1\nB2')
})

test('awk getline from file reads lines', async (t) => {
  /** @type {string[]} */
  const lines = []
  const enc = new TextEncoder()
  const io = {
    stdinLines: [],
    bytesToString(buf) {
      return buf ? new TextDecoder().decode(buf) : ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    readFile(p) {
      if (p === 'data.txt') return enc.encode('one\ntwo\n')
      return null
    }
  }
  await bareAwkRun(
    'BEGIN { while ((getline x < "data.txt") > 0) print x }',
    { fs: ' ', argc: 1, argv: ['awk'], environ: {} },
    io
  )
  t.is(lines.join('\n'), 'one\ntwo')
})

test('awk getline missing file returns 0 and does not throw', async (t) => {
  /** @type {string[]} */
  const lines = []
  const io = {
    stdinLines: [],
    bytesToString(buf) {
      return buf ? new TextDecoder().decode(buf) : ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    readFile() {
      return null
    }
  }
  await bareAwkRun(
    'BEGIN { print (getline x < "missing.txt"); print x }',
    { fs: ' ', argc: 1, argv: ['awk'], environ: {} },
    io
  )
  t.is(lines[0], '0')
  t.is(lines[1], '')
})

test('awk getline repeated reads preserve per-file cursor', async (t) => {
  /** @type {string[]} */
  const lines = []
  const enc = new TextEncoder()
  const io = {
    stdinLines: [],
    bytesToString(buf) {
      return buf ? new TextDecoder().decode(buf) : ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    readFile(p) {
      if (p === 'data.txt') return enc.encode('one\ntwo\nthree\n')
      return null
    }
  }
  await bareAwkRun(
    'BEGIN { getline a < "data.txt"; getline b < "data.txt"; print a; print b }',
    { fs: ' ', argc: 1, argv: ['awk'], environ: {} },
    io
  )
  t.is(lines[0], 'one')
  t.is(lines[1], 'two')
})

test('awk getline handles larger VFS input boundaries', async (t) => {
  /** @type {string[]} */
  const lines = []
  const big = 'x'.repeat(8192)
  const enc = new TextEncoder()
  const io = {
    stdinLines: [],
    bytesToString(buf) {
      return buf ? new TextDecoder().decode(buf) : ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    readFile(p) {
      if (p === 'big.txt') return enc.encode(`${big}\nend\n`)
      return null
    }
  }
  await bareAwkRun(
    'BEGIN { getline a < "big.txt"; getline b < "big.txt"; print length(a); print b }',
    { fs: ' ', argc: 1, argv: ['awk'], environ: {} },
    io
  )
  t.is(lines[0], '8192')
  t.is(lines[1], 'end')
})

test('sed Open Group Issue 7 golden fixture is present', async (t) => {
  const raw = await readFile(join(fixtureDir, 'sed-open-group-issue7.json'), 'utf8')
  const parsed = JSON.parse(raw)
  t.ok(Array.isArray(parsed.examples))
  t.ok(parsed.examples.length >= 3)
  t.ok(parsed.examples.some((e) => e.script === 's/[0-9][0-9]*/NUM/g'))
})

test('awk atan2 and unary minus', async (t) => {
  /** @type {string[]} */
  const lines = []
  const io = {
    stdinLines: [''],
    bytesToString() {
      return ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    async readFile() {
      return null
    }
  }
  await bareAwkRun(
    'BEGIN { print int(atan2(0,-1)*1000); print -3+1 }',
    { fs: ' ', argc: 1, argv: ['awk'], environ: {} },
    io
  )
  t.ok(lines[0].includes('3141') || lines[0] === '3141' || Number(lines[0]) > 3000)
  t.is(lines[1], '-2')
})

test('awk substr two-arg is suffix from start', async (t) => {
  /** @type {string[]} */
  const lines = []
  const io = {
    stdinLines: [''],
    bytesToString() {
      return ''
    },
    print(s) {
      lines.push(String(s).replace(/\n$/, ''))
    },
    async writeFile() {},
    async readFile() {
      return null
    }
  }
  await bareAwkRun(
    'BEGIN { print substr("abcdef", 3) }',
    { fs: ' ', argc: 1, argv: ['awk'], environ: {} },
    io
  )
  t.is(lines[0], 'cdef')
})