rose-ash/tests/playwright/pre-screen-sources.js

#!/usr/bin/env node
// pre-screen-sources.js — Identify hyperscript sources that hang the WASM parser
//
// For each unique _="..." source in hs-behavioral-data.js, spawns a child process
// that loads the WASM kernel + HS modules, then tries to compile. If it takes >3s,
// it's marked as hanging.
//
// Output:
//   tests/playwright/hs-safe-sources.json    — sources that compile OK (or error)
//   tests/playwright/hs-hanging-sources.json — sources that hang the parser/compiler
//
// Usage: node tests/playwright/pre-screen-sources.js

const { execFileSync } = require('child_process');
const fs = require('fs');
const path = require('path');

const PROJECT_ROOT = path.resolve(__dirname, '../..');
const DATA_FILE = path.join(__dirname, 'hs-behavioral-data.js');
const WORKER_FILE = path.join(__dirname, '_pre-screen-worker-timed.js');

// Extract unique sources
const data = require(DATA_FILE);
const sourcesSet = new Set();
for (const t of data) {
  const matches = t.html.matchAll(/_=['"]([^'"]+)['"]/g);
  for (const m of matches) sourcesSet.add(m[1]);
}
const allSources = [...sourcesSet];
console.log(`Found ${allSources.length} unique hyperscript sources`);

// Process in batches — each batch is a fresh child process
const BATCH_SIZE = 15;
const TIMEOUT_MS = 30000; // 30s for a batch of 15 (allows ~2s per source)

const safe = [];
const errors = [];
const hanging = [];
const timings = []; // {source, ms, status}

function testBatch(sources) {
  // Write sources to temp file for the worker
  const tmpFile = '/tmp/hs-batch-input.json';
  fs.writeFileSync(tmpFile, JSON.stringify(sources));

  try {
    const output = execFileSync(process.execPath, [WORKER_FILE], {
      timeout: TIMEOUT_MS,
      encoding: 'utf8',
      env: { ...process.env, HS_BATCH_FILE: tmpFile },
      cwd: PROJECT_ROOT,
    });

    // Parse results from worker stdout
    const lines = output.trim().split('\n');
    for (const line of lines) {
      if (line.startsWith('RESULT:')) {
        const { source, status, detail, ms } = JSON.parse(line.slice(7));
        timings.push({ source, status, ms });
        if (status === 'ok') {
          safe.push(source);
        } else if (status === 'error') {
          errors.push({ source, detail });
          safe.push(source); // errors are still "safe" — they don't hang
        } else {
          hanging.push(source);
        }
      }
    }
    return true;
  } catch (e) {
    if (e.killed || e.signal === 'SIGTERM') {
      // Whole batch timed out — need to bisect
      return false;
    }
    // Other error (crash) — treat all as hanging
    console.error(`  Batch crashed: ${e.message.slice(0, 200)}`);
    return false;
  }
}

function bisect(sources) {
  if (sources.length === 0) return;

  if (sources.length === 1) {
    // Single source that we know hangs
    console.log(`  HANG: ${sources[0].slice(0, 80)}`);
    hanging.push(sources[0]);
    return;
  }

  console.log(`  Bisecting ${sources.length} sources...`);

  // Try first half
  const mid = Math.ceil(sources.length / 2);
  const first = sources.slice(0, mid);
  const second = sources.slice(mid);

  if (!testBatch(first)) {
    bisect(first);
  }
  if (!testBatch(second)) {
    bisect(second);
  }
}

// Process batches
const totalBatches = Math.ceil(allSources.length / BATCH_SIZE);
for (let i = 0; i < allSources.length; i += BATCH_SIZE) {
  const batch = allSources.slice(i, i + BATCH_SIZE);
  const batchNum = Math.floor(i / BATCH_SIZE) + 1;
  process.stdout.write(`Batch ${batchNum}/${totalBatches} (${batch.length} sources)... `);

  if (testBatch(batch)) {
    console.log(`OK (${safe.length} safe, ${hanging.length} hanging so far)`);
  } else {
    console.log(`TIMEOUT — bisecting`);
    bisect(batch);
    console.log(`  After bisect: ${safe.length} safe, ${hanging.length} hanging`);
  }
}

// Write results
const safeFile = path.join(__dirname, 'hs-safe-sources.json');
const hangFile = path.join(__dirname, 'hs-hanging-sources.json');
const errFile = path.join(__dirname, 'hs-error-sources.json');

fs.writeFileSync(safeFile, JSON.stringify(safe, null, 2) + '\n');
fs.writeFileSync(hangFile, JSON.stringify(hanging, null, 2) + '\n');
fs.writeFileSync(errFile, JSON.stringify(errors, null, 2) + '\n');

console.log(`\nDone!`);
console.log(`  Safe (no hang): ${safe.length} (written to ${path.relative(PROJECT_ROOT, safeFile)})`);
console.log(`  Errors:         ${errors.length} (written to ${path.relative(PROJECT_ROOT, errFile)})`);
console.log(`  Hanging:        ${hanging.length} (written to ${path.relative(PROJECT_ROOT, hangFile)})`);

if (hanging.length > 0) {
  console.log(`\nHanging sources:`);
  for (const s of hanging) {
    console.log(`  - ${s}`);
  }
}

if (errors.length > 0) {
  console.log(`\nError sources:`);
  for (const e of errors) {
    console.log(`  - ${e.source}`);
    console.log(`    ${e.detail}`);
  }
}

// Timing summary — show slowest sources
if (timings.length > 0) {
  timings.sort((a, b) => b.ms - a.ms);
  console.log(`\nSlowest sources (top 20):`);
  for (let i = 0; i < Math.min(20, timings.length); i++) {
    const t = timings[i];
    console.log(`  ${t.ms}ms [${t.status}] ${t.source.slice(0, 80)}`);
  }
  const total = timings.reduce((s, t) => s + t.ms, 0);
  const avg = Math.round(total / timings.length);
  console.log(`\nTotal: ${total}ms, Avg: ${avg}ms, Max: ${timings[0].ms}ms`);

  // Flag anything over 3s as "slow" (potential near-hang)
  const slow = timings.filter(t => t.ms > 3000);
  if (slow.length > 0) {
    console.log(`\nWARNING: ${slow.length} sources took >3s:`);
    for (const t of slow) {
      console.log(`  ${t.ms}ms ${t.source}`);
    }
  }
}