fix(krawler): preserve URL query string and unify hooks registry

Author: easydime

packages/krawler/src/bin.js

@@ -1,10 +1,30 @@
 #!/usr/bin/env node
 
+import { spawnSync } from 'node:child_process'
+import { fileURLToPath } from 'node:url'
+
 // import cluster from 'cluster'
 import { Healthcheck } from './healthcheck.js'
 import { processOptions, cli } from './cli.js'
 import logger from './logger.js'
 
+// Re-exec under --conditions=development if not already set.
+// Without this, bin.js loads hooks via src/ while the user jobfile loads them via dist/index.mjs
+// (per package.json `exports.import`), giving two distinct hooks registries and breaking
+// custom hook registration. The development condition unifies both resolutions on src/.
+const hasDevCondition = process.execArgv.some(arg =>
+  arg === '--conditions=development' ||
+  (arg.startsWith('--conditions=') && arg.slice('--conditions='.length).split(',').includes('development'))
+)
+if (!hasDevCondition) {
+  const child = spawnSync(
+    process.execPath,
+    ['--conditions=development', fileURLToPath(import.meta.url), ...process.argv.slice(2)],
+    { stdio: 'inherit' }
+  )
+  process.exit(child.status ?? 1)
+}
+
 /* WIP: cluster mode, the main issue is that stores are created by a job before hook
    thus are not availalbe on workers not running the job but running only tasks
 if (cluster.isMaster) {

packages/krawler/src/tasks/tasks.http.js

@@ -27,11 +27,15 @@ export function getRequestParameters (options) {
     method: options.method || 'GET',
     headers: options.headers,
     body: options.body,
-    searchParams: buildSearchParams(queryParameters),
     cookieJar: options.jar,
     // Do not throw on non-2xx so the downstream consumer can read statusCode from the 'response' event
     throwHttpErrors: false
   }
+  // Only set searchParams when we actually have params; otherwise got would override
+  // any pre-existing query string already embedded in options.url.
+  if (Object.keys(queryParameters).length > 0) {
+    requestParameters.searchParams = buildSearchParams(queryParameters)
+  }
   if (options.timeout) {
     requestParameters.timeout = { request: options.timeout }
   }