Search result parsing broken - returns "Raw text (daemon mode)" for most engines

[youchi1]

When the daemon is running, stealth search skips result parsing for all engines and output formats. The search itself succeeds (page loads with results), but the parser falls back to a single "Raw text (daemon mode)" entry.

Reproduce

stealth daemon start
stealth search google "test query" -f json -n 5
stealth search bing "test query" -f json -n 5

Actual (daemon):

• Google: {"results": [{"title": "Raw text (daemon mode)", "content": ""}]}
• Bing: {"results": [{"title": "Raw text (daemon mode)", "content": "<entire raw page text>"}]}
• All engines return a single result with "title": "Raw text (daemon mode)" instead of parsed individual results

Without daemon, both Bing and Google parses correctly:

stealth daemon stop
stealth search bing "test query" -f json -n 5
# Returns properly parsed results: [{title, url, snippet}, ...]

Other formats (-f text, -f snapshot) are also affected - they return raw page dumps instead of extracted results.

Expected

Parsed individual results with title/url/snippet in daemon mode, same as without daemon.

Environment

• Version: 0.6.1

[Youhai020616]

Fixed in b96c9e9 🎉

Root cause: Daemon mode had no /extract route — search results fell back to raw text instead of running extractors server-side where the real page object exists.

Changes:

• daemon.js: added /extract route that runs extractors with the real page object
• client.js: added daemonExtract() client function
• search.js: daemon mode now calls /extract for proper result parsing

Fallback to raw text kept as safety net if extraction fails.

All 151 tests passing.