cayce
73ccdf7bdf
ci / lint-test-leaks (pull_request) Canceled after 0s
scanner (read-only): - sources from CAYCE_OPS_SCANNER_SOURCES JSON (rss + page types), stdlib only - keyword scoring -> risk_score; kind via heuristics (affiliate/partnership/content_gap) - idempotent: sha1 dedupe_key; repeated scans skip existing - every proposal audited (actor daemon:scanner); source errors recorded, never fatal executor (human gate): - approve/reject audited with operator identity; only proposed -> approved - AUTONOMOUS_MODE=true adds exactly one path: policy auto-approval (kinds allowlist + daily cap, actor auto:policy) — DECISIONS D5 - plugin registry by kind; unknown kinds fall back to log_action (D9) - retries with max_attempts, then failed; every transition audited - http_post plugin inert unless CAYCE_OPS_ENABLE_HTTP_ACTIONS=true tests: 13 new (dedupe, keyword filter, autonomy bounds, retry-then-fail, blocked http_post, fallback plugin) — 20 total passing
113 lines
3.6 KiB
Python
113 lines
3.6 KiB
Python
"""scanner tests — discovery, classification, dedupe. No network (injected fetch)."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import json
|
|
|
|
from cayceops.components.scanner import scanner
|
|
from cayceops.models import Proposal, ProposalStatus
|
|
|
|
RSS = b"""<?xml version="1.0"?>
|
|
<rss><channel>
|
|
<item><title>Acme launches affiliate program for hotels</title>
|
|
<link>https://acme.example.org/affiliates</link>
|
|
<description>Earn commission on every booking.</description></item>
|
|
<item><title>Weekly crypto digest</title>
|
|
<link>https://news.example.org/digest</link>
|
|
<description>Everything unrelated.</description></item>
|
|
</channel></rss>"""
|
|
|
|
PAGE = b"""<html><head><title>Partners</title></head><body>
|
|
<a href="https://pay.example.org/partner">Become a payment partner</a>
|
|
<a href="https://pay.example.org/docs">API docs</a>
|
|
</body></html>"""
|
|
|
|
|
|
def _fake_fetch(responses):
|
|
def fetch(url, cfg, timeout=20.0):
|
|
return responses[url]
|
|
|
|
return fetch
|
|
|
|
|
|
SOURCES = [
|
|
{"type": "rss", "url": "https://feeds.example.org/news.xml"},
|
|
{"type": "page", "url": "https://pay.example.org/partners"},
|
|
]
|
|
|
|
|
|
def _cfg(cfg, **kw):
|
|
return cfg(
|
|
CAYCE_OPS_SCANNER_SOURCES=json.dumps(SOURCES),
|
|
CAYCE_OPS_SCANNER_KEYWORDS="affiliate,partner,bitcoin,commission",
|
|
**kw,
|
|
)
|
|
|
|
|
|
def test_scan_files_proposals_with_kind_and_score(cfg, sess):
|
|
config = _cfg(cfg)
|
|
stats = scanner.scan(config, sess, fetch=_fake_fetch({
|
|
"https://feeds.example.org/news.xml": RSS,
|
|
"https://pay.example.org/partners": PAGE,
|
|
}))
|
|
|
|
assert stats["proposed"] == 3 # affiliate item + both page links (pages keep all)
|
|
proposals = sess.query(Proposal).all()
|
|
by_title = {p.title: p for p in proposals}
|
|
|
|
affiliate = by_title["Acme launches affiliate program for hotels"]
|
|
assert affiliate.kind == "affiliate"
|
|
assert affiliate.status == ProposalStatus.PROPOSED
|
|
assert affiliate.risk_score > 0
|
|
assert affiliate.dedupe_key
|
|
|
|
partner = by_title["Become a payment partner"]
|
|
assert partner.kind == "partnership"
|
|
|
|
|
|
def test_scan_is_idempotent_via_dedupe(cfg, sess):
|
|
config = _cfg(cfg)
|
|
scanner.scan(config, sess, fetch=_fake_fetch({
|
|
"https://feeds.example.org/news.xml": RSS,
|
|
"https://pay.example.org/partners": PAGE,
|
|
}))
|
|
stats2 = scanner.scan(config, sess, fetch=_fake_fetch({
|
|
"https://feeds.example.org/news.xml": RSS,
|
|
"https://pay.example.org/partners": PAGE,
|
|
}))
|
|
assert stats2["proposed"] == 0
|
|
assert stats2["skipped_existing"] == 3
|
|
|
|
|
|
def test_scan_filters_keywordless_feed_items(cfg, sess):
|
|
config = _cfg(cfg)
|
|
scanner.scan(config, sess, fetch=_fake_fetch({
|
|
"https://feeds.example.org/news.xml": RSS,
|
|
"https://pay.example.org/partners": PAGE,
|
|
}))
|
|
titles = {p.title for p in sess.query(Proposal).all()}
|
|
assert "Weekly crypto digest" not in titles # no keyword match in feed -> dropped
|
|
|
|
|
|
def test_scan_records_audit_for_each_proposal(cfg, sess):
|
|
from cayceops.models import AuditEntry
|
|
|
|
config = _cfg(cfg)
|
|
scanner.scan(config, sess, fetch=_fake_fetch({
|
|
"https://feeds.example.org/news.xml": RSS,
|
|
"https://pay.example.org/partners": PAGE,
|
|
}))
|
|
acts = [a.action for a in sess.query(AuditEntry).all()]
|
|
assert acts.count("propose") == 3
|
|
assert all(a.actor == "daemon:scanner" for a in sess.query(AuditEntry).all())
|
|
|
|
|
|
def test_scan_survives_source_error(cfg, sess):
|
|
def broken_fetch(url, cfg, timeout=20.0):
|
|
raise OSError("connection refused")
|
|
|
|
config = _cfg(cfg)
|
|
stats = scanner.scan(config, sess, fetch=broken_fetch)
|
|
assert stats["proposed"] == 0
|
|
assert len(stats["source_errors"]) == 2
|