upbeatBytes/tests/test_admin.py

import pytest
from fastapi.testclient import TestClient


def _make(tmp_path, monkeypatch, admin_email=""):
    db = tmp_path / "t.sqlite3"
    monkeypatch.setenv("GOODNEWS_DB", str(db))
    monkeypatch.setenv("GOODNEWS_PUBLIC_BASE_URL", "http://testserver")
    monkeypatch.setenv("GOODNEWS_ADMIN_EMAILS", admin_email)
    import importlib
    import goodnews.api as api
    importlib.reload(api)
    from goodnews.db import connect, init_db
    c = connect(str(db)); init_db(c)
    c.execute("INSERT INTO sources (id,name,feed_url,trust_score) VALUES (1,'S','http://s/f',5)")
    c.execute("INSERT INTO articles (id,source_id,canonical_url,title,url_hash) VALUES (1,1,'http://s/1','t1','h1')")
    c.execute("INSERT INTO article_scores (article_id,accepted,topic) VALUES (1,1,'science')")
    c.execute("INSERT INTO article_tags (article_id,tag) VALUES (1,'science')")
    c.commit(); c.close()
    return api.create_app(), api


def _signin(app, api, email):
    tc = TestClient(app)
    sent = {}
    import goodnews.email_send as es
    orig = es.send_magic_link
    es.send_magic_link = lambda to, link: sent.update(link=link)
    try:
        tc.post("/api/auth/email/start", json={"email": email})
        tc.post("/api/auth/email/verify", json={"token": sent["link"].split("token=")[1]})
    finally:
        es.send_magic_link = orig
    return tc


def test_admin_gating(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    assert TestClient(app).get("/api/admin/stats").status_code == 401   # anon
    nonadmin = _signin(app, api, "rando@x.com")
    assert nonadmin.get("/api/admin/stats").status_code == 403          # signed in, not admin
    assert nonadmin.get("/api/auth/me").json()["is_admin"] is False
    admin = _signin(app, api, "Boss@X.com")                              # case-insensitive match
    assert admin.get("/api/auth/me").json()["is_admin"] is True
    assert admin.get("/api/admin/stats").status_code == 200


def test_admin_stats_shape(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    admin = _signin(app, api, "boss@x.com")
    # log a couple of events
    admin.post("/api/events", json={"kind": "visit", "visitor": "v1"})
    admin.post("/api/events", json={"kind": "open", "article_id": 1, "visitor": "v1"})
    stats = admin.get("/api/admin/stats").json()
    assert set(stats) >= {"visitors", "returning", "once", "top_articles", "top_groupings", "top_topics", "shares", "daily"}
    assert stats["top_articles"][0]["id"] == 1 and stats["top_articles"][0]["opens"] == 1
    assert any(g["tag"] == "science" for g in stats["top_groupings"])


def _src(tc, sid=1):
    return next(s for s in tc.get("/api/admin/stats").json()["sources"] if s["id"] == sid)


def test_source_lifecycle_status_and_visibility(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    tc = _signin(app, api, "boss@x.com")
    # pause → status paused + active mirror 0
    assert tc.post("/api/admin/sources/1/status", json={"status": "paused"}).json()["active"] == 0
    assert _src(tc)["status"] == "paused" and _src(tc)["active"] == 0
    # retire → status retired, still active=0, articles stay visible
    tc.post("/api/admin/sources/1/status", json={"status": "retired"})
    assert _src(tc)["status"] == "retired" and _src(tc)["active"] == 0
    assert _src(tc)["content_visible"] == 1  # retire does NOT hide content
    # restore → active again, mirror 1
    tc.post("/api/admin/sources/1/status", json={"status": "active"})
    assert _src(tc)["status"] == "active" and _src(tc)["active"] == 1
    # hide content → feed excludes it; show → back
    tc.post("/api/admin/sources/1/visibility", json={"visible": False})
    assert _src(tc)["content_visible"] == 0
    from goodnews import queries
    import sqlite3, os
    c = sqlite3.connect(os.environ["GOODNEWS_DB"]); c.row_factory = sqlite3.Row
    assert queries.feed(c) == []  # hidden source's article drops out of the feed
    tc.post("/api/admin/sources/1/visibility", json={"visible": True})
    c2 = sqlite3.connect(os.environ["GOODNEWS_DB"]); c2.row_factory = sqlite3.Row
    assert len(queries.feed(c2)) == 1  # back in the feed
    # validation + 404
    assert tc.post("/api/admin/sources/1/status", json={"status": "bogus"}).status_code == 422
    assert tc.post("/api/admin/sources/999/status", json={"status": "paused"}).status_code == 404


def test_source_flag_and_gating(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    tc = _signin(app, api, "boss@x.com")
    tc.post("/api/admin/sources/1/review", json={"flag": True, "reason": "spammy lately"})
    assert _src(tc)["review_flag"] == 1 and _src(tc)["review_reason"] == "spammy lately"
    tc.post("/api/admin/sources/1/review", json={"flag": False})
    assert _src(tc)["review_flag"] == 0 and _src(tc)["review_reason"] is None
    anon = TestClient(app)
    assert anon.post("/api/admin/sources/1/status", json={"status": "paused"}).status_code == 401
    assert anon.post("/api/admin/sources/1/visibility", json={"visible": False}).status_code == 401
    assert anon.post("/api/admin/sources/1/review", json={"flag": True}).status_code == 401


def test_source_health_includes_metrics(tmp_path, monkeypatch):
    import sqlite3
    from goodnews import queries
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    import os
    c = sqlite3.connect(os.environ["GOODNEWS_DB"]); c.row_factory = sqlite3.Row
    sh = queries.source_health(c)
    s = sh[0]
    for key in ("active", "served", "accepted_total", "total_articles", "duplicates",
                "acceptance_rate", "duplicate_rate", "review_reason", "next_due_at"):
        assert key in s, f"missing {key}"
    assert s["served"] == 1 and s["acceptance_rate"] == 100


def test_admin_stats_days_param_clamped(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    tc = _signin(app, api, "boss@x.com")
    assert tc.get("/api/admin/stats?days=7").json()["days"] == 7
    assert tc.get("/api/admin/stats?days=90").json()["days"] == 90
    assert tc.get("/api/admin/stats?days=999").json()["days"] == 30   # clamped
    assert tc.get("/api/admin/stats").json()["days"] == 30            # default


def test_candidate_suggest_promote_paused(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    monkeypatch.setattr(api.feeds, "preview_feed",
                        lambda url, **k: {"url": url, "sampled": 5, "accepted": 4, "examples_accepted": ["A", "B"]})
    tc = _signin(app, api, "boss@x.com")
    cand = tc.post("/api/admin/candidates", json={"feed_url": "http://good/feed", "name": "Good Feed"}).json()
    assert cand["status"] == "suggested" and cand["preview"]["accepted"] == 4
    cid = cand["id"]
    assert any(c["id"] == cid for c in tc.get("/api/admin/candidates").json())
    # promote defaults to paused (active-on-approval off) — no mirror drift
    res = tc.post(f"/api/admin/candidates/{cid}/promote", json={}).json()
    assert res["source"]["status"] == "paused" and res["source"]["active"] == 0
    assert res["candidate"]["status"] == "promoted"
    assert any(s["name"] == "Good Feed" for s in tc.get("/api/admin/stats").json()["sources"])


def test_candidate_reject_and_gating(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    monkeypatch.setattr(api.feeds, "preview_feed", lambda url, **k: {"url": url, "sampled": 1, "accepted": 0})
    tc = _signin(app, api, "boss@x.com")
    cand = tc.post("/api/admin/candidates", json={"feed_url": "http://meh/feed"}).json()
    assert tc.post(f"/api/admin/candidates/{cand['id']}/reject").json()["status"] == "rejected"
    anon = TestClient(app)
    assert anon.get("/api/admin/candidates").status_code == 401
    assert anon.post("/api/admin/candidates", json={"feed_url": "http://x/f"}).status_code == 401
    assert anon.post("/api/admin/candidates/1/promote", json={}).status_code == 401


def test_safe_fetch_feed_blocks_ssrf():
    import pytest
    from goodnews.feeds import safe_fetch_feed
    for bad in ("http://127.0.0.1/x", "http://localhost/x", "file:///etc/passwd",
                "http://169.254.169.254/latest", "ftp://x/y"):
        with pytest.raises(RuntimeError):
            safe_fetch_feed(bad, timeout=2)


def test_export_sources_csv(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    tc = _signin(app, api, "boss@x.com")
    r = tc.get("/api/admin/export/sources.csv")
    assert r.status_code == 200 and r.headers["content-type"].startswith("text/csv")
    assert 'attachment; filename="sources.csv"' in r.headers["content-disposition"]
    lines = r.text.splitlines()
    assert lines[0].startswith("name,feed_url,homepage,status,visible,served")
    assert any("http://s/f" in ln for ln in lines[1:])   # the seeded source row
    assert TestClient(app).get("/api/admin/export/sources.csv").status_code == 401  # gated


def test_export_audience_csv(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    tc = _signin(app, api, "boss@x.com")
    tc.post("/api/events", json={"kind": "visit", "visitor": "v1"})
    r = tc.get("/api/admin/export/audience.csv?days=7")
    assert r.status_code == 200 and r.headers["content-type"].startswith("text/csv")
    body = r.text
    assert "metric,value" in body and "window_days,7" in body
    assert "date,visitors,opens" in body                  # daily time-series section
    assert TestClient(app).get("/api/admin/export/audience.csv").status_code == 401  # gated


def test_export_sources_csv_escapes_formula_injection(tmp_path, monkeypatch):
    import os, sqlite3
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    c = sqlite3.connect(os.environ["GOODNEWS_DB"])
    c.execute("UPDATE sources SET name = ?, review_flag = 1, review_reason = ? WHERE id = 1",
              ('=HYPERLINK("http://bad")', '+danger'))
    c.commit(); c.close()
    tc = _signin(app, api, "boss@x.com")
    body = tc.get("/api/admin/export/sources.csv").text
    assert "'=HYPERLINK" in body   # leading apostrophe defuses the formula (CSV may quote the cell)
    assert "'+danger" in body
    assert ",=HYPERLINK" not in body  # never written as a bare, evaluable formula cell


def test_source_check_preview_is_readonly(tmp_path, monkeypatch):
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    monkeypatch.setattr(api.feeds, "preview_feed", lambda url, **k: {"url": url, "sampled": 8, "accepted": 6})
    tc = _signin(app, api, "boss@x.com")
    before = _src(tc)
    r = tc.post("/api/admin/sources/1/preview").json()
    assert r["sampled"] == 8 and r["accepted"] == 6
    after = _src(tc)
    # read-only: no state/health change, no poll attempt recorded
    assert after["status"] == before["status"] and after["served"] == before["served"]
    assert after["last_success_at"] == before["last_success_at"] and after["next_due_at"] == before["next_due_at"]
    assert TestClient(app).post("/api/admin/sources/1/preview").status_code == 401  # gated
    assert tc.post("/api/admin/sources/999/preview").status_code == 404


def test_digest_toggle_and_unsubscribe(tmp_path, monkeypatch):
    import os, sqlite3
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    tc = _signin(app, api, "reader@x.com")
    assert tc.get("/api/auth/me").json()["digest_enabled"] is False
    assert tc.post("/api/account/digest", json={"enabled": True}).json()["digest_enabled"] is True
    assert tc.get("/api/auth/me").json()["digest_enabled"] is True
    c = sqlite3.connect(os.environ["GOODNEWS_DB"])
    uid, tok = c.execute("SELECT id, digest_unsub_token FROM users WHERE email='reader@x.com'").fetchone()
    c.close()
    assert tok  # token generated on opt-in
    # one-click unsubscribe: wrong token is rejected, right token disables
    assert "invalid" in TestClient(app).get(f"/api/digest/unsubscribe?u={uid}&t=nope").text.lower()
    assert "unsubscribed" in TestClient(app).get(f"/api/digest/unsubscribe?u={uid}&t={tok}").text.lower()
    assert tc.get("/api/auth/me").json()["digest_enabled"] is False
    # RFC 8058 one-click POST also disables
    tc.post("/api/account/digest", json={"enabled": True})
    assert TestClient(app).post(f"/api/digest/unsubscribe?u={uid}&t={tok}").json() == {"ok": True}
    assert tc.get("/api/auth/me").json()["digest_enabled"] is False
    assert TestClient(app).post("/api/account/digest", json={"enabled": True}).status_code == 401  # gated


def test_follows_and_following_feed(tmp_path, monkeypatch):
    import os, sqlite3
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    c = sqlite3.connect(os.environ["GOODNEWS_DB"])
    c.execute("INSERT INTO sources (id,name,feed_url) VALUES (2,'Other','http://o/f')")
    c.execute("INSERT INTO articles (id,source_id,canonical_url,title,url_hash) VALUES (2,2,'http://o/2','t2','h2')")
    c.execute("INSERT INTO article_scores (article_id,accepted,topic) VALUES (2,1,'tech')")
    c.commit(); c.close()
    tc = _signin(app, api, "reader@x.com")
    # no follows yet → empty following feed (not an error)
    assert tc.get("/api/feed?following=true").json()["count"] == 0
    # follow source 1 → only its article
    assert tc.post("/api/follows", json={"kind": "source", "value": "1"}).json()["ok"] is True
    assert any(x["value"] == "1" and x["name"] == "S" for x in tc.get("/api/follows").json())
    assert [i["id"] for i in tc.get("/api/feed?following=true").json()["items"]] == [1]
    # follow source 2 too → both
    tc.post("/api/follows", json={"kind": "source", "value": "2"})
    assert {i["id"] for i in tc.get("/api/feed?following=true").json()["items"]} == {1, 2}
    # follow tag works too (article 1 carries 'science')
    tc.post("/api/follows", json={"kind": "tag", "value": "Science"})  # normalized lower
    assert any(x["kind"] == "tag" and x["value"] == "science" for x in tc.get("/api/follows").json())
    # unfollow source 2 (DELETE via query) → back to {1}
    tc.delete("/api/follows?kind=source&value=2")
    assert {i["id"] for i in tc.get("/api/feed?following=true").json()["items"]} == {1}
    # anon: following feed empty, follows API gated, bad source 404
    assert TestClient(app).get("/api/feed?following=true").json()["count"] == 0
    assert TestClient(app).get("/api/follows").status_code == 401
    assert tc.post("/api/follows", json={"kind": "source", "value": "999"}).status_code == 404


def test_since_endpoint(tmp_path, monkeypatch):
    import os, sqlite3
    app, api = _make(tmp_path, monkeypatch, admin_email="boss@x.com")
    c = sqlite3.connect(os.environ["GOODNEWS_DB"])
    for aid, when in [(2, "2020-01-01 00:00:00"), (3, "2030-01-01 00:00:00")]:
        c.execute("INSERT INTO articles (id,source_id,canonical_url,title,url_hash,discovered_at) VALUES (?,1,?,?,?,?)",
                  (aid, f"http://s/{aid}", f"t{aid}", f"h{aid}", when))
        c.execute("INSERT INTO article_scores (article_id,accepted) VALUES (?,1)", (aid,))
    c.commit(); c.close()
    tc = TestClient(app)
    r = tc.get("/api/since?ts=2027-01-01T00:00:00Z").json()
    assert r["count"] == 1 and [i["id"] for i in r["items"]] == [3]   # only the post-2027 article
    assert tc.get("/api/since?ts=2099-01-01T00:00:00Z").json()["count"] == 0  # nothing newer
    assert tc.get("/api/since?ts=not-a-date").json()["count"] == 0            # invalid ts → quiet 0


def test_puzzle_endpoint(tmp_path, monkeypatch):
    import os
    import sqlite3
    from goodnews import games
    from goodnews.localtime import local_today
    app, api = _make(tmp_path, monkeypatch)
    tc = TestClient(app)
    r = tc.get("/api/puzzle/word?variant=5").json()
    assert r["game"] == "word" and r["variant"] == "5" and r["length"] == 5 and r["guesses"] == 6
    assert "answer" not in r  # the public puzzle response never carries the answer
    assert tc.get("/api/puzzle/word?variant=6").json()["guesses"] == 7
    assert tc.get("/api/puzzle/word?variant=9").status_code == 404
    assert tc.get("/api/puzzle/wordsearch").status_code == 404

    # server-adjudicated guessing (answer revealed only on solve / exhaustion)
    c = sqlite3.connect(os.environ["GOODNEWS_DB"]); c.row_factory = sqlite3.Row
    ans = games.generate_word_puzzle(c, local_today(), "5")["answer"]
    mid = tc.post("/api/puzzle/word/guess", json={"variant": "5", "guess": "xxxxx", "n": 1}).json()
    assert len(mid["colors"]) == 5 and mid["solved"] is False and mid["answer"] is None
    win = tc.post("/api/puzzle/word/guess", json={"variant": "5", "guess": ans, "n": 2}).json()
    assert win["solved"] is True and win["answer"] == ans and all(x == "correct" for x in win["colors"])
    last = tc.post("/api/puzzle/word/guess", json={"variant": "5", "guess": "xxxxx", "n": 6}).json()
    assert last["answer"] == ans  # exhausting guesses reveals it even when wrong
    # wrong length → 400
    assert tc.post("/api/puzzle/word/guess", json={"variant": "5", "guess": "toolong", "n": 1}).status_code == 400