"""

Tests for ingest.source_validator  -  neutrality rule

Every stored fact MUST carry a source_url whose host is an allowlisted .gov domain.

"""

import sqlite3

import tempfile

import os

import pytest

from ingest.source_validator import (

    ALLOWLISTED_HOSTS,

    SourceURLError,

    is_allowlisted,

    validate_source_url,

)

from ingest.db import get_connection, init_schema

class TestIsAllowlisted:

    def test_explicit_allowlist_hosts(self):

        for host in ALLOWLISTED_HOSTS:

            url = f"https://{host}/some/path"

            assert is_allowlisted(url), f"Expected {host!r} to be allowlisted"

    def test_any_gov_domain_is_allowed(self):

        gov_urls = [

            "https://www.usaspending.gov/award/CONT_AWD_N0001923C0001",

            "https://congress.gov/bill/119th-congress/house-bill/1",

            "https://fiscaldata.treasury.gov/datasets/debt-to-the-penny/",

            "https://clerk.house.gov/evs/2026/ROLL_001.xml",

            "https://www.senate.gov/legislative/LIS/roll_call_votes/vote1191/vote_119_1_00001.xml",

            "https://leg.colorado.gov/bills/hb25-1001",

            "https://legiscan.gov/somepage",

            "https://www.congress.gov/bill/119th-congress/senate-bill/99",

        ]

        for url in gov_urls:

            assert is_allowlisted(url), f"Expected .gov URL to be allowed: {url}"

    def test_non_gov_domains_rejected(self):

        bad_urls = [

            "https://www.nytimes.com/article/congress-bill",

            "https://propublica.org/data",

            "https://govtrack.us/congress/bills",

            "https://ballotpedia.org/H.R._1",

            "https://en.wikipedia.org/wiki/U.S._Congress",

            "https://politico.com/story",

            "https://api.example.com/bills",

            "",

        ]

        for url in bad_urls:

            assert not is_allowlisted(url), f"Expected {url!r} to be rejected"

    def test_empty_string_rejected(self):

        assert not is_allowlisted("")

    def test_none_like_string_rejected(self):

        assert not is_allowlisted("None")

    def test_subdomain_gov_allowed(self):

        assert is_allowlisted("https://api.congress.gov/v3/bill")

        assert is_allowlisted("https://api.fiscaldata.treasury.gov/services/api/fiscal_service/v1/accounting/od/debt_to_penny")

    def test_gov_in_path_not_in_host_rejected(self):

        assert not is_allowlisted("https://example.com/gov/bills")

class TestValidateSourceUrl:

    def test_valid_gov_url_does_not_raise(self):

        validate_source_url("https://www.usaspending.gov/award/CONT_AWD_TEST")

        validate_source_url("https://congress.gov/bill/119th-congress/house-bill/1")

    def test_invalid_url_raises_source_url_error(self):

        with pytest.raises(SourceURLError):

            validate_source_url("https://www.reuters.com/article/xyz")

    def test_empty_url_raises(self):

        with pytest.raises(SourceURLError):

            validate_source_url("")

    def test_non_gov_subdomain_raises(self):

        with pytest.raises(SourceURLError):

            validate_source_url("https://api.example.io/bills")

    def test_error_message_contains_host(self):

        with pytest.raises(SourceURLError, match="nytimes.com"):

            validate_source_url("https://www.nytimes.com/article/bill")

class TestDatabaseSourceUrlConstraint:

    """

    Integration test: simulate writing rows to SQLite and verify that any row

    whose source_url is not allowlisted would have been caught by validate_source_url

    before insertion.

    """

    def _make_db(self):

        tmp = tempfile.mktemp(suffix=".db")

        conn = get_connection(tmp)

        init_schema(conn)

        return conn, tmp

    def test_awards_with_valid_source_url_stored_successfully(self):

        conn, tmp = self._make_db()

        source_url = "https://www.usaspending.gov/award/CONT_AWD_TEST001"

        validate_source_url(source_url)

        conn.execute(

            "INSERT INTO awards (source, external_id, agency, recipient, amount, action_date, program, source_url) "

            "VALUES (?, ?, ?, ?, ?, ?, ?, ?)",

            ("usaspending", "TEST001", "DoD", "ACME Corp", 1_000_000, "2026-06-25", "contract", source_url),

        )

        conn.commit()

        row = conn.execute("SELECT source_url FROM awards WHERE external_id='TEST001'").fetchone()

        assert row is not None

        assert is_allowlisted(row["source_url"])

        conn.close()

        os.unlink(tmp)

    def test_awards_with_invalid_source_url_blocked(self):

        conn, tmp = self._make_db()

        bad_url = "https://www.politico.com/award/XYZ"

        with pytest.raises(SourceURLError):

            validate_source_url(bad_url)

            conn.execute(

                "INSERT INTO awards (source, external_id, agency, recipient, amount, action_date, program, source_url) "

                "VALUES (?, ?, ?, ?, ?, ?, ?, ?)",

                ("test", "BAD001", "Agency", "Bad Corp", 100, "2026-06-25", "contract", bad_url),

            )

        conn.close()

        os.unlink(tmp)

    def test_all_awards_from_usaspending_have_gov_source_url(self):

        """

        Simulate a batch of awards as the poller would produce them and confirm

        every source_url passes the allowlist check.

        """

        internal_ids = [

            "CONT_AWD_N0001923C0001",

            "CONT_AWD_W912HV23C0001",

            "ASST_NON_2023MU00009",

        ]

        for iid in internal_ids:

            url = f"https://www.usaspending.gov/award/{iid}"

            assert is_allowlisted(url), f"USAspending award URL failed allowlist: {url}"

    def test_bills_from_congress_have_gov_source_url(self):

        urls = [

            "https://www.congress.gov/bill/119th-congress/house-bill/1",

            "https://www.congress.gov/bill/119th-congress/senate-bill/99",

        ]

        for url in urls:

            assert is_allowlisted(url)

    def test_fiscal_from_treasury_has_gov_source_url(self):

        url = "https://fiscaldata.treasury.gov/datasets/debt-to-the-penny/debt-to-the-penny"

        assert is_allowlisted(url)

    def test_house_votes_have_gov_source_url(self):

        url = "https://clerk.house.gov/evs/2026/ROLL_001.xml"

        assert is_allowlisted(url)

    def test_senate_votes_have_gov_source_url(self):

        url = "https://www.senate.gov/legislative/LIS/roll_call_votes/vote1191/vote_119_1_00001.xml"

        assert is_allowlisted(url)