# ==============================================================================
# Author: Simon-Pierre Boucher <contact@spboucher.ai>
# File:   tests/test_vague3.py
# Desc:   Tests vague 3 — Twitch GQL public sans clé, stats YouTube sans clé
#         (page + InnerTube), découverte Wikipédia (catégories + Wikidata),
#         palmarès Apple Podcasts Canada
# ==============================================================================
import json
from datetime import datetime, timezone

from creaka.connectors.palmares_apple import feed_head_is_quebec
from creaka.connectors.twitch import gql_user_query, parse_gql_user
from creaka.connectors.wikipedia_qc import (accounts_from_claims, birth_year,
                                            is_deceased)
from creaka.connectors.youtube import ago_to_iso, parse_about, parse_channel_page
from creaka.connectors.youtube_recherche import is_quebec


# --- marqueurs QC (régression vague 3) ---------------------------------------------

def test_is_quebec_frontieres_de_mot():
    # « levis » matchait dans « television », « estrie » dans « tapestries »
    assert not is_quebec("Canadian television journalist")
    assert not is_quebec("She makes tapestries and pottery.")
    assert is_quebec("Humoriste de Lévis")
    assert is_quebec("Festival en Estrie chaque été")
    assert is_quebec("vlogueuse québécoise à Montréal")


# --- twitch (GQL public) ----------------------------------------------------------

_GQL_OK = {"data": {"user": {
    "id": "71092938", "login": "xqc", "displayName": "xQc",
    "description": "Streamer québécois.",
    "profileImageURL": "https://static-cdn.jtvnw.net/x-300x300.jpeg",
    "followers": {"totalCount": 12535966},
    "roles": {"isPartner": True, "isAffiliate": False},
    "lastBroadcast": {"startedAt": "2026-08-18T08:03:23Z",
                      "game": {"displayName": "Just Chatting"}}}}}


def test_twitch_parse_gql_user():
    info = parse_gql_user(_GQL_OK)
    assert info["followers"] == 12535966
    assert info["partner"] is True and info["affiliate"] is False
    assert info["bio"] == "Streamer québécois."
    assert info["avatar"].startswith("https://")
    assert info["last_stream"] == "2026-08-18T08:03:23Z"
    assert info["game"] == "Just Chatting"


def test_twitch_parse_gql_user_inexistant():
    # login libre/renommé : data.user = null → None (jamais d'exception)
    assert parse_gql_user({"data": {"user": None}}) is None
    assert parse_gql_user({}) is None


def test_twitch_gql_query_echappe_le_login():
    q = gql_user_query('haxor"} { evil')["query"]
    assert '\\"' in q  # json.dumps a échappé la tentative d'injection
    assert json.loads("{\"q\": " + q.split("login: ", 1)[1].split(")")[0] + "}")


# --- youtube sans clé --------------------------------------------------------------

def test_ago_to_iso():
    now = datetime(2026, 8, 18, tzinfo=timezone.utc)
    assert ago_to_iso("3 days ago", now) == "2026-08-15"
    assert ago_to_iso("Streamed 2 weeks ago", now) == "2026-08-04"
    assert ago_to_iso("1 year ago", now) == "2025-08-18"
    assert ago_to_iso("n'importe quoi") is None


_VIEW_MODEL = {
    "metadata": {"contentMetadataViewModel": {"metadataRows": [
        {"metadataParts": [{"text": {"content": "@chainetest"}}]},
        {"metadataParts": [{"text": {"content": "1.2M subscribers"}},
                           {"text": {"content": "266 videos"}}]},
    ]}},
    "banner": {"imageBannerViewModel": {"image": {"sources": [
        {"url": "https://yt3.example/banner.jpg", "width": 2120}]}}},
    "description": {"descriptionPreviewViewModel": {"rendererContext": {
        "commandContext": {"onTap": {"innertubeCommand": {
            "showEngagementPanelEndpoint": {"engagementPanel": {
                "continuationItemRenderer": {"continuationEndpoint": {
                    "continuationCommand": {"token": "TOK_ABOUT"}}}}}}}}}}},
}
_YTID = {
    "metadata": {"channelMetadataRenderer": {
        "title": "Chaîne Test", "externalId": "UC" + "a" * 22,
        "description": "Vlog québécois.",
        "avatar": {"thumbnails": [
            {"url": "https://yt3.example/avatar.jpg", "width": 160}]}}},
    "header": {"pageHeaderRenderer": {
        "content": {"pageHeaderViewModel": _VIEW_MODEL}}},
}
_PAGE = ("<html><script>var config = {\"INNERTUBE_API_KEY\":\"AIzaKEY\","
         "\"INNERTUBE_CONTEXT_CLIENT_VERSION\":\"2.2026\"};</script>"
         "<script>var ytInitialData = " + json.dumps(_YTID)
         + ";</script></html>")


def test_parse_channel_page():
    page = parse_channel_page(_PAGE)
    assert page["followers"] == 1_200_000
    assert page["videos"] == 266
    assert page["channel_id"] == "UC" + "a" * 22
    assert page["description"] == "Vlog québécois."
    assert page["avatar"] == "https://yt3.example/avatar.jpg"
    assert page["banner"] == "https://yt3.example/banner.jpg"
    assert page["about_token"] == "TOK_ABOUT"
    assert page["key"] == "AIzaKEY" and page["ver"] == "2.2026"


def test_parse_channel_page_illisible():
    assert parse_channel_page("<html>rien</html>") is None


def test_parse_about():
    text = ('{"aboutChannelViewModel":{"viewCountText":"687,115,991 views",'
            '"videoCountText":"266 videos","country":{"content":"Canada"}}}')
    about = parse_about(text)
    assert about == {"views": 687_115_991, "videos": 266, "country": "Canada"}
    # réponse sans panneau À propos (jeton périmé) → None, jamais de bruit
    assert parse_about('{"contents":{}}') is None


# --- wikipedia-qc ------------------------------------------------------------------

def _claim(value):
    return {"mainsnak": {"datavalue": {"value": value}}}


def test_accounts_from_claims():
    claims = {
        "P2003": [_claim("jadelavoie")],                 # instagram
        "P2397": [_claim("UC" + "b" * 22)],              # chaîne youtube
        "P856": [_claim("https://www.exemple.qc.ca/")],  # site officiel
    }
    accounts = {a.platform: a for a in accounts_from_claims(claims)}
    assert accounts["instagram"].handle == "jadelavoie"
    assert accounts["instagram"].signal == "base_publique"
    assert accounts["youtube"].url.endswith("/channel/UC" + "b" * 22)
    assert accounts["site-web"].handle == "exemple.qc.ca"
    assert all(a.confidence == 0.85 for a in accounts.values())


def test_birth_year_et_deces():
    claims = {"P569": [_claim({"time": "+1972-01-14T00:00:00Z"})]}
    assert birth_year(claims) == 1972
    assert not is_deceased(claims)
    assert is_deceased({"P570": [_claim({"time": "+2001-01-01T00:00:00Z"})]})
    assert birth_year({}) is None


# --- palmares-apple ----------------------------------------------------------------

_RSS_QC = ("""<rss><channel><title>Mon balado</title>
<language>fr-ca</language>
<description>Discussions à Montréal chaque semaine.</description>
<itunes:author>Équipe</itunes:author>
<item><title>Épisode 1</title></item></channel></rss>""")

_RSS_HORS_QC = ("""<rss><channel><title>The Daily</title>
<language>en-us</language>
<description>This is what the news should sound like.</description>
<item><title>Episode about montreal bagels</title></item></channel></rss>""")

# balado ANGLO qui mentionne Montréal dans sa description (dates de tournée) :
# la langue non francophone doit suffire à l'écarter (faux positif vague 3)
_RSS_TOURNEE = ("""<rss><channel><title>Comedy Show</title>
<language>en</language>
<description>Live shows in Toronto, Montreal and Vancouver!</description>
<item><title>Ep 1</title></item></channel></rss>""")


def test_feed_head_is_quebec_langue_et_marqueur_exiges():
    assert feed_head_is_quebec(_RSS_QC)
    # le marqueur dans un ÉPISODE ne suffit jamais (en-tête du canal seulement)
    assert not feed_head_is_quebec(_RSS_HORS_QC)
    # marqueur dans l'en-tête MAIS canal anglophone → jamais rattaché
    assert not feed_head_is_quebec(_RSS_TOURNEE)
    assert not feed_head_is_quebec("")