"""Qwen (Alibaba) — official blog RSS → ANNOUNCEMENT / RELEASE events; release posts followed for LLM extraction.

Almost every Qwen post announces a model (Qwen3-…, Qwen-Image, Qwen3Guard…), so items whose title names a model are treated as releases
and their article is queued with `llm_task: release_announcement` (the model passport is filled by the LLM stage, never guessed here).
"""
from __future__ import annotations

import re

from aiatlas.registry import org_ref
from aiatlas.sdk.connector import BaseConnector, Parsed, RunContext
from aiatlas.sdk.facts import Facts, Target
from aiatlas.sdk.fetch import FetchResult

from ._common import RELEASE_WORDS

FEED = "https://qwenlm.github.io/blog/index.xml"
MODEL_IN_TITLE = re.compile(r"\b(Qwen[\w.\-]*|QwQ[\w.\-]*|QVQ[\w.\-]*|Qwen-Image[\w.\-]*|Qwen-Agent|Wan[\w.\-]*)\b", re.IGNORECASE)
MAX_FOLLOW = 30


class QwenConnector(BaseConnector):
    name = "qwen"
    label = "Qwen — blog feed"
    description = "Qwen team blog (qwenlm.github.io) RSS: model release posts as RELEASE announcements queued for LLM extraction."
    source_key = "qwenlm.github.io"
    version = "1"
    parser_version = "1"
    interval_seconds = 3600
    min_interval_seconds = 1800
    rate_per_min = 10
    tier = 1
    priority = 0
    expected_min_records = 1
    concurrency = 2

    async def discover(self, ctx: RunContext) -> list[Target]:
        return [Target(url=FEED, doc_type="feed", key="feed", min_bytes=1000)]

    async def extract(self, ctx: RunContext, target: Target, res: FetchResult, parsed: Parsed) -> Facts:
        facts = Facts()
        org = org_ref("qwen")
        facts.entities.append(org)
        if parsed.kind != "feed":
            return facts
        for i, it in enumerate(parsed.feed_items):
            if not it.url or not it.title:
                continue
            names = sorted({m.group(1) for m in MODEL_IN_TITLE.finditer(it.title)})
            is_release = bool(names) or bool(RELEASE_WORDS.search(f"{it.title} {it.summary or ''}"))
            facts.event("ANNOUNCEMENT", "release" if is_release else "company", f"Qwen: {it.title}", entity=org, importance=2 if is_release else 1,
                        effective_at=it.published_at, dedupe_key=f"ANNOUNCEMENT:{it.url}", source_url=it.url,
                        meta={"source": "qwenlm.github.io/blog", "summary": (it.summary or "")[:300], "models_mentioned": names[:5], "is_release": is_release})
            if i < MAX_FOLLOW:
                facts.follow(it.url, doc_type="news", needs_llm=is_release, priority=1 if is_release else 3,
                             meta={"llm_task": "release_announcement", "published_at": it.published_at.isoformat() if it.published_at else None, "title": it.title,
                                   "models_mentioned": names[:5]})
        facts.document_title, facts.document_entity = "Qwen blog", org
        return facts


CONNECTORS = [QwenConnector]