news-mcp/news_mcp/enrichment/llm_enrich.py

from __future__ import annotations

from fnmatch import fnmatchcase
from typing import Any, Dict

from news_mcp.config import NEWS_ENTITY_BLACKLIST
from news_mcp.entity_normalize import normalize_entities
from news_mcp.llm import call_extraction, call_summary


def _matches_blacklist(value: str, blacklist=None) -> bool:
    patterns = [x.strip().lower() for x in (blacklist if blacklist is not None else NEWS_ENTITY_BLACKLIST) if x and x.strip()]
    key = str(value).strip().lower()
    if not key:
        return True
    return any(fnmatchcase(key, pattern) for pattern in patterns)


def _filter_entities(entities, blacklist=None):
    out = []
    for ent in entities or []:
        if _matches_blacklist(ent, blacklist=blacklist):
            continue
        out.append(ent)
    return out


async def classify_cluster_groq(cluster: Dict[str, Any]) -> Dict[str, Any]:
    parsed = await call_extraction(cluster)
    out = dict(cluster)
    topic = parsed.get("topic", cluster.get("topic"))
    if topic and _matches_blacklist(topic):
        topic = "other"
    out.update({
        "topic": topic,
        "entities": normalize_entities(_filter_entities(parsed.get("entities", []))),
        "sentiment": parsed.get("sentiment", "neutral"),
        "sentimentScore": parsed.get("sentimentScore"),
        "keywords": normalize_entities(_filter_entities(parsed.get("keywords", []))),
    })
    return out


async def summarize_cluster_groq(cluster: Dict[str, Any]) -> Dict[str, Any]:
    parsed = await call_summary(cluster)
    return parsed