121 lines
4.4 KiB
Python
121 lines
4.4 KiB
Python
"""催化事件存储与主题分数聚合。"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import logging
|
|
from datetime import datetime, timedelta
|
|
|
|
from sqlalchemy import text
|
|
|
|
from app.catalyst.mapper import analyze_catalyst
|
|
from app.catalyst.models import CatalystAnalysis, CatalystInput, ThemeCatalystScore
|
|
from app.db import tables
|
|
from app.db.database import get_db
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
async def ingest_catalyst(item: CatalystInput, use_llm: bool = True) -> CatalystAnalysis:
|
|
analysis = await analyze_catalyst(item, use_llm=use_llm)
|
|
await save_catalyst(analysis)
|
|
return analysis
|
|
|
|
|
|
async def save_catalyst(analysis: CatalystAnalysis) -> int:
|
|
async with get_db() as db:
|
|
result = await db.execute(
|
|
tables.catalysts_table.insert().values(
|
|
title=analysis.title,
|
|
summary=analysis.summary,
|
|
source=analysis.source,
|
|
url=analysis.url,
|
|
published_at=analysis.published_at,
|
|
catalyst_type=analysis.catalyst_type,
|
|
strength=analysis.strength,
|
|
freshness=analysis.freshness,
|
|
confidence=analysis.confidence,
|
|
raw_text=analysis.raw_text,
|
|
llm_reason=analysis.llm_reason,
|
|
is_active=True,
|
|
)
|
|
)
|
|
catalyst_id = int(result.inserted_primary_key[0])
|
|
if analysis.themes:
|
|
await db.execute(
|
|
tables.theme_catalysts_table.insert(),
|
|
[
|
|
{
|
|
"catalyst_id": catalyst_id,
|
|
"theme_id": theme.theme_id,
|
|
"theme_name": theme.theme_name,
|
|
"relevance": theme.relevance,
|
|
"reason": theme.reason,
|
|
}
|
|
for theme in analysis.themes
|
|
],
|
|
)
|
|
await db.commit()
|
|
return catalyst_id
|
|
|
|
|
|
async def get_recent_catalysts(limit: int = 30, hours: int = 72) -> list[dict]:
|
|
since = datetime.now() - timedelta(hours=hours)
|
|
async with get_db() as db:
|
|
result = await db.execute(
|
|
text(
|
|
"SELECT c.*, "
|
|
"GROUP_CONCAT(tc.theme_name || ':' || ROUND(tc.relevance, 0), ',') AS themes "
|
|
"FROM catalysts c "
|
|
"LEFT JOIN theme_catalysts tc ON tc.catalyst_id = c.id "
|
|
"WHERE c.is_active = 1 AND COALESCE(c.published_at, c.created_at) >= :since "
|
|
"GROUP BY c.id "
|
|
"ORDER BY COALESCE(c.published_at, c.created_at) DESC, c.id DESC "
|
|
"LIMIT :limit"
|
|
),
|
|
{"since": since, "limit": limit},
|
|
)
|
|
rows = result.mappings().all()
|
|
return [dict(row) for row in rows]
|
|
|
|
|
|
async def build_theme_catalyst_scores(hours: int = 72, limit: int = 20) -> list[ThemeCatalystScore]:
|
|
since = datetime.now() - timedelta(hours=hours)
|
|
async with get_db() as db:
|
|
rows = (
|
|
await db.execute(
|
|
text(
|
|
"SELECT tc.theme_id, tc.theme_name, "
|
|
"COUNT(*) AS catalyst_count, "
|
|
"SUM((c.strength * 0.45 + c.freshness * 0.25 + c.confidence * 0.15 + tc.relevance * 0.15)) AS raw_score, "
|
|
"GROUP_CONCAT(SUBSTR(COALESCE(tc.reason, c.summary, c.title), 1, 60), ' | ') AS reasons "
|
|
"FROM theme_catalysts tc "
|
|
"JOIN catalysts c ON c.id = tc.catalyst_id "
|
|
"WHERE c.is_active = 1 AND COALESCE(c.published_at, c.created_at) >= :since "
|
|
"GROUP BY tc.theme_id, tc.theme_name "
|
|
"ORDER BY raw_score DESC "
|
|
"LIMIT :limit"
|
|
),
|
|
{"since": since, "limit": limit},
|
|
)
|
|
).mappings().all()
|
|
|
|
scores = []
|
|
for row in rows:
|
|
raw = float(row.get("raw_score") or 0)
|
|
count = int(row.get("catalyst_count") or 0)
|
|
normalized = min(raw / max(count, 1), 100)
|
|
reasons = [
|
|
item.strip()
|
|
for item in str(row.get("reasons") or "").split("|")
|
|
if item.strip()
|
|
][:3]
|
|
scores.append(ThemeCatalystScore(
|
|
theme_id=row["theme_id"],
|
|
theme_name=row["theme_name"],
|
|
catalyst_score=round(normalized, 1),
|
|
catalyst_count=count,
|
|
top_reasons=reasons,
|
|
generated_by="catalyst_layer",
|
|
))
|
|
return scores
|