feat(agent): harden API-first service

This commit is contained in:
2026-09-16 21:12:47 +02:00
parent aa637270b7
commit c8b55fbd38
12 changed files with 839 additions and 162 deletions
+89 -74
View File
@@ -5,6 +5,7 @@ from __future__ import annotations
import json
import sqlite3
import threading
from dataclasses import dataclass, field
from pathlib import Path
@@ -66,11 +67,16 @@ class Retriever:
raise RuntimeError(
f"index fehlt ({self.db_path}) — zuerst 'python -m agent.cli ingest' ausführen"
)
self._con = sqlite3.connect(self.db_path)
# FastAPI fuehrt synchrone Endpunkte in Worker-Threads aus. Die
# Verbindung wird deshalb thread-uebergreifend verwendet; ein RLock
# serialisiert die kurzen SQLite-/Matrix-Abschnitte je Retriever.
self._lock = threading.RLock()
self._con = sqlite3.connect(self.db_path, check_same_thread=False)
self._con.row_factory = sqlite3.Row
self._client = client
self._owns_client = client is None
self._mat: np.ndarray | None = None
self._mat_norm: np.ndarray | None = None
self._mat_chunk_ids: list[int] | None = None
self._scope_cache: dict[str, set[int]] = {}
row = self._con.execute("SELECT MIN(stand), MAX(stand) FROM chunks").fetchone()
@@ -78,31 +84,33 @@ class Retriever:
self._stand_max = int((row[1] or "2026-01").replace("-", ""))
def close(self) -> None:
self._con.close()
if self._owns_client and self._client is not None:
self._client.close()
with self._lock:
self._con.close()
if self._owns_client and self._client is not None:
self._client.close()
# -- Index-Kennzahlen ---------------------------------------------------
def stats(self) -> dict:
n_chunks = self._con.execute("SELECT COUNT(*) FROM chunks").fetchone()[0]
n_entries = self._con.execute(
"SELECT COUNT(DISTINCT entry_id) FROM chunks"
).fetchone()[0]
n_vec = self._con.execute(
"SELECT COUNT(*) FROM vectors WHERE model = ?",
(self.cfg.embed_model,),
).fetchone()[0]
meta = dict(self._con.execute("SELECT key, value FROM meta").fetchall())
return {
"n_entries": n_entries,
"n_chunks": n_chunks,
"n_vectors": n_vec,
"dense_available": n_vec > 0 and not self.cfg.embed_off,
"stand_min": str(self._stand_min),
"stand_max": str(self._stand_max),
"built_at": meta.get("built_at"),
}
with self._lock:
n_chunks = self._con.execute("SELECT COUNT(*) FROM chunks").fetchone()[0]
n_entries = self._con.execute(
"SELECT COUNT(DISTINCT entry_id) FROM chunks"
).fetchone()[0]
n_vec = self._con.execute(
"SELECT COUNT(*) FROM vectors WHERE model = ?",
(self.cfg.embed_model,),
).fetchone()[0]
meta = dict(self._con.execute("SELECT key, value FROM meta").fetchall())
return {
"n_entries": n_entries,
"n_chunks": n_chunks,
"n_vectors": n_vec,
"dense_available": n_vec > 0 and not self.cfg.embed_off,
"stand_min": str(self._stand_min),
"stand_max": str(self._stand_max),
"built_at": meta.get("built_at"),
}
# -- Einzelverfahren ----------------------------------------------------
@@ -141,7 +149,12 @@ class Retriever:
if self.cfg.embed_off:
return {}
self._ensure_matrix()
if self._mat is None or len(self._mat) == 0:
if (
self._mat is None
or self._mat_norm is None
or self._mat_chunk_ids is None
or len(self._mat) == 0
):
return {}
if self._client is None:
self._client = OllamaClient(
@@ -166,7 +179,7 @@ class Retriever:
for i in order:
if len(out) >= limit:
break
cid = self._mat_chunk_ids[i]
cid = self._mat_chunk_ids[int(i)]
if scope_ids is not None and cid not in scope_ids:
continue
out[cid] = float(sims[i])
@@ -392,16 +405,17 @@ class Retriever:
deterministisch in Direktzahlung und Alternativen zerlegt. Das gilt
auch für den Offline-Retrieval-Eval, der keinen LLM-Planer aufruft.
"""
n = n_entries or self.cfg.context_blocks
latest_year = str(self._stand_max)[:4]
deterministic = decision_support_plan(question, default_year=latest_year)
if deterministic:
sub_queries, _qtype = deterministic
return self.search_multi(sub_queries, n_entries=n)
fused, bm_all, dn_all = self._fuse_queries(
[SubQuery(text=question)], self.cfg.candidate_pool
)
return self._context_from_fused(fused, n, bm_all, dn_all)
with self._lock:
n = n_entries or self.cfg.context_blocks
latest_year = str(self._stand_max)[:4]
deterministic = decision_support_plan(question, default_year=latest_year)
if deterministic:
sub_queries, _qtype = deterministic
return self.search_multi(sub_queries, n_entries=n)
fused, bm_all, dn_all = self._fuse_queries(
[SubQuery(text=question)], self.cfg.candidate_pool
)
return self._context_from_fused(fused, n, bm_all, dn_all)
def search_multi(
self, sub_queries: list, n_entries: int | None = None
@@ -412,44 +426,45 @@ class Retriever:
seiner besten Quelle hinein — sonst dominieren Einträge, die in
mehreren Sub-Queries mittelgut matchen). Temporal-Intent: kv-
Einträge im gefragten Geltungsjahr erhalten temporal_boost."""
n = n_entries or self.cfg.context_blocks
if (
len(sub_queries) == 1
and sub_queries[0].scope is None
and sub_queries[0].stand_year is None
):
return self.search(sub_queries[0].text, n_entries=n)
pool = self.cfg.candidate_pool
fused_total: dict[int, float] = {}
bm_all: set[int] = set()
dn_all: set[int] = set()
per_query_entries: list[list[str]] = []
for sq in sub_queries:
bm, dn, fused_q = self._search_one(sq, pool)
bm_all.update(bm)
dn_all.update(dn)
for cid, score in fused_q.items():
fused_total[cid] = fused_total.get(cid, 0.0) + score
rows_q = self._fetch_chunks(list(fused_q))
entry_best: dict[str, float] = {}
for cid, score in fused_q.items():
row = rows_q.get(cid)
if row is not None:
eid = row["entry_id"]
entry_best[eid] = max(entry_best.get(eid, 0.0), score)
per_query_entries.append(
[e for e, _ in sorted(entry_best.items(), key=lambda kv: -kv[1])]
with self._lock:
n = n_entries or self.cfg.context_blocks
if (
len(sub_queries) == 1
and sub_queries[0].scope is None
and sub_queries[0].stand_year is None
):
return self.search(sub_queries[0].text, n_entries=n)
pool = self.cfg.candidate_pool
fused_total: dict[int, float] = {}
bm_all: set[int] = set()
dn_all: set[int] = set()
per_query_entries: list[list[str]] = []
for sq in sub_queries:
bm, dn, fused_q = self._search_one(sq, pool)
bm_all.update(bm)
dn_all.update(dn)
for cid, score in fused_q.items():
fused_total[cid] = fused_total.get(cid, 0.0) + score
rows_q = self._fetch_chunks(list(fused_q))
entry_best: dict[str, float] = {}
for cid, score in fused_q.items():
row = rows_q.get(cid)
if row is not None:
eid = row["entry_id"]
entry_best[eid] = max(entry_best.get(eid, 0.0), score)
per_query_entries.append(
[e for e, _ in sorted(entry_best.items(), key=lambda kv: -kv[1])]
)
reserved: list[str] = []
for entries in per_query_entries:
taken = 0
for eid in entries:
if taken >= self.cfg.per_query_slots:
break
if eid not in reserved:
reserved.append(eid)
taken += 1
years = {sq.stand_year for sq in sub_queries if sq.stand_year}
return self._context_from_fused(
fused_total, n, bm_all, dn_all, stand_years=years, reserved=reserved
)
reserved: list[str] = []
for entries in per_query_entries:
taken = 0
for eid in entries:
if taken >= self.cfg.per_query_slots:
break
if eid not in reserved:
reserved.append(eid)
taken += 1
years = {sq.stand_year for sq in sub_queries if sq.stand_year}
return self._context_from_fused(
fused_total, n, bm_all, dn_all, stand_years=years, reserved=reserved
)