feat(mcp): tools stats_acheteur / stats_titulaire (agrégations)

This commit is contained in:
Colin Maudry
2026-07-09 20:30:16 +02:00
parent 11da1e9d8d
commit ee98aa8186
2 changed files with 126 additions and 2 deletions
+92 -1
View File
@@ -2,7 +2,7 @@
import re import re
from src.api.filters import FilterError, build_where from src.api.filters import FilterError, build_where
from src.db import count_marches, query_marches from src.db import aggregate_marches, count_marches, query_marches
from src.db import schema as duckdb_schema from src.db import schema as duckdb_schema
from src.mcp.serialization import to_json_records from src.mcp.serialization import to_json_records
from src.utils.data import DF_ACHETEURS, DF_TITULAIRES from src.utils.data import DF_ACHETEURS, DF_TITULAIRES
@@ -129,3 +129,94 @@ def search_organisations(
} }
for r in df.to_dicts() for r in df.to_dicts()
] ]
def _org_identite(org_type: str, org_id: str) -> dict:
df = query_marches(
where_sql=f'"{org_type}_id" = ?',
params=[org_id],
columns=[
f"{org_type}_id",
f"{org_type}_nom",
f"{org_type}_departement_nom",
f"{org_type}_commune_nom",
],
limit=1,
)
if df.height == 0:
return {"id": org_id, "nom": None, "departement": None, "commune": None}
r = df.row(0, named=True)
return {
"id": org_id,
"nom": r[f"{org_type}_nom"],
"departement": r[f"{org_type}_departement_nom"],
"commune": r[f"{org_type}_commune_nom"],
}
def compute_org_stats(org_type: str, org_id: str) -> dict:
"""Statistiques agrégées d'un acheteur ou titulaire."""
if org_type not in ORG_FRAMES:
raise ValueError(
f"type invalide: {org_type!r} (attendu 'acheteur' ou 'titulaire')"
)
other = "titulaire" if org_type == "acheteur" else "acheteur"
where_sql = f'"{org_type}_id" = ?'
params = [org_id]
identite = _org_identite(org_type, org_id)
totals = aggregate_marches(
select_sql='COUNT("uid") AS nb, COALESCE(SUM("montant"), 0) AS montant_total',
where_sql=where_sql,
params=params,
)
nb = int(totals["nb"][0])
if nb == 0:
return {
"identite": identite,
"nb_marches": 0,
"montant_total": 0,
"repartition_annuelle": [],
f"top_{other}s": [],
"top_cpv": [],
}
annuelle = aggregate_marches(
select_sql=(
"CAST(date_part('year', \"dateNotification\") AS INTEGER) AS annee, "
'COUNT("uid") AS nb_marches, '
'COALESCE(SUM("montant"), 0) AS montant_total'
),
where_sql=where_sql,
params=params,
group_by="date_part('year', \"dateNotification\")",
order_by="annee",
)
top_other = aggregate_marches(
select_sql=(
f'"{other}_id" AS id, any_value("{other}_nom") AS nom, '
'COUNT("uid") AS nb_marches, '
'COALESCE(SUM("montant"), 0) AS montant_total'
),
where_sql=where_sql,
params=params,
group_by=f'"{other}_id"',
order_by="nb_marches DESC",
limit=TOP_N,
)
top_cpv = aggregate_marches(
select_sql='"codeCPV" AS cpv, COUNT("uid") AS nb_marches',
where_sql=where_sql,
params=params,
group_by='"codeCPV"',
order_by="nb_marches DESC",
limit=TOP_N,
)
return {
"identite": identite,
"nb_marches": nb,
"montant_total": float(totals["montant_total"][0]),
"repartition_annuelle": to_json_records(annuelle),
f"top_{other}s": to_json_records(top_other),
"top_cpv": to_json_records(top_cpv),
}
+34 -1
View File
@@ -1,7 +1,12 @@
import pytest import pytest
import src.utils.search as search_mod import src.utils.search as search_mod
from src.mcp.queries import build_where_args, search_marches, search_organisations from src.mcp.queries import (
build_where_args,
compute_org_stats,
search_marches,
search_organisations,
)
from src.utils.data import DF_ACHETEURS from src.utils.data import DF_ACHETEURS
from src.utils.search import search_org from src.utils.search import search_org
@@ -80,3 +85,31 @@ def test_search_marches_no_match_is_empty():
def test_search_marches_bad_filter_returns_error(): def test_search_marches_bad_filter_returns_error():
result = search_marches(filtres_avances={"colonne_bidon__exact": "x"}) result = search_marches(filtres_avances={"colonne_bidon__exact": "x"})
assert "error" in result assert "error" in result
def test_compute_org_stats_acheteur_known():
stats = compute_org_stats("acheteur", "123")
assert stats["nb_marches"] >= 1
assert stats["montant_total"] == 10
assert stats["identite"]["id"] == "123"
assert stats["identite"]["nom"] == "ACHETEUR 1"
assert "top_titulaires" in stats
assert "top_cpv" in stats
# répartition annuelle dérivée de dateNotification (2025)
annees = [row["annee"] for row in stats["repartition_annuelle"]]
assert 2025 in annees
def test_compute_org_stats_titulaire_known():
stats = compute_org_stats("titulaire", "345")
assert stats["nb_marches"] >= 1
assert "top_acheteurs" in stats
def test_compute_org_stats_unknown_is_empty():
stats = compute_org_stats("acheteur", "inconnu-xyz")
assert stats["nb_marches"] == 0
assert stats["montant_total"] == 0
assert stats["repartition_annuelle"] == []
assert stats["top_titulaires"] == []
assert stats["top_cpv"] == []