feat(api): parsing des opérateurs d'agrégation (groupby/count/sum/avg/min/max) (#78)

src/api/filters.py

@@ -1,3 +1,4 @@
+from dataclasses import dataclass
 from datetime import date, datetime
 
 import polars as pl
@@ -20,6 +21,55 @@ OPERATORS = {
 
 RESERVED_PARAMS = {"page", "page_size", "columns", "count_results"}
 
+AGGREGATORS = {"groupby", "count", "sum", "avg", "min", "max"}
+AGG_SQL = {"count": "COUNT", "sum": "SUM", "avg": "AVG", "min": "MIN", "max": "MAX"}
+
+
+@dataclass
+class AggregationSpec:
+    select_sql: str
+    group_by_sql: str | None
+
+
+def parse_aggregators(
+    args: list[tuple[str, str]], schema: pl.Schema
+) -> "AggregationSpec | None":
+    """Détecte les drapeaux d'agrégation (`col__groupby`, `col__count`, ...).
+
+    Retourne None si aucun agrégateur. Sinon, construit les fragments SQL
+    `select_sql` et `group_by_sql` (noms de colonnes validés contre le schéma).
+    """
+    group_cols: list[str] = []
+    aggregates: list[tuple[str, str]] = []  # (operator, column)
+    has_agg = False
+
+    for key, _ in args:
+        parsed = _split_key(key)
+        if not parsed:
+            continue
+        col, op = parsed
+        if op not in AGGREGATORS:
+            continue
+        has_agg = True
+        if col not in schema:
+            raise FilterError(f"Colonne inconnue : {col!r}", field=key)
+        if op == "groupby":
+            group_cols.append(col)
+        else:
+            aggregates.append((op, col))
+
+    if not has_agg:
+        return None
+
+    select_parts = [f'"{c}"' for c in group_cols]
+    for op, col in aggregates:
+        select_parts.append(f'{AGG_SQL[op]}("{col}") AS "{col}__{op}"')
+
+    group_by_sql = ", ".join(f'"{c}"' for c in group_cols) if group_cols else None
+    return AggregationSpec(
+        select_sql=", ".join(select_parts), group_by_sql=group_by_sql
+    )
+
 
 class FilterError(ValueError):
     def __init__(self, message: str, field: str | None = None):
@@ -85,6 +135,9 @@ def build_where(
             raise FilterError(f"Paramètre non reconnu : {key}", field=key)
         col, op = parsed
 
+        if op in AGGREGATORS:
+            continue
+
         if op not in OPERATORS:
             raise FilterError(f"Opérateur inconnu : __{op}", field=key)
 

tests/api/test_filters.py

@@ -1,7 +1,7 @@
 import polars as pl
 import pytest
 
-from src.api.filters import FilterError, build_where
+from src.api.filters import AggregationSpec, FilterError, build_where, parse_aggregators
 
 SCHEMA = pl.Schema(
     {
@@ -151,3 +151,54 @@ def test_differs_filter_on_int():
     where, params, _ = build_where([("annee__differs", "2020")], SCHEMA)
     assert where == '"annee" IS DISTINCT FROM ?'
     assert params == [2020]
+
+
+def test_parse_aggregators_none_when_absent():
+    assert parse_aggregators([("uid__exact", "a")], SCHEMA) is None
+
+
+def test_parse_aggregators_groupby_and_count():
+    spec = parse_aggregators([("annee__groupby", ""), ("uid__count", "")], SCHEMA)
+    assert isinstance(spec, AggregationSpec)
+    assert spec.select_sql == '"annee", COUNT("uid") AS "uid__count"'
+    assert spec.group_by_sql == '"annee"'
+
+
+def test_parse_aggregators_multiple_aggregates():
+    spec = parse_aggregators(
+        [
+            ("annee__groupby", ""),
+            ("montant__sum", ""),
+            ("montant__avg", ""),
+            ("montant__min", ""),
+            ("montant__max", ""),
+        ],
+        SCHEMA,
+    )
+    assert spec.select_sql == (
+        '"annee", SUM("montant") AS "montant__sum", '
+        'AVG("montant") AS "montant__avg", '
+        'MIN("montant") AS "montant__min", '
+        'MAX("montant") AS "montant__max"'
+    )
+    assert spec.group_by_sql == '"annee"'
+
+
+def test_parse_aggregators_global_without_groupby():
+    spec = parse_aggregators([("uid__count", "")], SCHEMA)
+    assert spec.select_sql == 'COUNT("uid") AS "uid__count"'
+    assert spec.group_by_sql is None
+
+
+def test_parse_aggregators_unknown_column_raises():
+    with pytest.raises(FilterError):
+        parse_aggregators([("nope__count", "")], SCHEMA)
+
+
+def test_build_where_ignores_aggregator_flags():
+    where, params, _ = build_where(
+        [("annee__groupby", ""), ("uid__count", ""), ("montant__greater", "100")],
+        SCHEMA,
+    )
+    assert where == '"montant" >= ?'
+    assert params == [100.0]