feat: deterministic JSON schema inference (genson wrapper)

src/auto_reverse/doc/schema.py

@@ -0,0 +1,24 @@
+from __future__ import annotations
+
+from typing import Any
+
+from genson import SchemaBuilder
+
+
+class SchemaAccumulator:
+    """Accumulate JSON samples into a widening JSON Schema (genson-backed)."""
+
+    def __init__(self) -> None:
+        self._builder = SchemaBuilder()
+        self._count = 0
+
+    def add(self, value: Any) -> None:
+        self._builder.add_object(value)
+        self._count += 1
+
+    def schema(self) -> dict[str, Any] | None:
+        if self._count == 0:
+            return None
+        result: dict[str, Any] = self._builder.to_schema()
+        result.pop("$schema", None)
+        return result

tests/test_schema.py

@@ -0,0 +1,30 @@
+from auto_reverse.doc.schema import SchemaAccumulator
+
+
+def test_single_object_schema():
+    acc = SchemaAccumulator()
+    acc.add({"id": 1, "name": "Ada"})
+    schema = acc.schema()
+    assert schema["type"] == "object"
+    assert set(schema["properties"]) == {"id", "name"}
+
+
+def test_merge_widens_optional_fields():
+    acc = SchemaAccumulator()
+    acc.add({"id": 1, "name": "Ada"})
+    acc.add({"id": 2})  # name missing -> becomes optional
+    schema = acc.schema()
+    assert "id" in schema.get("required", [])
+    assert "name" not in schema.get("required", [])
+
+
+def test_array_schema():
+    acc = SchemaAccumulator()
+    acc.add([{"id": 1}, {"id": 2}])
+    schema = acc.schema()
+    assert schema["type"] == "array"
+    assert schema["items"]["type"] == "object"
+
+
+def test_empty_accumulator_returns_none():
+    assert SchemaAccumulator().schema() is None