Compact dynamic cache keys (#3395)

* Compact dynamic cache keys

* Use pydantic json for dynamic cache config

Author: koxudaxi

src/datamodel_code_generator/dynamic.py

@@ -8,6 +8,7 @@
 
 import ast
 import builtins
+import hashlib
 import itertools
 import json
 import sys
@@ -169,10 +170,13 @@ def _make_cache_key(schema: Mapping[str, Any], config: GenerateConfig) -> str |
     Returns None if the schema is not JSON-serializable.
     """
     try:
-        key_data = {"schema": dict(schema), "config": config.model_dump(mode="json", exclude_defaults=True)}
-        return json.dumps(key_data, sort_keys=True, separators=(",", ":"))
+        schema_json = json.dumps(dict(schema), sort_keys=True, separators=(",", ":"))
+        config_json = config.model_dump_json(exclude_defaults=True)
     except (TypeError, ValueError):
         return None
+    schema_digest = hashlib.sha256(schema_json.encode()).hexdigest()
+    config_digest = hashlib.sha256(config_json.encode()).hexdigest()
+    return f"{schema_digest}:{len(schema_json)}:{config_digest}:{len(config_json)}"
 
 
 def _detect_schema_input_file_type(input_: Mapping[str, Any]) -> InputFileType:

tests/main/test_dynamic_models.py

@@ -370,6 +370,21 @@ def test_non_serializable_schema_skips_cache() -> None:
     assert _make_cache_key(schema, make_config()) is None
 
 
+def test_dynamic_cache_key_does_not_retain_schema_payload() -> None:
+    """Test cache keys keep digests instead of the original schema text."""
+    from datamodel_code_generator import dynamic as dcg
+
+    unique_description = "schema-payload-marker-that-should-not-be-in-cache-key"
+    schema = make_object_schema({"name": {"type": "string", "description": unique_description}})
+
+    generate_dynamic_models(schema)
+
+    assert len(dcg._dynamic_models_cache) == 1
+    cache_key = next(iter(dcg._dynamic_models_cache))
+    assert unique_description not in cache_key
+    assert len(cache_key) < 180
+
+
 def test_cache_hit_inside_lock() -> None:
     """Test cache hit after acquiring lock (double-checked locking)."""
     from datamodel_code_generator import dynamic as dcg