cli/repology: dump raw cache

JSON serializer from Pydantic will be used, resulting in possibly better
perfomance.

find_work/cache.py

@@ -110,7 +110,7 @@ def _get_cache_path(cache_key: SupportsBytes) -> Path:
     return file.with_suffix(".json")
 
 
-def write_json_cache(data: Any, cache_key: SupportsBytes,
+def write_json_cache(data: Any, cache_key: SupportsBytes, *, raw: bool = False,
                      **kwargs: Any) -> None:
     """
     Write a JSON cache file in a temporary directory. Keyword arguments are
@@ -118,6 +118,7 @@ def write_json_cache(data: Any, cache_key: SupportsBytes,
 
     :param data: data to serialize
     :param cache_key: cache key object
+    :param raw: skip encoding and write raw data instead
     """
 
     cache = _get_cache_path(cache_key)
@@ -128,6 +129,9 @@ def write_json_cache(data: Any, cache_key: SupportsBytes,
 
     with open(cache, "wb") as file:
         try:
+            if raw:
+                cache.write_bytes(bytes(data))
+                return
             json.dump(data, file, **kwargs)
         except OSError:
             pass

find_work/cli/repology.py

@@ -12,7 +12,7 @@ import gentoopm
 import repology_client
 import repology_client.exceptions
 from gentoopm.basepm.atom import PMAtom
-from pydantic import RootModel, TypeAdapter
+from pydantic import TypeAdapter
 from repology_client.types import Package
 from sortedcontainers import SortedSet
 
@@ -43,15 +43,9 @@ def _projects_from_raw_json(raw_json: str | bytes) -> ProjectsMapping:
     return projects_adapter.validate_json(raw_json)
 
 
-def _projects_to_json(data: ProjectsMapping) -> dict[str, list]:
-    result: dict[str, list] = {}
-    for project, packages in data.items():
-        result[project] = []
-        for pkg in packages:
-            pkg_model = RootModel[Package](pkg)
-            pkg_dump = pkg_model.model_dump(mode="json", exclude_none=True)
-            result[project].append(pkg_dump)
-    return result
+def _projects_to_raw_json(data: ProjectsMapping) -> bytes:
+    projects_adapter = TypeAdapter(ProjectsMapping)
+    return projects_adapter.dump_json(data, exclude_none=True)
 
 
 def _collect_version_bumps(data: Iterable[set[Package]],
@@ -105,8 +99,8 @@ async def _outdated(options: Options) -> None:
             return
         options.say(Message.CACHE_WRITE)
         with dots():
-            json_data = _projects_to_json(data)
-            write_json_cache(json_data, options.cache_key)
+            raw_json_data = _projects_to_raw_json(data)
+            write_json_cache(raw_json_data, options.cache_key, raw=True)
 
     no_work = True
     for bump in _collect_version_bumps(data.values(), options):

tests/test_repology.py

@@ -2,7 +2,6 @@
 # SPDX-FileCopyrightText: 2024 Anna <cyber@sysrq.in>
 # No warranty
 
-import json
 from sortedcontainers import SortedSet
 from repology_client.types import Package
 
@@ -11,7 +10,7 @@ from find_work.cli import Options
 from find_work.cli.repology import (
     _collect_version_bumps,
     _projects_from_raw_json,
-    _projects_to_json,
+    _projects_to_raw_json,
 )
 
 
@@ -34,7 +33,7 @@ def test_projects_json_roundtrip():
             ),
         },
     }
-    assert data == _projects_from_raw_json(json.dumps(_projects_to_json(data)))
+    assert data == _projects_from_raw_json(_projects_to_raw_json(data))
 
 
 def test_collect_version_bumps():