[codex] Use Pyrefly for type checking

.github/workflows/ci.yml

@@ -58,7 +58,7 @@ jobs:
       - name: Sync dependencies
         run: uv sync --locked --dev
 
-      - name: Type check
+      - name: Pyrefly type check
         run: ./scripts/typecheck.sh
 
   test:

.pre-commit-config.yaml

@@ -7,8 +7,8 @@ repos:
         language: system
         always_run: true
         pass_filenames: false
-      - id: mypy
-        name: mypy
+      - id: pyrefly
+        name: pyrefly
         entry: ./scripts/typecheck.sh
         language: system
         always_run: true

pyproject.toml

@@ -19,8 +19,8 @@ gmaps-scraper = "gmaps_scraper.cli:main"
 
 [dependency-groups]
 dev = [
-  "mypy",
   "prek",
+  "pyrefly>=1.1.1",
   "ruff",
 ]
 
@@ -33,10 +33,11 @@ exclude-newer = "7 days"
 [tool.hatch.build.targets.wheel]
 packages = ["src/gmaps_scraper"]
 
-[tool.mypy]
-files = ["src"]
-python_version = "3.14"
-strict = true
+[tool.pyrefly]
+project-includes = ["src"]
+python-version = "3.14"
+preset = "strict"
+check-unannotated-defs = true
 
 [tool.ruff]
 line-length = 100

scripts/typecheck.sh

@@ -1,4 +1,8 @@
 #!/usr/bin/env bash
 set -euo pipefail
 
-uv run mypy
+if [[ "${GITHUB_ACTIONS:-}" == "true" ]]; then
+  uv run pyrefly check --output-format=github
+else
+  uv run pyrefly check --summarize-errors
+fi

src/gmaps_scraper/display_fields.py

@@ -299,7 +299,9 @@ def _repaired_display_diagnostics(
     repair_source: str,
 ) -> PlaceExtractionDiagnostics:
     existing = place.diagnostics
-    quality_flags = list(existing.quality_flags) if existing is not None else []
+    quality_flags: list[str] = (
+        list(existing.quality_flags) if existing is not None else []
+    )
     if "category_display_en" in repaired_fields:
         quality_flags = [
             flag for flag in quality_flags if flag != "needs_category_display_en"

src/gmaps_scraper/place_scraper.py

@@ -3317,7 +3317,7 @@ def _build_place_diagnostics(
 
     raw_sources = snapshot.get("field_sources")
     field_sources = {
-        key: str(value)
+        key: value
         for key, value in raw_sources.items()
         if isinstance(key, str)
         and isinstance(value, str)

src/gmaps_scraper/translation_memory.py

@@ -218,16 +218,20 @@ def _merge_learned_translation_memory(
     candidates: list[dict[str, object]],
 ) -> None:
     existing = _read_pending_payload(path)
-    entries = existing.setdefault("entries", [])
-    if not isinstance(entries, list):
+    raw_entries = existing.setdefault("entries", [])
+    if isinstance(raw_entries, list):
+        entries: list[object] = raw_entries
+    else:
         entries = []
         existing["entries"] = entries
     index: dict[tuple[object, object, tuple[object, ...]], dict[object, object]] = {}
     for entry in entries:
         if not isinstance(entry, dict):
             continue
-        field_kinds = entry.get("field_kinds")
-        if not isinstance(field_kinds, list):
+        raw_field_kinds = entry.get("field_kinds")
+        if isinstance(raw_field_kinds, list):
+            field_kinds: list[object] = raw_field_kinds
+        else:
             field_kinds = []
         index[(entry.get("source"), entry.get("target"), tuple(field_kinds))] = entry
     for candidate in candidates:
@@ -263,8 +267,10 @@ def _write_json_atomic(path: Path, payload: Mapping[str, object]) -> None:
 
 
 def _memory_entry_key(entry: Mapping[str, object]) -> tuple[object, object, tuple[object, ...]]:
-    field_kinds = entry.get("field_kinds")
-    if not isinstance(field_kinds, list):
+    raw_field_kinds = entry.get("field_kinds")
+    if isinstance(raw_field_kinds, list):
+        field_kinds: list[object] = raw_field_kinds
+    else:
         field_kinds = []
     return (entry.get("source"), entry.get("target"), tuple(field_kinds))