feat: add type checking for scripts and enhance project configuration

Author: Foster Guo

.pre-commit-config.yaml

@@ -46,6 +46,13 @@ repos:
         files: ^matcher_py/.*\.py$
         pass_filenames: false
 
+      - id: ty-check-scripts
+        name: ty check scripts
+        entry: bash -c 'cd matcher_rs/scripts && uv run ty check'
+        language: system
+        files: ^matcher_rs/scripts/.*\.py$
+        pass_filenames: false
+
       # matcher_java: clippy + checkstyle
       - id: clippy-matcher-java
         name: clippy matcher_java

Justfile

@@ -27,7 +27,7 @@ fmt-check:
 
 # -- Lint ----------------------------------------------------------------------
 
-lint: lint-rs lint-py lint-java lint-c
+lint: lint-rs lint-py lint-java lint-c lint-scripts
     cargo all-features clippy --workspace --all-targets -- -D warnings
     cargo doc --workspace --all-features --no-deps
 
@@ -59,6 +59,10 @@ lint-c:
     cargo fmt --all
     cargo clippy -- -D warnings
 
+[working-directory: 'matcher_rs/scripts']
+lint-scripts:
+    uv run ty check
+
 # -- Test ----------------------------------------------------------------------
 
 test: test-rs test-py test-java test-c

matcher_py/pyproject.toml

@@ -40,6 +40,9 @@ dev = ["pytest", "pip", "maturin"]
 [tool.uv]
 cache-keys = [{ file = "matcher_py.pyi"}, { file = "**/*.rs"}, { file = "**/*.toml"}, { file = "../matcher_rs/**/*.rs"}, { file = "../matcher_rs/**/*.toml"} ]
 
+[tool.ty.environment]
+python = ".venv"
+
 [tool.maturin]
 bindings = "pyo3"
 module-name = "matcher_py"

matcher_rs/scripts/bench_utils.py

@@ -141,9 +141,9 @@ def compare_result_maps(
     baseline: dict[str, float],
     candidate: dict[str, float],
     min_change_pct: float,
-) -> tuple[list[dict[str, object]], list[dict[str, object]], set[str], set[str]]:
-    regressions: list[dict[str, object]] = []
-    improvements: list[dict[str, object]] = []
+) -> tuple[list[dict[str, str | int | float]], list[dict[str, str | int | float]], set[str], set[str]]:
+    regressions: list[dict[str, str | int | float]] = []
+    improvements: list[dict[str, str | int | float]] = []
 
     shared_paths = sorted(set(baseline) & set(candidate))
     baseline_only = set(baseline) - set(candidate)
@@ -174,7 +174,7 @@ def compare_result_maps(
     return regressions, improvements, baseline_only, candidate_only
 
 
-def print_change_section(title: str, rows: Iterable[dict[str, object]]) -> None:
+def print_change_section(title: str, rows: Iterable[dict[str, str | int | float]]) -> None:
     rows = list(rows)
     print(title)
     if not rows:

matcher_rs/scripts/generate_process_map.py

@@ -216,10 +216,10 @@ def iter_scalar_chars() -> list[str]:
     ]
 
 
-def format_numeric_value(value: float) -> str:
+def format_numeric_value(value: int | float) -> str:
     if not math.isfinite(value):
         raise ValueError(f"unsupported numeric value: {value!r}")
-    if value.is_integer():
+    if isinstance(value, int) or value.is_integer():
         return str(int(value))
 
     fraction = Fraction(value).limit_denominator(MAX_FRACTION_DENOMINATOR)
@@ -514,15 +514,15 @@ def build_emoji_norm_map() -> dict[str, str]:
 # ---------------------------------------------------------------------------
 
 def render_mapping(mapping: dict[str, str]) -> str:
-    lines = [f"{key}\t{mapping[key]}" for key in sorted(mapping, key=ord)]
+    lines = [f"{key}\t{mapping[key]}" for key in sorted(mapping)]
     return "\n".join(lines) + "\n"
 
 
 def render_codepoints(codepoints: list[int]) -> str:
     return "\n".join(f"U+{codepoint:04X}" for codepoint in sorted(codepoints)) + "\n"
 
 
-def collect_outputs(root: Path) -> tuple[dict[Path, str], dict[str, object]]:
+def collect_outputs(root: Path) -> tuple[dict[Path, str], dict[str, str | dict[str, str] | list[str] | dict[str, int]]]:
     chars = iter_scalar_chars()
     process_map_dir = root / "matcher_rs" / "process_map"
 

matcher_rs/scripts/pyproject.toml

@@ -0,0 +1,10 @@
+[project]
+name = "matcher-scripts"
+version = "0.0.0"
+requires-python = ">=3.10"
+
+[dependency-groups]
+dev = ["plotly", "opencc", "pypinyin", "requests"]
+
+[tool.ty.environment]
+python = ".venv"

matcher_rs/scripts/visualize_dispatch.py

@@ -144,7 +144,7 @@ def make_winner_heatmap(rows: list[dict], mode: str, pat_cjk: int) -> go.Figure:
                 margin_row.append(0)
                 hover_row.append("")
                 continue
-            best_engine = max(throughputs, key=throughputs.get)
+            best_engine = max(throughputs, key=lambda k: throughputs[k])
             best_val = throughputs[best_engine]
             second_val = max(v for e, v in throughputs.items() if e != best_engine) if len(throughputs) > 1 else best_val
             margin = ((best_val - second_val) / second_val * 100) if second_val > 0 else 0

ty.toml

@@ -0,0 +1,2 @@
+[environment]
+root = ["matcher_py", "matcher_rs/scripts"]