refactor: Improve stats code and performance

src/griffe/cli.py

@@ -33,7 +33,6 @@
 from griffe.git import get_latest_tag, get_repo_root
 from griffe.loader import GriffeLoader, load, load_git
 from griffe.logger import get_logger
-from griffe.stats import _format_stats
 
 if TYPE_CHECKING:
     from griffe.extensions.base import Extensions, ExtensionType
@@ -411,7 +410,9 @@ def dump(
     elapsed = datetime.now(tz=timezone.utc) - started
 
     if stats:
-        logger.info(_format_stats({"time_spent_serializing": elapsed.microseconds, **loader.stats()}))
+        loader_stats = loader.stats()
+        loader_stats.time_spent_serializing = elapsed.microseconds
+        logger.info(loader_stats.as_text())
 
     return 0 if len(data_packages) == len(packages) else 1
 

src/griffe/loader.py

@@ -32,7 +32,7 @@
 from griffe.importer import dynamic_import
 from griffe.logger import get_logger
 from griffe.merger import merge_stubs
-from griffe.stats import stats
+from griffe.stats import Stats
 
 if TYPE_CHECKING:
     from griffe.enumerations import Parser
@@ -506,13 +506,16 @@ def resolve_module_aliases(
 
         return resolved, unresolved
 
-    def stats(self) -> dict:
+    def stats(self) -> Stats:
         """Compute some statistics.
 
         Returns:
             Some statistics.
         """
-        return {**stats(self), **self._time_stats}
+        stats = Stats(self)
+        stats.time_spent_visiting = self._time_stats["time_spent_visiting"]
+        stats.time_spent_inspecting = self._time_stats["time_spent_inspecting"]
+        return stats
 
     def _load_package(self, package: Package | NamespacePackage, *, submodules: bool = True) -> Module:
         top_module = self._load_module(package.name, package.path, submodules=submodules)

src/griffe/stats.py

@@ -3,172 +3,142 @@
 from __future__ import annotations
 
 from collections import defaultdict
-from typing import TYPE_CHECKING, Iterable, Union, cast
+from pathlib import Path
+from typing import TYPE_CHECKING
 
-from griffe.dataclasses import Class, Module
-from griffe.exceptions import BuiltinModuleError
+from griffe.enumerations import Kind
 
 if TYPE_CHECKING:
     from griffe.dataclasses import Alias, Object
     from griffe.loader import GriffeLoader
 
 
-def _direct(objects: Iterable[Object | Alias]) -> list[Object | Alias]:
-    return [obj for obj in objects if not obj.is_alias]
-
-
-def _n_modules(module: Module) -> int:
-    submodules = _direct(module.modules.values())
-    return len(submodules) + sum(_n_modules(cast(Module, mod)) for mod in submodules)
-
-
-def _n_classes(module_or_class: Module | Class) -> int:
-    submodules = _direct(module_or_class.modules.values())
-    subclasses = _direct(module_or_class.classes.values())
-    mods_or_classes = [mc for mc in (*submodules, *subclasses) if not mc.is_alias]
-    return len(subclasses) + sum(
-        _n_classes(cast(Union[Module, Class], mod_or_class)) for mod_or_class in mods_or_classes
-    )
-
-
-def _n_functions(module_or_class: Module | Class) -> int:
-    submodules = _direct(module_or_class.modules.values())
-    subclasses = _direct(module_or_class.classes.values())
-    functions = _direct(module_or_class.functions.values())
-    mods_or_classes = [*submodules, *subclasses]
-    return len(functions) + sum(
-        _n_functions(cast(Union[Module, Class], mod_or_class)) for mod_or_class in mods_or_classes
-    )
-
-
-def _n_attributes(module_or_class: Module | Class) -> int:
-    submodules = _direct(module_or_class.modules.values())
-    subclasses = _direct(module_or_class.classes.values())
-    attributes = _direct(module_or_class.attributes.values())
-    mods_or_classes = [*submodules, *subclasses]
-    return len(attributes) + sum(
-        _n_attributes(cast(Union[Module, Class], mod_or_class)) for mod_or_class in mods_or_classes
-    )
-
-
-def _merge_exts(exts1: dict[str, int], exts2: dict[str, int]) -> dict[str, int]:
-    for ext, value in exts2.items():
-        exts1[ext] += value
-    return exts1
-
-
-def _sum_extensions(exts: dict[str, int], module: Module) -> None:
-    current_exts = defaultdict(int)
-    try:
-        suffix = module.filepath.suffix  # type: ignore[union-attr]
-    except BuiltinModuleError:
-        current_exts[""] = 1
-    except AttributeError:
-        suffix = ""
-    else:
-        if suffix:
-            current_exts[suffix] = 1
-        for submodule in _direct(module.modules.values()):
-            _sum_extensions(current_exts, cast(Module, submodule))
-    _merge_exts(exts, current_exts)
-
-
-def stats(loader: GriffeLoader) -> dict:
-    """Return some loading statistics.
-
-    Parameters:
-        loader: The loader to compute stats from.
-
-    Returns:
-        Some statistics.
-    """
-    modules_by_extension = defaultdict(
-        int,
-        {
-            "": 0,
-            ".py": 0,
-            ".pyi": 0,
-            ".pyc": 0,
-            ".pyo": 0,
-            ".pyd": 0,
-            ".so": 0,
-        },
-    )
-    top_modules = loader.modules_collection.members.values()
-    for module in top_modules:
-        _sum_extensions(modules_by_extension, module)
-    n_lines = sum(len(lines) for lines in loader.lines_collection.values())
-    return {
-        "packages": len(top_modules),
-        "modules": len(top_modules) + sum(_n_modules(mod) for mod in top_modules),
-        "classes": sum(_n_classes(mod) for mod in top_modules),
-        "functions": sum(_n_functions(mod) for mod in top_modules),
-        "attributes": sum(_n_attributes(mod) for mod in top_modules),
-        "modules_by_extension": modules_by_extension,
-        "lines": n_lines,
-    }
-
-
-def _format_stats(stats: dict) -> str:
-    lines = []
-    packages = stats["packages"]
-    modules = stats["modules"]
-    classes = stats["classes"]
-    functions = stats["functions"]
-    attributes = stats["attributes"]
-    objects = sum((modules, classes, functions, attributes))
-    lines.append("Statistics")
-    lines.append("---------------------")
-    lines.append("Number of loaded objects")
-    lines.append(f"  Modules: {modules}")
-    lines.append(f"  Classes: {classes}")
-    lines.append(f"  Functions: {functions}")
-    lines.append(f"  Attributes: {attributes}")
-    lines.append(f"  Total: {objects} across {packages} packages")
-    per_ext = stats["modules_by_extension"]
-    builtin = per_ext[""]
-    regular = per_ext[".py"]
-    stubs = per_ext[".pyi"]
-    compiled = modules - builtin - regular - stubs
-    lines.append("")
-    lines.append(f"Total number of lines: {stats['lines']}")
-    lines.append("")
-    lines.append("Modules")
-    lines.append(f"  Builtin: {builtin}")
-    lines.append(f"  Compiled: {compiled}")
-    lines.append(f"  Regular: {regular}")
-    lines.append(f"  Stubs: {stubs}")
-    lines.append("  Per extension:")
-    for ext, number in sorted(per_ext.items()):
-        if ext:
-            lines.append(f"    {ext}: {number}")
-    visit_time = stats["time_spent_visiting"] / 1000
-    inspect_time = stats["time_spent_inspecting"] / 1000
-    total_time = visit_time + inspect_time
-    visit_percent = visit_time / total_time * 100
-    inspect_percent = inspect_time / total_time * 100
-    try:
-        visit_time_per_module = visit_time / regular
-    except ZeroDivisionError:
-        visit_time_per_module = 0
-    inspected_modules = builtin + compiled
-    try:
-        inspect_time_per_module = visit_time / inspected_modules
-    except ZeroDivisionError:
-        inspect_time_per_module = 0
-    lines.append("")
-    lines.append(
-        f"Time spent visiting modules ({regular}): "
-        f"{visit_time}ms, {visit_time_per_module:.02f}ms/module ({visit_percent:.02f}%)",
-    )
-    lines.append(
-        f"Time spent inspecting modules ({inspected_modules}): "
-        f"{inspect_time}ms, {inspect_time_per_module:.02f}ms/module ({inspect_percent:.02f}%)",
-    )
-    serialize_time = stats["time_spent_serializing"] / 1000
-    serialize_time_per_module = serialize_time / modules
-    lines.append(f"Time spent serializing: {serialize_time}ms, {serialize_time_per_module:.02f}ms/module")
-    return "\n".join(lines)
-
-
-__all__ = ["stats"]
+class Stats:
+    """Load statistics for a Griffe loader."""
+
+    def __init__(self, loader: GriffeLoader) -> None:
+        """Initialiwe the stats object.
+
+        Parameters:
+            loader: The loader to compute stats for.
+        """
+        self.loader = loader
+        modules_by_extension = defaultdict(
+            int,
+            {
+                "": 0,
+                ".py": 0,
+                ".pyi": 0,
+                ".pyc": 0,
+                ".pyo": 0,
+                ".pyd": 0,
+                ".so": 0,
+            },
+        )
+        top_modules = loader.modules_collection.members.values()
+        self.by_kind = {
+            Kind.MODULE: 0,
+            Kind.CLASS: 0,
+            Kind.FUNCTION: 0,
+            Kind.ATTRIBUTE: 0,
+        }
+        self.packages = len(top_modules)
+        self.modules_by_extension = modules_by_extension
+        self.lines = sum(len(lines) for lines in loader.lines_collection.values())
+        self.time_spent_visiting = 0
+        self.time_spent_inspecting = 0
+        self.time_spent_serializing = 0
+        for module in top_modules:
+            self._itercount(module)
+
+    def _itercount(self, root: Object | Alias) -> None:
+        if root.is_alias:
+            return
+        self.by_kind[root.kind] += 1
+        if root.is_module:
+            if isinstance(root.filepath, Path):
+                self.modules_by_extension[root.filepath.suffix] += 1
+            elif root.filepath is None:
+                self.modules_by_extension[""] += 1
+        for member in root.members.values():
+            self._itercount(member)
+
+    def as_text(self) -> str:
+        """Format the statistics as text.
+
+        Returns:
+            Text stats.
+        """
+        lines = []
+        packages = self.packages
+        modules = self.by_kind[Kind.MODULE]
+        classes = self.by_kind[Kind.CLASS]
+        functions = self.by_kind[Kind.FUNCTION]
+        attributes = self.by_kind[Kind.ATTRIBUTE]
+        objects = sum((modules, classes, functions, attributes))
+        lines.append("Statistics")
+        lines.append("---------------------")
+        lines.append("Number of loaded objects")
+        lines.append(f"  Modules: {modules}")
+        lines.append(f"  Classes: {classes}")
+        lines.append(f"  Functions: {functions}")
+        lines.append(f"  Attributes: {attributes}")
+        lines.append(f"  Total: {objects} across {packages} packages")
+        per_ext = self.modules_by_extension
+        builtin = per_ext[""]
+        regular = per_ext[".py"]
+        stubs = per_ext[".pyi"]
+        compiled = modules - builtin - regular - stubs
+        lines.append("")
+        lines.append(f"Total number of lines: {self.lines}")
+        lines.append("")
+        lines.append("Modules")
+        lines.append(f"  Builtin: {builtin}")
+        lines.append(f"  Compiled: {compiled}")
+        lines.append(f"  Regular: {regular}")
+        lines.append(f"  Stubs: {stubs}")
+        lines.append("  Per extension:")
+        for ext, number in sorted(per_ext.items()):
+            if ext:
+                lines.append(f"    {ext}: {number}")
+
+        visit_time = self.time_spent_visiting / 1000
+        inspect_time = self.time_spent_inspecting / 1000
+        total_time = visit_time + inspect_time
+        visit_percent = visit_time / total_time * 100
+        inspect_percent = inspect_time / total_time * 100
+
+        force_inspection = self.loader.force_inspection
+        visited_modules = 0 if force_inspection else regular
+        try:
+            visit_time_per_module = visit_time / visited_modules
+        except ZeroDivisionError:
+            visit_time_per_module = 0
+
+        inspected_modules = builtin + compiled + (regular if force_inspection else 0)
+        try:
+            inspect_time_per_module = inspect_time / inspected_modules
+        except ZeroDivisionError:
+            inspect_time_per_module = 0
+
+        lines.append("")
+        lines.append(
+            f"Time spent visiting modules ({visited_modules}): "
+            f"{visit_time}ms, {visit_time_per_module:.02f}ms/module ({visit_percent:.02f}%)",
+        )
+        lines.append(
+            f"Time spent inspecting modules ({inspected_modules}): "
+            f"{inspect_time}ms, {inspect_time_per_module:.02f}ms/module ({inspect_percent:.02f}%)",
+        )
+
+        serialize_time = self.time_spent_serializing / 1000
+        serialize_time_per_module = serialize_time / modules
+        lines.append(f"Time spent serializing: {serialize_time}ms, {serialize_time_per_module:.02f}ms/module")
+
+        return "\n".join(lines)
+
+
+stats = Stats
+"""Deprecated. Use `Stats` instead."""
+
+__all__ = ["Stats"]