feat: Rework import generation for stubs. (#50)

Closes #38 
Closes #24 

### Summary of Changes
- Reworked the import generation for Stubs. Imports should now be
correctly generated. (#38)
- Aliases are collected with mypy and are now used to resolve alias
origins (#24)

#### Other changes
- Added a "// Todo" for the generated Stubs, if an internal class is
used as a type.
- Added "// Todo Safe-DS does not support set types" if a set object is
used.
- If booleans where used in Literal Types the first letter was
capitalized in stubs, which it should not be.
- Removed the "where" part and lower limits of the Stubs for constrains

---------

Co-authored-by: megalinter-bot <[email protected]>

src/safeds_stubgen/api_analyzer/_api.py

@@ -261,6 +261,7 @@ class Parameter:
     id: str
     name: str
     is_optional: bool
+    # We do not support default values that aren't core classes or classes definied in the package we analyze.
     default_value: str | bool | int | float | None
     assigned_by: ParameterAssignment
     docstring: ParameterDocstring

src/safeds_stubgen/api_analyzer/_get_api.py

@@ -1,10 +1,13 @@
 from __future__ import annotations
 
 import logging
-from typing import TYPE_CHECKING
+from collections import defaultdict
+from pathlib import Path
 
 import mypy.build as mypy_build
 import mypy.main as mypy_main
+from mypy import nodes as mypy_nodes
+from mypy import types as mypy_types
 
 from safeds_stubgen.docstring_parsing import DocstringStyle, create_docstring_parser
 
@@ -13,11 +16,6 @@
 from ._ast_walker import ASTWalker
 from ._package_metadata import distribution, distribution_version, package_root
 
-if TYPE_CHECKING:
-    from pathlib import Path
-
-    from mypy.nodes import MypyFile
-
 
 def get_api(
     package_name: str,
@@ -29,17 +27,8 @@ def get_api(
     if root is None:
         root = package_root(package_name)
 
-    # Get distribution data
-    dist = distribution(package_name) or ""
-    dist_version = distribution_version(dist) or ""
-
-    # Setup api walker
-    api = API(dist, package_name, dist_version)
-    docstring_parser = create_docstring_parser(docstring_style)
-    callable_visitor = MyPyAstVisitor(docstring_parser, api)
-    walker = ASTWalker(callable_visitor)
-
     walkable_files = []
+    package_paths = []
     for file_path in root.glob(pattern="./**/*.py"):
         logging.info(
             "Working on file {posix_path}",
@@ -51,48 +40,151 @@ def get_api(
             logging.info("Skipping test file")
             continue
 
+        # Check if the current file is an init file
+        if file_path.parts[-1] == "__init__.py":
+            # if a directory contains an __init__.py file it's a package
+            package_paths.append(
+                file_path.parent,
+            )
+            continue
+
         walkable_files.append(str(file_path))
 
-    mypy_trees = _get_mypy_ast(walkable_files, root)
-    for tree in mypy_trees:
+    if not walkable_files:
+        raise ValueError("No files found to analyse.")
+
+    # Get distribution data
+    dist = distribution(package_name) or ""
+    dist_version = distribution_version(dist) or ""
+
+    # Get mypy ast and aliases
+    build_result = _get_mypy_build(walkable_files)
+    mypy_asts = _get_mypy_asts(build_result, walkable_files, package_paths, root)
+    aliases = _get_aliases(build_result.types, package_name)
+
+    # Setup api walker
+    api = API(dist, package_name, dist_version)
+    docstring_parser = create_docstring_parser(docstring_style)
+    callable_visitor = MyPyAstVisitor(docstring_parser, api, aliases)
+    walker = ASTWalker(callable_visitor)
+
+    for tree in mypy_asts:
         walker.walk(tree)
 
     return callable_visitor.api
 
 
-def _get_mypy_ast(files: list[str], root: Path) -> list[MypyFile]:
-    if not files:
-        raise ValueError("No files found to analyse.")
-
-    # Build mypy checker
+def _get_mypy_build(files: list[str]) -> mypy_build.BuildResult:
+    """Build a mypy checker and return the build result."""
     mypyfiles, opt = mypy_main.process_options(files)
-    opt.preserve_asts = True  # Disable the memory optimization of freeing ASTs when possible
-    opt.fine_grained_incremental = True  # Only check parts of the code that have changed since the last check
-    result = mypy_build.build(mypyfiles, options=opt)
 
-    # Check mypy data key root start
-    graphs = result.graph
-    graph_keys = list(graphs.keys())
-    root_path = str(root)
-
-    # Get the needed data from mypy. The __init__ files need to be checked first, since we have to get the
-    # reexported data for the packages first
-    results = []
-    init_results = []
-    for graph_key in graph_keys:
-        graph = graphs[graph_key]
-        graph_path = graph.abspath
-
-        if graph_path is None:  # pragma: no cover
-            raise ValueError("Could not parse path of a module.")
-
-        tree = graph.tree
-        if tree is None or root_path not in graph_path or not graph_path.endswith(".py"):
-            continue
+    # Disable the memory optimization of freeing ASTs when possible
+    opt.preserve_asts = True
+    # Only check parts of the code that have changed since the last check
+    opt.fine_grained_incremental = True
+    # Export inferred types for all expressions
+    opt.export_types = True
+
+    return mypy_build.build(mypyfiles, options=opt)
 
-        if graph_path.endswith("__init__.py"):
-            init_results.append(tree)
-        else:
-            results.append(tree)
 
-    return init_results + results
+def _get_mypy_asts(
+    build_result: mypy_build.BuildResult,
+    files: list[str],
+    package_paths: list[Path],
+    root: Path,
+) -> list[mypy_nodes.MypyFile]:
+    # Check mypy data key root start
+    parts = root.parts
+    graph_keys = list(build_result.graph.keys())
+    root_start_after = -1
+    for i in range(len(parts)):
+        if ".".join(parts[i:]) in graph_keys:
+            root_start_after = i
+            break
+
+    # Create the keys for getting the corresponding data
+    packages = [
+        ".".join(
+            package_path.parts[root_start_after:],
+        ).replace(".py", "")
+        for package_path in package_paths
+    ]
+
+    modules = [
+        ".".join(
+            Path(file).parts[root_start_after:],
+        ).replace(".py", "")
+        for file in files
+    ]
+
+    # Get the needed data from mypy. The packages need to be checked first, since we have
+    # to get the reexported data first
+    all_paths = packages + modules
+
+    asts = []
+    for path_key in all_paths:
+        tree = build_result.graph[path_key].tree
+        if tree is not None:
+            asts.append(tree)
+
+    return asts
+
+
+def _get_aliases(result_types: dict, package_name: str) -> dict[str, set[str]]:
+    aliases: dict[str, set[str]] = defaultdict(set)
+    for key in result_types:
+        if isinstance(key, mypy_nodes.NameExpr | mypy_nodes.MemberExpr | mypy_nodes.TypeVarExpr):
+            in_package = False
+            name = ""
+
+            if isinstance(key, mypy_nodes.NameExpr):
+                type_value = result_types[key]
+
+                if hasattr(type_value, "type") and getattr(type_value, "type", None) is not None:
+                    name = type_value.type.name
+                    in_package = package_name in type_value.type.fullname
+                elif hasattr(key, "name"):
+                    name = key.name
+                    fullname = ""
+
+                    if (
+                        hasattr(key, "node")
+                        and isinstance(key.node, mypy_nodes.TypeAlias)
+                        and isinstance(key.node.target, mypy_types.Instance)
+                    ):
+                        fullname = key.node.target.type.fullname
+                    elif isinstance(type_value, mypy_types.CallableType):
+                        bound_args = type_value.bound_args
+                        if bound_args and hasattr(bound_args[0], "type"):
+                            fullname = bound_args[0].type.fullname  # type: ignore[union-attr]
+                    elif hasattr(key, "node") and isinstance(key.node, mypy_nodes.Var):
+                        fullname = key.node.fullname
+
+                    if not fullname:
+                        continue
+
+                    in_package = package_name in fullname
+            else:
+                in_package = package_name in key.fullname
+                if in_package:
+                    type_value = result_types[key]
+                    name = key.name
+                else:
+                    continue
+
+            if in_package:
+                if isinstance(type_value, mypy_types.CallableType) and hasattr(type_value.bound_args[0], "type"):
+                    fullname = type_value.bound_args[0].type.fullname  # type: ignore[union-attr]
+                elif isinstance(type_value, mypy_types.Instance):
+                    fullname = type_value.type.fullname
+                elif isinstance(key, mypy_nodes.TypeVarExpr):
+                    fullname = key.fullname
+                elif isinstance(key, mypy_nodes.NameExpr) and isinstance(key.node, mypy_nodes.Var):
+                    fullname = key.node.fullname
+                else:  # pragma: no cover
+                    raise TypeError("Received unexpected type while searching for aliases.")
+
+                aliases[name].add(fullname)
+
+    return aliases