fix: Fixed the stubs generator (#108)

Closes #80 

### Summary of Changes
Fixed the stub generator, it should now run with the Library project.

To run the script, use the following command:
`safe-ds-stubgen -p "Library" -o "path/to/out" -nc -s
"path/to/Library/src" --docstyle NUMPYDOC`

---------

Co-authored-by: megalinter-bot <129584137+megalinter-bot@users.noreply.github.com>

docs/README.md

@@ -5,7 +5,7 @@
 [![codecov](https://codecov.io/gh/Safe-DS/Stub-Generator/branch/main/graph/badge.svg?token=UyCUY59HKM)](https://codecov.io/gh/Safe-DS/Stub-Generator)
 [![Documentation Status](https://readthedocs.org/projects/safe-ds-stub-generator/badge/?version=stable)](https://stubgen.safeds.com)
 
-Automated generation of [Safe-DS stubs](https://dsl.safeds.com/en/stable/language/stub-language/) for Python libraries.
+Automated generation of [Safe-DS stubs](https://dsl.safeds.com/en/stable/stub-language/) for Python libraries.
 
 ## Installation
 
@@ -29,7 +29,7 @@ options:
   -v, --verbose         show info messages
   -p PACKAGE, --package PACKAGE
                         The name of the package.
-  -s SRC, --src SRC     Directory containing the Python code of the package. If this is omitted, we try to locate the package with the given name in the current Python interpreter.
+  -s SRC, --src SRC     Source directory containing the Python code of the package.
   -o OUT, --out OUT     Output directory.
   --docstyle {PLAINTEXT,EPYDOC,GOOGLE,NUMPYDOC,REST}
                         The docstring style.

src/safeds_stubgen/api_analyzer/__init__.py

@@ -19,7 +19,7 @@
 from ._ast_visitor import result_name_generator
 from ._get_api import get_api
 from ._mypy_helpers import get_classdef_definitions, get_funcdef_definitions, get_mypyfile_definitions
-from ._package_metadata import distribution, distribution_version, package_root
+from ._package_metadata import distribution, distribution_version
 from ._types import (
     AbstractType,
     BoundaryType,
@@ -34,6 +34,7 @@
     TupleType,
     TypeVarType,
     UnionType,
+    UnknownType,
 )
 
 __all__ = [
@@ -58,7 +59,6 @@
     "LiteralType",
     "Module",
     "NamedType",
-    "package_root",
     "Parameter",
     "ParameterAssignment",
     "QualifiedImport",
@@ -68,6 +68,7 @@
     "TupleType",
     "TypeVarType",
     "UnionType",
+    "UnknownType",
     "VarianceKind",
     "WildcardImport",
 ]

src/safeds_stubgen/api_analyzer/_ast_visitor.py

@@ -938,6 +938,10 @@ def mypy_type_to_abstract_type(
                 # from the import information
                 missing_import_name = mypy_type.missing_import_name.split(".")[-1]  # type: ignore[union-attr]
                 name, qname = self._find_alias(missing_import_name)
+
+                if not qname:  # pragma: no cover
+                    return sds_types.UnknownType()
+
                 return sds_types.NamedType(name=name, qname=qname)
             else:
                 return sds_types.NamedType(name="Any", qname="typing.Any")
@@ -971,6 +975,10 @@ def mypy_type_to_abstract_type(
 
                 # if not, we check if it's an alias
                 name, qname = self._find_alias(mypy_type.name)
+
+                if not qname:  # pragma: no cover
+                    return sds_types.UnknownType()
+
                 return sds_types.NamedType(name=name, qname=qname)
 
         # Builtins
@@ -1001,7 +1009,8 @@ def mypy_type_to_abstract_type(
                 )
             else:
                 return sds_types.NamedType(name=type_name, qname=mypy_type.type.fullname)
-        raise ValueError("Unexpected type.")  # pragma: no cover
+
+        return sds_types.UnknownType()  # pragma: no cover
 
     def _find_alias(self, type_name: str) -> tuple[str, str]:
         module = self.__declaration_stack[0]
@@ -1010,27 +1019,17 @@ def _find_alias(self, type_name: str) -> tuple[str, str]:
         if not isinstance(module, Module):  # pragma: no cover
             raise TypeError(f"Expected module, got {type(module)}.")
 
-        name = ""
-        qname = ""
-        qualified_imports = module.qualified_imports
-        import_aliases = [qimport.alias for qimport in qualified_imports]
+        # First we check if it can be found in the imports
+        name, qname = self._search_alias_in_qualified_imports(module.qualified_imports, type_name)
+        if name and qname:
+            return name, qname
 
         if type_name in self.aliases:
             qnames: set = self.aliases[type_name]
             if len(qnames) == 1:
-                # We have to check if this is an alias from an import
-                import_name, import_qname = self._search_alias_in_qualified_imports(qualified_imports, type_name)
-
                 # We need a deepcopy since qnames is a pointer to the set in the alias dict
-                qname = import_qname if import_qname else deepcopy(qnames).pop()
-                name = import_name if import_name else qname.split(".")[-1]
-            elif type_name in import_aliases:
-                # We check if the type was imported
-                qimport_name, qimport_qname = self._search_alias_in_qualified_imports(qualified_imports, type_name)
-
-                if qimport_qname:
-                    qname = qimport_qname
-                    name = qimport_name
+                qname = deepcopy(qnames).pop()
+                name = qname.split(".")[-1]
             else:
                 # In this case some types where defined in multiple modules with the same names.
                 for alias_qname in qnames:
@@ -1041,14 +1040,9 @@ def _find_alias(self, type_name: str) -> tuple[str, str]:
                     if self.mypy_file is None:  # pragma: no cover
                         raise TypeError("Expected mypy_file (module information), got None.")
 
-                    if type_path == self.mypy_file.fullname:
+                    if self.mypy_file.fullname in type_path:
                         qname = alias_qname
                         break
-        else:
-            name, qname = self._search_alias_in_qualified_imports(qualified_imports, type_name)
-
-        if not qname:  # pragma: no cover
-            raise ValueError(f"It was not possible to find out where the alias {type_name} was defined.")
 
         return name, qname
 

src/safeds_stubgen/api_analyzer/_ast_walker.py

@@ -41,37 +41,28 @@ def __walk(self, node: MypyFile | ClassDef | Decorator | FuncDef | AssignmentStm
 
         self.__enter(node)
 
-        definitions: list = []
+        # Search nodes for more child nodes. Skip other not specified types, since we either get them through the
+        # ast_visitor, some other way or don't need to parse them at all
+        child_nodes = []
         if isinstance(node, MypyFile):
             definitions = get_mypyfile_definitions(node)
+            child_nodes = [
+                _def for _def in definitions if _def.__class__.__name__ in {"FuncDef", "ClassDef", "Decorator"}
+            ]
         elif isinstance(node, ClassDef):
             definitions = get_classdef_definitions(node)
-        elif isinstance(node, FuncDef):
+            child_nodes = [
+                _def
+                for _def in definitions
+                if _def.__class__.__name__ in {"AssignmentStmt", "FuncDef", "ClassDef", "Decorator"}
+            ]
+        elif isinstance(node, FuncDef) and node.name == "__init__":
             definitions = get_funcdef_definitions(node)
-
-        # Skip other types, since we either get them through the ast_visitor, some other way or
-        # don't need to parse them
-        child_nodes = [
-            _def
-            for _def in definitions
-            if _def.__class__.__name__
-            in {
-                "AssignmentStmt",
-                "FuncDef",
-                "ClassDef",
-                "Decorator",
-            }
-        ]
+            child_nodes = [_def for _def in definitions if _def.__class__.__name__ == "AssignmentStmt"]
 
         for child_node in child_nodes:
-            # Ignore global variables and function attributes if the function is an __init__
-            if isinstance(child_node, AssignmentStmt):
-                if isinstance(node, MypyFile):
-                    continue
-                if isinstance(node, FuncDef) and node.name != "__init__":
-                    continue
-
-            if isinstance(child_node, FuncDef) and isinstance(node, FuncDef):
+            # The '__mypy-replace' name is a mypy placeholer which we don't want to parse.
+            if getattr(child_node, "name", "") == "__mypy-replace":  # pragma: no cover
                 continue
 
             self.__walk(child_node, visited_nodes)

src/safeds_stubgen/api_analyzer/_get_api.py

@@ -14,33 +14,30 @@
 from ._api import API
 from ._ast_visitor import MyPyAstVisitor
 from ._ast_walker import ASTWalker
-from ._package_metadata import distribution, distribution_version, package_root
+from ._package_metadata import distribution, distribution_version
 
 if TYPE_CHECKING:
     from pathlib import Path
 
 
 def get_api(
-    package_name: str,
-    root: Path | None = None,
+    root: Path,
     docstring_style: DocstringStyle = DocstringStyle.PLAINTEXT,
     is_test_run: bool = False,
 ) -> API:
-    # Check root
-    if root is None:
-        root = package_root(package_name)
+    init_roots = _get_nearest_init_dirs(root)
+    if len(init_roots) == 1:
+        root = init_roots[0]
+
+    logging.info("Started gathering the raw package data with Mypy.")
 
     walkable_files = []
     package_paths = []
     for file_path in root.glob(pattern="./**/*.py"):
-        logging.info(
-            "Working on file {posix_path}",
-            extra={"posix_path": str(file_path)},
-        )
-
         # Check if the current path is a test directory
-        if not is_test_run and ("test" in file_path.parts or "tests" in file_path.parts):
-            logging.info("Skipping test file")
+        if not is_test_run and ("test" in file_path.parts or "tests" in file_path.parts or "docs" in file_path.parts):
+            log_msg = f"Skipping test file in {file_path}"
+            logging.info(log_msg)
             continue
 
         # Check if the current file is an init file
@@ -56,6 +53,9 @@ def get_api(
     if not walkable_files:
         raise ValueError("No files found to analyse.")
 
+    # Package name
+    package_name = root.stem
+
     # Get distribution data
     dist = distribution(package_name=package_name) or ""
     dist_version = distribution_version(dist=dist) or ""
@@ -77,6 +77,25 @@ def get_api(
     return callable_visitor.api
 
 
+def _get_nearest_init_dirs(root: Path) -> list[Path]:
+    all_inits = list(root.glob("./**/__init__.py"))
+    shortest_init_paths = []
+    shortest_len = -1
+    for init in all_inits:
+        path_len = len(init.parts)
+        if shortest_len == -1:
+            shortest_len = path_len
+            shortest_init_paths.append(init.parent)
+        elif path_len <= shortest_len:  # pragma: no cover
+            if path_len == shortest_len:
+                shortest_init_paths.append(init.parent)
+            else:
+                shortest_len = path_len
+                shortest_init_paths = [init.parent]
+
+    return shortest_init_paths
+
+
 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)

src/safeds_stubgen/api_analyzer/_package_metadata.py

@@ -1,15 +1,6 @@
 from __future__ import annotations
 
-import importlib
 from importlib.metadata import packages_distributions, version
-from pathlib import Path
-
-
-def package_root(package_name: str) -> Path:
-    path_as_string = importlib.import_module(package_name).__file__
-    if path_as_string is None:
-        raise AssertionError(f"Cannot find package root for '{path_as_string}'.")
-    return Path(path_as_string).parent
 
 
 def distribution(package_name: str) -> str | None:

src/safeds_stubgen/api_analyzer/_types.py

@@ -14,6 +14,8 @@ class AbstractType(metaclass=ABCMeta):
     @classmethod
     def from_dict(cls, d: dict[str, Any]) -> AbstractType:
         match d["kind"]:
+            case UnknownType.__name__:
+                return UnknownType.from_dict(d)
             case NamedType.__name__:
                 return NamedType.from_dict(d)
             case EnumType.__name__:
@@ -45,6 +47,21 @@ def from_dict(cls, d: dict[str, Any]) -> AbstractType:
     def to_dict(self) -> dict[str, Any]: ...
 
 
+@dataclass(frozen=True)
+class UnknownType(AbstractType):
+    @classmethod
+    def from_dict(cls, _: dict[str, Any]) -> UnknownType:
+        return UnknownType()
+
+    def to_dict(self) -> dict[str, str]:
+        return {"kind": self.__class__.__name__}
+
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, UnknownType):  # pragma: no cover
+            return NotImplemented
+        return True
+
+
 @dataclass(frozen=True)
 class NamedType(AbstractType):
     name: str
@@ -269,6 +286,11 @@ def to_dict(self) -> dict[str, Any]:
 
         return {"kind": self.__class__.__name__, "types": type_list}
 
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, ListType):  # pragma: no cover
+            return NotImplemented
+        return Counter(self.types) == Counter(other.types)
+
     def __hash__(self) -> int:
         return hash(frozenset(self.types))
 
@@ -315,6 +337,11 @@ def to_dict(self) -> dict[str, Any]:
             "return_type": self.return_type.to_dict(),
         }
 
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, CallableType):  # pragma: no cover
+            return NotImplemented
+        return Counter(self.parameter_types) == Counter(other.parameter_types) and self.return_type == other.return_type
+
     def __hash__(self) -> int:
         return hash(frozenset([*self.parameter_types, self.return_type]))
 
@@ -338,6 +365,11 @@ def to_dict(self) -> dict[str, Any]:
             "types": [t.to_dict() for t in self.types],
         }
 
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, SetType):  # pragma: no cover
+            return NotImplemented
+        return Counter(self.types) == Counter(other.types)
+
     def __hash__(self) -> int:
         return hash(frozenset(self.types))
 
@@ -353,6 +385,11 @@ def from_dict(cls, d: dict[str, Any]) -> LiteralType:
     def to_dict(self) -> dict[str, Any]:
         return {"kind": self.__class__.__name__, "literals": self.literals}
 
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, LiteralType):  # pragma: no cover
+            return NotImplemented
+        return Counter(self.literals) == Counter(other.literals)
+
     def __hash__(self) -> int:
         return hash(frozenset(self.literals))
 
@@ -390,6 +427,11 @@ def to_dict(self) -> dict[str, Any]:
 
         return {"kind": self.__class__.__name__, "types": type_list}
 
+    def __eq__(self, other: object) -> bool:
+        if not isinstance(other, TupleType):  # pragma: no cover
+            return NotImplemented
+        return Counter(self.types) == Counter(other.types)
+
     def __hash__(self) -> int:
         return hash(frozenset(self.types))