Add allow_external flag & Fix Tests

docs/introduction/advanced-settings.mdx

@@ -309,6 +309,19 @@ Controls import path overrides during import resolution.
 
 Enables and disables resolution of imports from `sys.path`.
 
+<Note>
+For this to properly work, you may also need to enable `allow_external`.
+</Note>
+
+## Flag: `allow_external`
+> **Default: `False`**
+
+Enables resolving imports, files, modules, and directories from outside of the repo path.
+
+<Warning>
+Turning this flag off may allow for bad actors to access files outside of the repo path! Use with caution!
+</Warning>
+
 ## Flag: `ts_dependency_manager`
 > **Default: `False`**
 

src/codegen/configs/models/codebase.py

@@ -21,6 +21,7 @@ def __init__(self, prefix: str = "CODEBASE", *args, **kwargs) -> None:
     import_resolution_paths: list[str] = Field(default_factory=lambda: [])
     import_resolution_overrides: dict[str, str] = Field(default_factory=lambda: {})
     py_resolve_syspath: bool = False
+    allow_external: bool = False
     ts_dependency_manager: bool = False
     ts_language_engine: bool = False
     v8_ts_engine: bool = False

src/codegen/sdk/codebase/codebase_context.py

@@ -381,7 +381,7 @@ def get_directory(self, directory_path: PathLike, create_on_missing: bool = Fals
         """
         # If not part of repo path, return None
         absolute_path = self.to_absolute(directory_path)
-        if not self.is_subdir(absolute_path):
+        if not self.is_subdir(absolute_path) and not self.config.allow_external:
             assert False, f"Directory {absolute_path} is not part of repo path {self.repo_path}"
             return None
 
@@ -611,7 +611,7 @@ def get_edges(self) -> list[tuple[NodeId, NodeId, EdgeType, Usage | None]]:
     def get_file(self, file_path: os.PathLike, ignore_case: bool = False) -> SourceFile | None:
         # If not part of repo path, return None
         absolute_path = self.to_absolute(file_path)
-        if not self.is_subdir(absolute_path):
+        if not self.is_subdir(absolute_path) and not self.config.allow_external:
             assert False, f"File {file_path} is not part of the repository path"
 
         # Check if file exists in graph

src/codegen/sdk/utils.py

@@ -339,3 +339,29 @@ def is_minified_js(content):
     except Exception as e:
         print(f"Error analyzing content: {e}")
         return False
+
+
+@contextmanager
+def use_cwd(path):
+    """Context manager that temporarily changes the current working directory.
+
+    Args:
+        path (str): The directory path to change to.
+
+    Yields:
+        str: The new current working directory.
+
+    Example:
+        ```python
+        with use_cwd("/path/to/directory"):
+            # Code here runs with the working directory set to '/path/to/directory'
+            ...
+        # Working directory is restored to the original
+        ```
+    """
+    old_cwd = os.getcwd()
+    try:
+        os.chdir(path)
+        yield path
+    finally:
+        os.chdir(old_cwd)

tests/unit/codegen/sdk/python/import_resolution/test_import_resolution.py

@@ -2,6 +2,7 @@
 
 from codegen.sdk.codebase.config import TestFlags
 from codegen.sdk.codebase.factory.get_session import get_codebase_session
+from codegen.sdk.utils import use_cwd
 
 if TYPE_CHECKING:
     from codegen.sdk.core.file import SourceFile
@@ -272,27 +273,31 @@ def func():
 """,
         },
     ) as codebase:
-        src_file: SourceFile = codebase.get_file("a/b/c/src.py")
-        consumer_file: SourceFile = codebase.get_file("consumer.py")
+        # Wrap CWD since we are using py_resolve_syspath
+        with use_cwd(tmpdir):
+            src_file: SourceFile = codebase.get_file("a/b/c/src.py")
+            consumer_file: SourceFile = codebase.get_file("consumer.py")
 
-        # Enable resolution via sys.path
-        codebase.ctx.config.py_resolve_syspath = True
+            # Enable resolution via sys.path
+            codebase.ctx.config.py_resolve_syspath = True
+            # Allow resolving files and modules outside of the repo path
+            codebase.ctx.config.allow_external = True
 
-        # =====[ Imports cannot be found without sys.path being set ]=====
-        assert len(consumer_file.imports) == 1
-        src_import: Import = consumer_file.imports[0]
-        src_import_resolution: ImportResolution = src_import.resolve_import()
-        assert src_import_resolution is None
+            # =====[ Imports cannot be found without sys.path being set ]=====
+            assert len(consumer_file.imports) == 1
+            src_import: Import = consumer_file.imports[0]
+            src_import_resolution: ImportResolution = src_import.resolve_import()
+            assert src_import_resolution is None
 
-        # Modify sys.path for this test only
-        monkeypatch.syspath_prepend("a")
+            # Modify sys.path for this test only
+            monkeypatch.syspath_prepend("a")
 
-        # =====[ Imports can be found with sys.path set and active ]=====
-        codebase.ctx.config.py_resolve_syspath = True
-        src_import_resolution = src_import.resolve_import()
-        assert src_import_resolution
-        assert src_import_resolution.from_file is src_file
-        assert src_import_resolution.imports_file is True
+            # =====[ Imports can be found with sys.path set and active ]=====
+            codebase.ctx.config.py_resolve_syspath = True
+            src_import_resolution = src_import.resolve_import()
+            assert src_import_resolution
+            assert src_import_resolution.from_file is src_file
+            assert src_import_resolution.imports_file is True
 
 
 def test_import_resolution_file_custom_resolve_path(tmpdir: str) -> None:
@@ -366,28 +371,32 @@ def func():
 """,
         },
     ) as codebase:
-        src_file: SourceFile = codebase.get_file("a/b/c/src.py")
-        consumer_file: SourceFile = codebase.get_file("consumer.py")
-
-        # Ensure we don't have overrites and enable syspath resolution
-        codebase.ctx.config.import_resolution_paths = []
-        codebase.ctx.config.py_resolve_syspath = True
-
-        # =====[ Import with sys.path set can be found ]=====
-        assert len(consumer_file.imports) == 1
-        # Modify sys.path for this test only
-        monkeypatch.syspath_prepend("a")
-        src_import: Import = consumer_file.imports[0]
-        src_import_resolution = src_import.resolve_import()
-        assert src_import_resolution
-        assert src_import_resolution.from_file.file_path == "a/c/src.py"
-
-        # =====[ Imports can be found with custom resolve over sys.path ]=====
-        codebase.ctx.config.import_resolution_paths = ["a/b"]
-        src_import_resolution = src_import.resolve_import()
-        assert src_import_resolution
-        assert src_import_resolution.from_file is src_file
-        assert src_import_resolution.imports_file is True
+        # Wrap CWD since we are using py_resolve_syspath
+        with use_cwd(tmpdir):
+            src_file: SourceFile = codebase.get_file("a/b/c/src.py")
+            consumer_file: SourceFile = codebase.get_file("consumer.py")
+
+            # Ensure we don't have overrites and enable syspath resolution
+            codebase.ctx.config.import_resolution_paths = []
+            codebase.ctx.config.py_resolve_syspath = True
+            # Allow resolving files and modules outside of the repo path
+            codebase.ctx.config.allow_external = True
+
+            # =====[ Import with sys.path set can be found ]=====
+            assert len(consumer_file.imports) == 1
+            # Modify sys.path for this test only
+            monkeypatch.syspath_prepend("a")
+            src_import: Import = consumer_file.imports[0]
+            src_import_resolution = src_import.resolve_import()
+            assert src_import_resolution
+            assert src_import_resolution.from_file.file_path == "a/c/src.py"
+
+            # =====[ Imports can be found with custom resolve over sys.path ]=====
+            codebase.ctx.config.import_resolution_paths = ["a/b"]
+            src_import_resolution = src_import.resolve_import()
+            assert src_import_resolution
+            assert src_import_resolution.from_file is src_file
+            assert src_import_resolution.imports_file is True
 
 
 def test_import_resolution_default_conflicts_overrite(tmpdir: str, monkeypatch) -> None:
@@ -412,34 +421,38 @@ def func():
 """,
         },
     ) as codebase:
-        src_file: SourceFile = codebase.get_file("a/src.py")
-        src_file_overrite: SourceFile = codebase.get_file("b/a/src.py")
-        consumer_file: SourceFile = codebase.get_file("consumer.py")
-
-        # Ensure we don't have overrites and enable syspath resolution
-        codebase.ctx.config.import_resolution_paths = []
-        codebase.ctx.config.py_resolve_syspath = True
-
-        # =====[ Default import works ]=====
-        assert len(consumer_file.imports) == 1
-        src_import: Import = consumer_file.imports[0]
-        src_import_resolution = src_import.resolve_import()
-        assert src_import_resolution
-        assert src_import_resolution.from_file is src_file
-
-        # =====[ Sys.path overrite has precedence ]=====
-        monkeypatch.syspath_prepend("b")
-        src_import_resolution = src_import.resolve_import()
-        assert src_import_resolution
-        assert src_import_resolution.from_file is not src_file
-        assert src_import_resolution.from_file is src_file_overrite
-
-        # =====[ Custom overrite has precedence ]=====
-        codebase.ctx.config.import_resolution_paths = ["b"]
-        src_import_resolution = src_import.resolve_import()
-        assert src_import_resolution
-        assert src_import_resolution.from_file is not src_file
-        assert src_import_resolution.from_file is src_file_overrite
+        # Wrap CWD since we are using py_resolve_syspath
+        with use_cwd(tmpdir):
+            src_file: SourceFile = codebase.get_file("a/src.py")
+            src_file_overrite: SourceFile = codebase.get_file("b/a/src.py")
+            consumer_file: SourceFile = codebase.get_file("consumer.py")
+
+            # Ensure we don't have overrites and enable syspath resolution
+            codebase.ctx.config.import_resolution_paths = []
+            codebase.ctx.config.py_resolve_syspath = True
+            # Allow resolving files and modules outside of the repo path
+            codebase.ctx.config.allow_external = True
+
+            # =====[ Default import works ]=====
+            assert len(consumer_file.imports) == 1
+            src_import: Import = consumer_file.imports[0]
+            src_import_resolution = src_import.resolve_import()
+            assert src_import_resolution
+            assert src_import_resolution.from_file is src_file
+
+            # =====[ Sys.path overrite has precedence ]=====
+            monkeypatch.syspath_prepend("b")
+            src_import_resolution = src_import.resolve_import()
+            assert src_import_resolution
+            assert src_import_resolution.from_file is not src_file
+            assert src_import_resolution.from_file is src_file_overrite
+
+            # =====[ Custom overrite has precedence ]=====
+            codebase.ctx.config.import_resolution_paths = ["b"]
+            src_import_resolution = src_import.resolve_import()
+            assert src_import_resolution
+            assert src_import_resolution.from_file is not src_file
+            assert src_import_resolution.from_file is src_file_overrite
 
 
 def test_import_resolution_init_wildcard(tmpdir: str) -> None: