CG-10827: Fix language detection bug in monorepos with mixed languages

src/codegen/git/utils/language.py

@@ -60,7 +60,7 @@
         raise ValueError(msg)
 
     # Initialize counters for each language
     language_counts = Counter()
     total_files = 0
 
     # Walk through the directory
@@ -124,7 +124,7 @@
         raise ValueError(msg)
 
     # Initialize counters for each language
     language_counts = Counter()
     total_files = 0
 
     # Initiate RepoOperator
@@ -132,8 +132,11 @@
     repo_config = RepoConfig.from_repo_path(repo_path=git_root)
     repo_operator = RepoOperator(repo_config=repo_config)
 
+    # Use the specified subfolder path for language detection if provided
+    subdirs = [base_path] if base_path else None
+
     # Walk through the directory
-    for rel_path, _ in repo_operator.iter_files(subdirs=[base_path] if base_path else None, ignore_list=GLOBAL_FILE_IGNORE_LIST):
+    for rel_path, _ in repo_operator.iter_files(subdirs=subdirs, ignore_list=GLOBAL_FILE_IGNORE_LIST):
         # Convert to Path object
         file_path = Path(git_root) / Path(rel_path)
 

src/codegen/sdk/codebase/config.py

@@ -46,23 +46,29 @@
         repo_path = os.path.abspath(path)
         git_root, base_path = split_git_path(repo_path)
         subdirectories = [base_path] if base_path else None
-        programming_language = programming_language or determine_project_language(repo_path)
+
+        # Only determine project language if not explicitly provided
+        detected_language = programming_language or determine_project_language(repo_path)
+
         repo_config = RepoConfig.from_repo_path(repo_path=git_root)
-        repo_config.language = programming_language
+        repo_config.language = detected_language
         repo_config.subdirectories = subdirectories
         # Create main project
         return cls(
             repo_operator=RepoOperator(repo_config=repo_config),
-            programming_language=programming_language,
+            programming_language=detected_language,
             base_path=base_path,
             subdirectories=subdirectories,
         )
 
     @classmethod
     def from_repo_operator(cls, repo_operator: RepoOperator, programming_language: ProgrammingLanguage | None = None, base_path: str | None = None) -> Self:
+        # Only determine project language if not explicitly provided
+        detected_language = programming_language or determine_project_language(repo_operator.repo_path)
+
         return cls(
             repo_operator=repo_operator,
-            programming_language=programming_language or determine_project_language(repo_operator.repo_path),
+            programming_language=detected_language,
             base_path=base_path,
             subdirectories=[base_path] if base_path else None,
         )

src/codegen/sdk/core/codebase.py

@@ -109,8 +109,8 @@
 TExport = TypeVar("TExport", bound="Export", default=Export)
 TSGlobalVar = TypeVar("TSGlobalVar", bound="Assignment", default=Assignment)
 PyGlobalVar = TypeVar("PyGlobalVar", bound="Assignment", default=Assignment)
 TSDirectory = Directory[TSFile, TSSymbol, TSImportStatement, TSGlobalVar, TSClass, TSFunction, TSImport]
 PyDirectory = Directory[PyFile, PySymbol, PyImportStatement, PyGlobalVar, PyClass, PyFunction, PyImport]
 
 
 @apidoc
@@ -200,22 +200,22 @@
         if repo_path is not None:
             main_project = ProjectConfig.from_path(
                 repo_path,
-                programming_language=ProgrammingLanguage(language.upper()) if language else None,
+                programming_language=ProgrammingLanguage(language.upper()) if isinstance(language, str) and language else language,
             )
             projects = [main_project]
         else:
             main_project = projects[0]
 
         # Initialize codebase
         self._op = main_project.repo_operator
         self.viz = VisualizationManager(op=self._op)
         self.repo_path = Path(self._op.repo_path)
         self.ctx = CodebaseContext(projects, config=config, secrets=secrets, io=io, progress=progress)
         self.console = Console(record=True, soft_wrap=True)
         if self.ctx.config.use_pink != PinkMode.OFF:
             import codegen_sdk_pink
 
             self._pink_codebase = codegen_sdk_pink.Codebase(self.repo_path)
 
     @noapidoc
     def __str__(self) -> str:
@@ -230,7 +230,7 @@
         yield "nodes", len(self.ctx.nodes)
         yield "edges", len(self.ctx.edges)
 
     __rich_repr__.angular = ANGULAR_STYLE
 
     @property
     @deprecated("Please do not use the local repo operator directly")
@@ -272,8 +272,8 @@
 
     @noapidoc
     def _symbols(self, symbol_type: SymbolType | None = None) -> list[TSymbol | TClass | TFunction | TGlobalVar]:
         matches: list[Symbol] = self.ctx.get_nodes(NodeType.SYMBOL)
         return [x for x in matches if x.is_top_level and (symbol_type is None or x.symbol_type == symbol_type)]
 
     # =====[ Node Types ]=====
     @overload
@@ -1392,7 +1392,7 @@
             logger.info("Initializing Codebase...")
             project = ProjectConfig.from_repo_operator(
                 repo_operator=repo_operator,
-                programming_language=ProgrammingLanguage(language.upper()) if language else None,
+                programming_language=ProgrammingLanguage(language.upper()) if isinstance(language, str) and language else language,
             )
             codebase = Codebase(projects=[project], config=config, secrets=secrets)
             logger.info("Codebase initialization complete")

tests/unit/codegen/sdk/codebase/test_language_detection.py

@@ -0,0 +1,74 @@
+import os
+import tempfile
+from pathlib import Path
+
+from codegen.sdk.codebase.config import ProjectConfig
+from codegen.shared.enums.programming_language import ProgrammingLanguage
+
+
+def test_explicit_language_respected():
+    """Test that explicitly provided language is respected and not overridden by detection."""
+    with tempfile.TemporaryDirectory() as tmp_dir:
+        # Create a temporary directory with more TypeScript files than Python files
+        ts_dir = Path(tmp_dir) / "ts"
+        py_dir = Path(tmp_dir) / "py"
+        ts_dir.mkdir()
+        py_dir.mkdir()
+
+        # Create TypeScript files
+        for i in range(5):
+            with open(ts_dir / f"file{i}.ts", "w") as f:
+                f.write(f"// TypeScript file {i}")
+
+        # Create fewer Python files
+        for i in range(2):
+            with open(py_dir / f"file{i}.py", "w") as f:
+                f.write(f"# Python file {i}")
+
+        # Initialize git repo
+        os.system(f"cd {tmp_dir} && git init && git config user.email 'test@example.com' && git config user.name 'Test User' && git add . && git commit -m 'Initial commit'")
+
+        # Test with explicit Python language
+        project_config = ProjectConfig.from_path(path=str(tmp_dir), programming_language=ProgrammingLanguage.PYTHON)
+
+        # Verify that the language is Python, not TypeScript (which would be detected based on file count)
+        assert project_config.programming_language == ProgrammingLanguage.PYTHON
+
+        # Test with explicit TypeScript language
+        project_config = ProjectConfig.from_path(
+            path=str(py_dir),  # Use Python directory
+            programming_language=ProgrammingLanguage.TYPESCRIPT,
+        )
+
+        # Verify that the language is TypeScript, not Python (which would be detected based on file count)
+        assert project_config.programming_language == ProgrammingLanguage.TYPESCRIPT
+
+
+def test_subfolder_language_detection():
+    """Test that language detection respects the specified subfolder."""
+    with tempfile.TemporaryDirectory() as tmp_dir:
+        # Create a temporary directory with TypeScript files in root and Python files in subfolder
+        ts_dir = Path(tmp_dir)
+        py_dir = Path(tmp_dir) / "python_only"
+        py_dir.mkdir()
+
+        # Create TypeScript files in root
+        for i in range(5):
+            with open(ts_dir / f"file{i}.ts", "w") as f:
+                f.write(f"// TypeScript file {i}")
+
+        # Create Python files in subfolder
+        for i in range(3):
+            with open(py_dir / f"file{i}.py", "w") as f:
+                f.write(f"# Python file {i}")
+
+        # Initialize git repo
+        os.system(f"cd {tmp_dir} && git init && git config user.email 'test@example.com' && git config user.name 'Test User' && git add . && git commit -m 'Initial commit'")
+
+        # Test with root path - should detect TypeScript
+        project_config = ProjectConfig.from_path(path=str(tmp_dir))
+        assert project_config.programming_language == ProgrammingLanguage.TYPESCRIPT
+
+        # Test with Python subfolder path - should detect Python
+        project_config = ProjectConfig.from_path(path=str(py_dir))
+        assert project_config.programming_language == ProgrammingLanguage.PYTHON