Add utils and tests

tests/test_parser.py

@@ -0,0 +1,93 @@
+import pytest
+from pathlib import Path
+from codechunk.core.parser import CodeParser, LANGUAGE_EXTENSIONS
+from codechunk.core.chunking import ParsedChunk, ChunkMetadata
+
+
+class TestCodeParser:
+    """Tests for CodeParser."""
+
+    def test_detect_language_python(self, tmp_path):
+        """Test Python language detection."""
+        parser = CodeParser()
+        file_path = tmp_path / "test.py"
+        assert parser.detect_language(file_path) == "python"
+
+    def test_detect_language_javascript(self, tmp_path):
+        """Test JavaScript language detection."""
+        parser = CodeParser()
+        file_path = tmp_path / "test.js"
+        assert parser.detect_language(file_path) == "javascript"
+
+    def test_detect_language_go(self, tmp_path):
+        """Test Go language detection."""
+        parser = CodeParser()
+        file_path = tmp_path / "test.go"
+        assert parser.detect_language(file_path) == "go"
+
+    def test_discover_files_python(self, python_project):
+        """Test file discovery in Python project."""
+        parser = CodeParser()
+        parser.discover_files(
+            python_project,
+            include_patterns=["*.py"],
+            exclude_patterns=[]
+        )
+
+        assert len(parser.files) >= 2
+        file_names = [f.name for f in parser.files]
+        assert "main.py" in file_names
+        assert "utils.py" in file_names
+
+    def test_parse_python_file(self, sample_python_code, tmp_path):
+        """Test parsing Python code."""
+        parser = CodeParser()
+        file_path = tmp_path / "test.py"
+        file_path.write_text(sample_python_code)
+
+        chunks = parser.parse_file(file_path)
+
+        assert len(chunks) >= 2
+
+        class_chunk = next((c for c in chunks if c.chunk_type == "class"), None)
+        assert class_chunk is not None
+        assert class_chunk.name == "DataProcessor"
+
+    def test_parse_javascript(self, sample_javascript_code, tmp_path):
+        """Test parsing JavaScript code."""
+        parser = CodeParser()
+        file_path = tmp_path / "test.js"
+        file_path.write_text(sample_javascript_code)
+
+        chunks = parser.parse_file(file_path)
+
+        assert len(chunks) >= 1
+
+    def test_parse_go(self, sample_go_code, tmp_path):
+        """Test parsing Go code."""
+        parser = CodeParser()
+        file_path = tmp_path / "test.go"
+        file_path.write_text(sample_go_code)
+
+        chunks = parser.parse_file(file_path)
+
+        assert len(chunks) >= 2
+
+
+class TestLanguageExtensions:
+    """Tests for language extension mappings."""
+
+    def test_all_extensions_mapped(self):
+        """Verify all expected extensions are mapped."""
+        expected_extensions = {
+            ".py": "python",
+            ".js": "javascript",
+            ".ts": "typescript",
+            ".go": "go",
+            ".rs": "rust",
+            ".java": "java",
+            ".cpp": "cpp",
+        }
+
+        for ext, lang in expected_extensions.items():
+            assert LANGUAGE_EXTENSIONS[ext] == lang