Add source code files (detectors, generators, CLI)

src/detectors/language.py

@@ -0,0 +1,111 @@
+"""Language detector for identifying programming languages."""
+
+from pathlib import Path
+from typing import Dict, List
+
+from src.data.patterns import LANGUAGE_INDICATOR_FILES
+from src.detectors.base import BaseDetector
+
+
+class LanguageDetector(BaseDetector):
+    """Detects programming languages used in a project."""
+
+    def __init__(self) -> None:
+        self.indicator_files = LANGUAGE_INDICATOR_FILES
+        self.priority = 100
+
+    def get_priority(self) -> int:
+        return self.priority
+
+    def detect(self, project_path: Path) -> List[str]:
+        """Detect programming languages from project files.
+
+        Args:
+            project_path: Path to the project directory.
+
+        Returns:
+            List of detected language names.
+        """
+        detected_languages = []
+
+        for language, indicators in self.indicator_files.items():
+            if self._detect_language(project_path, language, indicators):
+                detected_languages.append(language)
+
+        return detected_languages
+
+    def _detect_language(
+        self, project_path: Path, language: str, indicators: List[str]
+    ) -> bool:
+        """Check if a language is present based on indicators.
+
+        Args:
+            project_path: Project root directory.
+            language: Language name to check.
+            indicators: List of indicator files for the language.
+
+        Returns:
+            True if language is detected, False otherwise.
+        """
+        for indicator in indicators:
+            matches = list(project_path.rglob(indicator))
+            if matches:
+                return True
+
+        return False
+
+    def detect_from_content(self, project_path: Path) -> Dict[str, List[str]]:
+        """Detect languages from file content analysis.
+
+        Args:
+            project_path: Path to the project directory.
+
+        Returns:
+            Dict mapping languages to matching files.
+        """
+        language_files: Dict[str, List[str]] = {}
+
+        extension_map: Dict[str, List[str]] = {
+            ".py": ["python"],
+            ".js": ["nodejs"],
+            ".jsx": ["nodejs"],
+            ".ts": ["nodejs"],
+            ".tsx": ["nodejs"],
+            ".go": ["go"],
+            ".java": ["java"],
+            ".rs": ["rust"],
+            ".cs": ["csharp"],
+            ".cpp": ["cpp"],
+            ".c": ["cpp"],
+            ".h": ["cpp"],
+            ".rb": ["ruby"],
+            ".php": ["php"],
+            ".dart": ["dart"],
+            ".swift": ["swift"],
+            ".kt": ["kotlin"],
+            ".kts": ["kotlin"],
+            ".scala": ["scala"],
+            ".pl": ["perl"],
+            ".pm": ["perl"],
+            ".r": ["r"],
+            ".R": ["r"],
+            ".ex": ["elixir"],
+            ".exs": ["elixir"],
+            ".clj": ["clojure"],
+            ".lua": ["lua"],
+            ".hs": ["haskell"],
+        }
+
+        try:
+            for file_path in project_path.rglob("*"):
+                if file_path.is_file():
+                    suffix = file_path.suffix.lower()
+                    if suffix in extension_map:
+                        for lang in extension_map[suffix]:
+                            if lang not in language_files:
+                                language_files[lang] = []
+                            language_files[lang].append(str(file_path))
+        except (PermissionError, OSError):
+            pass
+
+        return language_files