Initial upload with CI/CD workflow

codesnap/core/language_detection.py

@@ -0,0 +1,129 @@
+"""Language detection module for CodeSnap."""
+
+import re
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Optional
+
+
+@dataclass
+class LanguageInfo:
+    """Information about a detected programming language."""
+
+    name: str
+    extensions: tuple[str, ...]
+    shebangs: tuple[str, ...]
+
+
+LANGUAGE_MAP: dict[str, LanguageInfo] = {
+    "python": LanguageInfo(
+        name="Python",
+        extensions=(".py", ".pyw", ".pyi"),
+        shebangs=("python", "python3", "pypy"),
+    ),
+    "javascript": LanguageInfo(
+        name="JavaScript",
+        extensions=(".js", ".mjs", ".cjs"),
+        shebangs=("node", "nodejs"),
+    ),
+    "typescript": LanguageInfo(
+        name="TypeScript",
+        extensions=(".ts", ".tsx"),
+        shebangs=(),
+    ),
+    "go": LanguageInfo(
+        name="Go",
+        extensions=(".go",),
+        shebangs=("go",),
+    ),
+    "rust": LanguageInfo(
+        name="Rust",
+        extensions=(".rs",),
+        shebangs=("rust", "rustc"),
+    ),
+    "java": LanguageInfo(
+        name="Java",
+        extensions=(".java",),
+        shebangs=(),
+    ),
+    "c": LanguageInfo(
+        name="C",
+        extensions=(".c", ".h"),
+        shebangs=(),
+    ),
+    "cpp": LanguageInfo(
+        name="C++",
+        extensions=(".cpp", ".cc", ".cxx", ".hpp", ".hxx"),
+        shebangs=(),
+    ),
+    "ruby": LanguageInfo(
+        name="Ruby",
+        extensions=(".rb", ".erb"),
+        shebangs=("ruby",),
+    ),
+    "php": LanguageInfo(
+        name="PHP",
+        extensions=(".php",),
+        shebangs=("php",),
+    ),
+}
+
+
+class LanguageDetector:
+    """Detects programming languages from file extensions and shebangs."""
+
+    def __init__(self) -> None:
+        self._ext_to_lang: dict[str, str] = {}
+        for lang_key, lang_info in LANGUAGE_MAP.items():
+            for ext in lang_info.extensions:
+                self._ext_to_lang[ext] = lang_key
+
+    def detect_from_path(self, path: Path) -> Optional[str]:
+        """Detect language from file path extension."""
+        suffix = path.suffix.lower()
+        return self._ext_to_lang.get(suffix)
+
+    def detect_from_content(self, content: str) -> Optional[str]:
+        """Detect language from file content shebang."""
+        first_line = content.split("\n")[0].strip()
+        if not first_line.startswith("#!"):
+            return None
+
+        shebang = first_line[2:].strip()
+        shebang_cmd = re.split(r"\s+", shebang)[0]
+
+        for lang_key, lang_info in LANGUAGE_MAP.items():
+            for shebang_pattern in lang_info.shebangs:
+                if shebang_cmd.endswith(shebang_pattern):
+                    return lang_key
+
+        return None
+
+    def detect(self, path: Path, content: Optional[str] = None) -> Optional[str]:
+        """Detect language using both path and content analysis."""
+        ext_result = self.detect_from_path(path)
+        if ext_result:
+            return ext_result
+
+        if content is not None:
+            return self.detect_from_content(content)
+
+        return None
+
+    def get_language_info(self, lang_key: str) -> Optional[LanguageInfo]:
+        """Get language information by key."""
+        return LANGUAGE_MAP.get(lang_key)
+
+    def get_supported_extensions(self) -> list[str]:
+        """Get all supported file extensions."""
+        return list(self._ext_to_lang.keys())
+
+    def get_supported_languages(self) -> list[str]:
+        """Get all supported language names."""
+        return [info.name for info in LANGUAGE_MAP.values()]
+
+
+def detect_language(path: Path, content: Optional[str] = None) -> Optional[str]:
+    """Convenience function to detect language of a file."""
+    detector = LanguageDetector()
+    return detector.detect(path, content)