Add config_auditor module files: cli, discovery, parsers

config_auditor/discovery.py

@@ -0,0 +1,119 @@
+from dataclasses import dataclass
+from pathlib import Path
+from typing import List, Optional
+
+
+@dataclass
+class ConfigFile:
+    path: Path
+    format: str
+    name: str
+
+
+class ConfigDiscovery:
+    PATTERNS = {
+        "json": [
+            "package.json",
+            "tsconfig*.json",
+            ".eslintrc*.json",
+            ".prettierrc*.json",
+            ".babelrc*.json",
+            "webpack.config.js",
+            "rollup.config.js",
+            "nest-cli.json",
+            "angular.json",
+            "pyproject.toml",
+            "poetry.toml",
+            "mypy.ini",
+            "pytest.ini",
+            "setup.cfg",
+            ".coveragerc",
+            "tox.ini",
+        ],
+        "yaml": [
+            ".gitlab-ci.yml",
+            ".github/**/*.yml",
+            ".github/**/*.yaml",
+            "docker-compose.yml",
+            "*.k8s.yaml",
+            "*.k8s.yml",
+            ".ansible.yml",
+            ".ansible/**/*.yml",
+        ],
+        "toml": [
+            "pyproject.toml",
+            "poetry.toml",
+            "Cargo.toml",
+            "PDM.toml",
+        ],
+    }
+
+    def __init__(self, max_depth: int = 3):
+        self.max_depth = max_depth
+
+    def discover(self, path: Path) -> List[ConfigFile]:
+        config_files = []
+
+        for format_patterns in self.PATTERNS.values():
+            for pattern in format_patterns:
+                matches = self._find_matches(path, pattern)
+                for match in matches:
+                    format_type = self._detect_format(match)
+                    if format_type:
+                        config_files.append(ConfigFile(
+                            path=match,
+                            format=format_type,
+                            name=match.name
+                        ))
+
+        return config_files
+
+    def _find_matches(self, base_path: Path, pattern: str) -> List[Path]:
+        matches = []
+
+        if "**" in pattern:
+            parts = pattern.split("**")
+            if len(parts) > 1:
+                search_pattern = parts[-1].lstrip("/")
+            else:
+                search_pattern = "*"
+            for p in base_path.rglob(search_pattern):
+                if p.is_file() and self._matches_pattern(p.name, search_pattern):
+                    matches.append(p)
+        elif "*" in pattern:
+            for p in base_path.rglob(pattern):
+                if p.is_file():
+                    matches.append(p)
+        else:
+            full_path = base_path / pattern
+            if full_path.exists():
+                matches.append(full_path)
+
+        return list(set(matches))
+
+    def _matches_pattern(self, filename: str, pattern: str) -> bool:
+        from fnmatch import fnmatch
+        return fnmatch(filename, pattern)
+
+    def _detect_format(self, path: Path) -> Optional[str]:
+        suffix = path.suffix.lower()
+        name = path.name.lower()
+
+        if suffix == ".json" or "eslintrc" in name or "babelrc" in name:
+            return "json"
+        elif suffix in (".yaml", ".yml"):
+            return "yaml"
+        elif suffix == ".toml" or name == "pyproject.toml":
+            return "toml"
+
+        if path.exists():
+            try:
+                content = path.read_text()
+                if content.strip().startswith("{") or content.strip().startswith("["):
+                    return "json"
+                elif content.strip().startswith("---") or ":" in content.split("\n")[0]:
+                    return "yaml"
+            except Exception:
+                pass
+
+        return None