Add rules, fixes, and llm modules

config_auditor/llm.py

@@ -0,0 +1,122 @@
+from abc import ABC, abstractmethod
+from typing import Any, Dict, Optional
+import json
+import requests
+
+
+class LLMClient(ABC):
+    @abstractmethod
+    def get_recommendation(self, config_type: str, issues: list) -> str:
+        pass
+
+
+class OllamaClient(LLMClient):
+    def __init__(self, endpoint: str = "http://localhost:11434", model: str = "llama3"):
+        self.endpoint = endpoint
+        self.model = model
+
+    def is_available(self) -> bool:
+        try:
+            response = requests.get(f"{self.endpoint}/api/tags", timeout=5)
+            return response.status_code == 200
+        except Exception:
+            return False
+
+    def get_recommendation(self, config_type: str, issues: list) -> str:
+        prompt = self._build_prompt(config_type, issues)
+
+        try:
+            response = requests.post(
+                f"{self.endpoint}/api/generate",
+                json={
+                    "model": self.model,
+                    "prompt": prompt,
+                    "stream": False
+                },
+                timeout=30
+            )
+
+            if response.status_code == 200:
+                result = response.json()
+                return result.get("response", "No recommendation available")
+        except Exception as e:
+            return f"LLM unavailable: {str(e)}. Falling back to rule-based suggestions."
+
+        return "Unable to get recommendation from LLM"
+
+    def _build_prompt(self, config_type: str, issues: list) -> str:
+        issues_text = "\n".join([f"- {i.get('message', '')}" for i in issues]) if issues else "No issues found"
+
+        prompt = f"""You are a configuration expert. Analyze the following {config_type} configuration issues and provide recommendations:
+
+Issues found:
+{issues_text}
+
+Please provide:
+1. A brief explanation of the issues
+2. Specific fix recommendations with code examples
+3. Best practices for this configuration type
+
+Format your response clearly with sections."""
+
+        return prompt
+
+
+class FallbackClient(LLMClient):
+    RECOMMENDATIONS = {
+        "deprecated-package": "Replace deprecated packages with their modern equivalents. Check the official documentation for migration guides.",
+        "outdated-version": "Consider updating dependencies to stable versions. Major version 0.x packages may have breaking changes.",
+        "missing-type-checking": "Enable strict mode in TypeScript for better type safety and improved developer experience.",
+        "missing-scripts": "Add standard scripts (test, build) to improve development workflow and CI/CD integration.",
+        "security-vulnerability": "Review security settings and ensure production configurations don't expose sensitive data.",
+    }
+
+    def get_recommendation(self, config_type: str, issues: list) -> str:
+        recommendations = []
+
+        for issue in issues:
+            category = issue.get("category", "")
+            if category in self.RECOMMENDATIONS:
+                recommendations.append(self.RECOMMENDATIONS[category])
+
+        if recommendations:
+            return "\n".join(recommendations)
+
+        return "Review the configuration for common best practices in your project type."
+
+
+class LLMProvider:
+    def __init__(self, endpoint: str = "http://localhost:11434", model: str = "llama3"):
+        self.endpoint = endpoint
+        self.model = model
+        self._client: Optional[LLMClient] = None
+
+    def get_client(self) -> LLMClient:
+        if self._client is None:
+            try:
+                client = OllamaClient(self.endpoint, self.model)
+                if client.is_available():
+                    self._client = client
+                else:
+                    self._client = FallbackClient()
+            except Exception:
+                self._client = FallbackClient()
+        return self._client
+
+    def get_recommendation(self, config_type: str, issues: list) -> str:
+        client = self.get_client()
+        return client.get_recommendation(config_type, issues)
+
+
+def load_config() -> Dict[str, Any]:
+    try:
+        with open("config.yaml", "r") as f:
+            import yaml
+            return yaml.safe_load(f) or {}
+    except FileNotFoundError:
+        return {}
+
+
+def get_llm_config() -> Dict[str, Any]:
+    config = load_config()
+    return config.get("llm", {"endpoint": "http://localhost:11434", "model": "llama3"})