import json import logging from typing import List, Dict, Any, Optional, Union # Ensure Union is imported logger = logging.getLogger(__name__) class BaseEndpoint: """所有端点对象的基类，可以包含一些通用属性或方法。""" def __init__(self, method: str, path: str): self.method = method self.path = path def to_dict(self) -> Dict[str, Any]: # 基类可以提供一个默认的 to_dict 实现或要求子类实现 raise NotImplementedError("Subclasses must implement to_dict") class YAPIEndpoint(BaseEndpoint): # Inherit from BaseEndpoint def __init__(self, data: Dict[str, Any], category_name: Optional[str] = None, category_id: Optional[int] = None): super().__init__(method=data.get("method", "GET").upper(), path=data.get("path", "")) self._raw_data = data self.title: str = data.get("title", "") self.desc: Optional[str] = data.get("desc") self._id: int = data.get("_id") self.project_id: int = data.get("project_id") self.catid: int = data.get("catid") self.req_params: List[Dict[str, Any]] = data.get("req_params", []) self.req_query: List[Dict[str, Any]] = data.get("req_query", []) self.req_headers: List[Dict[str, Any]] = data.get("req_headers", []) self.req_body_form: List[Dict[str, Any]] = data.get("req_body_form", []) self.req_body_type: Optional[str] = data.get("req_body_type") self.req_body_is_json_schema: bool = data.get("req_body_is_json_schema", False) self.req_body_other: Optional[str] = data.get("req_body_other") self.res_body_type: Optional[str] = data.get("res_body_type") self.res_body_is_json_schema: bool = data.get("res_body_is_json_schema", False) self.res_body: Optional[str] = data.get("res_body") self.status: str = data.get("status", "undone") self.api_opened: bool = data.get("api_opened", False) self.uid: int = data.get("uid") self.category_name = category_name self.category_id = category_id if category_id is not None else self.catid self._parsed_req_body_schema: Optional[Dict[str, Any]] = None if self.req_body_type == "json" and self.req_body_other and self.req_body_is_json_schema: try: self._parsed_req_body_schema = json.loads(self.req_body_other) except json.JSONDecodeError as e: logger.error(f"YAPIEndpoint (ID: {self._id}, Title: {self.title}): Failed to parse req_body_other as JSON during init: {e}. Content: {self.req_body_other[:200]}") self._parsed_res_body_schema: Optional[Dict[str, Any]] = None if self.res_body_type == "json" and self.res_body and self.res_body_is_json_schema: try: self._parsed_res_body_schema = json.loads(self.res_body) except json.JSONDecodeError as e: logger.error(f"YAPIEndpoint (ID: {self._id}, Title: {self.title}): Failed to parse res_body as JSON during init: {e}. Content: {self.res_body[:200]}") def to_dict(self) -> Dict[str, Any]: endpoint_dict = { "method": self.method, "path": self.path, "title": self.title, "summary": self.title, "description": self.desc or "", "operationId": f"{self.method.lower()}_{self.path.replace('/', '_').replace('{', '').replace('}', '')}_{self._id}", "tags": [self.category_name or str(self.catid)], "parameters": [], "requestBody": None, "responses": {}, "_source_format": "yapi", "_yapi_id": self._id, "_yapi_raw_data": self._raw_data # Keep raw data for debugging or deeper inspection if needed } # Path parameters from req_params for p_spec in self.req_params: param_name = p_spec.get("name") if not param_name: continue endpoint_dict["parameters"].append({ "name": param_name, "in": "path", "required": True, # Path parameters are always required "description": p_spec.get("desc", ""), "schema": {"type": "string", "example": p_spec.get("example", f"example_{param_name}")} }) # Query parameters from req_query for q_spec in self.req_query: param_name = q_spec.get("name") if not param_name: continue is_required = q_spec.get("required") == "1" # YAPI uses "1" for true param_schema = {"type": "string"} # Default to string, YAPI doesn't specify types well here if "example" in q_spec: param_schema["example"] = q_spec["example"] # Add other fields from YAPI query spec if needed (e.g., desc) endpoint_dict["parameters"].append({ "name": param_name, "in": "query", "required": is_required, "description": q_spec.get("desc", ""), "schema": param_schema }) # Header parameters from req_headers for h_spec in self.req_headers: param_name = h_spec.get("name") if not param_name or param_name.lower() == 'content-type': continue # Content-Type is handled by requestBody is_required = h_spec.get("required") == "1" default_value = h_spec.get("value") # YAPI uses 'value' for default/example header value param_schema = {"type": "string"} if default_value: if is_required: # If required, it's more like an example of what's expected param_schema["example"] = default_value else: # If not required, it's a default value param_schema["default"] = default_value endpoint_dict["parameters"].append({ "name": param_name, "in": "header", "required": is_required, "description": h_spec.get("desc", ""), "schema": param_schema }) # Request body if self.req_body_type == "json" and self._parsed_req_body_schema: endpoint_dict["requestBody"] = { "content": { "application/json": { "schema": self._parsed_req_body_schema } } } elif self.req_body_type == "form" and self.req_body_form: properties = {} required_form_params = [] for form_param in self.req_body_form: name = form_param.get("name") if not name: continue properties[name] = { "type": "string", # YAPI form params are typically strings, file uploads are different "description": form_param.get("desc","") } if form_param.get("example"): properties[name]["example"] = form_param.get("example") if form_param.get("required") == "1": required_form_params.append(name) endpoint_dict["requestBody"] = { "content": { "application/x-www-form-urlencoded": { "schema": { "type": "object", "properties": properties, "required": required_form_params if required_form_params else None # OpenAPI: omit if empty } } # YAPI also supports req_body_type = 'file', which would map to multipart/form-data # This example focuses on json and basic form. } } # Add other req_body_types if necessary (e.g., raw, file) # Responses # YAPI has a simpler response structure. We'll map its res_body to a default success response (e.g., 200 or 201). default_success_status = "200" if self.method == "POST": default_success_status = "201" # Common practice for POST success if self.res_body_type == "json" and self._parsed_res_body_schema: endpoint_dict["responses"][default_success_status] = { "description": "Successful Operation (from YAPI res_body)", "content": { "application/json": { "schema": self._parsed_res_body_schema } } } elif self.res_body_type == "json" and not self._parsed_res_body_schema and self.res_body: # Schema parsing failed but text exists endpoint_dict["responses"][default_success_status] = { "description": "Successful Operation (Schema parsing error, raw text might be available)", "content": {"application/json": {"schema": {"type": "object", "description": "Schema parsing failed for YAPI res_body."}}} # Placeholder } else: # No JSON schema, or other res_body_type endpoint_dict["responses"][default_success_status] = { "description": "Successful Operation (No specific schema provided in YAPI for this response)" } # Ensure there's always a default response if nothing specific was added if not endpoint_dict["responses"]: endpoint_dict["responses"]["default"] = {"description": "Default response from YAPI definition"} return endpoint_dict def __repr__(self): return f"" class SwaggerEndpoint(BaseEndpoint): # Inherit from BaseEndpoint def __init__(self, path: str, method: str, data: Dict[str, Any], global_spec: Dict[str, Any]): super().__init__(method=method.upper(), path=path) self._raw_data = data self._global_spec = global_spec # Store for $ref resolution self.summary: Optional[str] = data.get("summary") self.description: Optional[str] = data.get("description") self.operation_id: Optional[str] = data.get("operationId") self.tags: List[str] = data.get("tags", []) # Parameters, requestBody, responses are processed by to_dict def _resolve_ref(self, ref_path: str) -> Optional[Dict[str, Any]]: """Resolves a $ref path within the global OpenAPI/Swagger spec.""" if not ref_path.startswith("#/"): logger.warning(f"Unsupported $ref path: {ref_path}. Only local refs '#/...' are currently supported.") return None parts = ref_path[2:].split('/') # Remove '#/' and split current_level = self._global_spec try: for part in parts: # Decode URI component encoding if present (e.g. "~0" for "~", "~1" for "/") part = part.replace("~1", "/").replace("~0", "~") current_level = current_level[part] # It's crucial to return a copy if the resolved ref will be modified, # or ensure modifications happen on copies later. # For now, returning as is, assuming downstream processing is careful or uses copies. if isinstance(current_level, dict): return current_level # Potentially json.loads(json.dumps(current_level)) for a deep copy else: # Resolved to a non-dict, which might be valid for some simple refs but unusual for schemas logger.warning(f"$ref '{ref_path}' resolved to a non-dictionary type: {type(current_level)}. Value: {str(current_level)[:100]}") return {"type": "string", "description": f"Resolved $ref '{ref_path}' to non-dict: {str(current_level)[:100]}"} # Placeholder except (KeyError, TypeError, AttributeError) as e: logger.error(f"Failed to resolve $ref '{ref_path}': {e}", exc_info=True) return None def _process_schema_or_ref(self, schema_like: Any) -> Optional[Dict[str, Any]]: """ Processes a schema part, resolving $refs and recursively processing nested structures. Returns a new dictionary with resolved refs, or None if resolution fails badly. """ if not isinstance(schema_like, dict): if schema_like is None: return None logger.warning(f"Expected a dictionary for schema processing, got {type(schema_like)}. Value: {str(schema_like)[:100]}") return {"type": "string", "description": f"Schema was not a dict: {str(schema_like)[:100]}"} # Placeholder for non-dict schema # If it's a $ref, resolve it. if "$ref" in schema_like: return self._resolve_ref(schema_like["$ref"]) # This will be the new base schema_like # Create a copy to avoid modifying the original spec during processing processed_schema = schema_like.copy() # Recursively process 'properties' for object schemas if "properties" in processed_schema and isinstance(processed_schema["properties"], dict): new_properties = {} for prop_name, prop_schema in processed_schema["properties"].items(): resolved_prop = self._process_schema_or_ref(prop_schema) if resolved_prop is not None: # Only add if resolution was successful new_properties[prop_name] = resolved_prop # else: logger.warning(f"Failed to process property '{prop_name}' in {self.operation_id or self.path}") processed_schema["properties"] = new_properties # Recursively process 'items' for array schemas if "items" in processed_schema and isinstance(processed_schema["items"], dict): # 'items' should be a schema object resolved_items = self._process_schema_or_ref(processed_schema["items"]) if resolved_items is not None: processed_schema["items"] = resolved_items # else: logger.warning(f"Failed to process 'items' schema in {self.operation_id or self.path}") # Handle allOf, anyOf, oneOf by trying to merge or process them (simplistic merge for allOf) # This is a complex area of JSON Schema. This is a very basic attempt. if "allOf" in processed_schema and isinstance(processed_schema["allOf"], list): merged_all_of_props = {} merged_all_of_required = set() temp_schema_for_all_of = {"type": processed_schema.get("type", "object"), "properties": {}, "required": []} for sub_schema_data in processed_schema["allOf"]: resolved_sub_schema = self._process_schema_or_ref(sub_schema_data) if resolved_sub_schema and isinstance(resolved_sub_schema, dict): if "properties" in resolved_sub_schema: temp_schema_for_all_of["properties"].update(resolved_sub_schema["properties"]) if "required" in resolved_sub_schema and isinstance(resolved_sub_schema["required"], list): merged_all_of_required.update(resolved_sub_schema["required"]) # Copy other top-level keywords from the resolved_sub_schema if needed, e.g. description for key, value in resolved_sub_schema.items(): if key not in ["properties", "required", "type", "$ref", "allOf", "anyOf", "oneOf"]: if key not in temp_schema_for_all_of or temp_schema_for_all_of[key] is None: # prioritize existing temp_schema_for_all_of[key] = value if temp_schema_for_all_of["properties"]: processed_schema["properties"] = {**processed_schema.get("properties",{}), **temp_schema_for_all_of["properties"]} if merged_all_of_required: current_required = set(processed_schema.get("required", [])) current_required.update(merged_all_of_required) processed_schema["required"] = sorted(list(current_required)) del processed_schema["allOf"] # Remove allOf after processing # Copy other merged attributes back to processed_schema for key, value in temp_schema_for_all_of.items(): if key not in ["properties", "required", "type", "$ref", "allOf", "anyOf", "oneOf"]: if key not in processed_schema or processed_schema[key] is None: processed_schema[key] = value # anyOf, oneOf are harder as they represent choices. For now, we might just list them or pick first. # For simplicity in to_dict, we might not fully expand them but ensure refs inside are resolved. for keyword in ["anyOf", "oneOf"]: if keyword in processed_schema and isinstance(processed_schema[keyword], list): processed_sub_list = [] for sub_item in processed_schema[keyword]: resolved_sub = self._process_schema_or_ref(sub_item) if resolved_sub: processed_sub_list.append(resolved_sub) if processed_sub_list: # only update if some were resolved processed_schema[keyword] = processed_sub_list return processed_schema def to_dict(self) -> Dict[str, Any]: endpoint_data = { "method": self.method, "path": self.path, "summary": self.summary or "", "title": self.summary or self.operation_id or "", # Fallback for title "description": self.description or "", "operationId": self.operation_id or f"{self.method.lower()}_{self.path.replace('/', '_').replace('{', '').replace('}', '')}", "tags": self.tags, "parameters": [], "requestBody": None, "responses": {}, "_source_format": "swagger/openapi", "_swagger_raw_data": self._raw_data, # Keep raw for debugging "_global_api_spec_for_resolution": self._global_spec # For test cases that might need to resolve further } # Process parameters if "parameters" in self._raw_data and isinstance(self._raw_data["parameters"], list): for param_data_raw in self._raw_data["parameters"]: # Each param_data_raw could itself be a $ref or contain a schema that is a $ref processed_param_container = self._process_schema_or_ref(param_data_raw) if processed_param_container and isinstance(processed_param_container, dict): # If the parameter itself was a $ref, processed_param_container is the resolved object. # If it contained a schema that was a $ref, that nested schema should be resolved. # We need to ensure 'schema' key exists if 'in' is path, query, header if "schema" in processed_param_container and isinstance(processed_param_container["schema"], dict): # schema was present, process it further (it might have been already by _process_schema_or_ref if it was a complex object) # but if _process_schema_or_ref was called on param_data_raw which wasn't a ref itself, # the internal 'schema' ref might not have been re-processed with full context. # However, the recursive nature of _process_schema_or_ref should handle nested $refs. pass # Assume it's processed by the main call to _process_schema_or_ref on param_data_raw elif "content" in processed_param_container: # Parameter described by Content Object (OpenAPI 3.x) pass # Content object schemas should have been resolved by _process_schema_or_ref endpoint_data["parameters"].append(processed_param_container) # Process requestBody if "requestBody" in self._raw_data and isinstance(self._raw_data["requestBody"], dict): processed_rb = self._process_schema_or_ref(self._raw_data["requestBody"]) if processed_rb: endpoint_data["requestBody"] = processed_rb # Process responses if "responses" in self._raw_data and isinstance(self._raw_data["responses"], dict): for status_code, resp_data_raw in self._raw_data["responses"].items(): processed_resp = self._process_schema_or_ref(resp_data_raw) if processed_resp: endpoint_data["responses"][status_code] = processed_resp elif resp_data_raw: # If processing failed but raw exists, keep raw (though this is less ideal) endpoint_data["responses"][status_code] = resp_data_raw logger.warning(f"Kept raw response data for {status_code} due to processing failure for {self.operation_id or self.path}") if not endpoint_data["responses"]: # Ensure default response if none processed endpoint_data["responses"]["default"] = {"description": "Default response from Swagger/OpenAPI definition"} return endpoint_data def __repr__(self): return f"" class ParsedAPISpec: """解析后的API规范的通用基类""" def __init__(self, spec_type: str, endpoints: List[Union[YAPIEndpoint, SwaggerEndpoint]], spec: Dict[str, Any]): self.spec_type = spec_type self.endpoints = endpoints self.spec = spec # Store the original full spec dictionary, useful for $ref resolution if not pre-resolved class ParsedYAPISpec(ParsedAPISpec): """解析后的YAPI规范""" def __init__(self, endpoints: List[YAPIEndpoint], categories: List[Dict[str, Any]], spec: Dict[str, Any]): super().__init__(spec_type="yapi", endpoints=endpoints, spec=spec) self.categories = categories class ParsedSwaggerSpec(ParsedAPISpec): """解析后的Swagger/OpenAPI规范""" def __init__(self, endpoints: List[SwaggerEndpoint], tags: List[Dict[str, Any]], spec: Dict[str, Any]): super().__init__(spec_type="swagger", endpoints=endpoints, spec=spec) self.tags = tags class InputParser: """负责解析输入（如YAPI JSON）并提取API端点信息""" def __init__(self): self.logger = logging.getLogger(__name__) def parse_yapi_spec(self, file_path: str) -> Optional[ParsedYAPISpec]: self.logger.info(f"Parsing YAPI spec from: {file_path}") all_endpoints: List[YAPIEndpoint] = [] yapi_categories: List[Dict[str, Any]] = [] raw_spec_data_list: Optional[List[Dict[str, Any]]] = None # YAPI export is a list of categories try: with open(file_path, 'r', encoding='utf-8') as f: raw_spec_data_list = json.load(f) if not isinstance(raw_spec_data_list, list): self.logger.error(f"YAPI spec file {file_path} does not contain a JSON list as expected for categories.") return None for category_data in raw_spec_data_list: if not isinstance(category_data, dict): self.logger.warning(f"Skipping non-dictionary item in YAPI spec list: {str(category_data)[:100]}") continue cat_name = category_data.get("name") cat_id = category_data.get("_id", category_data.get("id")) # YAPI uses _id yapi_categories.append({"name": cat_name, "description": category_data.get("desc"), "id": cat_id}) for endpoint_data in category_data.get("list", []): if not isinstance(endpoint_data, dict): self.logger.warning(f"Skipping non-dictionary endpoint item in category '{cat_name}': {str(endpoint_data)[:100]}") continue try: yapi_endpoint = YAPIEndpoint(endpoint_data, category_name=cat_name, category_id=cat_id) all_endpoints.append(yapi_endpoint) except Exception as e_ep: self.logger.error(f"Error processing YAPI endpoint data (ID: {endpoint_data.get('_id', 'N/A')}, Title: {endpoint_data.get('title', 'N/A')}). Error: {e_ep}", exc_info=True) # The 'spec' for ParsedYAPISpec should be a dict representing the whole document. # Since YAPI export is a list of categories, we wrap it. yapi_full_spec_dict = {"yapi_categories": raw_spec_data_list} return ParsedYAPISpec(endpoints=all_endpoints, categories=yapi_categories, spec=yapi_full_spec_dict) except FileNotFoundError: self.logger.error(f"YAPI spec file not found: {file_path}") except json.JSONDecodeError as e: self.logger.error(f"Error decoding JSON from YAPI spec file {file_path}: {e}") except Exception as e: self.logger.error(f"An unexpected error occurred while parsing YAPI spec {file_path}: {e}", exc_info=True) return None def parse_swagger_spec(self, file_path: str) -> Optional[ParsedSwaggerSpec]: self.logger.info(f"Parsing Swagger/OpenAPI spec from: {file_path}") all_endpoints: List[SwaggerEndpoint] = [] swagger_tags: List[Dict[str, Any]] = [] raw_spec_data_dict: Optional[Dict[str, Any]] = None # Swagger/OpenAPI is a single root object try: with open(file_path, 'r', encoding='utf-8') as f: # TODO: Add YAML support if needed, e.g., using PyYAML raw_spec_data_dict = json.load(f) if not isinstance(raw_spec_data_dict, dict): self.logger.error(f"Swagger spec file {file_path} does not contain a JSON object as expected.") return None swagger_tags = raw_spec_data_dict.get("tags", []) paths = raw_spec_data_dict.get("paths", {}) for path, path_item_obj in paths.items(): if not isinstance(path_item_obj, dict): continue for method, operation_obj in path_item_obj.items(): # Common methods, can be extended if method.lower() not in ["get", "post", "put", "delete", "patch", "options", "head", "trace"]: continue # Skip non-standard HTTP methods or extensions like 'parameters' at path level if not isinstance(operation_obj, dict): continue try: # Pass the full raw_spec_data_dict for $ref resolution within SwaggerEndpoint swagger_endpoint = SwaggerEndpoint(path, method, operation_obj, global_spec=raw_spec_data_dict) all_endpoints.append(swagger_endpoint) except Exception as e_ep: self.logger.error(f"Error processing Swagger endpoint: {method.upper()} {path}. Error: {e_ep}", exc_info=True) return ParsedSwaggerSpec(endpoints=all_endpoints, tags=swagger_tags, spec=raw_spec_data_dict) except FileNotFoundError: self.logger.error(f"Swagger spec file not found: {file_path}") except json.JSONDecodeError as e: self.logger.error(f"Error decoding JSON from Swagger spec file {file_path}: {e}") except Exception as e: self.logger.error(f"An unexpected error occurred while parsing Swagger spec {file_path}: {e}", exc_info=True) return None