Xcode Diagnostics MCP Plugin

by leftspin
Verified
#!/usr/bin/env python3 """ Xcode Diagnostics MCP Plugin ---------------------------- Extracts and parses Xcode build errors and warnings from DerivedData logs. """ import os import glob import json import gzip import re import subprocess import sys import logging import uuid from pathlib import Path from typing import List, Dict, Any, Optional, Union, Callable from dataclasses import dataclass from datetime import datetime # Import MCP SDK try: from mcp.server import Server, StdioServerTransport from mcp.types import tool, Tool HAS_MCP_SDK = True except ImportError: HAS_MCP_SDK = False logging.warning("MCP SDK not installed. To install: pip install 'mcp'") # Set up logging logging.basicConfig( level=logging.DEBUG, filename='/tmp/xcode-mcp-debug.log', filemode='a', format='%(asctime)s - %(name)s - %(levelname)s - %(message)s' ) logger = logging.getLogger('xcode-diagnostics-mcp') @dataclass class DiagnosticIssue: """Represents a single diagnostic issue from Xcode build logs.""" type: str # 'error' or 'warning' message: str file_path: Optional[str] = None line_number: Optional[int] = None column: Optional[int] = None character_range: Optional[str] = None code: Optional[str] = None # Error/warning code if available notes: List[Dict[str, Any]] = None # Associated notes, suggestions, fixes def __post_init__(self): # Initialize notes as empty list if None if self.notes is None: self.notes = [] class XcodeDiagnostics: """Main class for extracting diagnostics from Xcode build logs.""" def __init__(self): self.derived_data_path = os.path.expanduser("~/Library/Developer/Xcode/DerivedData") def list_xcode_projects(self) -> List[Dict[str, Any]]: """ Lists all Xcode projects found in DerivedData directory. Returns: List of dictionaries containing project info """ if not os.path.exists(self.derived_data_path): return [] # Use os.scandir which is more efficient than os.listdir + os.path.join project_info = [] # Collect projects with modification times for sorting for entry in os.scandir(self.derived_data_path): if entry.is_dir(): dir_name = entry.name # Extract project name from directory name (ProjectName-hash format) parts = dir_name.split('-', 1) project_name = parts[0] if parts else dir_name # Check if it has Logs/Build directory build_logs_dir = os.path.join(entry.path, "Logs", "Build") has_build_logs = os.path.exists(build_logs_dir) # Get modification time for sorting try: # entry.stat() is more efficient than os.path.getmtime mtime = entry.stat().st_mtime except: mtime = 0 # Default to oldest if we can't get mtime # Store all the info we need project_info.append(( mtime, { "project_name": project_name, "directory_name": dir_name, "full_path": entry.path, "has_build_logs": has_build_logs, "last_modified": datetime.fromtimestamp(mtime).isoformat() } )) # Sort by modification time (most recent first) and extract just the project dictionaries project_info.sort(reverse=True, key=lambda x: x[0]) projects = [info[1] for info in project_info] return projects def get_latest_build_log(self, project_dir_name: str) -> Optional[str]: """ Get the path to the latest build log for a specific project. Args: project_dir_name: Directory name for the project in DerivedData Returns: Path to the latest build log file, or None if not found """ project_path = os.path.join(self.derived_data_path, project_dir_name) build_logs_dir = os.path.join(project_path, "Logs", "Build") if not os.path.exists(build_logs_dir): return None # List all .xcactivitylog files and sort by modification time log_files = glob.glob(f"{build_logs_dir}/*.xcactivitylog") if not log_files: return None log_files.sort(key=os.path.getmtime, reverse=True) return log_files[0] def extract_diagnostics(self, project_dir_name: str, include_warnings: bool = True) -> Dict[str, Any]: """ Extracts errors and warnings from the latest build log of a project. Args: project_dir_name: Directory name for the project in DerivedData include_warnings: Whether to include warnings (not just errors) Returns: Dictionary containing parsed diagnostics information """ log_file = self.get_latest_build_log(project_dir_name) if not log_file: return { "success": False, "message": f"No build logs found for project {project_dir_name}", "errors": [], "warnings": [] } logger.debug(f"Processing log file: {log_file}") logger.debug(f"Project directory: {project_dir_name}") # Use a direct grep approach first to check for common error patterns # This is a fallback mechanism to ensure we don't miss critical errors direct_errors = [] try: # Look directly for common error patterns using grep error_patterns = [ "error:", "variable already has a getter", "previous definition" ] for pattern in error_patterns: try: cmd = f"gunzip -c '{log_file}' | grep -A 2 -B 2 '{pattern}'" grep_result = subprocess.check_output(cmd, shell=True, encoding='latin-1', stderr=subprocess.DEVNULL) if pattern in grep_result: logger.debug(f"Found '{pattern}' in build log using direct grep") logger.debug(f"Context: {grep_result[:500]}...") # Log just the first 500 chars # Save this raw data to aid in debugging with open(f'/tmp/xcode-diagnostic-{pattern.replace(":", "")}.log', 'w') as f: f.write(grep_result) # For the getter error specifically, try to parse and create direct diagnostic issues if pattern == "variable already has a getter": lines = grep_result.split('\n') for line in lines: if "variable already has a getter" in line: # Try to extract file, line and column information getter_match = re.search(r'([^:]+\.swift):(\d+):(\d+): error: variable already has a getter', line) if getter_match: file_path = getter_match.group(1) line_number = int(getter_match.group(2)) column = int(getter_match.group(3)) # Create a direct diagnostic for this error direct_error = DiagnosticIssue( type="error", message="variable already has a getter", file_path=file_path, line_number=line_number, column=column ) # Look for the related note about previous definition for note_line in lines: if "previous definition of getter" in note_line and file_path in note_line: note_match = re.search(r'([^:]+\.swift):(\d+):(\d+): note: previous definition of getter here', note_line) if note_match: note_file = note_match.group(1) note_line_num = int(note_match.group(2)) note_column = int(note_match.group(3)) note = { "type": "note", "message": "previous definition of getter here", "file_path": note_file, "line_number": note_line_num, "column": note_column, "suggested_fix": None, "code_context": None } direct_error.notes.append(note) # Add to our direct errors list direct_errors.append(direct_error) logger.debug(f"Created direct diagnostic for getter error at {file_path}:{line_number}") except Exception as e: logger.debug(f"Error searching for '{pattern}': {str(e)}") except Exception as e: logger.debug(f"Error during direct grep search: {str(e)}") # Extract data from gzipped log file using our main parser issues = self._parse_log_file(log_file, include_warnings) # Add any direct errors we found to the main issues list if direct_errors: logger.debug(f"Adding {len(direct_errors)} direct errors to the issues list") # Merge direct errors with issues found by the main parser issues.extend(direct_errors) # Enhanced search for concurrency-related warnings if include_warnings: # Debug log about searching for concurrency warnings logger.debug(f"Looking for concurrency-related warnings in log file: {log_file}") # Search for concurrency-related phrases and patterns in the raw log concurrency_terms = [ "concurrency-safe", "nonisolated global", "Swift 6 language mode", "thread safety", "actor isolation", "sendable" ] # Use grep to search through the decompressed log (this gives us context) for term in concurrency_terms: try: # Use a more targeted grep approach cmd = f"gunzip -c '{log_file}' | grep -A 3 -B 3 '{term}'" grep_result = subprocess.check_output(cmd, shell=True, encoding='latin-1', stderr=subprocess.DEVNULL) if term in grep_result: logger.debug(f"Found '{term}' in build log using grep") logger.debug(f"Context: {grep_result[:500]}...") # Log just the first 500 chars except Exception as e: logger.debug(f"Error searching for '{term}': {str(e)}") # Process issues, ensuring notes are serialized correctly processed_errors = [] processed_warnings = [] for issue in issues: issue_dict = vars(issue) # Ensure notes are correctly processed if issue.notes: issue_dict["notes"] = issue.notes # This ensures notes are included if issue.type == 'error': processed_errors.append(issue_dict) elif issue.type == 'warning' and include_warnings: processed_warnings.append(issue_dict) # Search for any specific properties that might have concurrency warnings concurrency_properties_found = [] concurrency_warning_lines = [] # Check if the log file is readable if os.path.exists(log_file) and os.access(log_file, os.R_OK): try: # Use subprocess to decompress and search for specific concurrency-related patterns search_patterns = ["concurrency-safe", "global shared", "Swift 6", "actor isolation"] for pattern in search_patterns: try: cmd = f"gunzip -c '{log_file}' | grep -A 5 -B 5 '{pattern}'" search_result = subprocess.check_output(cmd, shell=True, encoding='latin-1', stderr=subprocess.DEVNULL) if pattern in search_result: concurrency_properties_found.append(pattern) concurrency_warning_lines.extend(search_result.split('\n')) logger.debug(f"Found concurrency pattern '{pattern}' in log") except subprocess.CalledProcessError: # grep returns exit code 1 if no matches, which raises CalledProcessError pass except Exception as e: logger.debug(f"Error searching for concurrency patterns: {str(e)}") return { "success": True, "log_file": log_file, "timestamp": datetime.fromtimestamp(os.path.getmtime(log_file)).isoformat(), "errors": processed_errors, "warnings": processed_warnings, "error_count": len(processed_errors), "warning_count": len(processed_warnings), "debug_info": { "concurrency_properties_found": concurrency_properties_found, "concurrency_warning_context": concurrency_warning_lines[:20] if concurrency_warning_lines else [], "parsing_info": { "patterns_used": [ "main_pattern", "alt_pattern", "swift_getter_pattern", "swift_error_pattern", "concurrency_pattern", "backup_concurrency_pattern" ], "direct_extraction_used": True, "getter_error_detected": "variable already has a getter" in str(concurrency_warning_lines) or any("variable already has a getter" in str(error) for error in processed_errors) } } } def _parse_log_file(self, log_file: str, include_warnings: bool) -> List[DiagnosticIssue]: """ Parses a .xcactivitylog file to extract error and warning information. Args: log_file: Path to the .xcactivitylog file include_warnings: Whether to include warnings Returns: List of DiagnosticIssue objects """ issues = [] try: # Use subprocess to decompress and search the log file cmd = f"gunzip -c '{log_file}' | strings" output = subprocess.check_output(cmd, shell=True, encoding='latin-1') # Additional debugging - save raw output to a file for analysis debug_log_file = '/tmp/xcode-diagnostic-raw.log' with open(debug_log_file, 'w') as f: f.write(output) logger.debug(f"Saved raw log output to {debug_log_file}") # Enhanced debug approach - search for all error/warning/note lines directly # This helps catch any diagnostic format we might miss with our patterns diagnostic_indicators = [": error:", ": warning:", ": note:"] for indicator in diagnostic_indicators: if indicator in output: line_indexes = [i for i, line in enumerate(output.split('\n')) if indicator in line] logger.debug(f"Found diagnostic indicator '{indicator}' in {len(line_indexes)} lines") # Direct extraction - immediately capture all these lines as issues # This ensures we don't miss any diagnostics, regardless of format lines = output.split('\n') for idx in line_indexes: line = lines[idx] # Try to extract information using a more permissive pattern simple_pattern = r'([^:]+):(\d+)(?::(\d+))?: (error|warning|note): (.+)' direct_match = re.search(simple_pattern, line) if direct_match: # Extract diagnostic information file_path = direct_match.group(1) line_number = int(direct_match.group(2)) column = int(direct_match.group(3)) if direct_match.group(3) else 1 issue_type = direct_match.group(4) message = direct_match.group(5).strip() # Skip warnings if not requested if issue_type == 'warning' and not include_warnings: continue # Skip notes as standalone issues (they'll be attached to their parent issues) if issue_type == 'note': continue logger.debug(f"Direct extraction: {file_path}:{line_number} - {issue_type}: {message}") # Look for context - the next line might be code code_context = None if idx + 1 < len(lines) and not any(ind in lines[idx+1] for ind in diagnostic_indicators): code_context = lines[idx+1].strip() # Create diagnostic instance # Notes will be handled separately during the main processing loop if issue_type in ['error', 'warning']: new_issue = DiagnosticIssue( type=issue_type, message=message, file_path=file_path, line_number=line_number, column=column, code=code_context ) # Check if this issue is already captured to avoid duplicates # Simple deduplication based on file, line, and message duplicate = False for existing in issues: if (existing.file_path == file_path and existing.line_number == line_number and existing.message == message): duplicate = True break if not duplicate: issues.append(new_issue) # Log a sample of diagnostic lines to help debug regex patterns sample_size = min(10, len(line_indexes)) for idx in line_indexes[:sample_size]: line = lines[idx] logger.debug(f"Sample diagnostic line: {line}") # Extract surrounding context for debugging start = max(0, idx - 3) end = min(len(lines), idx + 3) context = '\n'.join(lines[start:end]) logger.debug(f"Context for diagnostic at line {idx}:\n{context}") # Pattern for concurrency-related warnings in Swift files concurrency_pattern = r'([^:]+\.swift):(\d+):(\d+): (warning): (.+(?:concurrency-safe|global shared|Swift 6|nonisolated).+)' # Backup pattern for any type of concurrency warning backup_concurrency_pattern = r'([^:]+):(\d+):(\d+): (warning): (.+(?:concurrency|thread safety|isolation|actor).+)' # Pattern for Swift-specific errors like duplicate getters swift_getter_pattern = r'([^:]+\.swift):(\d+):(\d+): (error): (variable already has a getter)' # General Swift error pattern (more permissive than the main pattern) swift_error_pattern = r'([^:]+\.swift):(\d+):(\d+): (error|warning|note): (.+)' # Debug logging for checking raw lines from the log that contain diagnostics # This is to ensure we're capturing everything without relying on specific keywords swift_file_pattern = r'\.swift:' diagnostic_types = ["error:", "warning:", "note:"] for line in output.split('\n'): # Check if this looks like a diagnostic line for a Swift file if re.search(swift_file_pattern, line): for diag_type in diagnostic_types: if diag_type in line: logger.debug(f"Found Swift file diagnostic: {line}") # Check if our regex patterns will catch this if not re.search(r'([^:]+):(\d+):(\d+): (error|warning|note):', line): logger.debug("WARNING: This diagnostic might be missed by our regex patterns") # Split the output into lines for processing lines = output.split('\n') i = 0 # Store the source code context for diagnostic issues current_code_context = {} while i < len(lines): line = lines[i] # Extract errors and warnings using regex # Example formats: # /path/to/file.swift:10:15: error: use of unresolved identifier 'foo' # /path/to/file.swift:20:5: warning: variable declared but never used # Also handle other error formats like "Expected '{'" # And errors like "variable already has a getter" # Main pattern for standard errors/warnings with file:line:column format main_pattern = r'([^:]+):(\d+):(\d+): (error|warning|note): ([^\n]+)' # Alternative pattern for older format errors (no column) alt_pattern = r'([^:]+):(\d+): (error|warning|note): ([^\n]+)' # Pattern for concurrency-related warnings in Swift files (capturing full message) concurrency_pattern = r'([^:]+\.swift):(\d+):(\d+): (warning): (.+(?:concurrency-safe|global shared|Swift 6|nonisolated).+)' # Backup pattern for any type of concurrency warning that might be formatted differently backup_concurrency_pattern = r'([^:]+):(\d+):(\d+): (warning): (.+(?:concurrency|thread safety|isolation|actor).+)' match = re.search(main_pattern, line) if not match: # Try alternative pattern alt_match = re.search(alt_pattern, line) if alt_match: file_path = alt_match.group(1) line_number = int(alt_match.group(2)) column = 1 # Default column when not specified issue_type = alt_match.group(3) message = alt_match.group(4).strip() match = True # Set match to True to indicate we found something else: # Try the Swift getter specific pattern first getter_match = re.search(swift_getter_pattern, line) if getter_match: file_path = getter_match.group(1) line_number = int(getter_match.group(2)) column = int(getter_match.group(3)) issue_type = getter_match.group(4) message = getter_match.group(5).strip() match = True # Add a debug log for this case logger.debug(f"Found Swift getter error: {file_path}:{line_number}") logger.debug(f"Message: {message}") logger.debug(f"Original line: {line}") else: # Try the general Swift error pattern swift_match = re.search(swift_error_pattern, line) if swift_match: file_path = swift_match.group(1) line_number = int(swift_match.group(2)) column = int(swift_match.group(3)) issue_type = swift_match.group(4) message = swift_match.group(5).strip() match = True # Add a debug log for this case logger.debug(f"Found general Swift diagnostic: {file_path}:{line_number}") logger.debug(f"Type: {issue_type}, Message: {message}") logger.debug(f"Original line: {line}") else: # Try the concurrency pattern (more specific for Swift files) concurrency_match = re.search(concurrency_pattern, line) if concurrency_match: file_path = concurrency_match.group(1) line_number = int(concurrency_match.group(2)) column = int(concurrency_match.group(3)) issue_type = concurrency_match.group(4) message = concurrency_match.group(5).strip() match = True # Add a debug log for this case logger.debug(f"Found Swift concurrency warning: {file_path}:{line_number}") logger.debug(f"Message: {message}") logger.debug(f"Original line: {line}") else: # Try the backup concurrency pattern (for other concurrency issues) backup_match = re.search(backup_concurrency_pattern, line) if backup_match: file_path = backup_match.group(1) line_number = int(backup_match.group(2)) column = int(backup_match.group(3)) issue_type = backup_match.group(4) message = backup_match.group(5).strip() match = True # Add a debug log for this case logger.debug(f"Found backup concurrency warning: {file_path}:{line_number}") logger.debug(f"Message: {message}") logger.debug(f"Original line: {line}") else: # Generic fallback patterns for any errors/warnings that don't match our specific patterns # These will capture errors like "error: Multiple commands produce ..." without requiring file:line format fallback_error_pattern = r'^error: (.+)' fallback_warning_pattern = r'^warning: (.+)' fallback_error_match = re.search(fallback_error_pattern, line) if fallback_error_match: file_path = "unknown" # No file path in this format line_number = 0 # No line number column = 0 # No column issue_type = "error" message = fallback_error_match.group(1).strip() match = True logger.debug(f"Found generic error: {message}") logger.debug(f"Original line: {line}") else: fallback_warning_match = re.search(fallback_warning_pattern, line) if fallback_warning_match: file_path = "unknown" # No file path in this format line_number = 0 # No line number column = 0 # No column issue_type = "warning" message = fallback_warning_match.group(1).strip() match = True logger.debug(f"Found generic warning: {message}") logger.debug(f"Original line: {line}") if match: # For standard pattern if match is not True: # True means we matched the alternative pattern file_path = match.group(1) line_number = int(match.group(2)) column = int(match.group(3)) issue_type = match.group(4) message = match.group(5).strip() # We already set these variables for the alternative pattern # If this is a note, we'll handle it differently - we may need to associate it with a previous error if issue_type == 'note': # Look for an existing diagnostic that this note might be related to related_diagnostic = None for issue in issues: # Check if this note is about the same file if issue.file_path == file_path: # Add as a note to the most recent diagnostic from the same file related_diagnostic = issue if related_diagnostic: # This is a note that belongs to a previous diagnostic note = { "type": issue_type, "message": message, "file_path": file_path, "line_number": line_number, "column": column if 'column' in locals() else 1, "suggested_fix": None, "code_context": None } related_diagnostic.notes.append(note) # Move to the next line i += 1 continue else: # If we can't find a related diagnostic, treat it as a standalone issue # This ensures we don't miss anything issue_type = "error" # Convert note to error if standalone # Skip warnings if not requested if issue_type == 'warning' and not include_warnings: i += 1 continue # Create the diagnostic issue diagnostic = DiagnosticIssue( type=issue_type, message=message, file_path=file_path, line_number=line_number, column=column if 'column' in locals() else 1 ) # Look ahead for any associated notes, code context, and caret lines j = i + 1 # Store code context snippet if available if j < len(lines) and not re.search(main_pattern, lines[j]) and not lines[j].strip().startswith('/'): code_line = lines[j].rstrip() diagnostic.code = code_line current_code_context[f"{file_path}:{line_number}"] = code_line context_lines = [] while j < len(lines): note_line = lines[j] # Check for a note line - these typically follow the main diagnostic # Example: /path/to/file.swift:11:16: note: convert 'activityIdentifier' to a 'let' constant # Example: /path/to/file.swift:538:9: note: previous definition of getter here # We'll reuse the main pattern here since we've already added 'note' to it note_match = re.search(main_pattern, note_line) # Only consider it a note if it's actually a "note" type note_match = note_match if note_match and note_match.group(4) == 'note' else None # Debug any potential notes with important keywords to ensure we're capturing them if "previous definition" in note_line or "already has" in note_line: logger.debug(f"Important note found: {note_line}") # Force the pattern to match if it's a valid note but our main pattern missed it if not note_match and re.search(r'([^:]+\.swift):(\d+):(\d+): note:', note_line): logger.debug("Forcing note match for important diagnostic") # Custom extraction for this special case parts = re.search(r'([^:]+\.swift):(\d+):(\d+): note: ([^\n]+)', note_line) if parts: note_match = parts # Check for a fix-it line - these typically include suggested code changes # Example: static var activityIdentifier # ~~~ ^ # let fixit_pattern = r'\s*([\^~]+)\s*' fixit_match = re.search(fixit_pattern, note_line) # Check if this is a code context line code_context_pattern = r'^\s+.*$' # Indented line without file path code_match = re.search(code_context_pattern, note_line) and not fixit_match if note_match: # This is a note line note_file_path = note_match.group(1) note_line_number = int(note_match.group(2)) note_column = int(note_match.group(3)) note_type = note_match.group(4) note_message = note_match.group(5).strip() if len(note_match.groups()) >= 5 else "note" # Log when we found an important diagnostic like "previous definition" if "previous definition" in note_message or "already has" in note_message: logger.debug(f"Found important relationship: {note_message}") note = { "type": note_type, "message": note_message, "file_path": note_file_path, "line_number": note_line_number, "column": note_column, "suggested_fix": None, "code_context": None, "related_to_line": line_number, # Store which diagnostic line this note relates to "related_to_file": file_path # Store which file this note relates to } # Check for code context for this note context_key = f"{note_file_path}:{note_line_number}" if context_key in current_code_context: note["code_context"] = current_code_context[context_key] # Store relationship data for error tracking if "previous definition" in note_message: # This creates a link between the current error and a note's location logger.debug(f"Linking current diagnostic at {file_path}:{line_number} to previous definition at {note_file_path}:{note_line_number}") # Look ahead for suggested fix if j + 1 < len(lines): next_line = lines[j+1] # Check if next line is a caret line if re.search(fixit_pattern, next_line): # There's a fix-it indicator line j += 1 fixit_indicator = next_line note["fixit_indicator"] = fixit_indicator.rstrip() # And possibly a fix-it replacement line if j + 1 < len(lines) and not re.search(main_pattern, lines[j+1]) and not re.search(note_pattern, lines[j+1]): j += 1 fixit_replacement = lines[j].strip() note["suggested_fix"] = fixit_replacement diagnostic.notes.append(note) j += 1 elif fixit_match: # This is a caret line (^) indicating the position of the error # If there's no note line before it caret_line = note_line.rstrip() # If we haven't added the code context yet, do it now if not diagnostic.code and j > 0 and not re.search(main_pattern, lines[j-1]): diagnostic.code = lines[j-1].rstrip() # Store the caret line diagnostic.character_range = caret_line # Check if the next line is a fix suggestion if j + 1 < len(lines) and not re.search(main_pattern, lines[j+1]) and not re.search(fixit_pattern, lines[j+1]): j += 1 fix_suggestion = lines[j].strip() # Create an implicit note for the fix suggestion implicit_note = { "type": "implicit_fix", "message": "Fix suggestion", "file_path": file_path, "line_number": line_number, "column": column, "suggested_fix": fix_suggestion, "fixit_indicator": caret_line } diagnostic.notes.append(implicit_note) j += 1 elif code_match: # This is a code context line context_lines.append(note_line.rstrip()) # If we don't have code context yet for the main diagnostic, use this if not diagnostic.code: diagnostic.code = note_line.rstrip() j += 1 else: # This is not a related line, so we've reached the end of this diagnostic break # Add the diagnostic issue to our list issues.append(diagnostic) i = j # Move past all the notes we processed else: i += 1 except Exception as e: # If there's an error parsing, add a special "meta" error logger.exception(f"Error parsing log file: {str(e)}") issues.append(DiagnosticIssue( type="error", message=f"Error parsing log file: {str(e)}", file_path=log_file )) return issues # MCP Protocol Implementation using SDK if available if HAS_MCP_SDK: class XcodeDiagnosticsMcpServer: """MCP Server implementation using the official MCP SDK.""" def __init__(self): self.xcode = XcodeDiagnostics() self.server = Server( server_info={"name": "xcode-diagnostics", "version": "1.0.0"}, capabilities={ "tools": {}, } ) # Register tools self.setup_tools() def setup_tools(self): """Register all tools with the MCP server.""" @tool( name="get_xcode_projects", description="Lists all Xcode projects that have build logs in the DerivedData directory." ) async def get_xcode_projects(): """Lists all Xcode projects with build logs in DerivedData.""" projects = self.xcode.list_xcode_projects() return {"projects": projects} @tool( name="get_project_diagnostics", description="Gets diagnostic information (errors and warnings) from the latest build log of a specific project." ) async def get_project_diagnostics( project_dir_name: str, include_warnings: bool = True ): """ Gets diagnostic information from a specific project's build log. Args: project_dir_name: Directory name of the project in DerivedData include_warnings: Whether to include warnings in addition to errors """ return self.xcode.extract_diagnostics(project_dir_name, include_warnings) # Register the tools with the server self.server.tools.register(get_xcode_projects) self.server.tools.register(get_project_diagnostics) async def run(self): """Run the MCP server using stdio transport.""" logger.info("Starting Xcode Diagnostics MCP server with MCP SDK") transport = StdioServerTransport() await self.server.connect(transport) logger.info("MCP server connected and running") else: # Legacy MCP implementation for backward compatibility class McpServer: """ Implements the MCP protocol for the Xcode Diagnostics plugin. """ def __init__(self): self.methods = { "initialize": self.initialize, "shutdown": self.shutdown, "mcp.list_tools": self.list_tools, "mcp.call_tool": self.call_tool, "tools/list": self.list_tools, # New method path format "tools/call": self.call_tool, # New method path format "prompts/list": self.list_prompts, # New method path } self.tools = [ { "name": "get_xcode_projects", "description": "Lists all Xcode projects that have build logs in the DerivedData directory.", "inputSchema": { "type": "object", "properties": {}, "required": [] } }, { "name": "get_project_diagnostics", "description": "Gets diagnostic information (errors and warnings) from the latest build log of a specific project.", "inputSchema": { "type": "object", "properties": { "project_dir_name": { "type": "string", "description": "Directory name of the project in DerivedData (e.g. 'ProjectName-hash')" }, "include_warnings": { "type": "boolean", "description": "Whether to include warnings in addition to errors", "default": True } }, "required": ["project_dir_name"] } } ] self.tool_functions = { "get_xcode_projects": self._get_xcode_projects, "get_project_diagnostics": self._get_project_diagnostics } def _get_xcode_projects(self, params): """Implementation of get_xcode_projects for MCP.""" diagnostics = XcodeDiagnostics() projects = diagnostics.list_xcode_projects() return {"projects": projects} def _get_project_diagnostics(self, params): """Implementation of get_project_diagnostics for MCP.""" project_dir_name = params.get("project_dir_name") include_warnings = params.get("include_warnings", True) diagnostics = XcodeDiagnostics() return diagnostics.extract_diagnostics(project_dir_name, include_warnings) def initialize(self, params): """Handle initialize method, required by the MCP protocol.""" logger.info("Initializing MCP server with params: %s", params) capabilities = params.get("capabilities", {}) client_info = params.get("client_info", {}) # Respond with server information and capabilities return { "serverInfo": { "name": "xcode-diagnostics", "version": "1.0.0" }, "protocolVersion": "2024-11-05", "capabilities": { "tools": {}, "resources": {}, "prompts": {} } } def shutdown(self, params): """Handle shutdown method, required by the MCP protocol.""" logger.info("Shutting down MCP server") return {} def list_tools(self, params): """Handle tools/list and mcp.list_tools methods.""" return { "tools": self.tools } def list_prompts(self, params): """Handle prompts/list method.""" # No prompts in this implementation return { "prompts": [] } def call_tool(self, params): """Handle mcp.call_tool and tools/call methods.""" tool_name = params.get("name") tool_params = params.get("parameters", {}) # Also handle new format where arguments might be used instead of parameters if not tool_params and "arguments" in params: tool_params = params.get("arguments", {}) if tool_name not in self.tool_functions: return { "error": { "code": -32601, "message": f"Tool '{tool_name}' not found" } } try: result = self.tool_functions[tool_name](tool_params) # Format the result according to the latest MCP protocol return { "content": [ { "type": "text", "text": json.dumps(result) } ] } except Exception as e: logger.exception(f"Error calling tool {tool_name}") return { "error": { "code": -32000, "message": f"Tool execution error: {str(e)}" } } def handle_request(self, request): """ Handle a JSON-RPC request and return a response. """ logger.debug(f"Received request: {request}") request_id = request.get("id") method = request.get("method") params = request.get("params", {}) # Check for required fields if not method: return { "jsonrpc": "2.0", "id": request_id, "error": { "code": -32600, "message": "Invalid Request: missing method" } } # Handle method if method in self.methods: try: result = self.methods[method](params) return { "jsonrpc": "2.0", "id": request_id, "result": result } except Exception as e: logger.exception(f"Error handling method {method}") return { "jsonrpc": "2.0", "id": request_id, "error": { "code": -32000, "message": f"Server error: {str(e)}" } } else: return { "jsonrpc": "2.0", "id": request_id, "error": { "code": -32601, "message": f"Method '{method}' not found" } } def process_line(self, line): """ Process a line of input (JSON-RPC request) and return a response. """ try: request = json.loads(line) response = self.handle_request(request) return json.dumps(response) except json.JSONDecodeError: logger.error(f"Invalid JSON: {line}") return json.dumps({ "jsonrpc": "2.0", "id": None, "error": { "code": -32700, "message": "Parse error: invalid JSON" } }) def run(self): """ Run the MCP server, reading from stdin and writing to stdout. """ logger.info("Starting Xcode Diagnostics MCP server (legacy implementation)") try: for line in sys.stdin: line = line.strip() if line: response = self.process_line(line) print(response, flush=True) logger.debug(f"Sent response: {response}") except Exception as e: logger.exception("Fatal error in MCP server") sys.exit(1) # Function implementations outside the class for testing/debugging def get_xcode_projects(): """List all Xcode projects with build logs in DerivedData.""" diagnostics = XcodeDiagnostics() projects = diagnostics.list_xcode_projects() return json.dumps({"projects": projects}) def get_project_diagnostics(project_dir_name: str, include_warnings: bool = True): """Get diagnostic information from the latest build log of a project.""" diagnostics = XcodeDiagnostics() result = diagnostics.extract_diagnostics(project_dir_name, include_warnings) return json.dumps(result) # When run directly, start the MCP server if __name__ == "__main__": # Check if the --debug flag is passed if len(sys.argv) > 1 and sys.argv[1] == "--debug": # Run in debug mode - just print projects result = get_xcode_projects() print(result) else: # Run as an MCP server if HAS_MCP_SDK: # Use the MCP SDK implementation import asyncio server = XcodeDiagnosticsMcpServer() asyncio.run(server.run()) else: # Use the legacy implementation server = McpServer() server.run() # If we reach here, the server has stopped logger.info("MCP server stopped") sys.exit(0)