import json import sys import gzip import os from collections import defaultdict def fxprof_to_callgrind(profile_data): """ Converts Firefox Profiler data (in JSON format) to callgrind format, including file and line number information. """ # --- Phase 1: Parsing and Initialization --- if 'threads' not in profile_data or not profile_data['threads']: raise ValueError("Profile data contains no threads.") thread = None for t in profile_data['threads']: if t.get('samples') and t['samples'].get('stack'): thread = t break if not thread: raise ValueError("No thread with samples found in the profile.") print(f"Processing thread: {thread.get('name', 'Unnamed Thread')} (PID: {thread.get('pid')}, TID: {thread.get('tid')})", file=sys.stderr) # --- Enhanced Data Structures for Location Info --- graph = defaultdict(lambda: {'self_cost': 0.0, 'inclusive_cost': 0.0, 'calls': defaultdict(float)}) # Mappings for functions func_to_id = {} id_to_func_details = {} next_func_id = 1 # Mappings for files file_to_id = {} id_to_file = {} next_file_id = 1 def get_file_id(file_name): nonlocal next_file_id if file_name is None: return None if file_name not in file_to_id: file_id = next_file_id file_to_id[file_name] = file_id id_to_file[file_id] = file_name next_file_id += 1 return file_to_id[file_name] def get_func_id(func_name, file_id, line_num): nonlocal next_func_id # A function is uniquely identified by its name and file location func_key = (func_name, file_id) if func_key not in func_to_id: func_id = next_func_id func_to_id[func_key] = func_id id_to_func_details[func_id] = { 'name': func_name, 'file_id': file_id, 'line': line_num } next_func_id += 1 return func_to_id[func_key] # Pre-fetch tables for easier access stack_table = thread['stackTable'] frame_table = thread['frameTable'] func_table = thread['funcTable'] samples = thread['samples'] # Use shared stringArray if exists, else per-thread if 'shared' in profile_data and 'stringArray' in profile_data['shared']: string_table = profile_data['shared']['stringArray'] elif 'stringArray' in thread: string_table = thread['stringArray'] elif 'stringTable' in thread: string_table = thread['stringTable'] else: raise ValueError("No string table found in profile or thread") # --- Phase 2 & 3: Stack Reconstruction and Aggregation --- for sample_idx, stack_index in enumerate(samples['stack']): if stack_index is None: continue weight = 1 if 'weight' in samples and samples['weight'][sample_idx] is not None: weight = samples['weight'][sample_idx] current_stack_indices = [] curr = stack_index while curr is not None: frame_index = stack_table['frame'][curr] func_index = frame_table['func'][frame_index] # Extract function name name_index = func_table['name'][func_index] func_name = string_table[name_index] # Extract file name and line number file_name_index = func_table['fileName'][func_index] file_name = string_table[file_name_index] if file_name_index is not None else "Unknown File" line_num = func_table['lineNumber'][func_index] file_id = get_file_id(file_name) func_id = get_func_id(func_name, file_id, line_num) current_stack_indices.append(func_id) curr = stack_table['prefix'][curr] current_stack_indices.reverse() if not current_stack_indices: continue leaf_func_id = current_stack_indices[-1] graph[leaf_func_id]['self_cost'] += weight for func_id in current_stack_indices: graph[func_id]['inclusive_cost'] += weight for i in range(len(current_stack_indices) - 1): caller_id = current_stack_indices[i] callee_id = current_stack_indices[i+1] graph[caller_id]['calls'][callee_id] += weight # --- Phase 4: Proportional Cost Calculation --- total_incoming = defaultdict(float) for caller_id, data in graph.items(): for callee_id, count in data['calls'].items(): total_incoming[callee_id] += count # --- Phase 5: Generating the callgrind Output File --- output = [] # Header output.append("# callgrind format") output.append("version: 1") cmd_name = profile_data.get('meta', {}).get('product', 'UnknownApp') output.append(f"cmd: {cmd_name}") output.append("events: Samples") output.append("") # Definitions for file_id, file_name in id_to_file.items(): output.append(f"fl=({file_id}) {file_name}") for func_id, details in id_to_func_details.items(): output.append(f"fn=({func_id}) {details['name']}") output.append("") # Body for func_id, data in graph.items(): details = id_to_func_details[func_id] if details['file_id']: output.append(f"fl=({details['file_id']})") output.append(f"fn=({func_id})") line_num = details.get('line') or 1 output.append(f"{line_num} {int(data['self_cost'])}") for callee_id, call_count in data['calls'].items(): callee_details = id_to_func_details[callee_id] if callee_details['file_id'] and callee_details['file_id']!= details['file_id']: output.append(f"cfl=({callee_details['file_id']})") output.append(f"cfn=({callee_id})") # Using placeholder '1' for call site line number output.append(f"calls={int(call_count)} 1") if total_incoming[callee_id] == 0: edge_inclusive = graph[callee_id]['inclusive_cost'] else: fraction = call_count / total_incoming[callee_id] edge_inclusive = fraction * graph[callee_id]['inclusive_cost'] output.append(f"1 {int(edge_inclusive)}") output.append("") return "\n".join(output) if __name__ == "__main__": if len(sys.argv) != 2: print("Usage: python render_profile.py ", file=sys.stderr) sys.exit(1) input_filepath = sys.argv[1] try: # Check if file is gzipped based on extension if input_filepath.endswith('.json.gz'): with gzip.open(input_filepath, 'rt', encoding='utf-8') as f: profile = json.load(f) output_filepath = input_filepath.replace('.json.gz', '.capture') else: with open(input_filepath, 'r') as f: profile = json.load(f) output_filepath = input_filepath.replace('.json', '.capture') callgrind_output = fxprof_to_callgrind(profile) with open(output_filepath, 'w') as f: f.write(callgrind_output) print(f"Saved callgrind file to {output_filepath}") except FileNotFoundError: print(f"Error: Input file not found at '{input_filepath}'", file=sys.stderr) sys.exit(1) except (json.JSONDecodeError, ValueError) as e: print(f"Error processing profile file: {e}", file=sys.stderr) sys.exit(1)