Spaces:
Sleeping
Sleeping
File size: 10,245 Bytes
f59cf24 9a2b71a f59cf24 9a2b71a f59cf24 9a2b71a f59cf24 9a2b71a 6d5a8ce 9a2b71a f59cf24 9a2b71a f59cf24 9a2b71a f59cf24 9a2b71a f59cf24 6d5a8ce 9a2b71a 6d5a8ce 9a2b71a f59cf24 6d5a8ce f59cf24 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 |
# import json
# def analyze_code(language, code, tokenizer, model):
# messages = [
# {
# "role": "system",
# "content": (
# "You are a helpful and expert-level AI code reviewer and bug fixer. "
# "Your task is to analyze the given buggy code in the specified programming language, "
# "identify bugs (logical, syntax, runtime, etc.), and fix them. "
# "Return a JSON object with the following keys:\n\n"
# "1. 'bug_analysis': a list of objects, each containing:\n"
# " - 'line_number': the line number (approximate if needed)\n"
# " - 'error_message': a short name of the bug\n"
# " - 'explanation': short explanation of the problem\n"
# " - 'fix_suggestion': how to fix it\n"
# "2. 'corrected_code': the entire corrected code block.\n\n"
# "Respond with ONLY the raw JSON object, no extra commentary or markdown."
# )
# },
# {
# "role": "user",
# "content": f"💻 Language: {language}\n🐞 Buggy Code:\n```{language.lower()}\n{code.strip()}\n```"
# }
# ]
# inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
# attention_mask = (inputs != tokenizer.pad_token_id).long()
# outputs = model.generate(
# inputs,
# attention_mask=attention_mask,
# max_new_tokens=1024,
# do_sample=False,
# pad_token_id=tokenizer.eos_token_id,
# eos_token_id=tokenizer.eos_token_id
# )
# response = tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)
# # Try parsing response to JSON
# try:
# json_output = json.loads(response)
# return json_output
# except json.JSONDecodeError:
# print("⚠️ Could not decode response into JSON. Here's the raw output:\n")
# print(response)
# return None
# import json
# import logging
# import time
# import torch
# # Configure logging
# logger = logging.getLogger(__name__)
# def analyze_code(language, code, tokenizer, model):
# """
# Analyze code and return bug analysis with improved logging and error handling
# """
# start_time = time.time()
# logger.info(f"🔍 Starting analysis for {language} code ({len(code)} characters)")
# try:
# # Prepare messages
# messages = [
# {
# "role": "system",
# "content": (
# "You are a helpful and expert-level AI code reviewer and bug fixer. "
# "Your task is to analyze the given buggy code in the specified programming language, "
# "identify bugs (logical, syntax, runtime, etc.), and fix them. "
# "Return a JSON object with the following keys:\n\n"
# "1. 'bug_analysis': a list of objects, each containing:\n"
# " - 'line_number': the line number (approximate if needed)\n"
# " - 'error_message': a short name of the bug\n"
# " - 'explanation': short explanation of the problem\n"
# " - 'fix_suggestion': how to fix it\n"
# "2. 'corrected_code': the entire corrected code block.\n\n"
# "Respond with ONLY the raw JSON object, no extra commentary or markdown."
# )
# },
# {
# "role": "user",
# "content": f"💻 Language: {language}\n🐞 Buggy Code:\n```{language.lower()}\n{code.strip()}\n```"
# }
# ]
# logger.info("🔧 Applying chat template...")
# inputs = tokenizer.apply_chat_template(
# messages,
# add_generation_prompt=True,
# return_tensors="pt"
# ).to(model.device)
# attention_mask = (inputs != tokenizer.pad_token_id).long()
# logger.info(f"📏 Input length: {inputs.shape[1]} tokens")
# logger.info("🚀 Starting model generation...")
# generation_start = time.time()
# # Generate with more conservative settings
# with torch.no_grad(): # Ensure no gradients are computed
# outputs = model.generate(
# inputs,
# attention_mask=attention_mask,
# max_new_tokens=512, # Reduced from 1024 for faster inference
# do_sample=False,
# temperature=0.1, # Add temperature for more consistent output
# pad_token_id=tokenizer.eos_token_id,
# eos_token_id=tokenizer.eos_token_id,
# use_cache=True, # Enable KV cache for efficiency
# )
# generation_time = time.time() - generation_start
# logger.info(f"⚡ Generation completed in {generation_time:.2f} seconds")
# logger.info("📝 Decoding response...")
# response = tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)
# logger.info(f"📄 Response length: {len(response)} characters")
# logger.info(f"🔍 First 100 chars: {response[:100]}...")
# # Try parsing response to JSON
# logger.info("🔍 Attempting to parse JSON...")
# try:
# # Clean up response - remove any markdown formatting
# cleaned_response = response.strip()
# if cleaned_response.startswith('```json'):
# cleaned_response = cleaned_response[7:]
# if cleaned_response.startswith('```'):
# cleaned_response = cleaned_response[3:]
# if cleaned_response.endswith('```'):
# cleaned_response = cleaned_response[:-3]
# cleaned_response = cleaned_response.strip()
# json_output = json.loads(cleaned_response)
# total_time = time.time() - start_time
# logger.info(f"✅ Analysis completed successfully in {total_time:.2f} seconds")
# # Validate the JSON structure
# if not isinstance(json_output, dict):
# raise ValueError("Response is not a dictionary")
# if 'bug_analysis' not in json_output:
# logger.warning("⚠️ Missing 'bug_analysis' key, adding empty list")
# json_output['bug_analysis'] = []
# if 'corrected_code' not in json_output:
# logger.warning("⚠️ Missing 'corrected_code' key, adding original code")
# json_output['corrected_code'] = code
# return json_output
# except json.JSONDecodeError as e:
# logger.error(f"❌ JSON decode error: {e}")
# logger.error(f"📄 Raw response: {repr(response)}")
# # Return a fallback structure with the raw response
# fallback_response = {
# "bug_analysis": [{
# "line_number": 1,
# "error_message": "Analysis parsing failed",
# "explanation": "The AI model returned a response that couldn't be parsed as JSON",
# "fix_suggestion": "Please try again or check the code format"
# }],
# "corrected_code": code,
# "raw_output": response,
# "parsing_error": str(e)
# }
# return fallback_response
# except Exception as e:
# total_time = time.time() - start_time
# logger.error(f"❌ Analysis failed after {total_time:.2f} seconds: {str(e)}")
# logger.error(f"💥 Exception type: {type(e).__name__}")
# # Return error response
# return {
# "bug_analysis": [{
# "line_number": 1,
# "error_message": "Analysis failed",
# "explanation": f"An error occurred during analysis: {str(e)}",
# "fix_suggestion": "Please try again or contact support"
# }],
# "corrected_code": code,
# "error": str(e),
# "error_type": type(e).__name__
# }
# analyzer.py
# analyzer.py
import torch
import json
import time
import logging
# Configure logger
logger = logging.getLogger("CodeAnalyzer")
logger.setLevel(logging.INFO)
handler = logging.StreamHandler()
formatter = logging.Formatter("[%(asctime)s] [%(levelname)s] - %(message)s")
handler.setFormatter(formatter)
logger.addHandler(handler)
def analyze_code(tokenizer, model, language, code):
"""
Analyze and fix buggy code using CodeT5+ model with 'fix:' prompt prefix.
Works across multiple programming languages.
"""
start_time = time.time()
# Prepare prompt in CodeT5+ style
prompt = f"fix: {code.strip()}"
logger.info(f"🔍 Starting analysis for language: {language}")
logger.info(f"🧾 Prompt: {prompt[:80]}...")
try:
# Tokenize and generate response
inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=512).to(model.device)
with torch.no_grad():
output = model.generate(**inputs, max_new_tokens=1024)
# Decode output
response = tokenizer.decode(output[0], skip_special_tokens=True).strip()
elapsed = round(time.time() - start_time, 2)
logger.info(f"✅ Inference completed in {elapsed}s")
return {
"bug_analysis": [], # Optional: You could add heuristics here
"corrected_code": response
}
except Exception as e:
logger.error(f"❌ Error during analysis: {e}")
return {
"bug_analysis": [{
"line_number": 0,
"error_message": "Inference failed",
"explanation": str(e),
"fix_suggestion": "Try again with simpler code or retry later"
}],
"corrected_code": code
}
|