File size: 24,512 Bytes
4265aea |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 |
import argparse
import json
import os
import time
import glob
import logging
import sys
import traceback
from datetime import datetime
from pathlib import Path
from typing import List, Dict, Any, Optional, Tuple
def get_project_root() -> Path:
"""Get the project root directory."""
# Use the current working directory as the project root
return Path.cwd()
def ensure_directory(path: Path) -> None:
"""Ensure directory exists, create if it doesn't."""
path.mkdir(parents=True, exist_ok=True)
# Configure logging
log_dir = Path('test_result')
ensure_directory(log_dir)
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.StreamHandler(sys.stdout),
logging.FileHandler(log_dir / 'tokenizer_test.log')
]
)
logger = logging.getLogger(__name__)
class Tokenizer:
def __init__(self, tokenizer_path: str):
"""Initialize the EZ-Tokenizer with enhanced error handling and validation."""
try:
from tokenizers import Tokenizer as HFTokenizer
logger.info(f"Loading EZ-Tokenizer from {tokenizer_path}")
if not os.path.exists(tokenizer_path):
raise FileNotFoundError(f"EZ-Tokenizer file not found: {tokenizer_path}")
start_time = time.time()
self.tokenizer = HFTokenizer.from_file(tokenizer_path)
load_time = time.time() - start_time
self.vocab_size = self.tokenizer.get_vocab_size()
logger.info(f"EZ-Tokenizer loaded in {load_time:.2f} seconds. Vocabulary size: {self.vocab_size:,}")
# Run basic smoke tests
self._run_smoke_tests()
except Exception as e:
logger.error(f"Failed to initialize EZ-Tokenizer: {e}", exc_info=True)
logger.error(f"Failed to initialize tokenizer: {e}", exc_info=True)
raise
def _run_smoke_tests(self):
"""Run basic smoke tests to verify tokenizer functionality."""
test_cases = [
"Hello, world!",
"γγγ«γ‘γ―δΈη",
"μλ
νμΈμ",
"ΠΡΠΈΠ²Π΅Ρ, ΠΌΠΈΡ!",
"12345 !@#$%^&*()_+{}|:<>?",
""
]
logger.info("Running smoke tests...")
for text in test_cases:
try:
tokens = self.encode(text)
decoded = self.decode(tokens)
if text != decoded:
logger.warning(f"Roundtrip mismatch for {text!r} -> {decoded!r}")
except Exception as e:
logger.error(f"Smoke test failed for {text!r}: {e}")
raise
logger.info("Smoke tests completed successfully")
def encode(self, text: str, chunk_size: int = 10000) -> List[int]:
"""Encode text to token IDs with chunking for large inputs."""
try:
if not isinstance(text, str):
raise ValueError(f"Expected string, got {type(text).__name__}")
# Process in chunks if text is large
if len(text) <= chunk_size:
return self.tokenizer.encode(text).ids
# Process large text in chunks
tokens = []
for i in range(0, len(text), chunk_size):
chunk = text[i:i + chunk_size]
tokens.extend(self.tokenizer.encode(chunk).ids)
return tokens
except Exception as e:
logger.error(f"Encoding failed: {e}")
raise RuntimeError(f"Failed to encode text (length: {len(text)}): {e}")
def decode(self, token_ids: List[int], chunk_size: int = 10000) -> str:
"""Decode token IDs back to text with memory-efficient chunking."""
try:
if not token_ids:
return ""
if not all(isinstance(t, int) for t in token_ids):
raise ValueError("All token IDs must be integers")
# Process in chunks to prevent memory issues
if len(token_ids) <= chunk_size:
return self.tokenizer.decode(token_ids)
# Process large token sequences in chunks
chunks = []
for i in range(0, len(token_ids), chunk_size):
chunk = token_ids[i:i + chunk_size]
chunks.append(self.tokenizer.decode(chunk))
# Log progress periodically
if (i // chunk_size) % 10 == 0:
logger.info(f"Decoded {min(i + chunk_size, len(token_ids)):,}/{len(token_ids):,} tokens")
return "".join(chunks)
except Exception as e:
logger.error(f"Decoding failed: {e}")
raise RuntimeError(f"Failed to decode {len(token_ids)} tokens: {e}")
def get_vocab_size(self) -> int:
"""Return the size of the tokenizer's vocabulary."""
return self.vocab_size
def process_file_in_chunks(file_path: str, chunk_size: int = 1024 * 1024) -> str:
"""Read a file in chunks to avoid memory issues."""
chunks = []
try:
with open(file_path, 'r', encoding='utf-8', errors='replace') as f:
while True:
chunk = f.read(chunk_size)
if not chunk:
break
chunks.append(chunk)
return "".join(chunks)
except Exception as e:
logger.error(f"Error reading file {file_path}: {e}")
raise
def normalize_whitespace(text: str) -> str:
"""Normalize whitespace in code for more meaningful comparison."""
import re
# Replace all whitespace sequences with a single space
text = re.sub(r'\s+', ' ', text)
# Remove leading/trailing whitespace
return text.strip()
def calculate_token_metrics(original_tokens, decoded_tokens):
"""Calculate token-level accuracy metrics."""
min_len = min(len(original_tokens), len(decoded_tokens))
exact_matches = sum(1 for a, b in zip(original_tokens, decoded_tokens) if a == b)
return {
'token_accuracy': exact_matches / max(len(original_tokens), 1),
'token_precision': exact_matches / max(len(decoded_tokens), 1),
'token_recall': exact_matches / max(len(original_tokens), 1),
'token_f1': 2 * exact_matches / (len(original_tokens) + len(decoded_tokens))
if (len(original_tokens) + len(decoded_tokens)) > 0 else 0
}
def enhanced_char_metrics(original: str, decoded: str) -> dict:
"""Calculate enhanced character-level metrics."""
# Normalize both strings
norm_original = normalize_whitespace(original)
norm_decoded = normalize_whitespace(decoded)
# Calculate basic metrics
min_len = min(len(norm_original), len(norm_decoded))
max_len = max(len(norm_original), len(norm_decoded))
if max_len == 0:
return {
'char_accuracy': 1.0,
'char_similarity': 1.0,
'length_diff_ratio': 0.0
}
# Calculate matches
matches = sum(1 for a, b in zip(norm_original, norm_decoded) if a == b)
# Calculate similarity using Levenshtein distance if available
try:
from Levenshtein import ratio
similarity = ratio(norm_original, norm_decoded)
except ImportError:
similarity = matches / max_len if max_len > 0 else 1.0
return {
'char_accuracy': matches / max_len if max_len > 0 else 1.0,
'char_similarity': similarity,
'length_diff_ratio': abs(len(norm_original) - len(norm_decoded)) / max_len if max_len > 0 else 0.0
}
def validate_code_integrity(original: str, decoded: str) -> dict:
"""Validate code-specific integrity metrics."""
import ast
def can_parse(code: str) -> bool:
try:
ast.parse(code)
return True
except:
return False
original_parses = can_parse(original)
decoded_parses = can_parse(decoded)
return {
'original_parses': original_parses,
'decoded_parses': decoded_parses,
'both_parse': original_parses and decoded_parses
}
def calculate_metrics(original_text: str, decoded_text: str, tokens,
start_time: float, end_time: float) -> Dict[str, Any]:
"""Enhanced metrics calculation for tokenizer evaluation."""
# Basic metrics
token_count = len(tokens) if tokens else 0
char_count = len(original_text) if original_text else 0
process_time = max(end_time - start_time, 0.001) # Avoid division by zero
metrics = {
'tokens': token_count,
'chars': char_count,
'processing_time': process_time,
'tokens_per_second': token_count / process_time,
'chars_per_token': char_count / (token_count or 1) # Avoid division by zero
}
# Calculate rates
metrics.update({
'tokens_per_sec': len(tokens) / metrics['processing_time'],
'chars_per_sec': len(original_text) / metrics['processing_time']
})
# Enhanced character-level metrics
metrics.update(enhanced_char_metrics(original_text, decoded_text))
# Token-level metrics (if we have the original tokens)
if hasattr(tokens, 'tokens'): # If using tokenizers' Encoding object
original_tokens = tokens.tokens
decoded_tokens = tokenizer.encode(decoded_text).tokens
metrics.update(calculate_token_metrics(original_tokens, decoded_tokens))
# Code-specific validation for Python files
if original_text.strip().endswith('.py') or 'def ' in original_text or 'import ' in original_text:
metrics.update(validate_code_integrity(original_text, decoded_text))
return metrics
def print_metrics_summary(metrics: Dict[str, Any]):
"""Print a clean summary of the metrics."""
print("\n=== Tokenizer Test Results ===")
print(f"Processing Speed: {metrics.get('tokens_per_second', metrics.get('tokens_per_sec', 0)):,.0f} tokens/sec")
print(f"Characters per Token: {metrics.get('chars_per_token', 0):.2f}")
print(f"\nCharacter-Level Metrics:")
print(f" β’ Accuracy: {metrics.get('char_accuracy', 0)*100:.2f}%")
print(f" β’ Similarity: {metrics.get('char_similarity', 0)*100:.2f}%")
print(f" β’ Levenshtein Ratio: {metrics.get('levenshtein_ratio', 0)*100:.2f}%")
print(f"\nCode Integrity:")
print(f" β’ Original parses: {'β' if metrics.get('original_parses', False) else 'β'}")
print(f" β’ Decoded parses: {'β' if metrics.get('decoded_parses', False) else 'β'}")
print(f" β’ Both parse: {'β' if metrics.get('both_parse', False) else 'β'}")
def process_file(file_path: Path, tokenizer: Tokenizer, max_chunk_size: int = 100_000, sample_size: int = 100_000) -> Dict[str, Any]:
"""Process a single file in chunks and return metrics."""
try:
logger.info(f"\nProcessing file: {file_path}")
file_size = file_path.stat().st_size
logger.info(f"File size: {file_size / (1024*1024):.2f} MB")
# Initialize metrics
total_tokens = 0
total_chars = 0
total_time = 0
chunk_metrics = []
# Process file in chunks
total_read = 0
with open(file_path, 'r', encoding='utf-8', errors='replace') as f:
# Only read up to sample_size if specified
max_to_read = sample_size if sample_size > 0 else float('inf')
logger.info(f"Processing up to {max_to_read if max_to_read != float('inf') else 'all'} characters")
chunk = f.read(min(max_chunk_size, max_to_read - total_read))
total_read += len(chunk)
while chunk and total_read <= max_to_read:
if not chunk.strip():
chunk = f.read(max_chunk_size)
continue
# Process chunk
start_time = time.time()
try:
# Handle both tokenizer output formats (object with .ids or raw list)
tokens = tokenizer.encode(chunk)
token_ids = tokens.ids if hasattr(tokens, 'ids') else tokens
decoded_text = tokenizer.decode(token_ids)
except Exception as e:
logger.error(f"Error in tokenization: {e}")
# Skip this chunk if tokenization fails
chunk = f.read(max_chunk_size)
continue
end_time = time.time()
# Skip empty chunks
if not token_ids:
chunk = f.read(max_chunk_size)
continue
# Calculate metrics for this chunk
metrics = calculate_metrics(chunk, decoded_text, token_ids, start_time, end_time)
chunk_metrics.append(metrics)
# Update totals
total_tokens += len(token_ids)
total_chars += len(chunk)
total_time += (end_time - start_time)
# Log progress
if total_tokens % 1_000_000 == 0:
logger.info(f" Processed {total_tokens:,} tokens ({total_chars/1024/1024:.2f} MB)")
# Read next chunk (respecting sample size)
to_read = min(max_chunk_size, max_to_read - total_read)
if to_read <= 0:
# We've reached the sample size limit
break
chunk = f.read(to_read)
total_read += len(chunk)
# Calculate aggregate metrics
if not chunk_metrics:
logger.warning(f"No valid content found in file: {file_path}")
return None
# Calculate weighted averages based on token counts
total_weight = sum(m.get('tokens', 0) for m in chunk_metrics) or 1
avg_metrics = {
'chars_per_token': sum(m.get('chars_per_token', 0) * m.get('tokens', 0) for m in chunk_metrics) / total_weight,
'tokens_per_second': sum(m.get('tokens', 0) for m in chunk_metrics) / (total_time or 1),
'char_accuracy': sum(m.get('char_accuracy', 0) * m.get('tokens', 0) for m in chunk_metrics) / total_weight,
'tokens': total_tokens,
'chars': total_chars,
'processing_time': total_time,
'file_path': str(file_path)
}
# Log final metrics
logger.info(f" Total tokens: {total_tokens:,}")
logger.info(f" Total chars: {total_chars:,}")
logger.info(f" Avg chars/token: {avg_metrics['chars_per_token']:.2f}")
logger.info(f" Avg tokens/sec: {avg_metrics['tokens_per_second']:,.2f}")
return avg_metrics
except Exception as e:
logger.error(f"Error processing {file_path}: {e}")
logger.error(traceback.format_exc())
return None
def process_single_file(tokenizer: Tokenizer, file_path: str, sample_size: int = 0) -> Dict[str, Any]:
"""Process a single file and return metrics."""
logger.info(f"\nProcessing file: {file_path}")
try:
# Process file in chunks with sample size limit
metrics = process_file(file_path, tokenizer, sample_size=sample_size)
if not metrics:
logger.warning(f"Empty file or no valid content found: {file_path}")
return {}
# Add file info
metrics['file'] = os.path.basename(file_path)
metrics['file_size_mb'] = os.path.getsize(file_path) / (1024 * 1024)
# Log summary
logger.info(
f"Processed {metrics['file_size_mb']:.2f}MB: "
f"{metrics['tokens']:,} tokens, "
f"{metrics['chars_per_token']:.2f} chars/token, "
f"{metrics['tokens_per_second']:,.2f} tokens/sec"
)
# Print detailed metrics summary
print_metrics_summary(metrics)
return metrics
except Exception as e:
logger.error(f"Error processing {file_path}: {e}", exc_info=True)
return {'file': os.path.basename(file_path), 'error': str(e)}
def main():
# Set up default paths
project_root = get_project_root()
# Point to the root directory (one level up from Test_tokenizer)
root_dir = project_root.parent
default_tokenizer = root_dir / 'output' / 'tokenizer.json'
default_input = root_dir / 'Dataset' # Changed to look in root directory
default_output = root_dir / 'test_result' # Also put test results in root
# Ensure output directory exists
ensure_directory(default_output)
# Generate timestamp for output file
timestamp = datetime.now().strftime('%Y%m%d_%H%M%S')
default_output_file = default_output / f'test_results_{timestamp}.txt'
parser = argparse.ArgumentParser(description='Test tokenizer on code files')
parser.add_argument('--tokenizer', type=str, default=str(default_tokenizer),
help=f'Path to tokenizer.json file (default: {default_tokenizer})')
parser.add_argument('--input', type=str, default=str(default_input),
help=f'Input directory or file (default: {default_input})')
parser.add_argument('--output', type=str, default=str(default_output_file),
help=f'Output text file for results (default: {default_output_file})')
parser.add_argument('--sample', type=int, default=100000, help='Only process this many characters from each file (0 for full file)')
parser.add_argument('--max-files', type=int, default=10,
help='Maximum number of files to process (default: 10)')
parser.add_argument('--file-types', type=str, default='*',
help='Comma-separated list of file extensions to process (e.g., "py,js,json"). Default: all files')
args = parser.parse_args()
# Ensure output directory exists
output_dir = Path(args.output).parent
ensure_directory(output_dir)
# Initialize tokenizer
logger.info(f"Initializing tokenizer from {args.tokenizer}")
tokenizer = Tokenizer(args.tokenizer)
# Parse file types
file_extensions = []
if args.file_types != '*':
file_extensions = [ext.strip().lower() for ext in args.file_types.split(',')]
logger.info(f"Filtering by file extensions: {', '.join(file_extensions)}")
# Find input files
input_path = Path(args.input)
file_paths = []
if input_path.is_dir():
# Find all files in the input directory (recursively)
if file_extensions:
# If specific extensions are provided, only include those
for ext in file_extensions:
pattern = f'*.{ext.lstrip(".")}'
file_paths.extend(input_path.rglob(pattern))
else:
# Otherwise include all files
file_paths = list(input_path.rglob('*'))
# Filter out directories, hidden files, and ensure files exist
file_paths = [
f for f in file_paths
if f.is_file() and not f.name.startswith(('.', '_'))
]
# Sort files by size (smallest first) to process quicker files first
file_paths.sort(key=lambda x: x.stat().st_size)
logger.info(f"Found {len(file_paths)} files in {input_path}")
if file_paths:
logger.info(f"Sample files: {', '.join(f.name for f in file_paths[:min(5, len(file_paths))])}" +
('...' if len(file_paths) > 5 else ''))
else:
# Single file
file_paths = [input_path] if input_path.exists() else []
logger.info(f"Processing single file: {input_path}")
if not file_paths:
logger.warning(f"No files found in {input_path}")
return
# Process files
all_metrics = []
processed_count = 0
skipped_files = 0
# Get unique file paths (remove duplicates and sort)
unique_file_paths = []
seen_paths = set()
for file_path in file_paths:
abs_path = str(file_path.absolute())
if abs_path not in seen_paths:
seen_paths.add(abs_path)
unique_file_paths.append(file_path)
if len(unique_file_paths) < len(file_paths):
logger.info(f"Removed {len(file_paths) - len(unique_file_paths)} duplicate file paths")
# Limit to max_files if specified
if args.max_files > 0:
unique_file_paths = unique_file_paths[:args.max_files]
# Process each file
for file_path in unique_file_paths:
try:
if not file_path.exists():
logger.warning(f"File not found: {file_path}")
skipped_files += 1
continue
file_size_mb = os.path.getsize(file_path) / (1024 * 1024)
logger.info(f"\nProcessing: {file_path.name} ({file_size_mb:.2f} MB)")
# Process the file with sample option
metrics = process_single_file(tokenizer, file_path, args.sample)
if metrics:
all_metrics.append(metrics)
processed_count += 1
logger.info(f"Processed {processed_count}/{len(unique_file_paths)} files")
except Exception as e:
logger.error(f"Error processing {file_path}: {str(e)}")
skipped_files += 1
if skipped_files > 0:
logger.warning(f"Skipped {skipped_files} files due to errors")
# Calculate averages from all metrics
if all_metrics:
avg_metrics = {}
for key in all_metrics[0].keys():
if isinstance(all_metrics[0][key], (int, float)):
values = [r[key] for r in all_metrics if key in r]
if values:
avg_metrics[f'avg_{key}'] = sum(values) / len(values)
# Write results to file
with open(args.output, 'w', encoding='utf-8') as f:
f.write("=== Tokenizer Test Results ===\n")
f.write(f"Generated at: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n")
f.write(f"Tokenizer: {args.tokenizer}\n")
f.write(f"Input: {args.input}\n")
f.write(f"Sample size: {args.sample if args.sample > 0 else 'Full file'}\n\n")
f.write("=== Summary ===\n")
if all_metrics:
# Write aggregate metrics
for key, value in avg_metrics.items():
if isinstance(value, float):
f.write(f"{key}: {value:.4f}\n")
else:
f.write(f"{key}: {value}\n")
else:
f.write("No files were successfully processed\n")
# Write individual file results
f.write("\n=== File Details ===\n")
for result in all_metrics:
f.write(f"\nFile: {result.get('file', 'unknown')}\n")
for key, value in result.items():
if key != 'file':
if isinstance(value, float):
f.write(f" {key}: {value:.4f}\n")
else:
f.write(f" {key}: {value}\n")
logger.info(f"Results saved to {args.output}")
print(f"\nTest results saved to: {args.output}")
if all_metrics:
logger.info(f"\n=== Test Complete ===")
logger.info(f"Processed {processed_count} files")
logger.info(f"Average chars/token: {avg_metrics.get('avg_chars_per_token', 0):.2f}")
logger.info(f"Average tokens/sec: {avg_metrics.get('avg_tokens_per_sec', 0):,.0f}")
else:
logger.warning("No files were successfully processed")
if __name__ == "__main__":
try:
# Check for required dependencies
try:
import Levenshtein
except ImportError:
logger.warning("python-Levenshtein not found. Install with: pip install python-Levenshtein")
logger.warning("Falling back to basic similarity metrics")
main()
except KeyboardInterrupt:
logger.info("\nProcess interrupted by user")
sys.exit(1)
except Exception as e:
logger.error(f"An error occurred: {e}", exc_info=True)
sys.exit(1) |