Spaces:
Sleeping
Sleeping
File size: 16,115 Bytes
0a4529c |
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 |
# DEPENDENCIES
import re
import tiktoken
from typing import List
from typing import Optional
from config.models import TokenizerType
from config.settings import get_settings
from config.logging_config import get_logger
# Setup Logger and settings
logger = get_logger(__name__)
settings = get_settings()
class TokenCounter:
"""
Token counting utility with support for multiple tokenizers: Provides accurate token counts for chunking and context management
"""
def __init__(self, tokenizer_type: str = "cl100k_base"):
"""
Initialize token counter
Arguments:
----------
tokenizer_type { str } : Type of tokenizer to use
"""
self.tokenizer_type = tokenizer_type
self.logger = logger
# Validate tokenizer type
valid_tokenizers = [t.value for t in TokenizerType]
if tokenizer_type not in valid_tokenizers:
self.logger.warning(f"Invalid tokenizer type: {tokenizer_type}, using approximate")
self.tokenizer_type = TokenizerType.APPROXIMATE
self.tokenizer = None
return
# Initialize tokenizer
if (tokenizer_type != TokenizerType.APPROXIMATE):
try:
self.tokenizer = tiktoken.get_encoding(tokenizer_type)
self.logger.debug(f"Initialized tiktoken tokenizer: {tokenizer_type}")
except Exception as e:
self.logger.warning(f"Failed to load tiktoken: {repr(e)}, using approximation")
self.tokenizer = None
self.tokenizer_type = TokenizerType.APPROXIMATE
else:
self.tokenizer = None
def count_tokens(self, text: str) -> int:
"""
Count tokens in text
Arguments:
----------
text { str } : Input text
Returns:
--------
{ int } : Number of tokens
"""
if not text:
return 0
if self.tokenizer is not None:
# Use tiktoken for accurate counting
try:
tokens = self.tokenizer.encode(text)
return len(tokens)
except Exception as e:
self.logger.warning(f"Tokenizer error: {e}, falling back to approximation")
return self._approximate_token_count(text)
else:
# Use approximation
return self._approximate_token_count(text = text)
def _approximate_token_count(self, text: str) -> int:
"""
Approximate token count using multiple heuristics
"""
if not text:
return 0
# Method 1: Word-based estimation (accounts for subword tokenization)
words = text.split()
word_count = len(words)
# Method 2: Character-based estimation
char_count = len(text)
# Method 3: Hybrid approach with weighting
# - Short texts: more word-based (better for code/short docs)
# - Long texts: more character-based (better for prose)
if (char_count < 1000):
# Prefer word-based for short texts : Slightly higher for short texts
estimate = word_count * 1.33
else:
# Balanced approach for longer texts
word_estimate = word_count * 1.3
char_estimate = char_count / 4.0
estimate = (word_estimate + char_estimate) / 2
# Ensure reasonable bounds
min_tokens = max(1, word_count) # At least 1 token per word
max_tokens = char_count // 2 # At most 1 token per 2 chars
return max(min_tokens, min(int(estimate), max_tokens))
def encode(self, text: str) -> List[int]:
"""
Encode text to token IDs
Arguments:
----------
text { str } : Input text
Returns:
--------
{ list } : List of token IDs
"""
if self.tokenizer is None:
raise ValueError("Cannot encode with approximate tokenizer")
return self.tokenizer.encode(text)
def decode(self, tokens: List[int]) -> str:
"""
Decode token IDs to text
Arguments:
----------
tokens { list } : List of token IDs
Returns:
--------
{ str } : Decoded text
"""
if self.tokenizer is None:
raise ValueError("Cannot decode with approximate tokenizer")
return self.tokenizer.decode(tokens)
def truncate_to_tokens(self, text: str, max_tokens: int, suffix: str = "") -> str:
"""
Truncate text to maximum token count
Arguments:
----------
text { str } : Input text
max_tokens { int } : Maximum number of tokens
suffix { str } : Suffix to add (e.g., "...")
Returns:
--------
{ str } : Truncated text
"""
if self.tokenizer is not None:
# Use precise token-based truncation
tokens = self.encode(text)
if (len(tokens) <= max_tokens):
return text
# Account for suffix tokens
suffix_tokens = len(self.encode(suffix)) if suffix else 0
truncate_at = max_tokens - suffix_tokens
truncated_tokens = tokens[:truncate_at]
truncated_text = self.decode(truncated_tokens)
return truncated_text + suffix
else:
# Use character-based approximation
current_tokens = self.count_tokens(text = text)
if (current_tokens <= max_tokens):
return text
# Estimate character position
ratio = max_tokens / current_tokens
char_position = int(len(text) * ratio)
# Find nearest word boundary
truncated = text[:char_position]
last_space = truncated.rfind(' ')
if (last_space > 0):
truncated = truncated[:last_space]
return truncated + suffix
def split_into_token_chunks(self, text: str, chunk_size: int, overlap: int = 0) -> List[str]:
"""
Split text into chunks of approximately equal token count
Arguments:
----------
text { str } : Input text
chunk_size { int } : Target tokens per chunk
overlap { int } : Number of overlapping tokens between chunks
Returns:
--------
{ list } : List of text chunks
"""
if (overlap >= chunk_size):
raise ValueError("Overlap must be less than chunk_size")
if self.tokenizer is not None:
precise_chunks = self._split_precise(text = text,
chunk_size = chunk_size,
overlap = overlap,
)
return precise_chunks
else:
approximate_chunks = self._split_approximate(text = text,
chunk_size = chunk_size,
overlap = overlap,
)
return approximate_chunks
def _split_precise(self, text: str, chunk_size: int, overlap: int) -> List[str]:
"""
Split using precise token counts
"""
tokens = self.encode(text)
chunks = list()
start = 0
while (start < len(tokens)):
# Get chunk tokens
end = min(start + chunk_size, len(tokens))
chunk_tokens = tokens[start:end]
# Decode to text
chunk_text = self.decode(chunk_tokens)
chunks.append(chunk_text)
# Move to next chunk with overlap
start = end - overlap
# Avoid infinite loop
if ((start >= len(tokens)) or ((end == len(tokens)))):
break
return chunks
def _split_approximate(self, text: str, chunk_size: int, overlap: int) -> List[str]:
"""
Split using approximate token counts
"""
# Estimate characters per chunk : Rule = ~4 chars per token
chars_per_chunk = chunk_size * 4
overlap_chars = overlap * 4
chunks = list()
sentences = self._split_into_sentences(text = text)
current_chunk = list()
current_tokens = 0
for sentence in sentences:
sentence_tokens = self.count_tokens(text = sentence)
if (((current_tokens + sentence_tokens) > chunk_size) and current_chunk):
# Save current chunk
chunk_text = " ".join(current_chunk)
chunks.append(chunk_text)
# Start new chunk with overlap
if (overlap > 0):
# Keep last few sentences for overlap
overlap_text = chunk_text[-overlap_chars:] if len(chunk_text) > overlap_chars else chunk_text
current_chunk = [overlap_text, sentence]
current_tokens = self.count_tokens(text = " ".join(current_chunk))
else:
current_chunk = [sentence]
current_tokens = sentence_tokens
else:
current_chunk.append(sentence)
current_tokens += sentence_tokens
# Add final chunk
if current_chunk:
chunks.append(" ".join(current_chunk))
return chunks
@staticmethod
def _split_into_sentences(text: str) -> List[str]:
"""
Simple sentence splitter with better edge case handling
"""
if not text.strip():
return []
# Split on sentence boundaries
sentences = re.split(r'(?<=[.!?])\s+', text)
# Filter and clean
final_sentences = list()
for sentence in sentences:
sentence = sentence.strip()
if sentence:
# Handle abbreviations (basic)
if not any(sentence.endswith(abbr) for abbr in ['Dr.', 'Mr.', 'Mrs.', 'Ms.', 'etc.']):
final_sentences.append(sentence)
else:
# For abbreviations, keep with next sentence if possible
if final_sentences:
final_sentences[-1] += " " + sentence
else:
final_sentences.append(sentence)
return final_sentences
def get_token_stats(self, text: str) -> dict:
"""
Get comprehensive token statistics
Arguments:
----------
text { str } : Input text
Returns:
--------
{ dict } : Dictionary with statistics
"""
token_count = self.count_tokens(text = text)
char_count = len(text)
word_count = len(text.split())
stats = {"tokens" : token_count,
"characters" : char_count,
"words" : word_count,
"chars_per_token" : char_count / token_count if (token_count > 0) else 0,
"tokens_per_word" : token_count / word_count if (word_count > 0) else 0,
"tokenizer" : self.tokenizer_type,
}
return stats
def estimate_cost(self, text: str, cost_per_1k_tokens: float = 0.002) -> float:
"""
Estimate API cost for text.
Arguments:
----------
text { str } : Input text
cost_per_1k_tokens { float } : Cost per 1000 tokens (default: GPT-4 input)
Returns:
--------
{ float } : Estimated cost in dollars
"""
tokens = self.count_tokens(text = text)
cost = (tokens / 1000) * cost_per_1k_tokens
return round(cost, 6)
def batch_count_tokens(self, texts: List[str]) -> List[int]:
"""
Count tokens for multiple texts efficiently
Arguments:
----------
texts { list } : List of texts
Returns:
--------
{ list } : List of token counts
"""
token_counts = [self.count_tokens(text = text) for text in texts]
return token_counts
def find_token_boundaries(self, text: str, target_tokens: int) -> tuple[int, str]:
"""
Find character position that gives approximately target tokens
Arguments:
----------
text { str } : Input text
target_tokens { int } : Target number of tokens
Returns:
--------
{ tuple } : Tuple of (character_position, text_up_to_position)
"""
if self.tokenizer is not None:
tokens = self.encode(text)
if (len(tokens) <= target_tokens):
return len(text), text
target_tokens_subset = tokens[:target_tokens]
result_text = self.decode(target_tokens_subset)
return len(result_text), result_text
else:
# Approximate
total_tokens = self.count_tokens(text = text)
if (total_tokens <= target_tokens):
return len(text), text
ratio = target_tokens / total_tokens
char_pos = int(len(text) * ratio)
return char_pos, text[:char_pos]
# Global counter instance
_counter = None
def get_token_counter(tokenizer_type: str = "cl100k_base") -> TokenCounter:
"""
Get global token counter instance
Arguments:
----------
tokenizer_type { str } : Tokenizer type
Returns:
--------
{ TokenCounter } : TokenCounter instance
"""
global _counter
if _counter is None or _counter.tokenizer_type != tokenizer_type:
_counter = TokenCounter(tokenizer_type)
return _counter
# Convenience functions
def count_tokens(text: str, tokenizer_type: str = "cl100k_base") -> int:
"""
Quick token count
Arguments:
----------
text { str } : Input text
tokenizer_type { str } : Tokenizer type
Returns:
--------
{ int } : Token count
"""
counter = get_token_counter(tokenizer_type)
return counter.count_tokens(text)
def truncate_to_tokens(text: str, max_tokens: int, suffix: str = "...", tokenizer_type: str = "cl100k_base") -> str:
"""
Truncate text to max tokens
Arguments:
----------
text { str } : Input text
max_tokens { int } : Maximum tokens
suffix { str } : Suffix to add
tokenizer_type { str } : Tokenizer type
Returns:
---------
{ str } : Truncated text
"""
counter = get_token_counter(tokenizer_type)
return counter.truncate_to_tokens(text, max_tokens, suffix) |