Add object detection, tracking, mobile notifications, and search to any security camera.
""" CLIP tokenizer
Copied from https://github.com/openai/CLIP. Originally MIT License, Copyright (c) 2021 OpenAI.
"""
#https://github.com/mlfoundations/open_clip/blob/d3cdb734a2710feeb4c6307df037afa5f786a3e1/src/open_clip/tokenizer.py
import gzip
import html
import os
import random
import string
from functools import lru_cache, partial
from typing import Callable, List, Optional, Union
import numpy as np
import unicodedata
# https://stackoverflow.com/q/62691279
os.environ["TOKENIZERS_PARALLELISM"] = "false"
_nltk_init = False
DEFAULT_CONTEXT_LENGTH = 77 # default context length for OpenAI CLIP
def is_letter(ch):
return unicodedata.category(ch).startswith("L")
def is_number(ch):
return unicodedata.category(ch).startswith("N")
def unicode_clip_tokenize(text, special_tokens):
"""
Emulates the regex pattern from CLIP without using the 'regex' package:
special | 's | 't | 're | 've | 'm | 'll | 'd | letters | numbers | other chars
Returns a list of tokens exactly like CLIP regex would.
"""
tokens = []
i = 0
n = len(text)
special_sorted = sorted(special_tokens, key=len, reverse=True)
while i < n:
matched = False
for sp in special_sorted:
if text.startswith(sp, i):
tokens.append(sp)
i += len(sp)
matched = True
break
if matched:
continue
ch = text[i]
for cont in ["'s", "'t", "'re", "'ve", "'m", "'ll", "'d"]:
if text.startswith(cont, i):
tokens.append(cont)
i += len(cont)
matched = True
break
if matched:
continue
if is_letter(ch):
start = i
i += 1
while i < n and is_letter(text[i]):
i += 1
tokens.append(text[start:i])
continue
if is_number(ch):
tokens.append(ch)
i += 1
continue
if not ch.isspace():
start = i
i += 1
while (
i < n
and not text[i].isspace()
and not is_letter(text[i])
and not is_number(text[i])
):
i += 1
tokens.append(text[start:i])
continue
i += 1
return tokens
@lru_cache()
def default_bpe():
return os.path.join(os.path.dirname(os.path.abspath(__file__)), "bpe_simple_vocab_16e6.txt.gz")
@lru_cache()
def bytes_to_unicode():
"""
Returns list of utf-8 byte and a corresponding list of unicode strings.
The reversible bpe codes work on unicode strings.
This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
This is a significant percentage of your normal, say, 32K bpe vocab.
To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
And avoids mapping to whitespace/control characters the bpe code barfs on.
"""
bs = list(range(ord("!"), ord("~")+1))+list(range(ord("¡"), ord("¬")+1))+list(range(ord("®"), ord("ÿ")+1))
cs = bs[:]
n = 0
for b in range(2**8):
if b not in bs:
bs.append(b)
cs.append(2**8+n)
n += 1
cs = [chr(n) for n in cs]
return dict(zip(bs, cs))
def get_pairs(word):
"""Return set of symbol pairs in a word.
Word is represented as tuple of symbols (symbols being variable-length strings).
"""
pairs = set()
prev_char = word[0]
for char in word[1:]:
pairs.add((prev_char, char))
prev_char = char
return pairs
def basic_clean(text):
text = html.unescape(html.unescape(text))
return text.strip()
def whitespace_clean(text):
text = " ".join(text.split())
text = text.strip()
return text
def _clean_canonicalize(x):
# basic, remove whitespace, remove punctuation, lower case
return canonicalize_text(basic_clean(x))
def _clean_lower(x):
# basic, remove whitespace, lower case
return whitespace_clean(basic_clean(x)).lower()
def _clean_whitespace(x):
# basic, remove whitespace
return whitespace_clean(basic_clean(x))
def get_clean_fn(type: str):
if type == 'canonicalize':
return _clean_canonicalize
elif type == 'lower':
return _clean_lower
elif type == 'whitespace':
return _clean_whitespace
else:
assert False, f"Invalid clean function ({type})."
def canonicalize_text(
text,
*,
keep_punctuation_exact_string=None,
trans_punctuation: dict = str.maketrans("", "", string.punctuation),
):
"""Returns canonicalized `text` (lowercase and punctuation removed).
From: https://github.com/google-research/big_vision/blob/53f18caf27a9419231bbf08d3388b07671616d3d/big_vision/evaluators/proj/image_text/prompt_engineering.py#L94
Args:
text: string to be canonicalized.
keep_punctuation_exact_string: If provided, then this exact string kept.
For example providing '{}' will keep any occurrences of '{}' (but will
still remove '{' and '}' that appear separately).
"""
text = text.replace("_", " ")
if keep_punctuation_exact_string:
text = keep_punctuation_exact_string.join(
part.translate(trans_punctuation)
for part in text.split(keep_punctuation_exact_string)
)
else:
text = text.translate(trans_punctuation)
text = text.lower()
text = " ".join(text.split())
return text.strip()
class SimpleTokenizer(object):
def __init__(
self,
bpe_path: str = default_bpe(),
additional_special_tokens: Optional[List[str]] = None,
context_length: Optional[int] = DEFAULT_CONTEXT_LENGTH,
clean: str = 'lower',
reduction_mask: str = ''
):
self.byte_encoder = bytes_to_unicode()
self.byte_decoder = {v: k for k, v in self.byte_encoder.items()}
merges = gzip.open(bpe_path).read().decode("utf-8").split('\n')
merges = merges[1:49152-256-2+1]
merges = [tuple(merge.split()) for merge in merges]
vocab = list(bytes_to_unicode().values())
vocab = vocab + [v+'</w>' for v in vocab]
for merge in merges:
vocab.append(''.join(merge))
special_tokens = ['<start_of_text>', '<end_of_text>']
if additional_special_tokens:
special_tokens += additional_special_tokens
vocab.extend(special_tokens)
self.encoder = dict(zip(vocab, range(len(vocab))))
self.decoder = {v: k for k, v in self.encoder.items()}
self.bpe_ranks = dict(zip(merges, range(len(merges))))
self.cache = {t:t for t in special_tokens}
special = "|".join(special_tokens)
self.special_tokens_set = set(special_tokens)
self.vocab_size = len(self.encoder)
self.all_special_ids = [self.encoder[t] for t in special_tokens]
self.sot_token_id = self.all_special_ids[0]
self.eot_token_id = self.all_special_ids[1]
self.context_length = context_length
self.clean_fn = get_clean_fn(clean)
self.reduction_fn = get_reduction_mask_fn(reduction_mask) if reduction_mask else None
def bpe(self, token):
if token in self.cache:
return self.cache[token]
word = tuple(token[:-1]) + ( token[-1] + '</w>',)
pairs = get_pairs(word)
if not pairs:
return token+'</w>'
while True:
bigram = min(pairs, key = lambda pair: self.bpe_ranks.get(pair, float('inf')))
if bigram not in self.bpe_ranks:
break
first, second = bigram
new_word = []
i = 0
while i < len(word):
try:
j = word.index(first, i)
new_word.extend(word[i:j])
i = j
except Exception:
new_word.extend(word[i:])
break
if word[i] == first and i < len(word)-1 and word[i+1] == second:
new_word.append(first+second)
i += 2
else:
new_word.append(word[i])
i += 1
new_word = tuple(new_word)
word = new_word
if len(word) == 1:
break
else:
pairs = get_pairs(word)
word = ' '.join(word)
self.cache[token] = word
return word
def encode(self, text):
bpe_tokens = []
text = self.clean_fn(text)
for token in unicode_clip_tokenize(text, self.special_tokens_set):
token = ''.join(self.byte_encoder[b] for b in token.encode('utf-8'))
bpe_tokens.extend(self.encoder[bpe_token] for bpe_token in self.bpe(token).split(' '))
return bpe_tokens
def decode(self, tokens):
text = ''.join([self.decoder[token] for token in tokens])
text = bytearray([self.byte_decoder[c] for c in text]).decode('utf-8', errors="replace").replace('</w>', ' ')
return text
_tokenizer = SimpleTokenizer()