venv/lib/python3.11/site-packages/pip/_vendor/chardet/enums.py

   1 """
   2 All of the Enums that are used throughout the chardet package.
   3
   4 :author: Dan Blanchard (dan.blanchard@gmail.com)
   5 """
   6
   7 from enum import Enum, Flag
   8
   9
  10 class InputState:
  11     """
  12     This enum represents the different states a universal detector can be in.
  13     """
  14
  15     PURE_ASCII = 0
  16     ESC_ASCII = 1
  17     HIGH_BYTE = 2
  18
  19
  20 class LanguageFilter(Flag):
  21     """
  22     This enum represents the different language filters we can apply to a
  23     ``UniversalDetector``.
  24     """
  25
  26     NONE = 0x00
  27     CHINESE_SIMPLIFIED = 0x01
  28     CHINESE_TRADITIONAL = 0x02
  29     JAPANESE = 0x04
  30     KOREAN = 0x08
  31     NON_CJK = 0x10
  32     ALL = 0x1F
  33     CHINESE = CHINESE_SIMPLIFIED | CHINESE_TRADITIONAL
  34     CJK = CHINESE | JAPANESE | KOREAN
  35
  36
  37 class ProbingState(Enum):
  38     """
  39     This enum represents the different states a prober can be in.
  40     """
  41
  42     DETECTING = 0
  43     FOUND_IT = 1
  44     NOT_ME = 2
  45
  46
  47 class MachineState:
  48     """
  49     This enum represents the different states a state machine can be in.
  50     """
  51
  52     START = 0
  53     ERROR = 1
  54     ITS_ME = 2
  55
  56
  57 class SequenceLikelihood:
  58     """
  59     This enum represents the likelihood of a character following the previous one.
  60     """
  61
  62     NEGATIVE = 0
  63     UNLIKELY = 1
  64     LIKELY = 2
  65     POSITIVE = 3
  66
  67     @classmethod
  68     def get_num_categories(cls) -> int:
  69         """:returns: The number of likelihood categories in the enum."""
  70         return 4
  71
  72
  73 class CharacterCategory:
  74     """
  75     This enum represents the different categories language models for
  76     ``SingleByteCharsetProber`` put characters into.
  77
  78     Anything less than CONTROL is considered a letter.
  79     """
  80
  81     UNDEFINED = 255
  82     LINE_BREAK = 254
  83     SYMBOL = 253
  84     DIGIT = 252
  85     CONTROL = 251