�
�d� � �n � d dl mZmZ ddlmZ ddlmZ ddlmZm Z m
Z
ddl m Z m
Z
mZmZ G d� de� Zy )
� )�Optional�Union� )�
CharSetProber)�CodingStateMachine)�LanguageFilter� MachineState� ProbingState)� HZ_SM_MODEL�ISO2022CN_SM_MODEL�ISO2022JP_SM_MODEL�ISO2022KR_SM_MODELc � � � e Zd ZdZej
fdeddf� fd�
Zd � fd� Zede e
fd�� Z ede e
fd�� Z de
fd �Zd
eeef defd �Z� xZS )
�EscCharSetProberz�
This CharSetProber uses a "code scheme" approach for detecting encodings,
whereby easily recognizable escape or shift sequences are relied on to
identify these encodings.
� lang_filter�returnNc � �� t �| � |�� g | _ | j t j
z rP| j j
t t � � | j j
t t � � | j t j z r(| j j
t t � � | j t j z r(| j j
t t � � d| _ d | _ d | _ t" j$ | _ | j) � y )N)r r )�super�__init__� coding_smr r �CHINESE_SIMPLIFIED�appendr r r �JAPANESEr
�KOREANr �active_sm_count�_detected_charset�_detected_languager
� DETECTING�_state�reset)�selfr � __class__s ��3/usr/lib/python3/dist-packages/chardet/escprober.pyr zEscCharSetProber.__init__0 s� �� �
���[��1���� � � �n� ?� ?� ?� �N�N� !� !�"4�[�"A� B� �N�N� !� !�"4�5G�"H� I� � � �n� 5� 5� 5� �N�N� !� !�"4�5G�"H� I� � � �n� 3� 3� 3� �N�N� !� !�"4�5G�"H� I� �� �04�� �15�� �"�,�,�� � �
�
� � c �� �� t �| � � | j D ] }d|_ |j � � t | j � | _ d | _ d | _ y )NT)r r r �active�lenr r r )r! r r" s �r# r zEscCharSetProber.reset@ sT �� �
��
�� � � � �I� #�I� � �O�O� � � #�4�>�>� 2�� �!%�� �"&�� r$ c � � | j S �N�r �r! s r# � charset_namez EscCharSetProber.charset_nameI s � ��%�%�%r$ c � � | j S r) )r r+ s r# �languagezEscCharSetProber.languageM s � ��&�&�&r$ c �" � | j rdS dS )Ng�G�z��?g r* r+ s r# �get_confidencez EscCharSetProber.get_confidenceQ s � ��-�-�t�7�4�7r$ �byte_strc �
� |D ]� }| j D ]� }|j s�|j |� }|t j k( rQd|_ | xj
dz c_ | j
dk s�`t j | _ | j c c S |t j k( s��t j | _ |j � | _
|j | _ | j c c S �� | j S )NFr r )r r& �
next_stater �ERRORr r
�NOT_MEr �state�ITS_ME�FOUND_IT�get_coding_state_machiner r. r )r! r1 �cr � coding_states r# �feedzEscCharSetProber.feedT s� � �� &�A� !�^�^�
&� � �'�'� � (� 3� 3�A� 6� � �<�#5�#5�5�',�I�$��(�(�A�-�(��+�+�q�0�&2�&9�&9�� � #�z�z�)�!�\�%8�%8�8�".�"7�"7�D�K�-6�-O�-O�-Q�D�*�.7�.@�.@�D�+� �:�:�%�
&� &�" �z�z�r$ )r N)�__name__�
__module__� __qualname__�__doc__r �NONEr r �propertyr �strr, r. �floatr0 r �bytes� bytearrayr
r< �
__classcell__)r" s @r# r r ) s� �� �� 6D�5H�5H� �N� �T� � '� �&�h�s�m� &� �&� �'�(�3�-� '� �'�8�� 8��U�5�)�#3� 4� � � r$ r N)�typingr r �
charsetproberr �codingstatemachiner �enumsr r r
�escsmr r r
r r � r$ r# �<module>rN s, ��8 #� (� 2� =� =�� �=�}� =r$ |