3
� �X|. � @ s� d Z ddlZddlZejd�ZejdejejB ejB �Zejdej ej
B ejB ejB �Z ejdej �Z
G dd� de�ZG d d
� d
e�Zd8d d
�Zd9dd�Zd:dd�Zd;dd�Zdd� Zdd� Zdd� Zdd� Zd d � Zi Zd d � Z d d!� Z d"d#� Z d<d$d%�Z f fd&d'�Z G d(d)� d)e!�Z"d*d+� Z#d,d-� Z$d.d/� Z%ej&d=k �r~e'Z'e(Z(e)e*fZ+e*Z,d1Z-e.j/Z/e.j0Z0ddl1Z1ddl2Z2e1j1Z1e2j1Z3nFe4Z'e5Z(e)fZ+e)Z,d2Z-e.j6Z/e.j7Z0dd3l8m1Z1m3Z3m9Z9 G d4d5� d5e9�Z:d6d7� Z;dS )>z�
pygments.util
~~~~~~~~~~~~~
Utility functions.
:copyright: Copyright 2006-2017 by the Pygments team, see AUTHORS.
:license: BSD, see LICENSE for details.
� Nz[/\\ ]z�
(<\?.*?\?>)?\s*
<!DOCTYPE\s+(
[a-zA-Z_][a-zA-Z0-9]*
(?: \s+ # optional in HTML5
[a-zA-Z_][a-zA-Z0-9]*\s+
"[^"]*")?
)
[^>]*>
z<(.+?)(\s.*?)?>.*?</.+?>z\s*<\?xml[^>]*\?>c @ s e Zd ZdZdS )�
ClassNotFoundzCRaised if one of the lookup functions didn't find a matching class.N)�__name__�
__module__� __qualname__�__doc__� r r �/usr/lib/python3.6/util.pyr s r c @ s e Zd ZdS )� OptionErrorN)r r r r r r r r $ s r Fc C s@ | j ||�}|r|j� }||kr<td|djtt|��f ��|S )Nz%Value for option %s must be one of %sz, )�get�lowerr �join�map�str)�options�optnameZallowed�default�normcase�stringr r r �get_choice_opt( s r c C s| | j ||�}t|t�r|S t|t�r,t|�S t|t�sHtd||f ��n0|j� d
krXdS |j� dkrhd S td ||f ��d S )NzBInvalid type %r for option %s; use 1/0, yes/no, true/false, on/off�1�yes�true�onT�0�no�false�offFzCInvalid value %r for option %s; use 1/0, yes/no, true/false, on/off)r r r r )r r r r )r
�
isinstance�bool�int� string_typesr r )r r r r r r r � get_bool_opt2 s
r! c C s` | j ||�}yt|�S tk
r8 td||f ��Y n$ tk
rZ td||f ��Y nX d S )Nz=Invalid type %r for option %s; you must give an integer valuez>Invalid value %r for option %s; you must give an integer value)r
r � TypeErrorr �
ValueError)r r r r r r r � get_int_optF s r$ c C sH | j ||�}t|t�r |j� S t|ttf�r4t|�S td||f ��d S )Nz9Invalid type %r for option %s; you must give a list value)r
r r �split�list�tupler )r r r �valr r r � get_list_optT s
r) c C sR | j s
dS g }x4| j j� j� D ]"}|j� r>|jd|j� � q P q W dj|�j� S )N� � )r �strip�
splitlines�appendr �lstrip)�obj�res�liner r r �docstring_headline` s r3 c s � fdd�}� j |_ t|�S )zAReturn a static text analyser function that returns float values.c s\ y � | �}W n t k
r dS X |s*dS ytdtdt|���S ttfk
rV dS X d S )Ng g �?)� Exception�min�max�floatr# r" )�text�rv)�fr r � text_analysen s z%make_analysator.<locals>.text_analyse)r � staticmethod)r: r; r )r: r �make_analysatorl s r= c C s� | j d�}|dkr$| d|� j� }n| j� }|jd�r�y(dd� tj|dd� j� �D �d }W n tk
rr d S X tjd
| tj �}|j
|�dk r�d S d S )
a� Check if the given regular expression matches the last part of the
shebang if one exists.
>>> from pygments.util import shebang_matches
>>> shebang_matches('#!/usr/bin/env python', r'python(2\.\d)?')
True
>>> shebang_matches('#!/usr/bin/python2.4', r'python(2\.\d)?')
True
>>> shebang_matches('#!/usr/bin/python-ruby', r'python(2\.\d)?')
False
>>> shebang_matches('#!/usr/bin/python/ruby', r'python(2\.\d)?')
False
>>> shebang_matches('#!/usr/bin/startsomethingwith python',
... r'python(2\.\d)?')
True
It also checks for common windows executable file extensions::
>>> shebang_matches('#!C:\\Python2.4\\Python.exe', r'python(2\.\d)?')
True
Parameters (``'-f'`` or ``'--foo'`` are ignored so ``'perl'`` does
the same as ``'perl -e'``)
Note that this method automatically searches the whole string (eg:
the regular expression is wrapped in ``'^$'``)
�
r Nz#!c S s g | ]}|r|j d � r|�qS )�-)�
startswith)�.0�xr r r �
<listcomp>� s z#shebang_matches.<locals>.<listcomp>� � Fz^%s(\.(exe|cmd|bat|bin))?$T���) �findr r@ �
split_path_rer% r, �
IndexError�re�compile�
IGNORECASE�search)r8 �regex�indexZ
first_line�foundr r r �shebang_matches} s
rQ c C s< t j| �}|dkrdS |jd�}tj|tj�j|j� �dk S )z�Check if the doctype matches a regular expression (if present).
Note that this method only checks the first part of a DOCTYPE.
eg: 'html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"'
NFrD )�doctype_lookup_re�match�grouprJ rK �Ir, )r8 rN �mZdoctyper r r �doctype_matches� s
rW c C s
t | d�S )z3Check if the file looks like it has a html doctype.Zhtml)rW )r8 r r r �html_doctype_matches� s rX c C sl t j| �rdS t| �}yt| S tk
rf tj| �}|dk rDdS tj| dd� �dk }|t|< |S X dS )z2Check if a doctype exists or if we have some tags.TNi� )� xml_decl_rerS �hash�_looks_like_xml_cache�KeyErrorrR �tag_rerM )r8 �keyrV r9 r r r �looks_like_xml� s
r_ c C s d| d? d| d@ fS )Ni�� �
i � i� r )�cr r r �_surrogatepair� s rb c C s, || k rt d��| dk s |dk r(t d��tjdkrFdt| �t|�f S t| �\}}t|�\}}||kr�dt|�t|�t|�f S g }|jdt|�t|�||kr�t|�p�td�f � || d kr�|jd
t|d � t|d �td �td�� ||k�r|jdt|�td �t|�f � d d
j|� d S dS )zEReturns a regular expression string to match the given non-BMP range.zBad character rangei z+unirange is only defined for non-BMP rangesi�� z[%s-%s]z
(?:%s[%s-%s])z %s[%s-%s]i�� rE z[%s-%s][%s-%s]i � z(?:�|�)N)r# �sys�
maxunicode�unichrrb r. r )�a�bZahZalZbhZblZbufr r r �unirange� s*
&
rj c C s� g }d| d }d|d d }|j || d � |rXx\|D ]}|j || d � q<W n<x:|D ]2}t|d �}|j ||dd � |d d � q^W |j |d � d
j|�S )
z)Formats a sequence of strings for output.r+ � rE z = (�,�"NrD rd r> ���rF )r. �reprr ) Zvar_name�seq�rawZ indent_level�linesZ base_indentZ inner_indent�i�rr r r � format_lines s
&ru c C sB g }t � }x2| D ]*}||ks||kr&q|j|� |j|� qW |S )za
Returns a list with duplicates removed from the iterable `it`.
Order is preserved.
)�setr. �add)�itZ already_seenZlst�seenrs r r r �duplicates_removed s
rz c @ s e Zd ZdZdd� ZdS )�Futurez�Generic class to defer some work.
Handled specially in RegexLexerMeta, to support regex string construction at
first use.
c C s t �d S )N)�NotImplementedError)�selfr r r r
* s z
Future.getN)r r r r r
r r r r r{ $ s r{ c C st y| j d�} | dfS tk
rn y ddl}|j� }| j � } | |fS ttfk
rh | j d�} | dfS X Y nX dS )z�Decode *text* with guessed encoding.
First try UTF-8; this should fail for non-UTF-8 encodings.
Then try the preferred locale encoding.
Fall back to latin-1, which always works.
zutf-8r N�latin1)�decode�UnicodeDecodeError�locale�getpreferredencoding� LookupError)r8 r� Z prefencodingr r r � guess_decode. s
r� c C sD t |dd�r<y| j|j�} W n tk
r0 Y n X | |jfS t| �S )z�Decode *text* coming from terminal *term*.
First try the terminal encoding, if given.
Then try UTF-8. Then try the preferred locale encoding.
Fall back to latin-1, which always works.
�encodingN)�getattrr r� r� r� )r8 �termr r r �guess_decode_from_terminalC s
r� c C s" t | dd�r| jS ddl}|j� S )z7Return our best guess of encoding for the given *term*.r� Nr )r� r� r� r� )r� r� r r r �terminal_encodingT s r� � �ur* )�StringIO�BytesIO�
TextIOWrapperc @ s e Zd Zdd� ZdS )�UnclosingTextIOWrapperc C s | j � d S )N)�flush)r} r r r �closew s z UnclosingTextIOWrapper.closeN)r r r r� r r r r r� u s r� c s � fdd�}|S )z6Class decorator for creating a class with a metaclass.c sR | j j� }|jdd � |jdd � x |jdf �D ]}|j|� q0W � | j| j|�S )N�__dict__� __weakref__� __slots__)r� �copy�popr
r � __bases__)�clsZ orig_varsZ slots_var)� metaclassr r �wrapper} s
z add_metaclass.<locals>.wrapperr )r� r� r )r� r �
add_metaclass{ s r� )NF)N)N)N)Fr )r� r )<r rJ re rK rH �DOTALL� MULTILINE�VERBOSErR �UNICODErL r] rU rY r# r r4 r r r! r$ r) r3 r= rQ rW rX r[ r_ rb rj ru rz �objectr{ r� r� r� � version_inforg Zxranger Zunicoder Z text_typeZu_prefix�dictZ iteritemsZ
itervaluesr� Z cStringIOr� �chr�range�items�values�ior� r� r� r r r r �<module>
sf
-
'