File: //usr/lib/python3/dist-packages/pdfminer/__pycache__/psparser.cpython-38.pyc
U
�R�]�C � @ s� d dl Z d dlZd dlZddlmZ ddlmZ e�e�Z G dd� de
�Z G dd� de �Z G d d
� d
e �Z
G d d � d e �ZG d
d� de �ZG dd� de�ZG dd� de�ZG dd� de�ZG dd� de�Zee�Zee�ZejZejZed�Zed�Zed�Z ed�Z ed�Z ed �Z d d � Z d d � Z!e �"d!�Z#e �"d"�Z$e �"d#�Z%e �"d$�Z&e �"d%�Z'e �"d&�Z(e �"d'�Z)e �"d(�Z*e �"d%�Z+e �"d)�Z,e �"d*�Z-d+d,d-d.d/d0d1d2d3�Z.G d4d5� d5e�Z/G d6d7� d7e/�Z0dS )8� N� )�settings)�choplistc @ s e Zd ZdS )� PSExceptionN��__name__�
__module__� __qualname__� r
r
�3/usr/lib/python3/dist-packages/pdfminer/psparser.pyr s r c @ s e Zd ZdS )�PSEOFNr r
r
r
r r s r c @ s e Zd ZdS )�
PSSyntaxErrorNr r
r
r
r r
s r
c @ s e Zd ZdS )� PSTypeErrorNr r
r
r
r r s r c @ s e Zd ZdS )� PSValueErrorNr r
r
r
r r " s r c @ s e Zd ZdZdS )�PSObjectz0Base class for all PS or PDF-related data types.N)r r r �__doc__r
r
r
r r + s r c @ s e Zd ZdZdd� Zdd� ZdS )� PSLiteralaa A class that represents a PostScript literal.
Postscript literals are used as identifiers, such as
variable names, property names and dictionary keys.
Literals are case sensitive and denoted by a preceding
slash sign (e.g. "/Name")
Note: Do not create an instance of PSLiteral directly.
Always use PSLiteralTable.intern().
c C s
|| _ d S �N��name��selfr r
r
r �__init__A s zPSLiteral.__init__c C s | j }d| S �Nz/%rr r r
r
r �__repr__D s zPSLiteral.__repr__N�r r r r r r r
r
r
r r 4 s r c @ s e Zd ZdZdd� Zdd� ZdS )� PSKeywordaN A class that represents a PostScript keyword.
PostScript keywords are a dozen of predefined words.
Commands and directives in PostScript are expressed by keywords.
They are also used to denote the content boundaries.
Note: Do not create an instance of PSKeyword directly.
Always use PSKeywordTable.intern().
c C s
|| _ d S r r r r
r
r r W s zPSKeyword.__init__c C s | j }d| S r r r r
r
r r [ s zPSKeyword.__repr__Nr r
r
r
r r K s
r c @ s e Zd ZdZdd� Zdd� ZdS )�
PSSymbolTablez�A utility class for storing PSLiteral/PSKeyword objects.
Interned objects can be checked its identity with "is" operator.
c C s i | _ || _d S r ��dict�klass)r r r
r
r r i s zPSSymbolTable.__init__c C s. || j kr| j | }n| �|�}|| j |< |S r r )r r Zlitr
r
r �internn s
zPSSymbolTable.internN)r r r r r r! r
r
r
r r b s r � {� }� [� ]s <<s >>c C sR t | t�s&tjr td| f ��qN| }n(| j}tjrNzt|d�}W n Y nX |S )NzLiteral required: %r�utf-8) �
isinstancer r �STRICTr r �six�PY3�str��xr r
r
r � literal_name� s
r. c C s@ t | t�s$tjr td| ��q<| }n| j}tjr<t|dd�}|S )NzKeyword required: %rr& �ignore) r' r r r( r r r) r* r+ r, r
r
r � keyword_name� s
r0 s [\r\n]s \ss \Ss [0-9a-fA-F]s [#/%\[\]()<>{}\s]s [^\s0-9a-fA-F]s [0-9a-fA-F]{2}|.s [^0-9]s [()\134]s [0-7]� � �
� �
�( �) �\ )� b� t� n� f� r� (� )� \c @ s� e Zd ZdZdZdd� Zdd� Zdd� Zd d
� Zd d � Z d5dd�Z
dd� Z dd� Z dd� Z
dd� Zdd� Zdd � Zd d � Zd d � Zd!d"� Zd#d$� Zd%d&� Zd'd(� Zd)d*� Zd+d,� Zd-d.� Zd/d0� Zd1d2� Zd3d4� Z d
S )6� PSBaseParserzBMost basic PostScript parser that performs only tokenization.
i c C s || _ | �d� d S �Nr )�fp�seek�r rC r
r
r r � s
zPSBaseParser.__init__c C s d| j j| j| jf S )Nz<%s: %r, bufpos=%d>)� __class__r rC �bufpos�r r
r
r r � s zPSBaseParser.__repr__c C s d S r r
rH r
r
r �flush� s zPSBaseParser.flushc C s | � � d S r )rI rH r
r
r �close� s zPSBaseParser.closec C s | j | j S r )rG �charposrH r
r
r �tell� s zPSBaseParser.tellN�P c C sL | j �� }|s| j| j }| j �|� t�d|| j �|�� | j �|� d S )Nz poll(%d): %r)rC rL rG rK rD �log�info�read)r �pos�nZpos0r
r
r �poll� s
zPSBaseParser.pollc C sH t �d|� | j�|� || _d| _d| _| j| _d| _ d| _
g | _ dS )z0Seeks the parser to the given position.
zseek: %r� r N) rN �debugrC rD rG �bufrK � _parse_main�_parse1� _curtoken� _curtokenpos�_tokens�r rQ r
r
r rD � s zPSBaseParser.seekc C sH | j t| j�k rd S | j�� | _| j�| j�| _| js>td��d| _ d S )NzUnexpected EOFr ) rK �lenrV rC rL rG rP �BUFSIZr rH r
r
r �fillbuf� s zPSBaseParser.fillbufc C s� d}| j | j }d}| �� |rV| j| j| jd � }|dkr�||7 }| jd7 _q�t�| j| j�}|r�|| j| j|�d�� 7 }|�d�| _|dd� dkr�d }q�q�q|| j| jd� 7 }t| j�| _qt� d
||� ||fS ) z<Fetches a next line that ends either with \r or \n.
rT Fr �
r ���N�
Tznextline: %r, %r)
rG rK r_ rV �EOL�search�endr] rN rU )r ZlinebufZlineposZeol�c�mr
r
r �nextline� s* zPSBaseParser.nextlinec c s� | j �dd� | j �� }d}d|k r�|}td|| j �}| j �|� | j �|| �}|sZq�t|�d�|�d��}|dkr�|| }q ||d� | V |d|� }d}qZq dS )ziFetches a next line backword.
This is used to locate the trailers at the end of a file.
r � rT rb r` ra N)rC rD rL �maxr^ rP �rfind)r rQ rV Zprevpos�srR r
r
r � revreadlines s$
zPSBaseParser.revreadlinesc C s\ t �||�}|st|�S |�d�}|||d � }| j| | _|dkr\d| _| j| _|d S |dkrzd| _| j | _|d S |dks�|�
� r�|| _| j | _|d S |dkr�|| _| j | _|d S |�
� r�|| _| j| _|d S |dk�rd| _d| _| j| _|d S |d k�r"d| _| j| _|d S |d
k�rBd| _| j| _|d S | �t|�� |d S d S ) Nr r � %� /rT s -+� .r> � <� >)�NONSPCrd r] �startrG rZ rY �_parse_commentrX �_parse_literal�isdigit�
_parse_number� _parse_float�isalpha�_parse_keyword�paren�
_parse_string� _parse_wopen�
_parse_wclose�
_add_token�KWD�r rl �irg �jrf r
r
r rW ! sR
zPSBaseParser._parse_mainc C s | j �| j|f� d S r )r[ �appendrZ )r �objr
r
r r� M s zPSBaseParser._add_tokenc C sZ t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _| j| _|S rB )rc rd rY r] rt rW rX �r rl r� rg r� r
r
r ru Q s
zPSBaseParser._parse_commentc C s� t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _|||d � }|dkr|d| _| j| _|d S zt| jd�| _W n Y nX | � t
| j�� | j | _|S )Nr r � #rT r& ) � END_LITERALrd rY r] rt �hex�_parse_literal_hexrX r+ r� �LITrW r� r
r
r rv ] s$
zPSBaseParser._parse_literalc C sl |||d � }t �|�r>t| j�dk r>| j|7 _|d S | jr`| jt�t| jd��7 _| j| _ |S )Nr ri � )
�HEX�matchr] r� rY r) �int2byte�intrv rX �r rl r� rf r
r
r r� q s z PSBaseParser._parse_literal_hexc C s� t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _|||d � }|dkr�| j|7 _| j| _|d S z| �t| j�� W n t k
r� Y nX | j
| _|S )Nr r rp ) �
END_NUMBERrd rY r] rt ry rX r� r� �
ValueErrorrW r� r
r
r rx { s"
zPSBaseParser._parse_numberc C s� t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _z| �t| j�� W n tk
rv Y nX | j| _ |S rB )
r� rd rY r] rt r� �floatr� rW rX r� r
r
r ry � s
zPSBaseParser._parse_floatc C s� t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _| jdkr^d}n| jdkrnd}n
t| j�}| �|� | j| _|S )Nr s trueTs falseF) � END_KEYWORDrd rY r] rt r� r� rW rX �r rl r� rg r� �tokenr
r
r r{ � s
zPSBaseParser._parse_keywordc C s� t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _|||d � }|dkr|d| _| j| _|d S |dkr�| jd7 _| j|7 _|d S |dkr�| jd8 _| jr�| j|7 _|d S | � | j� | j
| _|d S )Nr r r@ rT r> r? ) �
END_STRINGrd rY r] rt �oct�_parse_string_1rX r| r� rW r� r
r
r r} � s.
zPSBaseParser._parse_stringc C s� |||d � }t �|�r>t| j�dk r>| j|7 _|d S | jrl| jt�t| jd��7 _| j| _ |S |t
kr�| jt�t
| �7 _| j| _ |d S )Nr � r1 ) �
OCT_STRINGr� r] r� rY r) r� r� r} rX �
ESC_STRINGr� r
r
r r� � s z PSBaseParser._parse_string_1c C s@ |||d � }|dkr4| � t� | j| _|d7 }n| j| _|S )Nr rq )r� �KEYWORD_DICT_BEGINrW rX �_parse_hexstringr� r
r
r r~ � s
zPSBaseParser._parse_wopenc C s6 |||d � }|dkr*| � t� |d7 }| j| _|S )Nr rr )r� �KEYWORD_DICT_ENDrW rX r� r
r
r r � s
zPSBaseParser._parse_wclosec C s~ t �||�}|s.| j||d � 7 _t|�S |�d�}| j|||� 7 _t�dd� t�d| j��}| �|� | j | _
|S )Nr c S s t �t| �d�d��S )Nr r� )r) r� r� �group)rg r
r
r �<lambda>� rT z/PSBaseParser._parse_hexstring.<locals>.<lambda>rT ) �END_HEX_STRINGrd rY r] rt �HEX_PAIR�sub�SPCr� rW rX r� r
r
r r� � s
z PSBaseParser._parse_hexstringc C s> | j s"| �� | �| j| j�| _q | j �d�}t�d|� |S )Nr z
nexttoken: %r)r[ r_ rX rV rK �poprN rU )r r� r
r
r � nexttoken� s zPSBaseParser.nexttoken)NrM ) r r r r r^ r r rI rJ rL rS rD r_ rh rm rW r� ru rv r� rx ry r{ r} r� r~ r r� r� r
r
r
r rA � s4
,
rA c @ sd e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dS )�
PSStackParserc C s t �| |� | �� d S r )rA r �resetrE r
r
r r � s zPSStackParser.__init__c C s g | _ d | _g | _g | _d S r )�context�curtype�curstack�resultsrH r
r
r r� s
zPSStackParser.resetc C s t �| |� | �� d S r )rA rD r� r\ r
r
r rD s zPSStackParser.seekc G s | j �|� d S r )r� �extend�r �objsr
r
r �push s zPSStackParser.pushc C s$ | j | d � }g | j | d �< |S r �r� )r rR r� r
r
r r� s zPSStackParser.popc C s | j }g | _ |S r r� r� r
r
r �popall s zPSStackParser.popallc G s8 zt �d|� W n t �d� Y nX | j�|� d S )Nzadd_results: %rz!add_results: (unprintable object))rN rU r� r� r� r
r
r � add_results s zPSStackParser.add_resultsc C s6 | j �|| j| jf� |g | _| _t�d||� d S )Nzstart_type: pos=%r, type=%r)r� r� r� r� rN rU )r rQ �typer
r
r �
start_type& s zPSStackParser.start_typec C sX | j |kr td| j |f ��dd� | jD �}| j�� \}| _ | _t�d|||� ||fS )NzType mismatch: %r != %rc S s g | ] \}}|�qS r
r
)�.0�_r� r
r
r �
<listcomp>/ s z*PSStackParser.end_type.<locals>.<listcomp>z"end_type: pos=%r, type=%r, objs=%r)r� r r� r� r� rN rU )r r� r� rQ r
r
r �end_type, s
zPSStackParser.end_typec C s d S r r
)r rQ r� r
r
r �
do_keyword4 s zPSStackParser.do_keywordc C s& | j �s�| �� \}}t|tjtttjtjt f�rB| �
||f� �n�|t krZ| � |d� �n~|t
kr�z| �
| �d�� W n tk
r� tjr�� Y nX �n@|tkr�| � |d� �n(|tk�r2zT| �d�\}}t|�d dkr�td| ��tdd� td|�D ��}| �
||f� W n tk
�r. tj�r*� Y nX n�|tk�rJ| � |d� n�|tk�r�z| �
| �d�� W n tk
�r� tj�r�� Y nX nLt|t��r�t� d ||| j � | � ||� n t� d
||| j � | � ||� � | j �r�q q | �!� q | j �"d�}zt� d |� W n t� d � Y nX |S )
z�Yields a list of objects.
Returns keywords, literals, strings, numbers, arrays and dictionaries.
Arrays and dictionaries are represented as Python lists and dictionaries.
�a�dri r z Invalid dictionary construct: %rc s s& | ] \}}|d k rt |�|fV qd S r )r. )r� �k�vr
r
r � <genexpr>W s z+PSStackParser.nextobject.<locals>.<genexpr>�pz&do_keyword: pos=%r, token=%r, stack=%rz)unknown token: pos=%r, token=%r, stack=%rznextobject: %rz nextobject: (unprintable object))#r� r� r' r) Z
integer_typesr� �boolZ string_typesZ binary_typer r� �KEYWORD_ARRAY_BEGINr� �KEYWORD_ARRAY_ENDr� r r r( r� r� r] r
r r �KEYWORD_PROC_BEGIN�KEYWORD_PROC_ENDr rN rU r� r� �errorr� rI r� )r rQ r� r� r� r� r
r
r �
nextobject7 s^
zPSStackParser.nextobjectN)r r r r r� rD r� r� r� r� r� r� r� r� r
r
r
r r� � s r� )1�reZloggingr) � r Zutilsr Z getLoggerr rN � Exceptionr r r
r r �objectr r r r ZPSLiteralTableZPSKeywordTabler! r� r� r� r� r� r� r� r� r. r0 �compilerc r� rs r� r� r� r� r� r� r� r� r� rA r� r
r
r
r �<module> sT
Q