File: //usr/lib/python3/dist-packages/pdfminer/__pycache__/converter.cpython-38.pyc
U
�R�]^P � @ sv d dl Z d dlZddlmZ ddlmZ ddlmZ ddlmZ ddlm Z ddlm
Z
dd lm Z dd
lm Z dd lm
Z
dd lmZ dd
lmZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ d dlZe �e �Z G dd� de�Z G dd� de �Z G dd � d e �Z G d d � d e �Z!G d d � d e �Z"G d!d"� d"e �Z#dS )#� N� )�
PDFTextDevice)�PDFUnicodeNotDefined)� LTContainer)�LTPage)�LTText)�LTLine)�LTRect)�LTCurve)�LTFigure)�LTImage)�LTChar)�
LTTextLine)� LTTextBox)�LTTextBoxVertical)� LTTextGroup)�apply_matrix_pt)� mult_matrix)�enc)�bbox2str)�utilsc @ s^ e Zd Zddd�Zdd� Zdd� Zd d
� Zd d � Zd
d� Zdd� Z dd� Z
dd� Z dd� Z dS )�PDFLayoutAnalyzerr Nc C s" t �| |� || _|| _g | _d S �N)r �__init__�pageno�laparams�_stack��self�rsrcmgrr r � r �4/usr/lib/python3/dist-packages/pdfminer/converter.pyr # s
zPDFLayoutAnalyzer.__init__c C s` |j \}}}}t|||f�\}}t|||f�\}}ddt|| �t|| �f}t| j|�| _d S )Nr )�mediaboxr �absr r �cur_item)r �page�ctm�x0�y0�x1�y1r" r r r! �
begin_page* s z PDFLayoutAnalyzer.begin_pagec C sl | j rttt| j ����t| jt�s6ttt| j����| jd k rN| j� | j� | j
d7 _
| � | j� d S )Nr ) r �AssertionError�str�len�
isinstancer$ r �typer Zanalyzer �receive_layout)r r% r r r! �end_page2 s
zPDFLayoutAnalyzer.end_pagec C s( | j �| j� t||t|| j��| _d S r )r �appendr$ r r r& )r �name�bbox�matrixr r r! � begin_figure; s z PDFLayoutAnalyzer.begin_figurec C s@ | j }t| j t�s$ttt| j ����| j�� | _ | j �|� d S r ) r$ r/ r r, r- r0 r �pop�add)r �_Zfigr r r! �
end_figure@ s
z PDFLayoutAnalyzer.end_figurec C sR t | jt�s ttt| j����t||| jj| jj| jj | jj
f�}| j� |� d S r ) r/ r$ r r, r- r0 r r' r( r) r* r9 )r r4 �stream�itemr r r! � render_imageG s �� z PDFLayoutAnalyzer.render_imagec C s d� dd� |D ��}|dkr�|d \}}} |d \}}
} t| j|| f�\}} t| j|
| f�\}
} ||
ksp| | kr�| j�t|j|| f|
| f||||j|j�� d S |dk�r�|d \}}} |d \}}
} |d \}} }
|d \}}}t| j|| f�\}} t| j|
| f�\}
} t| j| |
f�\} }
t| j||f�\}}||
k�rZ| |
k�rZ| |k�rZ|| k�s�| | k�r�|
| k�r�|
|k�r�||k�r�| j�t |j|| | |
f||||j|j�� d S g }|D ]@}t
dt |�d�D ](}|� t| j|| ||d f�� �q̐q�| j�t
|j|||||j|j�� d S )
N� c s s | ]}|d V qdS )r Nr )�.0�xr r r! � <genexpr>P s z/PDFLayoutAnalyzer.paint_path.<locals>.<genexpr>Zmlr r Zmlllh� � )�joinr r& r$ r9 r � linewidthZscolor�ncolorr �ranger. r3 r
)r �gstate�stroke�fill�evenodd�path�shaper: r' r( r) r* Zx2Zy2Zx3Zy3Zpts�p�ir r r! �
paint_pathO sf �
(���� �* �z PDFLayoutAnalyzer.paint_pathc
C s� z*|� |�} t| tj�s(ttt| ����W n tk
rJ | �||�} Y nX |� |�}
|�
|�} t |||||| |
| ||�
} | j �
| � | jS r )Z to_unichrr/ �sixZ text_typer, r- r0 r �handle_undefined_charZ
char_widthZ char_dispr
r$ r9 Zadv)
r r6 �font�fontsizeZscalingZrise�cid�ncs� graphicstate�textZ textwidthZtextdispr= r r r! � render_chars s
z PDFLayoutAnalyzer.render_charc C s t �d||� d| S )Nzundefined: %r, %rz(cid:%d))�log�info)r rT rV r r r! rS s z'PDFLayoutAnalyzer.handle_undefined_charc C s d S r r �r Zltpager r r! r1 � s z PDFLayoutAnalyzer.receive_layout)r N)
�__name__�
__module__� __qualname__r r+ r2 r7 r; r> rQ rZ rS r1 r r r r! r ! s
$ r c @ s& e Zd Zd dd�Zdd� Zdd� ZdS )
�PDFPageAggregatorr Nc C s t j| |||d� d | _d S )N�r r )r r �resultr r r r! r � s zPDFPageAggregator.__init__c C s
|| _ d S r �rc r] r r r! r1 � s z PDFPageAggregator.receive_layoutc C s | j S r rd �r r r r! �
get_result� s z PDFPageAggregator.get_result)r N)r^ r_ r` r r1 rf r r r r! ra � s
ra c @ s e Zd Zddd�ZdS )� PDFConverter�utf-8r Nc C s� t j| |||d� || _|| _t| jd�rFd| jjkr>d| _q�d| _nfdd l}t| j|j �rdd| _nHt| j|j
�rzd| _n2z| j� d� d| _W n t k
r� d| _Y nX d S )Nrb �mode�bTFr � é)
r r �outfp�codec�hasattrri � outfp_binary�ior/ �BytesIO�StringIO�write� TypeError)r r rl rm r r rp r r r! r � s$
zPDFConverter.__init__)rh r N)r^ r_ r` r r r r r! rg � s rg c @ s6 e Zd Zddd�Zdd� Zd d
� Zd d � Zd
d� ZdS )�
TextConverterrh r NFc C s& t j| |||||d� || _|| _d S �N)rm r r )rg r �
showpageno� imagewriter)r r rl rm r r rw rx r r r! r � s zTextConverter.__init__c C s4 t �|| jd�}tjr$| jr$|�� }| j�|� d S )N�ignore) r Zcompatible_encode_methodrm rR ZPY3ro �encoderl rs �r rY r r r! �
write_text� s
zTextConverter.write_textc s: � �fdd�� �j r$��d|j � � |� ��d� d S )Nc sp t | t�r | D ] }� |� qnt | t�r6��| �� � t | t�rL��d� n t | t�rl�jd k rl�j�| � d S )N�
) r/ r r r| �get_textr r rx � export_image�r= �child��renderr r r! r� � s
z,TextConverter.receive_layout.<locals>.renderzPage %s
� )rw r| �pageidr] r r� r! r1 � s
z TextConverter.receive_layoutc C s | j d krd S t�| ||� d S r )rx rg r> )r r4 r< r r r! r> � s
zTextConverter.render_imagec C s d S r r )r rI rJ rK rL rM r r r! rQ � s zTextConverter.paint_path)rh r NFN)r^ r_ r` r r| r1 r> rQ r r r r! ru � s �
ru c
@ s� e Zd Zddddddd�Zddd �Zd
d d d d
dddd dddd�ddif dd�Zdd� Zdd� Zdd� Zd d � Z d d � Z
d d!� Z d"d#� Z d$d%� Z
d3d'd(�Zd)d*� Zd+d,� Zd-d.� Zd/d0� Zd1d2� Zd S )4�
HTMLConverterZyellowZmagentaZcyanZredZblackZgray)�figure�textline�textbox� textgroup�curver% Zblue)r� �charrh r Ng �?ZnormalT�2 r )r� r% r� c C s� t j| |||||d� || _|| _|| _| | _|
| _| | _|
| _|| _ | rf| j�
| j � | j �
| j � | j| _
d | _g | _| �� d S rv )rg r �scale� fontscale�
layoutmoderw �
pagemarginrx � rect_colors� text_colors�update� RECT_COLORS� TEXT_COLORS�_yoffset�_font�
_fontstack� write_header)r r rl rm r r r� r� r� rw r� rx �debugr� r� r r r! r � s" zHTMLConverter.__init__c C s" | j r|�| j �}| j�|� d S r �rm rz rl rs r{ r r r! rs s zHTMLConverter.writec C s: | � d� | jr"| � d| j � n
| � d� | � d� d S )Nz
<html><head>
zA<meta http-equiv="Content-Type" content="text/html; charset=%s">
z5<meta http-equiv="Content-Type" content="text/html">
z</head><body>
)rs rm re r r r! r� s
zHTMLConverter.write_headerc C s4 | � dd�dd� td| j�D �� � | � d� d S )Nz8<div style="position:absolute; top:0px;">Page: %s</div>
z, c s s | ]}d ||f V qdS )z<a href="#%s">%s</a>Nr )r@ rP r r r! rB s z-HTMLConverter.write_footer.<locals>.<genexpr>r z</body></html>
)rs rE rH r re r r r! � write_footer s
�
zHTMLConverter.write_footerc C s | � t|d �� d S r )rs r r{ r r r! r| $ s zHTMLConverter.write_textc
C sP | j �|�}|d k rL| �d|||| j | j| | j || j || j f � d S )Nzn<span style="position:absolute; border: %s %dpx solid; left:%dpx; top:%dpx; width:%dpx; height:%dpx;"></span>
)r� �getrs r� r� )r �color� borderwidthrA �y�w�hr r r! �
place_rect( s ��zHTMLConverter.place_rectc C s | � |||j|j|j|j� d S r )r� r' r* �width�height)r r� r� r= r r r! � place_border2 s zHTMLConverter.place_borderc
C sX | j d k rT| j �|�}| �dt|d �||| j | j| | j || j || j f � d S )Nzd<img src="%s" border="%d" style="position:absolute; left:%dpx; top:%dpx;" width="%d" height="%d" />
)rx r rs r r� r� )r r= r� rA r� r� r� r4 r r r! � place_image6 s
��zHTMLConverter.place_imagec C s` | j �|�}|d k r\| �d||| j | j| | j || j | j f � | �|� | �d� d S )NzP<span style="position:absolute; color:%s; left:%dpx; top:%dpx; font-size:%dpx;">z</span>
)r� r� rs r� r� r� r| )r r� rY rA r� �sizer r r! �
place_text@ s (�
zHTMLConverter.place_textFc C sR | j �| j� d | _| �d||||| j | j| | j || j || j f � d S )Nzv<div style="position:absolute; border: %s %dpx solid; writing-mode:%s; left:%dpx; top:%dpx; width:%dpx; height:%dpx;">)r� r3 r� rs r� r� )r r� r� rA r� r� r� Z writing_moder r r! � begin_divI s ��zHTMLConverter.begin_divc C s. | j d k r| �d� | j�� | _ | �d� d S )N�</span>z</div>)r� rs r� r8 )r r� r r r! �end_divS s
zHTMLConverter.end_divc C s\ ||f}|| j krN| j d k r&| �d� | �dt|�|| j | j f � || _ | �|� d S )Nr� z.<span style="font-family: %s; font-size:%dpx">)r� rs r r� r� r| )r rY �fontnamerU rT r r r! �put_textZ s
�
zHTMLConverter.put_textc C s | � d� d S )Nz<br>�rs re r r r! � put_newlinee s
zHTMLConverter.put_newlinec s: ��fdd��� ��fdd�� � |� � j �j7 _ d S )Nc s. t | t�r*� �dd| � | D ] }�|� q d S )Nr� r )r/ r r� r� �r �
show_groupr r! r� j s
z0HTMLConverter.receive_layout.<locals>.show_groupc
s� t | t�r�� j| j7 _��dd| � �jr`��d�j| j �j � ��d| j| jf � | D ] }� |� qd| j d k r�| j D ] }�|� q��nt | t
�r���dd| � �n�t | t �r��� dd| j
| j| j| j� | D ] }� |� qڈ�d� �n�t | t��r"��| d| j
| j| j| j� �n��jdk�r�t | t��r^��dd| � | D ]}� |� �qLn�t | t��r���d d| � ��d t| jd �| j
| jd
� | D ]}� |� �q�n6t | t��r���d d| � ��d | �� | j
| j| j� n�t | t��r | D ]}� |� �q��jd k�r��� � n�t | t��rj�� d d| j
| j| j| j| � � � | D ]}� |� �qN��d � n>t | t��r��� | �� | j | j� nt | t ��r���!| �� � d S )
Nr% r z*<div style="position:absolute; top:%dpx;">z <a name="%s">Page %s</a></div>
r� r� �exactr� r� � r� Zloose)"r/ r r� r* r� rw rs r� r� �groupsr
r r� r' r� r� r� r r� r� r r r� r- �indexr
r~ r� r� Zget_writing_moder� r� r r| )r= r� �group�r� r r� r r! r� q sh
�
� z,HTMLConverter.receive_layout.<locals>.render)r� r� r] r r� r! r1 i s
5z HTMLConverter.receive_layoutc C s | � � d S r �r� re r r r! �close� s zHTMLConverter.close)F)r^ r_ r` r� r� r rs r� r� r| r� r� r� r� r� r� r� r� r1 r� r r r r! r� � sF � � �
Ar� c @ sP e Zd Ze�d�Zddd�Zdd � Zd
d � Zd d
� Z dd� Z
dd� Z dd� Z dS )� XMLConverterz [ - - - ]rh r NFc C s. t j| |||||d� || _|| _| �� d S rv )rg r rx � stripcontrolr� )r r rl rm r r rx r� r r r! r � s
zXMLConverter.__init__c C s" | j r|�| j �}| j�|� d S r r� r{ r r r! rs � s zXMLConverter.writec C s0 | j r| �d| j � n
| �d� | �d� d S )Nz%<?xml version="1.0" encoding="%s" ?>
z<?xml version="1.0" ?>
z<pages>
)rm rs re r r r! r� � s
zXMLConverter.write_headerc C s | � d� d S )Nz </pages>
r� re r r r! r� � s
zXMLConverter.write_footerc C s( | j r| j�d|�}| �t|d �� d S )Nr? )r� �CONTROL�subrs r r{ r r r! r| � s zXMLConverter.write_textc s* ��fdd��� ��fdd�� � |� d S )Nc sd t | t�r&� �d| jt| j�f � n:t | t�r`� �dt| j� � | D ] }�|� qH� �d� d S )Nz <textbox id="%d" bbox="%s" />
z<textgroup bbox="%s">
z
</textgroup>
)r/ r rs r� r r5 r r� r� r r! r� � s
�
z/XMLConverter.receive_layout.<locals>.show_groupc s� t | t�rz��d| jt| j�| jf � | D ] }� |� q,| jd k rl��d� | jD ] }�|� qT��d� ��d� �nLt | t�r���d| j t| j�f � �n$t | t
�rʈ�d| j t| j�f � �n�t | t �r���d| j t| j�| � � f � �n�t | t
��r@��d| jt| j�f � | D ]}� |� �q"��d � �n�t | t��r���d
t| j� � | D ]}� |� �qd��d � �nDt | t��r�d }t | t��r�d
}��d| jt| j�|f � | D ]}� |� �q�d� n�t | t��r4��dt| jd �t| j�| jj| jj| jf � ��| �� � ��d� n�t | t ��rT��d| �� � nrt | t ��r��j d k �r��j � | �}��dt|d �| j | j!f � n��d| j | j!f � nd�s�t"t#d| f���d S )Nz%<page id="%s" bbox="%s" rotate="%d">
z <layout>
z
</layout>
z</page>
z"<line linewidth="%d" bbox="%s" />
z"<rect linewidth="%d" bbox="%s" />
z+<curve linewidth="%d" bbox="%s" pts="%s"/>
z <figure name="%s" bbox="%s">
z
</figure>
z<textline bbox="%s">
z </textline>
r? z wmode="vertical"z <textbox id="%d" bbox="%s"%s>
z </textbox>
zD<text font="%s" bbox="%s" colourspace="%s" ncolour="%s" size="%.3f">z</text>
z<text>%s</text>
z*<image src="%s" width="%d" height="%d" />
z!<image width="%d" height="%d" />
FZ Unhandled)$r/ r rs r� r r5 �rotater� r rF r r
Zget_ptsr r4 r r r r� r
r r� rW rX rG r� r| r~ r r rx r r� r� r, r- )r= r� r� Zwmoder4 r� r r! r� � s�
�
�
�
�
� � �� �
�z+XMLConverter.receive_layout.<locals>.renderr r] r r� r! r1 � s <zXMLConverter.receive_layoutc C s | � � d S r r� re r r r! r� s zXMLConverter.close)rh r NNF)
r^ r_ r` �re�compiler� r rs r� r� r| r1 r� r r r r! r� � s
�
Kr� )$Zloggingr� Z pdfdevicer Zpdffontr Zlayoutr r r r r r
r r r
r r r r r r r r r r? rR Z getLoggerr^ r[ r ra rg ru r� r� r r r r! �<module> s<
h 1 K