File: //lib/python3/dist-packages/pdfminer/__pycache__/layout.cpython-38.pyc
U
�R�]'X � @ s d dl mZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ ddlmZ dd lm Z dd
lm
Z
d d l Z G d d
� d
e �Z
G dd� de �ZG dd� de �ZG dd� de �ZG dd� de�ZG dd� de�ZG dd� de�ZG dd� de�ZG d d � d e�ZG d d � d ee�ZG d d!� d!ee�ZG d"d#� d#e�ZG d$d%� d%e�ZG d&d'� d'ee�ZG d(d)� d)e�ZG d*d+� d+e�Z G d,d-� d-e�Z G d.d/� d/e�Z G d0d1� d1e �Z G d2d3� d3e �Z G d4d5� d5e�Z!G d6d7� d7e!�Z"G d8d9� d9e!�Z#G d:d;� d;e�Z$G d<d=� d=e$�Z%G d>d?� d?e$�Z&d S )@� )�SortedListWithKey� )�INF)�Plane)� get_bound)�uniq)�fsplit)�bbox2str)�
matrix2str��apply_matrix_ptNc @ s e Zd Zddd�Zdd� ZdS )�
IndexAssignerr c C s
|| _ d S �N��index)�selfr � r �1/usr/lib/python3/dist-packages/pdfminer/layout.py�__init__ s zIndexAssigner.__init__c C sD t |t�r"| j|_| jd7 _n t |t�r@|D ]}| �|� q0d S �Nr )�
isinstance� LTTextBoxr � LTTextGroup�run)r �obj�xr r r r s
zIndexAssigner.runN)r )�__name__�
__module__� __qualname__r r r r r r r
s
r
c @ s e Zd Zd
dd�Zdd� Zd S ) �LAParams� �?� @皙�����?Fc C s. || _ || _|| _|| _|| _|| _|| _d S r )� line_overlap� char_margin� line_margin� word_margin�
boxes_flow�detect_vertical� all_texts)r r# r$ r% r& r' r( r) r r r r $ s zLAParams.__init__c C s d| j | j| j| jf S )NzM<LAParams: char_margin=%.1f, line_margin=%.1f, word_margin=%.1f all_texts=%r>)r$ r% r& r) �r r r r �__repr__5 s �zLAParams.__repr__N)r r! r r" r FF�r r r r r+ r r r r r " s �
r c @ s e Zd Zdd� ZdS )�LTItemc C s dS )z Perform the layout analysis.Nr �r �laparamsr r r �analyze> s zLTItem.analyzeN�r r r r0 r r r r r- < s r- c @ s e Zd Zdd� Zdd� ZdS )�LTTextc C s d| j j| �� f S )Nz<%s %r>)� __class__r �get_textr* r r r r+ G s �zLTText.__repr__c C s t �d S r ��NotImplementedErrorr* r r r r4 K s zLTText.get_textN)r r r r+ r4 r r r r r2 E s r2 c @ s| e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
d� Z dd� Z
dd� Z dd� Z dd� Z
dd� Zdd� Zdd � Zd S ) � LTComponentc C s t �| � | �|� d S r )r- r �set_bbox�r �bboxr r r r S s
zLTComponent.__init__c C s d| j jt| j�f S )Nz<%s %s>)r3 r r r: r* r r r r+ X s �zLTComponent.__repr__c C s t �d S r ��
ValueError�r �_r r r �__lt__] s zLTComponent.__lt__c C s t �d S r r; r= r r r �__le___ s zLTComponent.__le__c C s t �d S r r; r= r r r �__gt__a s zLTComponent.__gt__c C s t �d S r r; r= r r r �__ge__c s zLTComponent.__ge__c C sB |\}}}}|| _ || _|| _|| _|| | _|| | _|| _d S r )�x0�y0�x1�y1�width�heightr: )r r: rC rD rE rF r r r r8 f s
zLTComponent.set_bboxc C s | j dkp| jdkS �Nr )rG rH r* r r r �is_emptyq s zLTComponent.is_emptyc C s2 t |t�sttt|����|j| jko0| j|jkS r )r r7 �AssertionError�str�typerC rE �r r r r r � is_hoverlapt s zLTComponent.is_hoverlapc C sN t |t�sttt|����| �|�r(dS tt| j|j �t| j |j ��S d S rI �
r r7 rK rL rM rO �min�absrC rE rN r r r � hdistancex s
zLTComponent.hdistancec C sN t |t�sttt|����| �|�rFtt| j|j �t| j |j ��S dS d S rI rP rN r r r �hoverlap s
"zLTComponent.hoverlapc C s2 t |t�sttt|����|j| jko0| j|jkS r )r r7 rK rL rM rD rF rN r r r � is_voverlap� s zLTComponent.is_voverlapc C sN t |t�sttt|����| �|�r(dS tt| j|j �t| j |j ��S d S rI �
r r7 rK rL rM rU rQ rR rD rF rN r r r � vdistance� s
zLTComponent.vdistancec C sN t |t�sttt|����| �|�rFtt| j|j �t| j |j ��S dS d S rI rV rN r r r �voverlap� s
"zLTComponent.voverlapN)r r r r r+ r? r@ rA rB r8 rJ rO rS rT rU rW rX r r r r r7 Q s r7 c @ s e Zd Zddd�Zdd� ZdS )�LTCurveFNc C s> t �| t|�� || _|| _|| _|| _|| _|| _|| _ d S r )
r7 r r �pts� linewidth�stroke�fill�evenodd�stroking_color�non_stroking_color)r r[ rZ r\ r] r^ r_ r` r r r r � s zLTCurve.__init__c C s d� dd� | jD ��S )N�,c s s | ]}d | V qdS )z %.3f,%.3fNr )�.0�pr r r � <genexpr>� s z"LTCurve.get_pts.<locals>.<genexpr>)�joinrZ r* r r r �get_pts� s zLTCurve.get_pts)FFFNN)r r r r rf r r r r rY � s
rY c @ s e Zd Zddd�ZdS )�LTLineFNc
C s t �| |||g|||||� d S r �rY r ) r r[ Zp0Zp1r\ r] r^ r_ r` r r r r � s zLTLine.__init__)FFFNN�r r r r r r r r rg � s rg c @ s e Zd Zddd�ZdS )�LTRectFNc
C s@ |\}} }
} t �| ||| f|
| f|
| f|| fg|||||� d S r rh ) r r[ r: r\ r] r^ r_ r` rC rD rE rF r r r r � s 0zLTRect.__init__)FFFNNri r r r r rj � s rj c @ s e Zd Zdd� Zdd� ZdS )�LTImagec C sn t �| |� || _|| _|�d�|�d�f| _|�d�| _|�dd�| _|�d�| _t | jt
�sj| jg| _d S )N)�WZWidth)�HZHeight)ZIMZ ImageMask)ZBPCZBitsPerComponentr )ZCSZ
ColorSpace) r7 r �name�streamZget_any�srcsizeZ imagemask�bitsZ
colorspacer �list)r rn ro r: r r r r � s �
zLTImage.__init__c C s d| j j| jt| j�| jf S �Nz<%s(%s) %s %r>)r3 r rn r r: rp r* r r r r+ � s
��zLTImage.__repr__Nr, r r r r rk � s
rk c @ s e Zd Zdd� Zdd� ZdS )�LTAnnoc C s
|| _ d S r ��_text)r �textr r r r � s zLTAnno.__init__c C s | j S r ru r* r r r r4 � s zLTAnno.get_textN�r r r r r4 r r r r rt � s rt c @ s, e Zd Zdd� Zdd� Zdd� Zdd� Zd S )
�LTCharc C s� t �| � || _|| _|j| _| | _|
| _|| | | _|�� r�|� � | } |\} }
| d krf| d } n | | d } d|
| d }
| }|
| }||| j f}|| |f}n6|�
� | }|� � | }|| }d|f}| j|| f}| j\}}}}}}d|| | k �o|| dk| _ t
| j|�\}}t
| j|�\} } | |k �rH| | }} | |k �r\| | }} t�| ||| | f� |�� �r�| j| _n| j| _d S )Nr g����MbP?i� r )r2 r rv �matrix�fontname�ncs� graphicstate�advZ is_verticalZ get_widthZ
get_heightZ get_descentZuprightr r7 rG �sizerH ) r rz ZfontZfontsizeZscalingZriserw Z textwidthZtextdispr| r} rG ZvxZvyZtxZtyZbllZburrH Zdescent�a�b�c�d�e�frC rD rE rF r r r r � sH
zLTChar.__init__c C s, d| j jt| j�t| j�| j| j| �� f S )Nz(<%s %s matrix=%s font=%r adv=%s text=%r>) r3 r r r: r
rz r{ r~ r4 r* r r r r+ s ��zLTChar.__repr__c C s | j S r ru r* r r r r4 s zLTChar.get_textc C s dS )z<Returns True if two characters can coexist in the same line.Tr rN r r r �
is_compatible s zLTChar.is_compatibleN)r r r r r+ r4 r� r r r r ry � s -ry c @ s< e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd d � Zd
S )� LTContainerc C s t �| |� g | _d S r )r7 r �_objsr9 r r r r % s zLTContainer.__init__c C s
t | j�S r )�iterr� r* r r r �__iter__* s zLTContainer.__iter__c C s
t | j�S r )�lenr� r* r r r �__len__- s zLTContainer.__len__c C s | j �|� d S r )r� �appendrN r r r �add0 s zLTContainer.addc C s |D ]}| � |� qd S r )r� )r �objsr r r r �extend4 s zLTContainer.extendc C s | j D ]}|�|� qd S r )r� r0 )r r/ r r r r r0 9 s
zLTContainer.analyzeN) r r r r r� r� r� r� r0 r r r r r� # s r� c @ s e Zd Zdd� Zdd� ZdS )�LTExpandableContainerc C s t �| t
t
t t f� d S r )r� r r r* r r r r C s z LTExpandableContainer.__init__c C sJ t �| |� | �t| j|j�t| j|j�t| j|j�t| j|j�f� d S r ) r� r� r8 rQ rC rD �maxrE rF rN r r r r� G s �zLTExpandableContainer.addN)r r r r r� r r r r r� A s r� c @ s e Zd Zdd� Zdd� ZdS )�LTTextContainerc C s t �| � t�| � d S r )r2 r r� r* r r r r R s
zLTTextContainer.__init__c C s d� dd� | D ��S )N� c s s | ]}t |t�r|�� V qd S r )r r2 r4 �rb r r r r rd X s
z+LTTextContainer.get_text.<locals>.<genexpr>)re r* r r r r4 W s zLTTextContainer.get_textNrx r r r r r� P s r� c @ s, e Zd Zdd� Zdd� Zdd� Zdd� Zd S )
�
LTTextLinec C s t �| � || _d S r )r� r r& �r r& r r r r _ s
zLTTextLine.__init__c C s d| j jt| j�| �� f S )Nz
<%s %s %r>)r3 r r r: r4 r* r r r r+ d s
��zLTTextLine.__repr__c C s t �| |� t�| td�� d S )N�
)r� r0 r� r� rt r. r r r r0 i s zLTTextLine.analyzec C s t �d S r r5 )r �plane�ratior r r �find_neighborsn s zLTTextLine.find_neighborsN)r r r r r+ r0 r� r r r r r� ] s r� c @ s$ e Zd Zdd� Zdd� Zdd� ZdS )�LTTextLineHorizontalc C s t �| |� t
| _d S r )r� r r �_x1r� r r r r t s z LTTextLineHorizontal.__init__c C s\ t |t�rD| jrD| jt|j|j� }| j|j| k rDt� | t
d�� |j | _t � | |� d S �N� )
r ry r& r� rG rH r� rC r� r� rt rE r� �r r Zmarginr r r r� y s zLTTextLineHorizontal.addc s@ |�j � |��j�j� �j�j� f�}� �fdd�|D �S )Nc sV g | ]N}t |t�rt|j�j �� k rt|j�j �� k sNt|j�j �� k r|�qS r )r r� rR rH rC rE r� �r� r r r �
<listcomp>� s
�z7LTTextLineHorizontal.find_neighbors.<locals>.<listcomp>)rH �findrC rD rE rF �r r� r� r� r r� r r� � s
"z#LTTextLineHorizontal.find_neighborsN�r r r r r� r� r r r r r� r s r� c @ s$ e Zd Zdd� Zdd� Zdd� ZdS )�LTTextLineVerticalc C s t �| |� t | _d S r )r� r r �_y0r� r r r r � s zLTTextLineVertical.__init__c C s\ t |t�rD| jrD| jt|j|j� }|j| | jk rDt� | t
d�� |j | _t � | |� d S r� )
r ry r& r� rG rH rF r� r� r� rt rD r� r� r r r r� � s zLTTextLineVertical.addc s@ |�j � |��j� �j�j� �jf�}� �fdd�|D �S )Nc sV g | ]N}t |t�rt|j�j �� k rt|j�j �� k sNt|j�j �� k r|�qS r )r r� rR rG rD rF r� r� r r r� � s
�z5LTTextLineVertical.find_neighbors.<locals>.<listcomp>)rG r� rC rD rE rF r� r r� r r� � s
"z!LTTextLineVertical.find_neighborsNr� r r r r r� � s r� c @ s e Zd Zdd� Zdd� ZdS )r c C s t �| � d| _d S )N���)r� r r r* r r r r � s
zLTTextBox.__init__c C s d| j j| jt| j�| �� f S rs )r3 r r r r: r4 r* r r r r+ � s ��zLTTextBox.__repr__Nr, r r r r r � s r c @ s e Zd Zdd� Zdd� ZdS )�LTTextBoxHorizontalc C s" t �| |� | jjdd� d� d S )Nc S s | j S r )rF �r r r r �<lambda>� � z-LTTextBoxHorizontal.analyze.<locals>.<lambda>��key�r r0 r� �sortr. r r r r0 � s zLTTextBoxHorizontal.analyzec C s dS )Nzlr-tbr r* r r r �get_writing_mode� s z$LTTextBoxHorizontal.get_writing_modeN�r r r r0 r� r r r r r� � s r� c @ s e Zd Zdd� Zdd� ZdS )�LTTextBoxVerticalc C s" t �| |� | jjdd� d� d S )Nc S s | j S r )rE r� r r r r� � r� z+LTTextBoxVertical.analyze.<locals>.<lambda>r� r� r. r r r r0 � s zLTTextBoxVertical.analyzec C s dS )Nztb-rlr r* r r r r� � s z"LTTextBoxVertical.get_writing_modeNr� r r r r r� � s r� c @ s e Zd Zdd� ZdS )r c C s t �| � | �|� d S r )r� r r� )r r� r r r r � s
zLTTextGroup.__init__Nri r r r r r � s r c @ s e Zd Zdd� ZdS )�LTTextGroupLRTBc s& t �| � � | jj� fdd�d� d S )Nc s& d� j | j d� j | j| j S r )r' rC rD rF r� �r/ r r r� � s �z)LTTextGroupLRTB.analyze.<locals>.<lambda>r� �r r0 r� r� r. r r� r r0 � s zLTTextGroupLRTB.analyzeNr1 r r r r r� � s r� c @ s e Zd Zdd� ZdS )�LTTextGroupTBRLc s& t �| � � | jj� fdd�d� d S )Nc s( d� j | j| j d� j | j S r )r' rC rE rF r� r� r r r� � s �z)LTTextGroupTBRL.analyze.<locals>.<lambda>r� r� r. r r� r r0 � s zLTTextGroupTBRL.analyzeNr1 r r r r r� � s r� c @ s4 e Zd Zdd� Zdd� Zdd� Zdd� Zd d
� Zd S ) �LTLayoutContainerc C s t �| |� d | _d S r )r� r �groupsr9 r r r r � s zLTLayoutContainer.__init__c c s� d }d }|D �]j}|d k �rt|� |�oj|�|�ojt|j|j�|j |�|�k oj|�|�t|j|j�|j k }|j
o�|� |�o�|� |�o�t|j|j�|j |� |�k o�|�
|�t|j|j�|j k }|r�t|t�s�|r�t|t�r�|�|� n�|d k r�|V d }nv|�r*|�s*t|j�}|�|� |�|� nJ|�rV|�sVt|j�}|�|� |�|� n t|j�}|�|� |V d }|}q |d k�r�t|j�}|�|� |V d S r )r� rU rQ rH r# rX rS r� rG r$ r( rO rT rW r r� r� r� r& )r r/ r� Zobj0�line�obj1ZhalignZvalignr r r �
group_objects� sn
�������������
z LTLayoutContainer.group_objectsc c s� t | j�}|�|� i }|D ]�}|�||j�}||kr8q g }|D ]&}|�|� ||kr@|�|�|�� q@t|t�rzt � } nt
� } t |�D ]}
| � |
� | ||
< q�q t
� } |D ]8}||kr�q�|| } | | kr�q�| � | � | �� s�| V q�d S r )r r: r� r� r% r� �popr r� r� r� r r� �setrJ ) r r/ �linesr� �boxesr� Z neighbors�membersr� �boxr Zdoner r r �group_textlines@ s:
z!LTLayoutContainer.group_textlinesc s� |st t||f���dd� }� fdd�}dd� }t|d�}tt|��D ]D}|| }t|d t|��D ]$} || }
|�d |||
�||
f� qdqFt| j�� � �|� |�r�|� d �\} } }}
| d kr�|||
�r�|�d| ||
f� q�t
|t t f�s�t
|
t t f��r
t ||
g�}
n t
||
g�}
� �|� � �|
� ||
g��fd
d �|D �}|D ]}|�|� �qH� D ] }|�d ||
|�|
|f� �q^� �|
� q�t� �dk�s�t tt� ����t� �S ) Nc S s` t | j|j�}t | j|j�}t| j|j�}t| j|j�}|| || | j| j |j|j S )a� A distance function between two TextBoxes.
Consider the bounding rectangle for obj1 and obj2.
Return its area less the areas of obj1 and obj2,
shown as 'www' below. This value may be negative.
+------+..........+ (x1, y1)
| obj1 |wwwwwwwwww:
+------+www+------+
:wwwwwwwwww| obj2 |
(x0, y0) +..........+------+
)rQ rC rD r� rE rF rG rH )r� �obj2rC rD rE rF r r r �distb s
z/LTLayoutContainer.group_textboxes.<locals>.distc s\ t | j|j�}t | j|j�}t| j|j�}t| j|j�}t� �||||f��}|�| |f�S )zECheck if there's any other object between obj1 and obj2.
) rQ rC rD r� rE rF r� r� �
difference)r� r� rC rD rE rF r� )r� r r �isanyt s z0LTLayoutContainer.group_textboxes.<locals>.isanyc S s | \}}}}||fS r r )�tr� r� r> r r r �key_obj~ s z2LTLayoutContainer.group_textboxes.<locals>.key_objr� r r c s0 g | ](\}}}}|� ks |� kr||||f�qS r r )rb r� r� r� r� )�removedr r r� � s
�z5LTLayoutContainer.group_textboxes.<locals>.<listcomp>)rK rL r �ranger� r� r r: r� r� r r� r� r� �removerr )r r/ r� r� r� r� Zdists�ir� �jr� r� r� �groupZ to_remove�r�otherr )r� r� r �group_textboxes_ sB
�
z!LTLayoutContainer.group_textboxesc C s� t dd� | �\}}|D ]}|�|� q|s.d S t| �||��}t dd� |�\}}|D ]}|�|� qTt| �||��}d|jkr�|jdkr�|r�| �||�| _t� }| jD ]} | �|� |� | � q�|j
dd� d� ndd � }
|j
|
d� || | | _ d S )
Nc S s
t | t�S r )r ry r� r r r r� � r� z+LTLayoutContainer.analyze.<locals>.<lambda>c S s | � � S r )rJ r� r r r r� � r� r� r c S s | j S r r �r� r r r r� � r� r� c S s, t | t�rd| j | jfS d| j| jfS d S )Nr r )r r� rE rD rC r� r r r �getkey� s
z)LTLayoutContainer.analyze.<locals>.getkey) r r0 rr r� r� r' r� r� r
r r� r� ) r r/ ZtextobjsZ otherobjsr Z textlinesZemptiesZ textboxesZassignerr� r� r r r r0 � s*
zLTLayoutContainer.analyzeN)r r r r r� r� r� r0 r r r r r� � s
H Br� c @ s$ e Zd Zdd� Zdd� Zdd� ZdS )�LTFigurec sf || _ � | _|\}}}}t� fdd�||f|| |f||| f|| || ffD ��}t�| |� d S )Nc 3 s | ]\}}t � ||f�V qd S r r )rb rc �q�rz r r rd � s �z$LTFigure.__init__.<locals>.<genexpr>)rn rz r r� r )r rn r: rz r �y�w�hr r� r r � s *� zLTFigure.__init__c C s" d| j j| jt| j�t| j�f S )Nz<%s(%s) %s matrix=%s>)r3 r rn r r: r
rz r* r r r r+ � s
��zLTFigure.__repr__c C s |j s
d S t�| |� d S r )r) r� r0 r. r r r r0 � s zLTFigure.analyzeN)r r r r r+ r0 r r r r r� � s r� c @ s e Zd Zddd�Zdd� ZdS )�LTPager c C s t �| |� || _|| _d S r )r� r �pageid�rotate)r r� r: r� r r r r � s zLTPage.__init__c C s d| j j| jt| j�| jf S )Nz<%s(%r) %s rotate=%r>)r3 r r� r r: r� r* r r r r+ � s
��zLTPage.__repr__N)r r, r r r r r� � s
r� )'Zsortedcontainersr Zutilsr r r r r r r
r Zsix�objectr
r r- r2 r7 rY rg rj rk rt ry r� r� r� r� r� r� r r� r� r r� r� r� r� r� r r r r �<module> sH J
?
S