File: //usr/lib/python3/dist-packages/pdfminer/__pycache__/cmapdb.cpython-38.pyc
U
�R�]�/ � @ s� d Z ddlZddlZddlZddlZz ddlZW n ek
rL ddlZY nX ddlZddl Z ddl
m Z ddl
m Z ddl
m
Z
ddl
mZ ddl
mZ dd l
mZ dd
lmZ dd lmZ dd lmZ ddlZe �e�ZG d
d� de�ZG dd� de �Z G dd� de �Z G dd� de �Z G dd� de �Z G dd� de �Z!G dd� de �Z"G dd � d e!�Z#G d d � d e �Z$G d d � d e!�Z%G d!d"� d"e �Z&G d#d$� d$e �Z'd%d&� Z(ed'k�r�e�)e(ej*�� dS )(a Adobe character mapping (CMap) support.
CMaps provide the mapping between character codes and Unicode
code-points to character ids (CIDs).
More information is available on the Adobe website:
http://opensource.adobe.com/wiki/display/cmap/CMap+Resources
� N� )�
PSStackParser)�
PSSyntaxError)�PSEOF)� PSLiteral)� literal_name)�KWD)� name2unicode)�choplist)�nunpackc @ s e Zd ZdS )� CMapErrorN��__name__�
__module__� __qualname__� r r �1/usr/lib/python3/dist-packages/pdfminer/cmapdb.pyr % s r c @ s@ e Zd ZdZdd� Zdd� Zdd� Zdd � Zd
d � Zd d
� Z dS )�CMapBaser c K s |� � | _d S �N)�copy�attrs��self�kwargsr r r �__init__- s
zCMapBase.__init__c C s | j �dd�dkS )N�WModer �r �get�r r r r � is_vertical1 s zCMapBase.is_verticalc C s || j |< d S r )r )r �k�vr r r �set_attr4 s
zCMapBase.set_attrc C s d S r r )r �code�cidr r r � add_code2cid8 s zCMapBase.add_code2cidc C s d S r r �r r$ r# r r r �add_cid2unichr; s zCMapBase.add_cid2unichrc C s d S r r �r �cmapr r r �use_cmap> s zCMapBase.use_cmapN)
r r r �debugr r r" r% r' r* r r r r r ) s r c @ s> e Zd Zdd� Zdd� Zdd� Zdd� Zejd d fd
d �Z d S ) �CMapc K s t j| f|� i | _d S r )r r �code2cidr r r r r D s z
CMap.__init__c C s d| j �d� S )Nz
<CMap: %s>�CMapNamer r r r r �__repr__I s z
CMap.__repr__c s8 t |t�sttt|����� fdd�� � | j|j� d S )Nc sB t �|�D ]2\}}t|t�r4i }|| |< � ||� q
|| |< q
d S r )�six� iteritems�
isinstance�dict)Zdst�srcr r! �d�r r r r O s
zCMap.use_cmap.<locals>.copy)r2 r, �AssertionError�str�typer- r( r r6 r r* L s z
CMap.use_cmapc c sV t �d| |� | j}t�|�D ]2}||krJ|| }t|t�rP|V | j}q | j}q d S )Nzdecode: %r, %r)�logr+ r- r0 Z iterbytesr2 �int)r r# r5 �ir r r �decodeZ s
z CMap.decodeNc C sf |d kr| j }d}tt�|��D ]@\}}||f }t|t�rP|�d||f � q | j|||d� q d S )Nr zcode %r = cid %d
)�outr- r# )r- �sortedr0 r1 r2 r; �write�dump)r r> r- r# r r! �cr r r rA g s
z CMap.dump)
r r r r r/ r* r= �sys�stdoutrA r r r r r, B s
r, c @ s e Zd Zdd� ZdS )� IdentityCMapc C s( t |�d }|r t�d| |�S dS d S )N� z>%dHr ��len�structZunpack�r r# �nr r r r= v s zIdentityCMap.decodeN�r r r r= r r r r rE t s rE c @ s e Zd Zdd� ZdS )�IdentityCMapBytec C s$ t |�}|r t�d| |�S dS d S )Nz>%dBr rG rJ r r r r= � s zIdentityCMapByte.decodeNrL r r r r rM ~ s rM c @ s2 e Zd Zdd� Zdd� Zdd� Zejfdd�Zd S )
�
UnicodeMapc K s t j| f|� i | _d S r )r r �
cid2unichrr r r r r � s zUnicodeMap.__init__c C s d| j �d� S )Nz<UnicodeMap: %s>r. r r r r r r/ � s zUnicodeMap.__repr__c C s t �d| |� | j| S )Nzget_unichr: %r, %r)r: r+ rO )r r$ r r r �
get_unichr� s zUnicodeMap.get_unichrc C s0 t t�| j��D ]\}}|�d||f � qd S )Nzcid %d = unicode %r
)r? r0 r1 rO r@ )r r> r r! r r r rA � s zUnicodeMap.dumpN) r r r r r/ rP rC rD rA r r r r rN � s rN c @ s e Zd Zdd� ZdS )�FileCMapc C s� t |t�rt |t�s,ttt|�t|�f���| j}|d d� D ].}t|�}||kr\|| }q>i }|||< |}q>t|d �}|||< d S )N���)r2 r8 r; r7 r9 r- �ord)r r# r$ r5 rB �tr r r r% � s ,
zFileCMap.add_code2cidN)r r r r% r r r r rQ � s rQ c @ s e Zd Zdd� ZdS )�FileUnicodeMapc C s| t |t�sttt|����t |t�r6t|j�| j|< nBt |t �rT|�
dd�| j|< n$t |t�rpt � |�| j|< nt
|��d S )NzUTF-16BE�ignore)r2 r; r7 r8 r9 r r �namerO �bytesr= r0 Zunichr� TypeErrorr& r r r r' � s
z FileUnicodeMap.add_cid2unichrN)r r r r' r r r r rU � s rU c @ s e Zd Zdd� ZdS )�PyCMapc C s* t j| |d� |j| _|jr&d| jd< d S �N)r. r r )r, r ZCODE2CIDr- Z IS_VERTICALr )r rW �moduler r r r � s
zPyCMap.__init__N�r r r r r r r r rZ � s rZ c @ s e Zd Zdd� ZdS )� PyUnicodeMapc C s2 t j| |d� |r&|j| _d| jd< n|j| _d S r[ )rN r Z CID2UNICHR_VrO r Z CID2UNICHR_H)r rW r\ �verticalr r r r � s zPyUnicodeMap.__init__Nr] r r r r r^ � s r^ c @ sJ e Zd Zi Zi ZG dd� de�Zedd� �Zedd� �Z ed dd ��Z
d
S ) �CMapDBc @ s e Zd ZdS )zCMapDB.CMapNotFoundNr
r r r r � CMapNotFound� s ra c C s� |� dd�}d| }t�d|� tj�dd�tj�tj�t �d�f}|D ]V}tj�||�}tj�
|�rHt � |�}z"tt|�dt�|�� ��W � S |�
� X qHt�|��d S ) N� � z %s.pickle.gzz loading: %rZ CMAP_PATHz/usr/share/pdfminer/r) r )�replacer: �info�os�environr �path�join�dirname�__file__�exists�gzip�open�closer9 r8 �pickle�loads�readr` ra )�klassrW �filenameZ
cmap_pathsZ directoryrh Zgzfiler r r �
_load_data� s �
" zCMapDB._load_datac C s� |dkrt dd�S |dkr$t dd�S |dkr6tdd�S |dkrHtdd�S z | j| W S tk
rh Y nX | �|�}t||� | j|< }|S )Nz
Identity-Hr )r z
Identity-Vr ZOneByteIdentityHZOneByteIdentityV)rE rM � _cmap_cache�KeyErrorru rZ )rs rW �datar) r r r �get_cmap� s
zCMapDB.get_cmapFc sZ z| j � | W S tk
r$ Y nX | �d� �� � �fdd�dD � | j �< }|| S )Nz
to-unicode-%sc s g | ]}t �� |��qS r )r^ )�.0r! �rx rW r r �
<listcomp> s z*CMapDB.get_unicode_map.<locals>.<listcomp>)FT)� _umap_cacherw ru )rs rW r_ Zumapsr r{ r �get_unicode_map s zCMapDB.get_unicode_mapN)F) r r r rv r} r ra � classmethodru ry r~ r r r r r` � s
r` c @ s� e Zd Zdd� Zdd� Zed�Zed�Zed�Zed�Z ed �Z
ed
�Z ed �Z ed �Z
ed
�Zed�Zed�Zed�Zed�Zed�Zed�Zed�Zdd� ZdS )�
CMapParserc C s t �| |� || _d| _d S )NT)r r r) �_in_cmap)r r) �fpr r r r s zCMapParser.__init__c C s& z | � � W n tk
r Y nX d S r )Z
nextobjectr r r r r �run s
zCMapParser.runs begincmaps endcmaps usecmaps defs begincodespaceranges endcodespaceranges
begincidranges endcidranges begincidchars
endcidchars beginbfranges
endbfranges beginbfchars endbfchars beginnotdefranges endnotdefrangec C s� || j kr d| _| �� d S || jkr0d| _d S | js:d S || jkr�z,| �d�\\}}\}}| j�t|�|� W n t k
r� Y nX d S || j
kr�z*| �d�\\}}| j� t �
t|��� W n( t k
r� Y n t jk
r� Y nX d S || jk�r| �� d S || jk�r | �� d S || jk�r4| �� d S || jk�rFdd� | �� D �}td|�D ]�\}} }
t|t��r\t| t��r\t|
t��r\t|�t| �k�r��q\|d d� } | d d� } | | k�rƐq\|dd � }
| dd � }t|
�}t|�}t|
�}t|| d �D ]6}| t�d || �| d � }| j� ||
| � �q�q\d S || j k�r^| �� d S || j k�r�d
d� | �� D �}td|�D ]4\}
}t|t��r�t|
t��r�| j� |t|
�� �q�d S || j k�r�| �� d S || j k�r�d d� | �� D �}td|�D ]�\}} }t|t!��r t| t!��r t|�t| �k�r8�q t|�}t| �}t|t"��r�t|| d �D ] }| j�#|| || � �qdnp|dd � }t|�}|d d� }t|�}t|| d �D ]6}|t�d || �| d � }| j�#|| |� �q��q d S || j$k�r| �� d S || j%k�rvd d� | �� D �}td|�D ]4\}
}t|
t!��r<t|t!��r<| j�#t|
�|� �q<d S || j&k�r�| �� d S || j'k�r�| �� d S | �(||f� d S )
NTFrF r c S s g | ] \}}|�qS r r �rz �__�objr r r r| U s z)CMapParser.do_keyword.<locals>.<listcomp>� ���z>Lc S s g | ] \}}|�qS r r r� r r r r| l s c S s g | ] \}}|�qS r r r� r r r r| v s c S s g | ] \}}|�qS r r r� r r r r| � s ))�KEYWORD_BEGINCMAPr� Zpopall�KEYWORD_ENDCMAP� KEYWORD_DEF�popr) r" r r �KEYWORD_USECMAPr* r` ry ra �KEYWORD_BEGINCODESPACERANGE�KEYWORD_ENDCODESPACERANGE�KEYWORD_BEGINCIDRANGE�KEYWORD_ENDCIDRANGEr
r2 r8 r; rH r �rangerI Zpackr% �KEYWORD_BEGINCIDCHAR�KEYWORD_ENDCIDCHAR�KEYWORD_BEGINBFRANGE�KEYWORD_ENDBFRANGErX �listr' �KEYWORD_BEGINBFCHAR�KEYWORD_ENDBFCHAR�KEYWORD_BEGINNOTDEFRANGE�KEYWORD_ENDNOTDEFRANGE�push)r �pos�token�_r r! ZcmapnameZobjs�s�er$ ZsprefixZeprefixZsvarZevar�s1Ze1Zvlenr< �xr# �var�base�prefixr r r �
do_keyword- s�
��
� zCMapParser.do_keywordN)r r r r r� r r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r� r r r r r� s& r� c C sH | dd � }|D ]2}t |d�}t� }t||��� |�� |�� qd S )Nr �rb)rn rU r� r� ro rA )�argv�args�fnamer� r) r r r �main� s
r� �__main__)+�__doc__rC rf �os.pathrm ZcPicklerp � ImportErrorrI ZloggingZpsparserr r r r r r Z
encodingdbr Zutilsr
r r0 Z getLoggerr r: � Exceptionr �objectr r, rE rM rN rQ rU rZ r^ r` r� r� �exitr� r r r r �<module> sJ
2
7