Mini Shell
3
s� ^ k � @ s� d dl Z d dlZd dlZd dlZd dlZd dlZd dlmZ ej ddddddg�Z
d g iZ d
Z ej
d��Zddddd�Zi Zi Zi Zi Zi ZG dd� d�ZG dd� d�ZG dd � d �ZG d d � d �ZG d d � d �ZG d!d"� d"ee�ZG d#d$� d$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!d/d0� Z"d1d2� Z#d3d4� Z$d5d6� Z%d7d8� Z&d9d:� Z'd;d<� Z(d�d>d?�Z)d@dA� Z*d�dCdD�Z+dEdF� Z,d�dHdI�Z-d�dJdK�Z.d�dLdM�Z/d�dNdO�Z0d�dPdQ�Z1d�dRdS�Z2d�dTdU�Z3d�dVdW�Z4dXZ5d�dZd[�Z6d�d\d]�Z7d�d^d_�Z8d�d`da�Z9d�dbdc�Z:d�ddde�Z;d�dfdg�Z<dhdi� Z=djdk� Z>dldm� Z?d�dndo�Z@d�dqdr�ZAG dsdt� dt�ZBeB� ZCeDduk�r�d dlEZEeA� eEjF� dS )�� N)�ContentHandler�Locale�language�script� territory�variant�encoding�data_files_readz/usr/share/langtablez^(?P<language>[a-z]{2,3}z(?=$|@z,|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))z|_[A-Z0-9]{2,3}(?=$|@)z))z (?:_(?P<script>[A-Z][a-z]{3})z)){0,1}z (?:_(?P<territory>[A-Z0-9]{2,3})ZLatnZCyrlZDeva)�latinZiqtelif�cyrillicZ
devanagaric @ s e Zd Zddd�ZdS )�territory_db_itemNc C s4 || _ || _|| _|| _|| _|| _|| _|| _d S )N)�names�scripts�locales� languages� keyboards� inputmethods� consolefonts� timezones) �selfr
r r r r r r r � r � /usr/lib/python3.6/langtable.py�__init__� s zterritory_db_item.__init__)NNNNNNNN)�__name__�
__module__� __qualname__r r r r r r � s r c @ s e Zd Zddd�ZdS )�language_db_itemNc C sF || _ || _|| _|| _|| _|| _|| _|| _| | _|
| _ | | _
d S )N) �iso639_1�
iso639_2_t�
iso639_2_br
r r � territoriesr r r r ) r r r r r
r r r r r r r r r r r � s zlanguage_db_item.__init__) NNNNNNNNNNN)r r r r r r r r r � s r c @ s e Zd Zddd�ZdS )�keyboard_db_itemNTc C s" || _ || _|| _|| _|| _d S )N)� description�ascii�commentr r )r r" r# r r r$ r r r r � s
zkeyboard_db_item.__init__)NTNNN)r r r r r r r r r! � s r! c @ s e Zd Zddd�ZdS )�timezone_db_itemNc C s
|| _ d S )N)r
)r r
r r r r � s ztimezone_db_item.__init__)N)r r r r r r r r r% � s r% c @ s e Zd Zddd�ZdS )�timezoneIdPart_db_itemNc C s
|| _ d S )N)r
)r r
r r r r � s z timezoneIdPart_db_item.__init__)N)r r r r r r r r r& � s r& c @ s e Zd ZdZdd� Zdd� ZdS )�LangtableContentHandlerz�
A base class inherited from the xml.sax.handler.ContentHandler class
providing handling for SAX events produced when parsing the langtable data
files.
c C s
d | _ d S )N)�_save_to)r r r r r � s z LangtableContentHandler.__init__c C s> | j dkrdS t| | j �}|r(|| }n|}t| | j |� dS )z Handler for the text data event.N)r( �getattr�setattr)r ZcontentZ old_valueZ new_valuer r r �
characters� s
z"LangtableContentHandler.charactersN)r r r �__doc__r r+ r r r r r' � s r' c s8 e Zd ZdZ� fdd�Zdd� Zdd� Zdd � Z� ZS )
�TerritoriesContentHandlerzFHandler for SAX events produced when parsing the territories.xml file.c sZ t t| �j� d | _d | _d | _d | _d | _d | _d | _ d | _
d | _ d | _ d | _
d | _d S )N)�superr- r � _territoryId�_item_id�
_item_rank�
_item_name�_names�_scripts�_locales�
_languages�
_keyboards�
_inputmethods�
_consolefonts�
_timezones)r )� __class__r r r � s z"TerritoriesContentHandler.__init__c C s� |dkrJt � | _t � | _t � | _t � | _t � | _t � | _t � | _t � | _n>|dkrZd| _ n.|dkrjd | _ n |d krzd
| _ n|dkr�d| _ d S )Nr � territoryIdr/ �
languageId�scriptId�localeId�
keyboardId�
inputmethodId�
consolefontId�
timezoneIdr0 �trNamer2 �rankr1 )r= r> r? r@ rA rB rC )
�dictr3 r4 r5 r6 r7 r8 r9 r: r( )r �name�attrsr r r � startElement� s$
z&TerritoriesContentHandler.startElementc
C s� d | _ |dkrzt| j| j| j| j| j| j| j| j d�t
t | j �< d | _ d | _d | _d | _d | _d | _d | _d | _d | _ �nF|dkr�| j
| jt | j�< | j� �n |dkr�t| j�| jt | j�< | j� n�|dkr�t| j�| jt | j�< | j� n�|dk�rt| j�| jt | j�< | j� n�|dk�rDt| j�| jt | j�< | j� n||dk�rnt| j�| jt | j�< | j� nR|d k�r�t| j�| jt | j�< | j� n(|d
k�r�t| j�| j t | j�< | j� d S ) Nr )r
r r r r r r r rG r �localer �keyboard� inputmethod� consolefont�timezone)r( r r3 r4 r5 r6 r7 r8 r9 r: �_territories_db�strr/ r2 r0 � _clear_item�intr1 )r rG r r r �
endElement sX
z$TerritoriesContentHandler.endElementc C s d | _ d | _d | _d S )N)r0 r2 r1 )r r r r rQ E s z%TerritoriesContentHandler._clear_item) r r r r, r rI rS rQ �
__classcell__r r )r; r r- � s
5r- c s8 e Zd ZdZ� fdd�Zdd� Zdd� Zdd � Z� ZS )
�KeyboardsContentHandlerzDHandler for SAX events produced when parsing the keyboards.xml file.c sB t t| �j� d | _d | _d | _d | _d | _d | _d | _ d | _
d S )N) r. rU r � _keyboardId� _description�_ascii�_commentr0 r1 r6 � _territories)r )r; r r r M s z KeyboardsContentHandler.__init__c C s| |dkrt � | _t � | _n^|dkr*d| _nN|dkr:d| _n>|dkrJd| _n.|dkrZd | _n |dkrjd | _n|d
krxd| _d S )NrK r@ rV r" rW r# rX r$ rY r= r< r0 rE r1 )r= r< )rF r6 rZ r( )r rG rH r r r rI ^ s
z$KeyboardsContentHandler.startElementc C s� d | _ |dkr^t| j| jdk| j| j| jd�tt| j �< d | _ d | _d | _d | _d | _d | _nN|dkr�t
| j �| jt| j �< | j
� n&|dkr�t
| j �| jt| j �< | j
� d S )NrK �True)r" r# r$ r r r r )r( r! rW rX rY r6 rZ �
_keyboards_dbrP rV rR r1 r0 rQ )r rG r r r rS s s(
z"KeyboardsContentHandler.endElementc C s d | _ d | _d S )N)r0 r1 )r r r r rQ � s z#KeyboardsContentHandler._clear_item) r r r r, r rI rS rQ rT r r )r; r rU J s
rU c s8 e Zd ZdZ� fdd�Zdd� Zdd� Zdd � Z� ZS )
�LanguagesContentHandlerzDHandler for SAX events produced when parsing the languages.xml file.c sr t t| �j� d | _d | _d | _d | _d | _d | _d | _ d| _
d | _ d | _ d | _
d | _d | _d | _d | _d | _d S )NF)r. r] r � _languageId� _iso639_1� _iso639_2_t� _iso639_2_br0 r1 r2 � _in_namesr3 r4 r5 rZ r7 r8 r9 r: )r )r; r r r � s" z LanguagesContentHandler.__init__c C s� |dkrJt � | _t � | _t � | _t � | _t � | _t � | _t � | _t � | _n�|dkrb| j rbd| _
n�|dkrrd| _
nt|dkr�d| _
nd|dkr�d | _
nT|d
kr�d | _ nD|dkr�d| _
n4|dkr�| j r�d| _
n |dkr�d| _
n|dkr�d| _
d S )Nr r= r^ ziso639-1r_ z
iso639-2-tr` z
iso639-2-bra r
Tr> r? r< r@ rA rB rC r0 rD r2 rE r1 )r> r? r< r@ rA rB rC ) rF r3 r4 r5 rZ r7 r8 r9 r: rb r( )r rG rH r r r rI � s8
z$LanguagesContentHandler.startElementc
C s� d | _ |dkr�t| j| j| j| j| j| j| j| j | j
| j | j d� t
t| j�< d | _d | _d | _d | _d | _d | _d | _d | _d | _ d | _
d | _ d | _ �nZ|dkr�d| _�nH|dkr�| j| jt| j�< | j� �n"|dkr�t| j�| jt| j�< | j� n�|dk�r"t| j�| jt| j�< | j� n�|dk�rLt| j�| jt| j�< | j� n�|d k�rvt| j�| j t| j�< | j� n||d
k�r�t| j�| j
t| j�< | j� nR|d k�r�t| j�| j t| j�< | j� n(|d k�r�t| j�| j t| j�< | j� d S )
Nr ) r r r r
r r r r r r r r
FrG r rJ r rK rL rM rN )r( r r_ r` ra r3 r4 r5 rZ r7 r8 r9 r: �
_languages_dbrP r^ rb r2 r0 rQ rR r1 )r rG r r r rS � sh
z"LanguagesContentHandler.endElementc C s d | _ d | _d | _d S )N)r0 r2 r1 )r r r r rQ s z#LanguagesContentHandler._clear_item) r r r r, r rI rS rQ rT r r )r; r r] � s
$?r] c s8 e Zd ZdZ� fdd�Zdd� Zdd� Zdd � Z� ZS )
�TimezonesContentHandlerzDHandler for SAX events produced when parsing the timezones.xml file.c s* t t| �j� d | _d | _d | _d | _d S )N)r. rd r � _timezoneIdr0 r2 r3 )r )r; r r r s
z TimezonesContentHandler.__init__c C sD |dkrt � | _n.|dkr"d| _n |dkr2d| _n|dkr@d| _d S )NrN rC re r= r0 rD r2 )rF r3 r( )r rG rH r r r rI ) s
z$TimezonesContentHandler.startElementc C sX d | _ |dkr2t| jd�tt| j�< d | _d | _n"|dkrT| j| jt| j�< | j� d S )NrN )r
rG ) r( r% r3 �
_timezones_dbrP re r2 r0 rQ )r rG r r r rS 9 s z"TimezonesContentHandler.endElementc C s d | _ d | _d S )N)r0 r2 )r r r r rQ K s z#TimezonesContentHandler._clear_item) r r r r, r rI rS rQ rT r r )r; r rd s
rd c s8 e Zd ZdZ� fdd�Zdd� Zdd� Zdd � Z� ZS )
� TimezoneIdPartsContentHandlerzJHandler for SAX events produced when parsing the timezoneidparts.xml file.c s* t t| �j� d | _d | _d | _d | _d S )N)r. rg r �_timezoneIdPartIdr0 r2 r3 )r )r; r r r R s
z&TimezoneIdPartsContentHandler.__init__c C sD |dkrt � | _n.|dkr"d| _n |dkr2d| _n|dkr@d| _d S )N�timezoneIdPart�timezoneIdPartIdrh r= r0 rD r2 )rF r3 r( )r rG rH r r r rI ^ s
z*TimezoneIdPartsContentHandler.startElementc C sX d | _ |dkr2t| jd�tt| j�< d | _d | _n"|dkrT| j| jt| j�< | j� d S )Nri )r
rG ) r( r& r3 �_timezoneIdParts_dbrP rh r2 r0 rQ )r rG r r r rS n s z(TimezoneIdPartsContentHandler.endElementc C s d | _ d | _d S )N)r0 r2 )r r r r rQ � s z)TimezoneIdPartsContentHandler._clear_item) r r r r, r rI rS rQ rT r r )r; r rg O s
rg c C s� | j d� | j d� �xltt�D �]^}| j d� | j d| d � t| j}| j d� x>t|�D ]2}| j dF| d d
|| jd d � d
d � q`W | j d� t| j}| j d� xFt|j� dd� d�D ].\}}| j dG| d d t|� d d � q�W | j d� t| j}| j d� xHt|j� d d� d�D ]0\}}| j dH| d d t|� d d � �q0W | j d!� t| j } | j d"� xHt| j� d#d� d�D ]0\}
}| j dI|
d d t|� d d% � �q�W | j d&� t| j
} | j d'� xHt| j� d(d� d�D ]0\} }| j dJ| d+ d t|� d d, � �q W | j d-� t| j }
| j d.� xHt|
j� d/d� d�D ]0\}}| j dK| d2 d t|� d d3 � �qhW | j d4� t| j }| j d5� xHt|j� d6d� d�D ]0\}}| j dL| d9 d t|� d d: � �q�W | j d;� t| j
}| j d<� xHt|j� d=d� d�D ]0\}}| j dM| d@ d t|� d dA � �q8W | j dB� | j dC� q W | j dD� dES )Nz
Only for internal use
z'<?xml version="1.0" encoding="UTF-8"?>
z<territories>
z <territory>
z <territoryId>z</territoryId>
z <names>
z <name>z <languageId>z
</languageId>z<trName>�&z&z </trName>z</name>
z
</names>
z <scripts>
c S s d| d | d fS )N� r ���r )�xr r r �<lambda>� s z)_write_territories_file.<locals>.<lambda>)�keyz <script>z
<scriptId>z </scriptId>z<rank>z</rank>z
</script>
z </scripts>
z <locales>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <locale>z
<localeId>z </localeId>z
</locale>
z </locales>
z <languages>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <language>z </language>
z </languages>
z <keyboards>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <keyboard>z <keyboardId>z
</keyboardId>z </keyboard>
z </keyboards>
z <inputmethods>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <inputmethod>z<inputmethodId>z</inputmethodId>z</inputmethod>
z </inputmethods>
z <consolefonts>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <consolefont>z<consolefontId>z</consolefontId>z</consolefont>
z </consolefonts>
z <timezones>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <timezone>z <timezoneId>z
</timezoneId>z </timezone>
z </timezones>
z </territory>
z</territories>
Nz <name><languageId>z <script><scriptId>z <locale><localeId>z <language><languageId>z <keyboard><keyboardId>z" <inputmethod><inputmethodId>z" <consolefont><consolefontId>z <timezone><timezoneId>)�write�sortedrO r
�replacer �itemsrP r r r r r r )�filer< r
rG r r> rE r r? r r= r r@ r rA r rB r rC r r r �_write_territories_file� sp
.
&
(
(
(
(
(
(
rw c C s� | j d� | j d� �x�tt�D �]�}| j d� | j d| d � | j dtt| j� d � | j dtt| j� d � | j d
tt| j� d � t| j}| j d � x6t|�D ]*}| j dL| d d || d d � q�W | j d� t| j}| j d� xHt|j � dd� d�D ]0\}}| j dM| d d t|� d d � �qW | j d � t| j
}| j d � xHt|j � d d� d�D ]0\}}| j dN| d# d t|� d d$ � �q~W | j d%� t| j } | j d&� xHt| j � d'd� d�D ]0\}
}| j dO|
d* d t|� d d+ � �q�W | j d,� t| j } | j d-� xHt| j � d.d� d�D ]0\} }| j dP| d1 d t|� d d2 � �qNW | j d3� t| j
}
| j d4� xHt|
j � d5d� d�D ]0\}}| j dQ| d8 d t|� d d9 � �q�W | j d:� t| j}| j d;� xHt|j � d<d� d�D ]0\}}| j dR| d? d t|� d d@ � �q W | j dA� t| j}| j dB� xHt|j � dCd� d�D ]0\}}| j dS| dF d t|� d dG � �q�W | j dH� | j dI� q W | j dJ� dKS )Tz
Only for internal use
z'<?xml version="1.0" encoding="UTF-8"?>
z <languages>
z
<language>
z <languageId>z</languageId>
z <iso639-1>z </iso639-1>
z <iso639-2-t>z</iso639-2-t>
z <iso639-2-b>z</iso639-2-b>
z <names>
z <name>z <languageId>z
</languageId>z<trName>z </trName>z</name>
z
</names>
z <scripts>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z'_write_languages_file.<locals>.<lambda>)rq z <script>z
<scriptId>z </scriptId>z<rank>z</rank>z
</script>
z </scripts>
z <locales>
c S s d| d | d fS )Nrm r rn r )ro r r r rp � s z <locale>z
<localeId>z </localeId>z
</locale>
z </locales>
z <territories>
c S s d| d | d fS )Nrm r rn r )ro r r r rp s z <territory>z
<territoryId>z</territoryId>z
</territory>
z </territories>
z <keyboards>
c S s d| d | d fS )Nrm r rn r )ro r r r rp s z <keyboard>z <keyboardId>z
</keyboardId>z </keyboard>
z </keyboards>
z <inputmethods>
c S s d| d | d fS )Nrm r rn r )ro r r r rp s z <inputmethod>z<inputmethodId>z</inputmethodId>z</inputmethod>
z </inputmethods>
z <consolefonts>
c S s d| d | d fS )Nrm r rn r )ro r r r rp s z <consolefont>z<consolefontId>z</consolefontId>z</consolefont>
z </consolefonts>
z <timezones>
c S s d| d | d fS )Nrm r rn r )ro r r r rp & s z <timezone>z <timezoneId>z
</timezoneId>z </timezone>
z </timezones>
z </language>
z
</languages>
Nz <name><languageId>z <script><scriptId>z <locale><localeId>z <territory><territoryId>z <keyboard><keyboardId>z" <inputmethod><inputmethodId>z" <consolefont><consolefontId>z <timezone><timezoneId>)rr rs rc rP r r r r
r ru r r r r r r )rv r= r
rG r r> rE r r? r r< r r@ r rA r rB r rC r r r �_write_languages_file� sv
&
(
(
(
(
(
(
(
rx c C s� | j d� | j d� �x`tt�D �]R}| j d� | j d| d � | j dt| j d � | j dtt| j� d � t| jd
kr�| j d t| j d � t| j}| j d
� xFt|j� dd� d�D ].\}}| j d!| d d t|� d d � q�W | j d� t| j }| j d� xHt|j� dd� d�D ]0\}}| j d"| d d t|� d d � �q,W | j d � | j d � q W | j d � d
S )#z
Only for internal use
z'<?xml version="1.0" encoding="UTF-8"?>
z <keyboards>
z
<keyboard>
z <keyboardId>z</keyboardId>
z <description>z</description>
z <ascii>z </ascii>
Nz
<comment>z </comment>
z <languages>
c S s d| d | d fS )Nrm r rn r )ro r r r rp @ s z'_write_keyboards_file.<locals>.<lambda>)rq z <language>z <languageId>z
</languageId>z<rank>z</rank>z </language>
z </languages>
z <territories>
c S s d| d | d fS )Nrm r rn r )ro r r r rp I s z <territory>z
<territoryId>z</territoryId>z
</territory>
z </territories>
z </keyboard>
z
</keyboards>
z <language><languageId>z <territory><territoryId>)
rr rs r\ r" rP r# r$ r ru r )rv r@ r r= rE r r< r r r �_write_keyboards_file1 s0
&
(
ry c C s� | j d� | j d� x�tt�D ]�}| j d� | j d| d � t| j}| j d� x6t|�D ]*}| j d| d d
|| d d � q\W | j d
� | j d� q W | j d� dS )z
Only for internal use
z'<?xml version="1.0" encoding="UTF-8"?>
z <timezones>
z
<timezone>
z <timezoneId>z</timezoneId>
z <names>
z <name>z <languageId>z
</languageId>z<trName>z </trName>z</name>
z
</names>
z </timezone>
z
</timezones>
Nz <name><languageId>)rr rs rf r
)rv rC r
rG r r r �_write_timezones_fileT s
&
rz c C s� | j d� | j d� x�tt�D ]�}| j d� | j d| d � t| j}| j d� x6t|�D ]*}| j d| d d
|| d d � q\W | j d
� | j d� q W | j d� dS )z
Only for internal use
z'<?xml version="1.0" encoding="UTF-8"?>
z<timezoneIdParts>
z <timezoneIdPart>
z <timezoneIdPartId>z</timezoneIdPartId>
z <names>
z <name>z <languageId>z
</languageId>z<trName>z </trName>z</name>
z
</names>
z </timezoneIdPart>
z</timezoneIdParts>
Nz <name><languageId>)rr rs rk r
)rv rj r
rG r r r �_write_timezoneIdParts_filej s
&
r{ c C s2 t jjj� }|j|_|j|_|j|_ |j
| � dS )zo
Only for internal use. Parses a given file object with a given SAX handler
using an expat parser.
N) �xmlZparsersZexpatZ ParserCreaterI ZStartElementHandlerrS ZEndElementHandlerr+ ZCharacterDataHandlerZ ParseFile)rv � sax_handler�parserr r r � _expat_parse� s
r c C s x�t jjt jjt jjt��d�t jjtd�fD ]�}t jj|| �}t jj|�r�t|dd��,}t j
d| � t ||� t d j
|� W dQ R X dS t jj|| d �}t jj|�r.tj|dd��,}t j
d| � t ||� t d j
|� W dQ R X dS q.W t j
d� dS ) z
Only for internal use
�data�rb)�modezreading file=%sr Nz.gzzno readable file found.)�os�path�join�dirname�realpath�__file__�_DATADIR�isfile�open�logging�infor �_INFO�append�gzip)�filenamer} �dirr� rv r r r �
_read_file� s$
r� c
- C s t | d�� }tjd| � t|� W dQ R X t |d�� }tjd| � t|� W dQ R X t |d�� }tjd| � t|� W dQ R X t |d�� }tjd| � t|� W dQ R X t |d�� }tjd| � t|� W dQ R X t |d�� } tjd| � t| � W dQ R X dS )z
Only for internal use
�wzwriting territories file=%sNzwriting languages file=%szwriting keyboards file=%szwriting timezones file=%sz writing timezoneidparts file=%s)r� r� r� rw rx ry rz r{ )
ZterritoriesfilenameZlanguagesfilenameZkeyboardsfilenameZtimezonesfilenameZtimezoneidpartsfilenameZterritoriesfileZ
languagesfileZ
keyboardsfileZ
timezonesfileZtimezoneidpartsfiler r r � _write_files� s& r� Tc sF g }x<t � � fdd�|d�D ]"}� | dkr |j|� | g� q W |S )Nc s � j | �| fS )N)�get)ro )rF r r rp � s z,_dictionary_to_ranked_list.<locals>.<lambda>)rq �reverser )rs r� )rF r� Z sorted_list�itemr )rF r �_dictionary_to_ranked_list� s
r� c C s t tdd� | ��S )Nc S s | d S )Nr r )ro r r r rp � s z&_ranked_list_to_list.<locals>.<lambda>)�list�map)� ranked_listr r r �_ranked_list_to_list� s r� �� c C sb t | �dks| S xLtdt | �d �D ]6}| | d | |d d |kr$| d|d � } P q$W | S )Nrm r )�len�range)r� Zcut_off_factor�ir r r �_make_ranked_list_concise� s r� c C s� d}d}d}d}d}| r�| j d�}| j d�}|dkrf||krf| |d |� }| d|� | |d� } n$|dkr�| |d d� }| d|� } | r�| j� j d�}|dk r�| j� j d�}|dkr�d }| d|� } | r�| dkr�d
}d}d }d} | �r8xJtD ]B} | j| t| �} | jdt| �r�t| }| jdt| d�} q�W | �rr| j d�}|dk�rr| |d d� j� }| d|� } | �r�tj| �}
|
�r�|
jd�}|
jd��r�|
jd�}|
jd��r�|
jd�}nt j
d| � t |||||d�S )u�
Parses a locale name in glibc or CLDR format and returns
language, script, territory, variant, and encoding
:param localeId: The name of the locale
:type localeId: string
:return: The parts of the locale: language, script, territory, variant, encoding
:rtype: A namedtuple of strings
Locale(language=string,
script=string,
territory=string,
variant=string,
encoding=string)
It replaces glibc names for scripts like “latin”
with the iso-15924 script names like “Latn”.
I.e. these inputs all give the same result:
“sr_latin_RS”
“sr_Latn_RS”
“sr_RS@latin”
“sr_RS@Latn”
Examples:
>>> parse_locale('de_DE')
Locale(language='de', script='', territory='DE', variant='', encoding='')
>>> parse_locale('de_DE.UTF-8')
Locale(language='de', script='', territory='DE', variant='', encoding='UTF-8')
>>> parse_locale('de_DE.utf8')
Locale(language='de', script='', territory='DE', variant='', encoding='utf8')
>>> parse_locale('de_DE@euro')
Locale(language='de', script='', territory='DE', variant='EURO', encoding='')
>>> parse_locale('de_DE.ISO-8859-15')
Locale(language='de', script='', territory='DE', variant='', encoding='ISO-8859-15')
>>> parse_locale('de_DE.ISO-8859-15@euro')
Locale(language='de', script='', territory='DE', variant='EURO', encoding='ISO-8859-15')
>>> parse_locale('de_DE.iso885915@euro')
Locale(language='de', script='', territory='DE', variant='EURO', encoding='iso885915')
>>> parse_locale('gez_ER.UTF-8@abegede')
Locale(language='gez', script='', territory='ER', variant='ABEGEDE', encoding='UTF-8')
>>> parse_locale('ar_ER.UTF-8@saaho')
Locale(language='ar', script='', territory='ER', variant='SAAHO', encoding='UTF-8')
>>> parse_locale('zh_Hant_TW')
Locale(language='zh', script='Hant', territory='TW', variant='', encoding='')
>>> parse_locale('zh_TW')
Locale(language='zh', script='', territory='TW', variant='', encoding='')
>>> parse_locale('es_419')
Locale(language='es', script='', territory='419', variant='', encoding='')
>>> parse_locale('sr_latin_RS')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='')
>>> parse_locale('sr_Latn_RS')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='')
>>> parse_locale('sr_RS@latin')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='')
>>> parse_locale('sr_RS@Latn')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='')
>>> parse_locale('sr_RS.UTF-8@latin')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='UTF-8')
>>> parse_locale('ca_ES')
Locale(language='ca', script='', territory='ES', variant='', encoding='')
>>> parse_locale('ca_ES.UTF-8')
Locale(language='ca', script='', territory='ES', variant='', encoding='UTF-8')
>>> parse_locale('ca_ES_VALENCIA')
Locale(language='ca', script='', territory='ES', variant='VALENCIA', encoding='')
>>> parse_locale('ca_Latn_ES_VALENCIA')
Locale(language='ca', script='Latn', territory='ES', variant='VALENCIA', encoding='')
>>> parse_locale('ca_ES.UTF-8@valencia')
Locale(language='ca', script='', territory='ES', variant='VALENCIA', encoding='UTF-8')
>>> parse_locale('ca_ES@valencia')
Locale(language='ca', script='', territory='ES', variant='VALENCIA', encoding='')
>>> parse_locale('en_US_POSIX')
Locale(language='en', script='', territory='US', variant='POSIX', encoding='')
>>> parse_locale('POSIX')
Locale(language='en', script='', territory='US', variant='POSIX', encoding='')
>>> parse_locale('C')
Locale(language='en', script='', territory='US', variant='POSIX', encoding='')
>>> parse_locale('C.UTF-8')
Locale(language='en', script='', territory='US', variant='POSIX', encoding='UTF-8')
� �.�@r rm Nz @valenciaZ _VALENCIA�VALENCIA�C�POSIX� en_US_POSIX�enZUSr r r z&localeId contains invalid locale id=%s)r r r r r )r� r� r� ) �find�lower�upper�_glibc_script_idsrt �endswith�_cldr_locale_pattern�match�groupr� r� r ) r? r r r r r Z dot_indexZat_indexZvalencia_indexrq r� r r r � parse_locale� sh k
r� r� c C s t | �}|jdkr,t|j|j|jd|jd�}|jdkrPtd|j|j|j|jd�}|j r�|r�|tkrlt| }t|j||j|j|jd�}|j r�|r�t|j|j||j|jd�}|jdko�|jo�|j �r|jdkr�t|jd
|j|j|jd�}n&|jdk�rt|jd|j|j|jd�}|S )u�
Parses languageId and if it contains a valid ICU locale id,
returns the values for language, script, and territory found
in languageId instead of the original values given.
Before parsing, it replaces glibc names for scripts like “latin”
with the iso-15924 script names like “Latn”, both in the
languageId and the scriptId parameter. I.e. language id like
“sr_latin_RS” is accepted as well and treated the same as
“sr_Latn_RS”.
Examples:
>>> _parse_and_split_languageId(languageId='de_DE')
Locale(language='de', script='', territory='DE', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='de_DE.UTF-8')
Locale(language='de', script='', territory='DE', variant='', encoding='UTF-8')
>>> _parse_and_split_languageId(languageId='de_DE.utf8')
Locale(language='de', script='', territory='DE', variant='', encoding='utf8')
>>> _parse_and_split_languageId(languageId='de_DE@euro')
Locale(language='de', script='', territory='DE', variant='EURO', encoding='')
>>> _parse_and_split_languageId(languageId='de_DE.ISO-8859-15')
Locale(language='de', script='', territory='DE', variant='', encoding='ISO-8859-15')
>>> _parse_and_split_languageId(languageId='de_DE.ISO-8859-15@euro')
Locale(language='de', script='', territory='DE', variant='EURO', encoding='ISO-8859-15')
>>> _parse_and_split_languageId(languageId='de_DE.iso885915@euro')
Locale(language='de', script='', territory='DE', variant='EURO', encoding='iso885915')
>>> _parse_and_split_languageId(languageId='gez_ER.UTF-8@abegede')
Locale(language='gez', script='', territory='ER', variant='ABEGEDE', encoding='UTF-8')
>>> _parse_and_split_languageId(languageId='ar_ER.UTF-8@saaho')
Locale(language='ar', script='', territory='ER', variant='SAAHO', encoding='UTF-8')
>>> _parse_and_split_languageId(languageId='zh_Hant_TW')
Locale(language='zh', script='Hant', territory='TW', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='zh_TW')
Locale(language='zh', script='Hant', territory='TW', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='zh_Hans_CN')
Locale(language='zh', script='Hans', territory='CN', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='zh_CN')
Locale(language='zh', script='Hans', territory='CN', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='es_419')
Locale(language='es', script='', territory='419', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='sr_latin_RS')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='sr_Latn_RS')
Locale(language='sr', script='Latn', territory='RS', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='ca_ES')
Locale(language='ca', script='', territory='ES', variant='', encoding='')
>>> _parse_and_split_languageId(languageId='ca_ES.UTF-8')
Locale(language='ca', script='', territory='ES', variant='', encoding='UTF-8')
>>> _parse_and_split_languageId(languageId='ca_ES_VALENCIA')
Locale(language='ca_ES_VALENCIA', script='', territory='ES', variant='VALENCIA', encoding='')
>>> _parse_and_split_languageId(languageId='ca_Latn_ES_VALENCIA')
Locale(language='ca_ES_VALENCIA', script='Latn', territory='ES', variant='VALENCIA', encoding='')
>>> _parse_and_split_languageId(languageId='ca_Latn_ES_valencia')
Locale(language='ca_ES_VALENCIA', script='Latn', territory='ES', variant='VALENCIA', encoding='')
>>> _parse_and_split_languageId(languageId='ca_ES.UTF-8@valencia')
Locale(language='ca_ES_VALENCIA', script='', territory='ES', variant='VALENCIA', encoding='UTF-8')
>>> _parse_and_split_languageId(languageId='ca_ES@valencia')
Locale(language='ca_ES_VALENCIA', script='', territory='ES', variant='VALENCIA', encoding='')
>>> _parse_and_split_languageId(languageId='ca_Latn_ES@valencia')
Locale(language='ca_ES_VALENCIA', script='Latn', territory='ES', variant='VALENCIA', encoding='')
>>> _parse_and_split_languageId(languageId='ca_Latn_ES@VALENCIA')
Locale(language='ca_ES_VALENCIA', script='Latn', territory='ES', variant='VALENCIA', encoding='')
r� r� )r r r r r r� Zca_ES_VALENCIA�zh�cmn�CN�SGZHans�HK�MO�TWZHant)r� r� )r� r� )r� r� r� )r� r r r r r r r� )r= r> r< rJ r r r �_parse_and_split_languageIds sR Y
r� c C s t |||d�}|j}|j}|j}| tk�r |rf|rf|rf|d | d | }|t| jkrft| j| S |r�|r�|d | }|t| jkr�t| j| S |r�|r�|d | }|t| jkr�t| j| S |r�|}|t| jkr�t| j| S |o�dt| jk�r t| jd S dS )u* Query translations of territory names
:param territoryId: identifier for the territory
:type territoryId: string
:param languageIdQuery: identifier for the language used in the result
:type languageIdQuery: string
:param scriptIdQuery: identifier for the script used in the result
:type scriptIdQuery: string
:param territoryIdQuery: identifier for the territory used in the result
:type territoryIdQuery: string
:param fallback: Whether a fallback to English should be returned if the
name cannot be found in the requested language.
:type fallback: Boolean
:rtype: string
**Examples:**
Switzerland is called “Schweiz” in German:
>>> print(territory_name(territoryId="CH", languageIdQuery="de"))
Schweiz
And it is called “Svizzera” in Italian:
>>> print(territory_name(territoryId="CH", languageIdQuery="it"))
Svizzera
And it is called “スイス” in Japanese:
>>> print(territory_name(territoryId="CH", languageIdQuery="ja"))
スイス
)r= r> r< �_r� r� )r� r r r rO r
)r< �languageIdQuery�
scriptIdQuery�territoryIdQuery�fallbackrJ �icuLocaleIdQueryr r r �territory_name� s4 !
r� c
C s" | sdS d}t | ||d�}|j} |j}|j}t |||d�} | j}| j}| j}|s\| }|}|}| of|of|�rP| d | d | }|tk�rP|r�|r�|r�|d | d | }
|
t| jkr�t| j|
S |r�|r�|d | }
|
t| jkr�t| j|
S |o�|�r(|d | }
|
t| jk�r(t| j|
S |�rP|}
|
t| jk�rPt| j|
S | �r�|�r�| d | }|tk�r�t||||d�} |�r�|�r�|�r�|d | d | }
|
t| jk�r�t| j|
} | �r�| d | d S | S |�r0|�r0|d | }
|
t| jk�r0t| j|
} | �r,| d | d S | S |�r�|�r�|d | }
|
t| jk�r�t| j|
} | �r|| d | d S | S |�r�|}
|
t| jk�r�t| j|
} | �r�| d | d S | S | �r|�r| d | }|tk�r�|�r(|�r(|�r(|d | d | }
|
t| jk�r(t| j|
S |�r^|�r^|d | }
|
t| jk�r^t| j|
S |�r�|�r�|d | }
|
t| jk�r�t| j|
S |�r�|}
|
t| jk�r�t| j|
S |jd
k�rt| |||d�} t||||d�} | �r| �r| d | d S | }|tk�r�|�rV|�rV|�rV|d | d | }
|
t| jk�rVt| j|
S |�r�|�r�|d | }
|
t| jk�r�t| j|
S |�r�|�r�|d | }
|
t| jk�r�t| j|
S |�r�|}
|
t| jk�r�t| j|
S |�r |�r |tk�r d t| jk�r t| jd S dS ) u� Query translations of language names
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:param languageIdQuery: identifier for the language used in the result
:type languageIdQuery: string
:param scriptIdQuery: identifier for the script used in the result
:type scriptIdQuery: string
:param territoryIdQuery: identifier for the territory used in the result
:type territoryIdQuery: string
:param fallback: Whether a fallback to English should be returned if the
name cannot be found in the requested language.
:type fallback: Boolean
:rtype: string
**Examples:**
>>> print(language_name(languageId="sr"))
српски
I.e. the endonym for “Serbian” in the default Cyrillic script is
“српски”.
If the script “Cyrl” is supplied as well, the name of the
script is added for clarity:
>>> print(language_name(languageId="sr", scriptId="Cyrl"))
српски (Ћирилица)
And in Latin script the endonym is:
>>> print(language_name(languageId="sr", scriptId="Latn"))
srpski (Latinica)
And “Serbian” translated to English is:
>>> print(language_name(languageId="sr", languageIdQuery="en"))
Serbian
And with adding the script information:
>>> print(language_name(languageId="sr", scriptId="Cyrl", languageIdQuery="en"))
Serbian (Cyrillic)
>>> print(language_name(languageId="sr", scriptId="Latn", languageIdQuery="en"))
Serbian (Latin)
>>> print(language_name(languageId="de_DE", languageIdQuery="en"))
German (Germany)
>>> print(language_name(languageId="es_419", languageIdQuery="en"))
Spanish (Latin America)
>>> print(language_name(languageId="ca_ES"))
català (Espanya)
>>> print(language_name(languageId="ca_ES.UTF-8"))
català (Espanya)
>>> print(language_name(languageId="ca_ES@valencia"))
valencià (Espanya)
>>> print(language_name(languageId="ca_ES.utf8@valencia"))
valencià (Espanya)
>>> print(language_name(languageId="ca_ES.utf8@valencia"))
valencià (Espanya)
>>> print(language_name(languageId="ca_ES.utf8@valencia", languageIdQuery='de'))
Valencianisch (Spanien)
>>> print(language_name(languageId="ca_ES.utf8@valencia", languageIdQuery='en'))
Valencian (Spain)
r� )r= r> r< r� )r< r� r� r� z (�)r� )r= r� r� r� r� )r� ) r� r r r rc r
r� r �
language_name)
r= r> r< r� r� r� r� � icuLocaleIdrJ Z localeQueryr� ZcnameZlnamer r r r� 4 s� O
r� c C s� | o|s dS | j d�}g }xj|D ]b}|tkr8|j|� q |t| jkrdt| j| }|r�|j|� q |dkr |jdd�}|j|� q W t|�t|�kr�dj|�S dS )z�Query translation of timezone IDs by querying translations
for each part of the ID seperately and putting the results together
r� �/r� r� � )�splitrk r� r
rt r� r� )rC r� ZtimezoneId_partsZ
part_namesZtimezoneId_partrG r r r � _timezone_name_from_id_parts s$
r� c C sH | o|s dS | t kr0|t | jkr0t | j| S t| |d�}|rD|S dS )z<
Internal helper function to translate timezone IDs
r� )rC r� )rf r
r� )rC r� Zname_from_partsr r r �_timezone_name s
r� c C s� t |||d�}|j}|j}|j}|rP|rP|rPt| |d | d | d�}|rP|S |rt|rtt| |d | d�}|rt|S |r�|r�t| |d | d�}|r�|S |r�t| |d�}|r�|S | S )u Query translations of timezone IDs
:param timezoneId: identifier for the time zone
:type timezoneId: string
:param languageIdQuery: identifier for the language used in the result
:type languageIdQuery: string
:param scriptIdQuery: identifier for the script used in the result
:type scriptIdQuery: string
:param territoryIdQuery: identifier for the territory used in the result
:type territoryId: string
:rtype: string
**Examples:**
>>> print(timezone_name(timezoneId='US/Pacific', languageIdQuery='ja'))
アメリカ合衆国/太平洋時間
If no translation can be found, the timezone ID is returned
unchanged:
>>> print(timezone_name(timezoneId='Pacific/Pago_Pago', languageIdQuery='xxx'))
Pacific/Pago_Pago
)r= r> r< r� )rC r� )r� r r r r� )rC r� r� r� rJ rG r r r �
timezone_name, s> r� c C s^ | sdS t | �t d�kr"| jd�} x6tD ].}x(t| jD ]}| t| j| kr8|S q8W q(W dS )u� Query the territoryId from a translated name of a territory.
:param territoryName: the translated name of a language
:type territoryName: string
:rtype: string
The translated name given should be a Python Unicode string or an
UTF-8 encoded string.
The translated name can be in any language. But there will be only
a result if the translation matches exactly.
**Examples:**
>>> territoryId("India")
'IN'
>>> territoryId("भारत")
'IN'
>>> territoryId("インド")
'IN'
>>> territoryId("Latin America")
'419'
>>> territoryId("Latinoamérica")
'419'
r� zUTF-8)�type�decoderO r
)Z
territoryNamer< r� r r r r< d s
r< c C s" | sdS t | �t d�kr"| jd�} x>tD ]6}x0t| jD ]"}| j� t| j| j� kr8|S q8W q(W tjdtjtjB �}|j | �}|�r |j
d�}|j
d�}x�tD ]z}xtt| jD ]f}|j� t| j| j� kr�xFt D ]>}x8t | jD ]*}|j� t | j| j� kr�|d | S q�W q�W q�W q�W dS )u) Query the languageId from a translated name of a language.
:param languageName: the translated name of a language
:type languageName: string
:rtype: string
The translated name given should be a Python Unicode string or an
UTF-8 encoded string.
The translated name can be in any language. But there will be only
a result if the translation matches exactly.
**Examples:**
>>> languageId("Marathi")
'mr'
>>> languageId("मराठी")
'mr'
>>> languageId("マラーティー語")
'mr'
r� zUTF-8z?^(?P<language_name>[^()]+)[\s]+[(](?P<territory_name>[^()]+)[)]r� r� r� ) r� r� rc r
r� �re�compile� MULTILINE�UNICODE�searchr� rO ) Z languageNamer= r� Zlanguage_territory_patternr� r� r� r< ZicuLocaleId_territoryr r r r= � s.
r= i@B Fc C s i }d}t |||d�}|j}|j}|j}|rf|rf|rf|d | d | tkrf|d | d | }d}nN|r�|r�|d | tkr�|d | }n(|r�|r�|d | tkr�|d | }d}d}|tk�r@x|t| jD ]n}t| j| dkr�||k�rt| j| ||< n*|| t| j| 9 < || t9 < || |9 < q�W d} |tk�r�| �r�x�t| jD ]r}t| j| dk�rb||k�r�t| j| ||< n*|| t| j| 9 < || t9 < || | 9 < �qbW t|�}
| �r�t |
�}
|�r�|
S t
|
�S dS ) u� List suitable glibc locales
:param concise: if True, return only to highly ranked results
:type concise: boolean
:param show_weights: Also return the weights used in the ranking
:type show_weights: boolean
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:rtype: a list of strings
**Examples:**
List the suitable locales for the language “German”:
>>> list_locales(languageId="de")
['de_DE.UTF-8', 'de_AT.UTF-8', 'de_CH.UTF-8', 'de_IT.UTF-8', 'de_LI.UTF-8', 'de_BE.UTF-8', 'de_LU.UTF-8']
So this returns a list of locales for German. These lists are
sorted in order of decreasing likelyhood, i.e. the most common
value comes first.
One can also list the possible locales for the territory “Switzerland”:
>>> list_locales(territoryId="CH")
['de_CH.UTF-8', 'fr_CH.UTF-8', 'it_CH.UTF-8', 'wae_CH.UTF-8']
If one knows both, the language “German” and the territory
“Switzerland”, the result is unique:
>>> list_locales(languageId="de", territoryId="CH")
['de_CH.UTF-8']
F)r= r> r< r� T�d r rm N) r� r r r rc r � extra_bonusrO r� r� r� ) �concise� show_weightsr= r> r< Zranked_locales�
skipTerritoryrJ �language_bonus�territory_bonusr� r r r � list_locales� sP '$
r� c C s� i }d}t |||d�}|j}|j}|j}|r2|gS |rZ|rZ|d | tkrZ|d | }d}d}|tkr�xzt| jD ]l} t| j| dkrr| |kr�t| j| || < n*|| t| j| 9 < || t9 < || |9 < qrW d}
|tko�| �rxx�t| jD ]r} t| j| dk�r| |k�r8t| j| || < n*|| t| j| 9 < || t9 < || |
9 < �qW t|�} | �r�t | �} |�r�| S t
| �S dS ) u_ List scripts used for a language and/or in a territory
:param concise: if True, return only to highly ranked results
:type concise: boolean
:param show_weights: Also return the weights used in the ranking
:type show_weights: boolean
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:rtype: a list of strings
Returns a list of ISO-15924 script ids:
https://en.wikipedia.org/wiki/ISO_15924
**Examples:**
List the suitable scripts for the language “Serbian”:
>>> list_scripts(languageId="sr")
['Cyrl', 'Latn']
So this returns a list of scripts which are in use for
Serbian. These lists are sorted in order of decreasing likelyhood,
i.e. the most common value comes first.
List the suitable scripts for the language “Punjabi”:
>>> list_scripts(languageId="pa")
['Guru', 'Arab']
One can also list the possible scripts for a territory like
“Pakistan”:
>>> list_scripts(territoryId="PK")
['Arab']
If one knows both, the language “Punjabi” and the territory
“Pakistan” or “India”, one can find out which script is the
preferred one:
>>> list_scripts(languageId="pa", territoryId="PK")
['Arab']
So the preferred script for Punjabi in Pakistan is “Arab”
>>> list_scripts(languageId="pa", territoryId="IN")
['Guru', 'Arab']
and the preferred script for Punjabi in India is “Guru”.
F)r= r> r< r� Tr� r rm N) r� r r r rc r r� rO r� r� r� ) r� r� r= r> r< Zranked_scriptsr� rJ r� r r� r� r r r � list_scripts sJ 8
r� c C s i }d}t |||d�}|j}|j}|j}|rf|rf|rf|d | d | tkrf|d | d | }d}nR|r�|r�|d | tkr�|d | }d}n(|r�|r�|d | tkr�|d | }d}d}|tk�rDx|t| jD ]n} t| j| dkr�| |k�rt| j| || < n*|| t| j| 9 < || t9 < || |9 < q�W d}
|tk�r�| �r�x�t| jD ]r} t| j| dk�rf| |k�r�t| j| || < n*|| t| j| 9 < || t9 < || |
9 < �qfW t|�} | �r�t | �} |�r�| S t
| �S dS ) u� List suitable input methods
:param concise: if True, return only to highly ranked results
:type concise: boolean
:param show_weights: Also return the weights used in the ranking
:type show_weights: boolean
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:rtype: a list of strings
**Examples:**
List the suitable input methods for the language “Japanese”:
>>> list_inputmethods(languageId="ja")
['ibus/kkc', 'ibus/anthy']
So this returns a list of input methods for Japanese. These lists are
sorted in order of decreasing likelyhood, i.e. the most common
value comes first.
One can also list the possible input methods for the territory “Japan”:
>>> list_inputmethods(territoryId="JP")
['ibus/kkc', 'ibus/anthy']
F)r= r> r< r� Tr� r rm N) r� r r r rc r r� rO r� r� r� ) r� r� r= r> r< Zranked_inputmethodsr� rJ r� rL r� r� r r r �list_inputmethodsv sR $
r� c C s� i }d}t |||d�}|j}|j}|j}|rf|rf|rf|d | d | tkrf|d | d | }d}nN|r�|r�|d | tkr�|d | }n(|r�|r�|d | tkr�|d | }d}d}|tk�r@x|t| jD ]n} t| j| dkr�| |k�rt| j| || < n*|| t| j| 9 < || t9 < || |9 < q�W d}
|tk�r�x�t| jD ]r} t| j| dk�rZ| |k�r�t| j| || < n*|| t| j| 9 < || t9 < || |
9 < �qZW t|�} | �r�t | �} |�r�| S t
| �S dS )u� List likely X11 keyboard layouts
:param concise: if True, return only to highly ranked results
:type concise: boolean
:param show_weights: Also return the weights used in the ranking
:type show_weights: boolean
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:rtype: a list of strings
**Examples:**
Listing likely X11 keyboard layouts for “German”:
>>> list_keyboards(languageId="de")
['de(nodeadkeys)', 'de(deadacute)', 'at(nodeadkeys)', 'ch', 'be(oss)']
Listing likely X11 keyboard layouts for “Switzerland”:
>>> list_keyboards(territoryId="CH")
['ch', 'ch(fr)', 'it']
When specifying both “German” *and* “Switzerland”, the
returned X11 keyboard layout is unique:
>>> list_keyboards(languageId="de", territoryId="CH")
['ch']
F)r= r> r< r� Trm r N) r� r r r rc r r� rO r� r� r� ) r� r� r= r> r< Zranked_keyboardsr� rJ r� rK r� r� r r r �list_keyboards� sP !$
r� c C s� i }d}t |||d�}|j}|j}|j}|rf|rf|rf|d | d | tkrf|d | d | }d}nN|r�|r�|d | tkr�|d | }n(|r�|r�|d | tkr�|d | }d}d}|tk�r@x|t| jD ]n} t| j| dkr�| |k�rt| j| || < n*|| t| j| 9 < || t9 < || |9 < q�W d}
|tk�r�x�t| jD ]r} t| j| dk�rZ| |k�r�t| j| || < n*|| t| j| 9 < || t9 < || |
9 < �qZW t|�} | �r�t | �} |�r�| S t
| �S dS ) a; List likely Linux Console fonts
:param concise: if True, return only to highly ranked results
:type concise: boolean
:param show_weights: Also return the weights used in the ranking
:type show_weights: boolean
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:rtype: a list of strings
**Examples:**
Listing likely console fonts for English:
>>> list_consolefonts(languageId="en")
['eurlatgr']
Listing likely console fonts for Greek:
>>> list_consolefonts(languageId="el")
['eurlatgr', 'iso07u-16', 'LatGrkCyr-8x16']
Listing likely console fonts for Greece:
>>> list_consolefonts(territoryId="GR")
['eurlatgr', 'iso07u-16', 'LatGrkCyr-8x16']
Listing likely console fonts for Greek in Greece:
list_consolefonts(languageId="el", territoryId="GR")
['eurlatgr']
Listing likely console fonts for Greek in a non-Greek country like
the UK (the language has higher weight):
>>> list_consolefonts(languageId="el", territoryId="GB")
['eurlatgr']
Listing likely console fonts for Russian in Russia:
>>> list_consolefonts(languageId="ru", territoryId="RU")
['latarcyrheb-sun16']
Listing likely console fonts for Russian in a non-Russian country like
the UK (the language has higher weight):
>>> list_consolefonts(languageId="ru", territoryId="GB")
['latarcyrheb-sun16', 'eurlatgr']
F)r= r> r< r� Tr� r rm N) r� r r r rc r r� rO r� r� r� ) r� r� r= r> r< Zranked_consolefontsr� rJ r� rM r� r� r r r �list_consolefonts sP 7$
r� c C s� i }d}t |||d�}|j}|j}|j}|rf|rf|rf|d | d | tkrf|d | d | }d}nN|r�|r�|d | tkr�|d | }n(|r�|r�|d | tkr�|d | }d}d}|tk�r@x|t| jD ]n} t| j| dkr�| |k�rt| j| || < n*|| t| j| 9 < || t9 < || |9 < q�W d}
|tk�r�x�t| jD ]r} t| j| dk�rZ| |k�r�t| j| || < n*|| t| j| 9 < || t9 < || |
9 < �qZW t|�} | �r�t | �} |�r�| S t
| �S dS ) av List likely timezones
:param concise: if True, return only to highly ranked results
:type concise: boolean
:param show_weights: Also return the weights used in the ranking
:type show_weights: boolean
:param languageId: identifier for the language
:type languageId: string
:param scriptId: identifier for the script
:type scriptId: string
:param territoryId: identifier for the territory
:type territoryId: string
:rtype: a list of strings
**Examples:**
>>> list_timezones(territoryId="DE")
['Europe/Berlin']
>>> list_timezones(languageId="de")
['Europe/Berlin', 'Europe/Vienna', 'Europe/Zurich', 'Europe/Brussels', 'Europe/Luxembourg']
>>> list_timezones(territoryId="CH")
['Europe/Zurich']
>>> list_timezones(languageId="fr", territoryId="CH")
['Europe/Zurich']
>>> list_timezones(languageId="fr")
['Europe/Paris', 'America/Montreal', 'Europe/Brussels', 'Europe/Zurich', 'Europe/Luxembourg']
The territory gets more weight than the language:
>>> list_timezones(languageId="ja", territoryId="CH")
['Europe/Zurich', 'Asia/Tokyo']
F)r= r> r< r� Trm r r� N) r� r r r rc r r� rO r� r� r� ) r� r� r= r> r< Zranked_timezonesr� rJ r� rN r� r� r r r �list_timezonesr sP %$
r� c C s | t krt | jS dS )a� Check whether a keyboard layout supports ASCII
:param keyboardId: identifier for the keyboard
:type keyboardId: string
:rtype: string
Returns True if the keyboard layout with that id can be used to
type ASCII, returns false if the keyboard layout can not be used
to type ASCII or if typing ASCII with that keyboard layout is
difficult.
**Examples:**
>>> supports_ascii("jp")
True
>>> supports_ascii("ru")
False
T)r\ r# )r@ r r r �supports_ascii� s
r� c C s ddl } | jd�d jS )z%
Return version of langtable
r N� langtable)�
pkg_resources�require�version)r� r r r r� � s r� c C s� ddl } | jd�d j}| jd�d j}| jd�d j}td| � td| � td| � tdtjjt � � tdt
d � dS )
z)
Print some info about langtable
r Nr� zProject name: = %sz
Version: = %szModule path: = %szLoaded from: %szData files read: %sr ) r� r� � project_namer� � module_path�printr� r� r� r� r� )r� r� r� r� r r r r� � s r� c C s> t j| �}|r6d|jd�fd|jd�fd|jd�fgS g S dS )z2
Internal test function, do not use this.
r r r N)r� r� r� )r? r� r r r �_test_cldr_locale_pattern� s
(r� c C s� t t|�d tt| |d�� d t|� d tt| |d�� d d tt| |||d�� d t|� d tt| |d�� d t|� d tt| |d�� d d tt| |||d�� � dS )z2
Internal test function, do not use this.
z: )r� r= �
)r� r< z +: )r� r= r> r< N)r� rP �reprr� r� )r� r= r> r< r r r �_test_language_territory� s �r� � /dev/nullc C sb t j}| rt j}t j|dd|d� tdt� � tdt� � tdt� � tdt� � tdt � � d S ) Nr� z%(levelname)s: %(message)s)r� �filemode�format�levelzterritories.xmlz
languages.xmlz
keyboards.xmlz
timezones.xmlztimezoneidparts.xml)
r� �INFO�DEBUGZ basicConfigr� r- r] rU rd rg )�debugZ logfilenameZ log_levelr r r �_init s r� c @ s e Zd Zdd� Zdd� ZdS )�__ModuleInitializerc C s
t � d S )N)r� )r r r r r " s z __ModuleInitializer.__init__c C s d S )Nr )r r r r �__del__& s z__ModuleInitializer.__del__N)r r r r r� r r r r r� ! s r� �__main__z ^(?P<language>[a-z]{2,3}(?=$|@zJ^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))z`^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)zb^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))z^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})z�^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})(?=$|@z�^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})(?=$|@|_[A-Z0-9]{2,3}(?=$|@)z�^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})(?=$|@|_[A-Z0-9]{2,3}(?=$|@))){0,1}z�^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})(?=$|@|_[A-Z0-9]{2,3}(?=$|@))){0,1}(?:_(?P<territory>[A-Z0-9]{2,3})z�^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})(?=$|@|_[A-Z0-9]{2,3}(?=$|@))){0,1}(?:_(?P<territory>[A-Z0-9]{2,3})(?=$|@z�^(?P<language>[a-z]{2,3}(?=$|@|_[A-Z][a-z]{3}(?=$|@|_[A-Z0-9]{2,3}(?=$|@))|_[A-Z0-9]{2,3}(?=$|@)))(?:_(?P<script>[A-Z][a-z]{3})(?=$|@|_[A-Z0-9]{2,3}(?=$|@))){0,1}(?:_(?P<territory>[A-Z0-9]{2,3})(?=$|@)){0,1})T)r� )r� r� r� )NNNNT)NNNNNNT)NN)NN)NNNN)r� )r� )TFNNN)TFNNN)TFNNN)TFNNN)TFNNN)TFNNN)N)FNNN)Fr� )Gr� r� r� r� � collectionsZxml.parsers.expatr| Zxml.sax.handlerr �
namedtupler r� r� r� r� r� rO rc r\ rf rk r r r! r% r& �objectr' r- rU r] rd rg rw rx ry rz 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� r� r� Z
__module_initr ZdoctestZtestmodr r r r �<module>a s� jJ 55UX#
&
<
S
8
)
2
S
b
L
M
c
Q