mò
v
Lc @ s d Z d Z d Z d Z d k Z d k Z d k l Z l Z d k Z d f d „ ƒ YZ d „ Z
e
d ƒ Z e
d
ƒ Z e
d ƒ Z
e
d ƒ Z e
d
ƒ Z e
d ƒ Z d „ Z d „ Z d d „ Z d „ Z d d „ Z d „ Z d „ Z g d d „ Z g d d „ Z e i d ƒ Z d „ Z d „ Z d „ Z d d „ Z e i! d ƒ Z" d „ Z# e$ d j o– d k% l& Z& d! k' l( Z( xy e& d" D]i Z) e( e) ƒ oV e* e) d# ƒ Z+ e+ i, d$ ƒ Z- g Z. e e- e. ƒ Z/ e/ o
e) Ge/ GHqöe0 e. ƒ GHqqWn d S(% sa
$Id$
This module deals with detecting XML encodings, using both BOMs and
explicit declarations.
s' Joseph Walton s
$Revision$s Copyright (c) 2004 Joseph WaltonN( s ObscureEncodings NonstdEncodingt FailingCodecc B s t Z d „ Z d d „ Z RS( Nc C s
| | _ d S( N( t namet self( R R ( ( tD /web/script/rubys/feedvalidator.org/src/feedvalidator/xmlEncoding.pyt __init__ s t strictc C s t d | i d ƒ ‚ d S( Ns No codec available for s& in this installation of FeedValidator( t UnicodeErrorR R ( R t txtt errors( ( R t fail s ( t __name__t
__module__R R ( ( ( R R s c C s, y t i | ƒ SWn t | ƒ i Sn Xd S( N( t codecst
getdecodert codecR R ( R ( ( R R
s s UTF-32BEs UTF-32LEs UTF-16BEs UTF-16LEt IBM037s
ISO-8859-1c C s‘ | i d d | ƒ d } | i d d | ƒ d } d } xB | | | !D]3 } | d j o | d d d } qF | d 7} qF W| d 7} | | f S( Ns
i i s i ( R t countt idxt rowt rfindt lnt columnt c( R R R R R R ( ( R t _position+ s
c C s | i d d ƒ i d d ƒ S( Ns
s
s
( R t replace( R ( ( R t _normaliseNewlines7 s c C s6 | o | \ | i d <| i d
(as of 2004-03-27)?s US-ASCIIs
ISO-8859-1s EUC-JPs
ISO-8859-2s ISO-8859-15s
ISO-8859-7s KOI8-Rt SHIFT_JISs WINDOWS-1250s WINDOWS-1251s WINDOWS-1252s WINDOWS-1254s WINDOWS-1255s WINDOWS-1256t GB18030N( R1 R0 R( ( R0 ( ( R t isCommone s s UTF-8c C s¡ | d } | d j o) t | d t d d d d g | ƒ } n^| d j o) t | d t d d d d g | ƒ } n(| d j p
| d j o t d
ƒ ‚ nþ| d d j o/ t | d t d
d d d g | d d
ƒ} n¾| d d j o/ t | d t
d
d d d g | d d
ƒ} n~| d d j o& t | d t d g | d d ƒ} nG| d j o( t | t d d d d d g | ƒ } n| d j o( t | t d d d d d g | ƒ } nÝ | d j o( t | t d d
d d d g | ƒ } n¨ | d j o( t | t
d d
d d d g | ƒ } ns | d j o' t t
t | ƒ d ƒ ƒ p d& } n? | d j o% t | t d! d" d# d$ g | ƒ } n
| d% f } | S(' sÈ This is the logic from appendix F.1 of the XML 1.0 specification.
Pass in the start of a document (>= 256 octets), and receive the encoding to
use, or None if there is a problem with the document.i t þÿs UTF-32s ISO-10646-UCS-4t CSUCS4s UCS-4s ÿþ t ÿþs þÿ s3 Unable to process UCS-4 with unusual octet orderingi s þÿs UTF-16s ISO-10646-UCS-2t CSUNICODEs UCS-2R. s ÿþi s s UTF-8t R? R@ RA RC RB R R R- R R, R/ R1 R4 RD RE t compileRF R$ RO R) R[ t
getencoderRb Rc R
t syst argvt os.patht isfileR0 t opent ft readt lt logR% t repr(' R% Rb R$ R1 Rp R RF RE RO R R, R[ Rj RU RC RT R Ru R
Rg R4 Rn R R/ Rc R RB Ri Rh R& R? Rr RD R> Rt R) R@ R0 RA ( ( R t ? sX
)
E