U
'1eG
@ s d Z dZzBdZddlZdd ZejddddedddZdddi fdd ZW n ek
rb dZY nX G d
d d Z e Z d Z
dZ dZ d
Z
ee
ZdZdddgZdd ZddefddZereZdd ZdZdd ZddefddZd d Zd+d d Zd,d d!Zed"kredd# ddl Z ddl Z dd$l m Z ddl Z e Z dZ!e j"dd D ]fZ#e j$%e#sje&d% nHe'e#Z(e() Z*W 5 Q R X e&d&e#d'ee* ee*dd# e!d7 Z!qNe!re&d(e e d)d* dS )-at Very simple and fast XML parser, used for intra-paragraph text.
Devised by Aaron Watters in the bad old days before Python had fast
parsers available. Constructs the lightest possible in-memory
representation; parses most files we have seen in pure python very
quickly.
The output structure is the same as the one produced by pyRXP,
our validating C-based parser, which was written later. It will
use pyRXP if available.
This is used to parse intra-paragraph markup.
Example parse::
text in xml
( "this",
{"type": "xml"},
[ "text ",
("b", None, ["in"], None),
" xml"
]
None )
{ 0: "this"
"type": "xml"
1: ["text ",
{0: "b", 1:["in"]},
" xml"]
}
Ie, xml tag translates to a tuple:
(name, dictofattributes, contentlist, miscellaneousinfo)
where miscellaneousinfo can be anything, (but defaults to None)
(with the intention of adding, eg, line number information)
special cases: name of "" means "top level, no containing tag".
Top level parse always looks like this::
("", list, None, None)
contained text of None means
In order to support stuff like::
AT THE MOMENT & ETCETERA ARE IGNORED. THEY MUST BE PROCESSED
IN A POST-PROCESSING STEP.
PROLOGUES ARE NOT UNDERSTOOD. OTHER STUFF IS PROBABLY MISSING.
Nc C s t | d S )N)print)s r ;/tmp/pip-unpacked-wheel-109iniqw/reportlab/lib/rparsexml.pywarnCB> s r z string input)ZErrorOnValidityErrorsZNoNoDTDWarningZExpandCharacterEntitiesZExpandGeneralEntitiesr ZsrcNameZ
ReturnUTF8c C s* |t _t j| f|}|r |p(dd |gd fS )N ) pyRXP_parsereoCBparse)ZxmlTextoneOutermostTagr
entityReplacerZ parseOptspr r r parsexmlI s r c @ s e Zd Zedd ZdS ) smartDecodec s0 t d dd l fdd}t|tj_|| S )Ninitialr c s. t | tr| S | }td | |d S )Nfinalencoding)
isinstancestrdetectr decode)r Zcddchardetr r __call__U s
z&smartDecode.__call__..__call__)r r staticmethodr __class__r )r r r r r r Q s
zsmartDecode.__call__N)__name__
__module__ __qualname__r r r r r r r P s r r )z<<)z>>)z&&c C s> g }| D ]0}d|kr.t D ]\}}|||}q|| q|S )Nr# ) replacelistreplaceappend)Z contentListresulteoldnewr r r unEscapeContentListf s r+ c C s4 t r tdt| |d\}}|r,|d d S |S dS )z.official interface: discard unused cursor infoz)pyRXP not found, fallback parser disabled)r
r N) RequirePyRXP ImportError parsexml0)xmltextr r
r
r' cursorr r r parsexmlSimpleo s r2 c C s t | d }t|S )Nr)openreadr )filenamerawr r r parseFile| s r8 c C s d}d}|dkr| d|}|dk r&q|d }d}|D ]L}t|}| ||| |kr6d}| d|}|dk rztd| |d }q6|dkrd}q|S )zBskip any prologue found after cursor, return index of rest of text)z!DOCTYPEz?xmlz!--Nr! r r r" zcan't close prologue %r)findlen
ValueError) textr1 Zprologue_elementsdoneZ openbracketZpastfoundr( ler r r
skip_prologue s&
r@ c 1 C s t | } t}d } }}|dk r(| } |}| d|} | | d | d }
d} | dk r|dk r| |d g}|rz||}||||ft| fS tdt| ||d g } |dk rt }}
t| |}n | dk rtdt| dd n|
d
krp| | | d d krp| d }| t|}|dk rJtd
t| ||d t }| || g}|tt }d} nz|
dkr| | | d dkr| d| d }|| k rtdt| ||d |d }| | dkrtdt| ||d d|d fS | d| }|dk }|d }| d }| || }d|krp|d dkr\|dd }d} | }
|
}|}nzd|krd}|st|d
dd rd}|dkr| d|}|d }|dk }| || }|st|d
dd rd}q|rtdt| | | d |}| |d dkrH|d }|dd }d} | }|
d}|d }|
}|d }
|
}|d }|d d |d< i }} d} t|} | | k r|| } | d } | } | d dkrtd t| d| dd kr.| | krtd t| || } | d } d | | f } q| } |
}!|!d }"| dt|" }#|# }#z|#d }$|#d }%W n" td t|#| |!f Y nX |$|% krdksn |$|% krd!krn n |#dd }#|#| |< |"}q| dk r| }| dk r| d|}&|&|k rp|
tkrLd} | |d }'t| }|'rl| |' n td"|
t| ||d f q| |&d dkrj| d|&}(|(|&k rtd#t| |&|&d | |&d |( })|)
}*|*d }+|
|+kr@| d| },t|,
d$}-| d| },t|,
d$}.td%|.|-t|
t|+t| ||d& f | ||& }'|'r\| |' |(d }d} n@| ||& }'|'r| |' t | |&d|d'\}/}|/r| |/ q|r|r||}||||f}0|0|fS )(zsimple recursive descent xml parser...
return (dictionary, endcharacter)
special case: comment returns (None, endcharacter)Nr! r r zno tags at non-toplevel %s z.non top level entry should be at start tag: %s
z![ r zunclosed CDATA %sz!- z
text in xml ]]>
just testing brackets feature
rr )rt ru r r r test s rv __main__ru rn z!!!!! no file at {f!r}zparsing z |t|=z timed at z.2fz secs.)r )r )+__doc__r- Z simpleparseZpyRXPUr Parserr r r. r rQ ZNAMEKEYZ CONTENTSKEYrU r: ZLENCDATAMARKERrT r$ r+ r2 r8 verboser@ r/ ri rt rv r sysosro Z reportlabrs seenargvfpathisfiler r4 _fr5 rb r r r r sr 7
i