IR-edZddlZddlZddlZddlmZgdZejdZddZ e Z dd l m Z e j Z n #e$rYnwxYwejdd Zd Zd ZdS)z8 This module includes a fast iterator-based XML parser. N)data)get_xml_iteratorget_xml_encoding xml_readlinesc#4Kt|r|VdStj|d5}tjdr |jVn(t|tj r|Vn |jVddddS#1swxYwYdS)a Returns a function suitable for streaming input, or a file object. This function is only useful if passing off to C code where: - If it's a real file object, we want to use it as a real C file object to avoid the Python overhead. - If it's not a real file object, it's much handier to just have a Python function to call. This is somewhat quirky behavior, of course, which is why it is private. For a more useful version of similar behavior, see `astropy.utils.misc.get_readable_fileobj`. Parameters ---------- fd : object May be: - a file object. If the file is uncompressed, this raw file object is returned verbatim. Otherwise, the read method is returned. - a function that reads from a stream, in which case it is returned verbatim. - a file path, in which case it is opened. Again, like a file object, if it's uncompressed, a raw file object is returned, otherwise its read method. - an object with a :meth:`read` method, in which case that method is returned. Returns ------- fd : context-dependent See above. Nbinaryencodingwin) callablerget_readable_fileobjsysplatform startswithread isinstanceioFileIO)fdnew_fds .startSsA 4 8&:TU V    GGGcd|djjffdS)NF)rjoinstriprr)r r"r#r$s rendz_fast_iterparse..endYsU  ##%%)6+EF       r&TFr() xml.parsersrr r ParserCreatespecified_attributesStartElementHandlerEndElementHandlerrCharacterDataHandlerParse) r buffersizerrr%r+r2rr"r#r$s @@@r_fast_iterparser4HsW!!!!!! B<<w E D           ! !F"&F!&F"F"&+F LE 4  D   dE !!!HtJ   E"dOOOr&) _iterparserFc#Kt|5}|rt|}nt|}t|VddddS#1swxYwYdS)af Returns an iterator over the elements of an XML file. The iterator doesn't ever build a tree, so it is much more memory and time efficient than the alternative in ``cElementTree``. Parameters ---------- source : path-like, readable file-like, or callable Handle that contains the data or function that reads it. If a function or callable object, it must directly read from a stream. Non-callable objects must define a ``read`` method. Returns ------- parts : iterator The iterator returns 4-tuples (*start*, *tag*, *data*, *pos*): - *start*: when `True` is a start element event, otherwise an end element event. - *tag*: The name of the element - *data*: Depends on the value of *event*: - if *start* == `True`, data is a dictionary of attributes - if *start* == `False`, data is a string containing the text content of the element - *pos*: Tuple (*line*, *col*) indicating the source of the event. N)r_slow_iterparser4iter)source_debug_python_based_parserrcontexts rrrsJ ) 0 0B % *%b))GG%b))G7mm s3AAAct|5}t|\}}}}|r|dkrtd dddn #1swxYwY|jdpdS)a Determine the encoding of an XML file by reading its header. Parameters ---------- source : path-like, readable file-like, or callable Handle that contains the data or function that reads it. If a function or callable object, it must directly read from a stream. Non-callable objects must define a ``read`` method. Returns ------- encoding : str xmlzInvalid XML fileNr zutf-8)rnextOSErrorget)r:iteratorr%tagrposs rrrs & ! !.X $XsD# .u ,-- -%............... 48J   *7*s,A  A A ct|}tj||5}|d|}dddn #1swxYwY|S)a Get the lines from a given XML file. Correctly determines the encoding and always returns unicode. Parameters ---------- source : path-like, readable file-like, or callable Handle that contains the data or function that reads it. If a function or callable object, it must directly read from a stream. Non-callable objects must define a ``read`` method. Returns ------- lines : list of unicode r rN)rrr seek readlines)r:r input xml_liness rrrs  ''H "6H = = =& 1 OO%% &&&&&&&&&&&&&&& s*AA #A )r)F)__doc__ contextlibrr astropy.utilsr__all__contextmanagerrr4r8r(r6 IterParser ImportErrorrrrr&rrRs   C C C 3"3"3"l****^" !,OO   D  ))))X+++0s >AA