nojhan/weboob-devel
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

move select() in parser

Browse source
This commit is contained in:
Romain Bignon 2011-04-08 19:39:57 +02:00
commit 9afb301ebe
30 changed files with 197 additions and 197 deletions
Download patch file Download diff file Expand all files Collapse all files

81
weboob/tools/parsers/lxmlparser.py
View file

@ -21,50 +21,10 @@
import lxml.html
from .iparser import IParser
from ..browser import BrokenPageError
__all__ = ['LxmlHtmlParser', 'select', 'SelectElementException']
class SelectElementException(Exception):
pass
def select(element, selector, nb=None, method='cssselect'):
"""
Select one or many elements from an element, using lxml cssselect by default.
Raises SelectElementException if not found.
@param element [obj] element on which to apply selector
@param selector [str] CSS or XPath expression
@param method [str] (cssselect|xpath)
@param nb [int] number of elements expected to be found.
Use None for undefined number, and 'many' for 1 to infinite.
@return one or many Element
"""
if method == 'cssselect':
results = element.cssselect(selector)
if nb is None:
return results
elif isinstance(nb, basestring) and nb == 'many':
if results is None or len(results) == 0:
raise SelectElementException('Element not found with selector "%s"' % selector)
elif len(results) == 1:
raise SelectElementException('Only one element found with selector "%s"' % selector)
else:
return results
elif isinstance(nb, int) and nb > 0:
if results is None:
raise SelectElementException('Element not found with selector "%s"' % selector)
elif len(results) < nb:
raise SelectElementException('Not enough elements found (%d expected) with selector "%s"' % (nb, selector))
else:
return results[0] if nb == 1 else results
else:
raise Exception('Unhandled value for kwarg "nb": %s' % nb)
else:
raise NotImplementedError('Only cssselect method is implemented for the moment')
__all__ = ['LxmlHtmlParser']
class LxmlHtmlParser(IParser):
@ -83,3 +43,40 @@ class LxmlHtmlParser(IParser):
def tostring(self, element):
return lxml.html.tostring(element, encoding=unicode)
@classmethod
def select(cls, element, selector, nb=None, method='cssselect'):
"""
Select one or many elements from an element, using lxml cssselect by default.
Raises BrokenPageError if not found.
@param element [obj] element on which to apply selector
@param selector [str] CSS or XPath expression
@param method [str] (cssselect|xpath)
@param nb [int] number of elements expected to be found.
Use None for undefined number, and 'many' for 1 to infinite.
@return one or many Element
"""
if method == 'cssselect':
results = element.cssselect(selector)
if nb is None:
return results
elif isinstance(nb, basestring) and nb == 'many':
if results is None or len(results) == 0:
raise BrokenPageError('Element not found with selector "%s"' % selector)
elif len(results) == 1:
raise BrokenPageError('Only one element found with selector "%s"' % selector)
else:
return results
elif isinstance(nb, int) and nb > 0:
if results is None:
raise BrokenPageError('Element not found with selector "%s"' % selector)
elif len(results) < nb:
raise BrokenPageError('Not enough elements found (%d expected) with selector "%s"' % (nb, selector))
else:
return results[0] if nb == 1 else results
else:
raise Exception('Unhandled value for kwarg "nb": %s' % nb)
else:
raise NotImplementedError('Only cssselect method is implemented for the moment')