TXT Output: Table support

This commit is contained in:
Kovid Goyal 2009-09-27 12:00:39 -06:00
parent 0428cf9bd6
commit a49fe89309

View File

@ -26,12 +26,18 @@ BLOCK_TAGS = [
'h5', 'h5',
'h6', 'h6',
'li', 'li',
'tr',
] ]
BLOCK_STYLES = [ BLOCK_STYLES = [
'block', 'block',
] ]
SPACE_TAGS = [
'span',
'td',
]
class TXTMLizer(object): class TXTMLizer(object):
def __init__(self, log): def __init__(self, log):
@ -170,6 +176,10 @@ class TXTMLizer(object):
if not end.endswith(u'\n\n') and hasattr(elem, 'text') and elem.text != None and elem.text.strip() != '': if not end.endswith(u'\n\n') and hasattr(elem, 'text') and elem.text != None and elem.text.strip() != '':
text.append(u'\n\n') text.append(u'\n\n')
if tag in SPACE_TAGS:
if not end.endswith('u ') and hasattr(elem, 'text') and elem.text != None and elem.text.strip() != '':
text.append(u' ')
# Process tags that contain text. # Process tags that contain text.
if hasattr(elem, 'text') and elem.text != None and elem.text.strip() != '': if hasattr(elem, 'text') and elem.text != None and elem.text.strip() != '':
text.append(elem.text) text.append(elem.text)