def fill_in_the_blanks(): req=urlopen('http://web.mit.edu/puzzle/www/2013/coinheist.com/get_smart/fill_in_the_blanks/index.html') data=req.read().decode() req.close() tables = data.split('</table>') grids=[parse_html_table(t,styleattr='class',styledict={'b':0}) for t in tables] to_openpyxl_multi(grids[:-1]).save('fill_in_the_blanks.xlsx')
def read_html(data,args): try: data=data.decode() except UnicodeDecodeError: raise NotRecognized if '<div class="bk"' in data: return parse_html_grid(data) elif '<table' in data: if not args.color_attribute: raise NotRecognized('HTML contains a table, but --color-attribute not specified. Try specifying --color-attribute and --color-value-dark.') return parse_html_table(data,styleattr=args.color_attribute,styledict={args.color_value_dark:0}) else: raise NotRecognized
def read_html(data, args): try: data = data.decode() except UnicodeDecodeError: raise NotRecognized if '<div class="bk"' in data: return parse_html_grid(data) elif '<table' in data: if not args.color_attribute: raise NotRecognized( 'HTML contains a table, but --color-attribute not specified. Try specifying --color-attribute and --color-value-dark.' ) return parse_html_table(data, styleattr=args.color_attribute, styledict={args.color_value_dark: 0}) else: raise NotRecognized