Dekker1/Fourmi
Archived
1
0
Fork 0
Code

added code to recognize various table formats

Browse Source
This commit is contained in:
RTB 2014-05-07 17:09:42 +02:00
parent 9c80f291b6
commit 95e24f9c44
1 changed files with 22 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
FourmiCrawler/sources/NIST.py
View File

@ -25,6 +25,28 @@ class NIST(Source):
log.msg('NIST symbol: |%s|, name: |%s|' % (symbol, name),
level=log.DEBUG)
for tables in sel.xpath('//table[@class="data"]'):
if tables.xpath('@summary').extract()[0] == 'One dimensional data':
log.msg('NIST table: Aggregrate data', level=log.DEBUG)
elif tables.xpath('tr/th="Initial Phase"').extract()[0] == '1':
log.msg('NIST table; Enthalpy/entropy of phase transition',
level=log.DEBUG)
elif tables.xpath('tr[1]/td'):
log.msg('NIST table: Horizontal table', level=log.DEBUG)
elif (tables.xpath('@summary').extract()[0] ==
'Antoine Equation Parameters'):
log.msg('NIST table: Antoine Equation Parameters',
level=log.DEBUG)
elif len(tables.xpath('tr[1]/th')) == 5:
log.msg('NIST table: generic 5 columns', level=log.DEBUG)
# Symbol (unit) Temperature (K) Method Reference Comment
elif len(tables.xpath('tr[1]/th')) == 4:
log.msg('NIST table: generic 4 columns', level=log.DEBUG)
# Symbol (unit) Temperature (K) Reference Comment
else:
log.msg('NIST table: NOT SUPPORTED', level=log.WARNING)
continue #Assume unsupported
def new_compound_request(self, compound):
return Request(url=self.website[:-1] + self.search % compound,
callback=self.parse)