edpibu/edt-parser
Archived
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Fixed multirow cells

Browse source
This commit is contained in:
Edgar P. Burkhart 2020-09-21 07:30:57 +02:00
parent 95c9183bdf
commit 436a334fd0
Signed by: edpibu
GPG key ID: 9833D3C5A25BD227
1 changed files with 9 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
parser.py
View file

@ -43,6 +43,7 @@ class GoogleSheetsCalParser(HTMLParser):
self.column = -1 self.column = -1
self.nextCol = -1 self.nextCol = -1
self.rowspan = 0 self.rowspan = 0
self.multirows = {}
self.date = {} self.date = {}
import priv import priv
@ -64,6 +65,9 @@ class GoogleSheetsCalParser(HTMLParser):
self.row += 1 self.row += 1
self.nextCol = -1 self.nextCol = -1
elif tag == 'td': elif tag == 'td':
while self.nextCol in self.multirows.keys() and self.multirows[self.nextCol][1] > 0:
self.nextCol += self.multirows[self.nextCol][0]
self.column = self.nextCol self.column = self.nextCol
if 'colspan' in dAttrs.keys(): if 'colspan' in dAttrs.keys():
self.nextCol += int(dAttrs['colspan']) self.nextCol += int(dAttrs['colspan'])
@ -73,6 +77,8 @@ class GoogleSheetsCalParser(HTMLParser):
self.inCell = True self.inCell = True
if 'rowspan' in dAttrs.keys(): if 'rowspan' in dAttrs.keys():
self.rowspan = int(dAttrs['rowspan']) self.rowspan = int(dAttrs['rowspan'])
self.multirows[self.column] = [self.nextCol -
self.column, self.rowspan]
else: self.rowspan = 1 else: self.rowspan = 1
elif (self.row - 2) % 12 == 1: elif (self.row - 2) % 12 == 1:
self.inDate = True self.inDate = True
@ -83,6 +89,9 @@ class GoogleSheetsCalParser(HTMLParser):
if tag == 'td': if tag == 'td':
self.inCell = False self.inCell = False
self.inDate = False self.inDate = False
elif tag == 'tr':
for mr in self.multirows.values():
mr[1] -= 1
elif tag == 'tbody': elif tag == 'tbody':
self.inTable = False self.inTable = False