import re
f = open("calendar.html")
data = f.read()
f.close()
tableMatch = re.search(
"instruction and finals.*?" # prefix
"(?P.*?)" # body
"", # suffix
data,
re.DOTALL)
body = tableMatch.group("body")
#print body
rows = re.findall(
"(.*?)",
body,
re.DOTALL | re.MULTILINE)
#print rows
fieldPattern = re.compile("(.*?)", re.DOTALL | re.MULTILINE)
for row in rows:
fields = fieldPattern.findall(row)
if len(fields) == 3:
date = re.sub("\s+", " ", re.sub("<.*?>", "", fields[1]))
name = re.sub("\s+", " ", re.sub("<.*?>", "", fields[0]))
print date, name