def handle_data(self, data):
# print('tag '+data) #DEBUG
if self.is_header > 0:
self.is_header += 1
# print('Pippo '+str(self.is_header)) # DEBUG
data = data.strip()
# needed to manage multiple teacher lines
is_append = False
if self.immolation:
# count the 4 lines of data for the lesson (subject, teacher, rooms, address)
if self.dataLine <= 3:
# pass #DEBUG
if data == '/':
self.dataLine -= 1
is_append = True
if self.dataLine == 0:
self.lesson.subject = data.upper()
print(self.lesson.subject) #DEBUG
if self.dataLine == 1:
self.lesson.teacher = self.lesson.teacher + data.upper()
if is_append == True:
self.dataLine -= 1
is_append = False
print(self.lesson.teacher) #DEBUG
if self.dataLine == 2:
self.lesson.rooms = data.upper()
print(self.lesson.rooms) #DEBUG
if self.dataLine == 3:
self.lesson.address = data.upper()
print(self.lesson.address) #DEBUG
# print(data.upper()) # DEBUG
self.dataLine += 1
# if is the last line
if self.dataLine == 4:
print("--") # DEBUG
self.lesson.semesterStartDate = self.semesterStartDate
self.lesson.semesterEndDate = self.semesterEndDate
self.dataLine = 0
self.lesson.persist()
# reset the entity
self.lesson = Lesson()
if self.is_header == 9: # find the correct line
# print(data.upper()) #DEBUG
# extract the data information and foreach date create the datetime object
line = data.split(sep="-")
self.semesterStartDate = datetime.datetime.strptime(
line[0], ": %A %d %B %Y ")
self.semesterEndDate = datetime.datetime.strptime(
line[1], " %A %d %B %Y")
# print(self.semesterStartDate) #DEBUG
# print(self.semesterEndDate) #DEBUG
self.is_header = 0
def __init__(self): # Override here
HTMLParser.__init__(self)
# todo: add documentation
self.immolation = False
# is the line counter for parsing the header data
self.is_header = 0
# is the time of the firts hour of lessons. Mandatory to calculate the start hour from offsets
self.firstHour = datetime.datetime.strptime("08:30", "%H:%M")
# the lesson database entity
self.lesson = Lesson()
self.dataLine = 0
self.semesterStartDate = datetime.datetime
self.semesterEndDate = datetime.datetime
def __init__(self): # Override here
HTMLParser.__init__(self)
# todo: add documentation
self.immolation = False
# is the line counter for parsing the header data
self.is_header = 0
# is the time of the firts hour of lessons. Mandatory to calculate the start hour from offsets
self.firstHour = datetime.datetime.strptime("08:30", "%H:%M")
# the lesson database entity
self.lesson = Lesson()
self.dataLine = 0
self.semesterStartDate = datetime.datetime
self.semesterEndDate = datetime.datetime
def handle_data(self, data):
# print('tag '+data) #DEBUG
if self.is_header > 0:
self.is_header += 1
# print('Pippo '+str(self.is_header)) # DEBUG
data = data.strip()
# needed to manage multiple teacher lines
is_append = False
if self.immolation:
# count the 4 lines of data for the lesson (subject, teacher, rooms, address)
if self.dataLine <= 3:
# pass #DEBUG
if data == '/':
self.dataLine -= 1
is_append = True
if self.dataLine == 0:
self.lesson.subject = data.upper()
print(self.lesson.subject) #DEBUG
if self.dataLine == 1:
self.lesson.teacher = self.lesson.teacher+data.upper()
if is_append == True:
self.dataLine -=1
is_append = False
print(self.lesson.teacher) #DEBUG
if self.dataLine == 2:
self.lesson.rooms = data.upper()
print(self.lesson.rooms) #DEBUG
if self.dataLine == 3:
self.lesson.address = data.upper()
print(self.lesson.address) #DEBUG
# print(data.upper()) # DEBUG
self.dataLine += 1
# if is the last line
if self.dataLine == 4:
print("--") # DEBUG
self.lesson.semesterStartDate = self.semesterStartDate
self.lesson.semesterEndDate = self.semesterEndDate
self.dataLine = 0
self.lesson.persist()
# reset the entity
self.lesson = Lesson()
if self.is_header == 9: # find the correct line
# print(data.upper()) #DEBUG
# extract the data information and foreach date create the datetime object
line = data.split(sep="-")
self.semesterStartDate = datetime.datetime.strptime(line[0], ": %A %d %B %Y ")
self.semesterEndDate = datetime.datetime.strptime(line[1], " %A %d %B %Y")
# print(self.semesterStartDate) #DEBUG
# print(self.semesterEndDate) #DEBUG
self.is_header = 0
class MLStripper(HTMLParser):
def __init__(self): # Override here
HTMLParser.__init__(self)
# todo: add documentation
self.immolation = False
# is the line counter for parsing the header data
self.is_header = 0
# is the time of the firts hour of lessons. Mandatory to calculate the start hour from offsets
self.firstHour = datetime.datetime.strptime("08:30", "%H:%M")
# the lesson database entity
self.lesson = Lesson()
self.dataLine = 0
self.semesterStartDate = datetime.datetime
self.semesterEndDate = datetime.datetime
def handle_starttag(self, tag, attrs):
if tag == 'table':
for attr in attrs:
if attr[0] == 'class' and attr[1] == 'cellTabs':
break
if attr[0] == 'id':
self.immolation = True
# Hours coded from 0 to 9 as the nine hours avalaible in a standard lesson day from 08:30 to 09:30
# Day coded form 0 to 6 where 0 is monday and 6 is sunday.
coords = attr[1].split(sep='_') # Format as follow: DAY_HOUR
self.lesson.day = coords[1]
# calculate the correct start hour for the lesson
dt = self.firstHour + datetime.timedelta(hours=int(coords[2]))
self.lesson.hour = dt.time()
# print(self.lesson.hour) #DEBUG
# print(coords) #DEBUG
elif tag == 'td':
for attr in attrs:
# find the header with the information regarding the semester
if attr[1] == 'ttTitleTD':
# count the lines of the header
self.is_header += 1
# print('Pippo '+str(self.is_header)) #DEBUG
def handle_endtag(self, tag):
if tag == 'table': # or tag == 'br' #DEBUG
self.immolation = False
def handle_data(self, data):
# print('tag '+data) #DEBUG
if self.is_header > 0:
self.is_header += 1
# print('Pippo '+str(self.is_header)) # DEBUG
data = data.strip()
# needed to manage multiple teacher lines
is_append = False
if self.immolation:
# count the 4 lines of data for the lesson (subject, teacher, rooms, address)
if self.dataLine <= 3:
# pass #DEBUG
if data == '/':
self.dataLine -= 1
is_append = True
if self.dataLine == 0:
self.lesson.subject = data.upper()
print(self.lesson.subject) #DEBUG
if self.dataLine == 1:
self.lesson.teacher = self.lesson.teacher+data.upper()
if is_append == True:
self.dataLine -=1
is_append = False
print(self.lesson.teacher) #DEBUG
if self.dataLine == 2:
self.lesson.rooms = data.upper()
print(self.lesson.rooms) #DEBUG
if self.dataLine == 3:
self.lesson.address = data.upper()
print(self.lesson.address) #DEBUG
# print(data.upper()) # DEBUG
self.dataLine += 1
# if is the last line
if self.dataLine == 4:
print("--") # DEBUG
self.lesson.semesterStartDate = self.semesterStartDate
self.lesson.semesterEndDate = self.semesterEndDate
self.dataLine = 0
self.lesson.persist()
# reset the entity
self.lesson = Lesson()
if self.is_header == 9: # find the correct line
# print(data.upper()) #DEBUG
# extract the data information and foreach date create the datetime object
line = data.split(sep="-")
self.semesterStartDate = datetime.datetime.strptime(line[0], ": %A %d %B %Y ")
self.semesterEndDate = datetime.datetime.strptime(line[1], " %A %d %B %Y")
# print(self.semesterStartDate) #DEBUG
# print(self.semesterEndDate) #DEBUG
self.is_header = 0
class MLStripper(HTMLParser):
def __init__(self): # Override here
HTMLParser.__init__(self)
# todo: add documentation
self.immolation = False
# is the line counter for parsing the header data
self.is_header = 0
# is the time of the firts hour of lessons. Mandatory to calculate the start hour from offsets
self.firstHour = datetime.datetime.strptime("08:30", "%H:%M")
# the lesson database entity
self.lesson = Lesson()
self.dataLine = 0
self.semesterStartDate = datetime.datetime
self.semesterEndDate = datetime.datetime
def handle_starttag(self, tag, attrs):
if tag == 'table':
for attr in attrs:
if attr[0] == 'class' and attr[1] == 'cellTabs':
break
if attr[0] == 'id':
self.immolation = True
# Hours coded from 0 to 9 as the nine hours avalaible in a standard lesson day from 08:30 to 09:30
# Day coded form 0 to 6 where 0 is monday and 6 is sunday.
coords = attr[1].split(
sep='_') # Format as follow: DAY_HOUR
self.lesson.day = coords[1]
# calculate the correct start hour for the lesson
dt = self.firstHour + datetime.timedelta(
hours=int(coords[2]))
self.lesson.hour = dt.time()
# print(self.lesson.hour) #DEBUG
# print(coords) #DEBUG
elif tag == 'td':
for attr in attrs:
# find the header with the information regarding the semester
if attr[1] == 'ttTitleTD':
# count the lines of the header
self.is_header += 1
# print('Pippo '+str(self.is_header)) #DEBUG
def handle_endtag(self, tag):
if tag == 'table': # or tag == 'br' #DEBUG
self.immolation = False
def handle_data(self, data):
# print('tag '+data) #DEBUG
if self.is_header > 0:
self.is_header += 1
# print('Pippo '+str(self.is_header)) # DEBUG
data = data.strip()
# needed to manage multiple teacher lines
is_append = False
if self.immolation:
# count the 4 lines of data for the lesson (subject, teacher, rooms, address)
if self.dataLine <= 3:
# pass #DEBUG
if data == '/':
self.dataLine -= 1
is_append = True
if self.dataLine == 0:
self.lesson.subject = data.upper()
print(self.lesson.subject) #DEBUG
if self.dataLine == 1:
self.lesson.teacher = self.lesson.teacher + data.upper()
if is_append == True:
self.dataLine -= 1
is_append = False
print(self.lesson.teacher) #DEBUG
if self.dataLine == 2:
self.lesson.rooms = data.upper()
print(self.lesson.rooms) #DEBUG
if self.dataLine == 3:
self.lesson.address = data.upper()
print(self.lesson.address) #DEBUG
# print(data.upper()) # DEBUG
self.dataLine += 1
# if is the last line
if self.dataLine == 4:
print("--") # DEBUG
self.lesson.semesterStartDate = self.semesterStartDate
self.lesson.semesterEndDate = self.semesterEndDate
self.dataLine = 0
self.lesson.persist()
# reset the entity
self.lesson = Lesson()
if self.is_header == 9: # find the correct line
# print(data.upper()) #DEBUG
# extract the data information and foreach date create the datetime object
line = data.split(sep="-")
self.semesterStartDate = datetime.datetime.strptime(
line[0], ": %A %d %B %Y ")
self.semesterEndDate = datetime.datetime.strptime(
line[1], " %A %d %B %Y")
# print(self.semesterStartDate) #DEBUG
# print(self.semesterEndDate) #DEBUG
self.is_header = 0
