Scrape data from SchoolAutomate portals

sa_scrape.py

#!/usr/bin/env python3

import requests
import getpass
import html.parser

class LoginFormParser(html.parser.HTMLParser):
    def __init__(self):
        html.parser.HTMLParser.__init__(self)

        self.data = {}

    def handle_starttag(self, tag, attrs):
        if(tag != 'input'):
            return

        try:
            tag_type = [item for item in attrs if 'type' in item][0][1]
            if(tag_type != 'hidden'):
                return

            tag_name = [item for item in attrs if 'name' in item][0][1]
            tag_value = [item for item in attrs if 'value' in item][0][1]

            if tag_name in {'user_id', 'password', 'body_color', 'welcome_url', 'page_url', 'login_type'}:
                self.data[tag_name] = tag_value

        except IndexError as e:
            pass

    def handle_endtag(self, tag):
        pass

    def handle_data(self, data):
        pass

class ChecklistParser(html.parser.HTMLParser):
    def __init__(self):
        html.parser.HTMLParser.__init__(self)

        self.data = {}

        self.current_tag = None
        self.parse_tag = None

    def handle_starttag(self, tag, attrs):
        self.current_tag = tag

    def handle_endtag(self, tag):
        pass

    def handle_data(self, data):
        if self.parse_tag == None:
            if 'Student ID' in data:
                self.parse_tag = 'student_id'
            elif 'Student name' in data:
                self.parse_tag = 'student_name'
            elif 'Course/Major' in data:
                self.parse_tag = 'course'
            elif 'Total units required for this course' in data:
                self.parse_tag = 'units_required'
            elif 'Year' in data:
                self.parse_tag = 'year'
            elif 'Status' in data:
                self.parse_tag = 'status'
            elif 'Total units taken' in data:
                self.parse_tag = 'units_taken'
            elif 'GWA' in data:
                self.parse_tag = 'gwa'
        else:
            tags = {'student_name', 'course', 'units_required', 'year', 'status', 'units_taken', 'gwa'}

            if (self.parse_tag == 'student_id') and (self.current_tag == 'b'):
                self.data[self.parse_tag] = data
                self.parse_tag = None
            elif (self.parse_tag in tags) and (self.current_tag == 'strong'):
                self.data[self.parse_tag] = data
                self.parse_tag = None

def portal_login(session, username, password):
    request = session.get('http://gti-binan.dyndns.org:8339/PARENTS_STUDENTS/main_files/parents_student_main_page_rightFrame.jsp')

    parser = LoginFormParser()
    parser.feed(request.text)

    data = {}
    data[parser.data['user_id']] = username
    data['user_id'] = parser.data['user_id']
    data[parser.data['password']] = password
    data['password'] = parser.data['password']
    data['is_secured'] = 1
    data['x'] = 0
    data['y'] = 0
    data['body_color'] = parser.data['body_color']
    data['welcome_url'] = parser.data['welcome_url']
    data['page_url'] = parser.data['page_url']
    data['login_type'] = parser.data['login_type']

    request = session.post('http://gti-binan.dyndns.org:8339/commfile/login.jsp', data=data)

    return ('welcome_stud.jsp' in request.text)

def portal_logout(session):
    pass

def scrape_checklist(session):
    request = session.get('http://gti-binan.dyndns.org:8339/PARENTS_STUDENTS/acad_performance/stud_cur_residency_eval.jsp')

    parser = ChecklistParser()
    parser.feed(request.text)

    return parser.data

def scrape_graderelease(session):
    pass

def main():
    username = input('Username: ')
    password = getpass.getpass()

    session = requests.Session()

    if portal_login(session, username, password):
        data_checklist = scrape_checklist(session)

        print('Student ID: ' + data_checklist['student_id'])
        print('Student Name: ' + data_checklist['student_name'])
        print('Course/Major: ' + data_checklist['course'])
        print('Units Required: ' + data_checklist['units_required'])
        print('Year: ' + data_checklist['year'])
        print('Status: ' + data_checklist['status'])
        print('Units Taken: ' + data_checklist['units_taken'])
        print('GWA: ' + data_checklist['gwa'])

        scrape_graderelease(session)
    else:
        print('Login failed!')

if __name__ == '__main__':
    main()