py_spider/glut.py

from dataclasses import dataclass, asdict
from enum import Enum, auto
import requests
from lxml import etree

class Term(Enum):
    SPRING = auto()
    AUTUMN = auto()
    ALL = auto()

    def __repr__(self):
        return self.name.lower()


class GLUTAcademic:
    class LoginFailedError(Exception):
        pass

    FINAL_GRADE_DICT = {
        '优秀': 95,
        '良': 85,
        '中': 75,
        '及格': 65,
        '不及格': 40,
        '': 0,
    }

    @dataclass
    class Score:
        year: int
        term: Term
        department: str
        course_number: int
        course_name: str
        course_sequence: int
        instructor: str
        final_grade: int
        gpa: float
        credits: float
        hours: float
        assessment_method: str
        course_attribute: str
        notes: str
        exam_type: str
        retake_flag: str
        course_requirements: str
        course_category: str
        coefficient: str
        second_degree_minor: str
        pass_flag: str

        def as_dict(self):
            d = asdict(self)
            d['term'] = repr(self.term)
            return d

        def __str__(self):
            return (f"Score(学年={self.year}, 学期={self.term}, 开课院系={self.department}, "
                    f"课程号={self.course_number}, 课程名={self.course_name}, 课序号={self.course_sequence}, "
                    f"主讲教师={self.instructor}, 总评={self.final_grade}, 绩点={self.gpa}, "
                    f"学分={self.credits}, 学时={self.hours}, 考核方式={self.assessment_method}, "
                    f"选课属性={self.course_attribute}, 备注={self.notes}, 考试性质={self.exam_type}, "
                    f"是否缓考={self.retake_flag}, 课程要求={self.course_requirements}, "
                    f"课程类别={self.course_category}, 系数={self.coefficient}, "
                    f"二学位辅修={self.second_degree_minor}, 及格标志={self.pass_flag})")

    def __init__(self, username: int, password: str):
        self.session = requests.Session()
        self.username = username
        self.password = password
        self.login_status = False
        self.base_url = 'https://jw.glut.edu.cn'

    def login(self):
        url = f"{self.base_url}/academic/j_acegi_security_check"\
            f"?j_username={self.username}&j_password={self.password}&j_captcha=undefined"
        res = self.session.get(url)
        if res.status_code == 200 and res.headers.get('X-Frame-Options') != 'DENY':
            self.login_status = True
        else:
            raise self.LoginFailedError('login error maybe username or password is wrong')

    def get_cookies(self):
        if not self.login_status:
            return None
        return '; '.join([f'{key}={value}' for key, value in self.session.cookies.items()])

    def get_scores_raw(self, year: int = 2024, term: Term = Term.ALL):
        url = f"{self.base_url}/academic/manager/score/studentOwnScore.do"
        headers = {'Content-Type': 'application/x-www-form-urlencoded'}
        term_dict = {
            Term.SPRING: "1",
            Term.AUTUMN: "2",
            Term.ALL: ""
        }

        data = {
            'year': year - 2013 + 33,
            'term': term_dict[term],
            'prop': '',
            'groupName': '',
            'para': '0',
            'sortColumn': '',
            'Submit': '查询'
        }
        res = self.session.post(url, data=data, headers=headers)

        if res.status_code != 200 or res.headers.get('X-Frame-Options') == 'DENY':
            raise RuntimeError("Failed to get scores")
        return res.content

    def parse_scores(self, raw_score: bytes) -> list[Score]:
        tree = etree.fromstring(raw_score, parser=etree.HTMLParser(encoding='utf-8'))
        table = tree.xpath('//table[@class="datalist"]')
        if not table or len(table) != 1:
            err_msg = f"Failed to parse scores {table}"
            with open('error.html', 'wb') as f:
                f.write(raw_score)
            raise RuntimeError(err_msg)
        table = table[0]
        tr_list = table.xpath('tr')[1:]
        return [self.parse_scope(tr.xpath('td/text()')) for tr in tr_list]

    def parse_scope(self, texts: list[str]) -> Score:
        texts = [t.strip() for t in texts]
        final_grade = self.FINAL_GRADE_DICT.get(texts[7], -1)
        if final_grade == -1:
            try:
                final_grade = float(texts[7])
            except ValueError as exc:
                raise ValueError('Final grade is not a int number') from exc

        term_map = {
            '春': Term.SPRING,
            '秋': Term.AUTUMN
        }
        term = term_map.get(texts[1])
        if term is None:
            raise ValueError(f'term error maybe 春 or 秋 but got {texts[1]}')

        return self.Score(
            year = int(texts[0]),
            term = term,
            department= texts[2],
            course_number = int(texts[3]),
            course_name = texts[4],
            course_sequence = int(texts[5]),
            instructor = texts[6],
            final_grade = final_grade,
            gpa = float(texts[8]),
            credits = float(texts[9]),
            hours = float(texts[10]),
            assessment_method = texts[11],
            course_attribute = texts[12],
            notes = texts[13],
            exam_type = texts[14],
            retake_flag = texts[15],
            course_requirements = texts[16],
            course_category = texts[17],
            coefficient = texts[18],
            second_degree_minor = texts[19],
            pass_flag = texts[20]
        )

    def get_scores(self, year: int = 2024, term: Term = Term.ALL) -> list[Score]:
        raw_score = self.get_scores_raw(year, term)
        return self.parse_scores(raw_score)

    def calculate_scores(self, scores: list[Score]) -> float:
        if not scores:
            return 0.0
        total_credits = 0
        total_grades = 0
        for i in scores:
            if i.course_attribute != '必修' or i.course_name.startswith('体育'):
                continue
            total_credits += i.credits
            total_grades += i.final_grade * i.credits
        return total_grades / total_credits