prase.py

from lxml import etree

def open_html(filepath):
    with open(filepath, "r", encoding="utf-8") as f:
        html = f.read()
    return html

def prase(html_str):
    html = etree.HTML(html_str)
    table = html.xpath("/html/body/form/table[@id='T1']/tr")[2:]

    course = []
    semester = ""
    for tr in table:
        if (len(tr.xpath("td"))) == 1:
            if len(tr.xpath("td/div")) == 1:
                semester = tr.xpath("td/div/text()")[0]
        else:
            if len(tr.xpath("td/div/font")) > 0:
                a_course = {}
                a_course['id'] = tr.xpath("td/div/font/text()")[0]
                a_course['name'] = tr.xpath("td[2]/div/font/text()")[0]
                a_course['grade'] = tr.xpath("td[3]/div/font/text()")[0]
                a_course['credit'] = tr.xpath("td[4]/div/font/text()")[0]
                a_course['grade_point'] = tr.xpath("td[5]/div/font/text()")[0]
                a_course['is_passed'] = tr.xpath("td[6]/div/font/text()")[0]
                a_course['is_normal'] = tr.xpath("td[7]/div/font/text()")[0]
                if  len(tr.xpath("td[8]/div/font/text()")) == 1:
                    a_course['is_elective'] = tr.xpath("td[8]/div/font/text()")[0]
                else:
                    a_course['is_elective'] = tr.xpath("td[8]/div/font/font/text()")[0]

                a_course['date'] = tr.xpath("td[9]/div/font/text()")[0]
                a_course['semester'] = semester
                course.append(a_course)

    student = {}

    informations = html.xpath("/html/body/form/table[@id='T1']/tr[1]/td/font/text()")
    student['id'] = informations[1].strip()
    student['name'] = informations[2].strip()
    student['department']= informations[3].strip()
    student['major']= informations[4].strip()

    grades = html.xpath("/html/body/form/table[@id='T1']/tr[2]/td")[0]
    student['all_grade'] = grades.xpath("font[1]/text()")[0]
    student['all_credit'] = grades.xpath("font[2]/text()")[0][1:]
    student['unpassed_credit'] = grades.xpath("font[3]/text()")[0]
    student['unpassed_count'] = grades.xpath("font[4]/text()")[0]
    student['course'] = course

    return student


if __name__ == "__main__":
    print(
        prase(
            open_html("./1754200.html")
        )
    )