HD Course

spider.py

# coding=utf8

import requests


class HDCourse:

    def __init__(self, id, title, teacher_name, videos):
        self.id = id
        self.title = title
        self.teacher_name = teacher_name
        self.videos = videos

    def __repr__(self):
        desc = u"%s (%s) %s 个视频" % \
                 (self.title, self.teacher_name, len(self.videos))
        return desc.encode('utf8')

    def download_videos(self):
        for video in self.videos:
            vid = video['video_id']
            video_title = video['title']
            uri = "https://course.hundun.cn/oss_video_url?app_role=yxs" \
                    "&clarity=0&clientType=iPhone%206%20Plus_iOS_10.0.2" \
                    "&device_type=phone&isVerify=0&net=wifi&" \
                    "versionName=1.11.6&video_id=%s" % vid
            res = requests.get(uri).json()
            play_uri = res['video_info']['url']
            print('=' * 20)
            print('start download %s' % video_title)
            r = requests.get(play_uri)
            with open(("%s.mp4" % video_title), "wb") as code:
                code.write(r.content)

    @classmethod
    def get_by_id(cls, cid):
        uri = "https://course.hundun.cn/get_course_detail?app_role=yxs" \
                "&clientType=iPhone%206%20Plus_iOS_10.0.2&course_id=%s" \
                "&device_type=phone&isVerify=0&list_type=&net=wifi&unionid=" \
                "&versionName=1.11.6" % cid
        res = requests.get(uri).json()
        course_detail = res['course_detail']
        course_meta = course_detail['course_meta']
        videos = course_detail['i18n_directory'][0]['directory']
        return cls(cid,
                   course_meta['title'],
                   course_meta['teacher_name'],
                   videos)

    @classmethod
    def get_courses(cls, page=0):
        uri = "https://course.hundun.cn/get_course_list?app_role=yxs" \
                "&card_id=card_main&clientType=iPhone%206%20Plus_iOS_10.0.2" \
                "&device_type=phone&isVerify=0&net=wifi&order_type=time" \
                "&page=%s&versionName=1.11.6" % page
        res = requests.get(uri).json()
        return [cls.get_by_id(cr['course_id']) for cr in res['course_list']]