コード例 #1
0
ファイル: crawl_reviews.py プロジェクト: Epinilla/planner
def main():
    reviews = json.loads(open("reviews.json", "r").read())

    Review.objects.filter(is_crawled=True).delete()
    total_added = 0
    for review in reviews:
        code = review["course"]
        course_query = CourseCode.objects.filter(code=code)
        if course_query.exists():
            r = Review()
            r.course = course_query[0].course
            r.rating = review["rating"]
            r.grade = review["grade"]
            r.text = review["text"].replace("<div>", "").replace("</div>", "")
            r.is_crawled = True
            r.created_at = str_to_timestamp(review["created_at"])
            r.updated_at = str_to_timestamp(review["updated_at"])
            r.author = DEFAULT_USER
            r.save()
            total_added += 1
            if total_added % 50 == 0:
                print "Added %s reviews" % total_added