pastExam/load.py

import os, argparse, json, re
from datetime import date

from app import ipfs_client, db, Paper

def check_length(x, limit=30, allow_null=False):
    return (x and len(x) <= limit) or (allow_null and not x)

if __name__ == "__main__":
    # parse arguments
    parser = argparse.ArgumentParser(description='add multiple learning resources simultaneously')
    parser.add_argument('path',
        type = str,
        nargs = 1,
        help = 'the root path to the resources')
    args = parser.parse_args()
    path = args.path[0]
    if not os.path.isdir(path):
        raise Exception(f'{path} is not a directory, which should have been the root path to the resources.')

    # load resources
    for course in os.listdir(path):
        course_path = os.path.join(path, course)
        if os.path.isdir(course_path):
            for d in os.listdir(course_path):
                dir_path = os.path.join(course_path, d)
                if os.path.isdir(dir_path):
                    print(f'item {dir_path}')
                    fm = None
                    for r in os.listdir(dir_path):
                        res_path = os.path.join(dir_path, r)
                        if r == 'format.json':
                            with open(res_path, 'r') as f:
                                fm = json.load(f)
                            jsonExistence = True
                    if fm is None:
                        raise Exception(f'there\'s no format.json in {dir_path}.')
                    if len(os.listdir(dir_path)) == 1:
                        raise Exception(f'there\'s no normal file in {dir_path}.')
                    if fm['course'] != course:
                        raise Exception('the course name in json {} is not equal to the directory name {}'.format(fm['course'], course))

                    # check length
                    if not check_length(fm['author']):
                        raise Exception('the length of the name of author is too long: {}'.format(fm['author']))
                    if not check_length(fm['teacher']):
                        raise Exception('the length of the name of teacher is too long: {}'.format(fm['teacher']))
                    if not check_length(fm['note'], 200, True):
                        raise Exception('the legnth of the name of note is too long: {}'.format(fm['note']))

                    # add directory and immediate files in the directory except format.json
                    res = ipfs_client.add(dir_path,
                        recursive = True,
                        pattern = re.compile(r'^(?!format.json$)'))

                    file_hash = ''
                    for r in res:
                        if r.get('Name') == d:
                            file_hash = r.get('Hash')
                    if file_hash == '':
                        raise Exception('the directory is not put into IPFS, something wrong happened.')
                    
                    if not Paper.query.filter_by(file_hash = file_hash).count():
                        paper = Paper(
                                    course = fm['course'],
                                    teacher = fm['teacher'],
                                    year = fm['year'],
                                    notes = fm['note'],
                                    anon = fm['author'] == '匿名',
                                    author = fm['author'],
                                    create_date = date.today(),
                                    file_hash = file_hash
                                )
                        db.session.add(paper)
                        db.session.commit()

                        print(f'item {dir_path} is added with hash {file_hash}')
                    else:
                        print(f'{dir_path} has been added with hash {file_hash} before')