tiyn/stud.ip-crawler
1
0
Fork 0
mirror of https://github.com/tiyn/stud.ip-crawler.git synced 2025-10-31 11:11:18 +01:00
Code Issues Projects Releases Wiki Activity

bugfixes: now all courses in the curr semester get downloaded

Browse Source 
- limit for courses was set to low to display current courses
- unused variable docs was removed from local function
This commit is contained in:
TiynGER
2020-11-10 11:51:56 +01:00
parent 719c89e7cf
commit 92ee51e56c
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/crawler.py
View File

@@ -67,7 +67,7 @@ class Crawler:
self.create_dir(base_dir) self.create_dir(base_dir)
curr_courses = self.studip.get_curr_courses( curr_courses = self.studip.get_curr_courses(
self.studip.get_uid(), self.studip.get_curr_semester()) self.studip.get_uid(), self.studip.get_curr_semester())
log.info('crawling all current courses' + curr_courses) log.info('crawling all current courses')
os.chdir(base_dir) os.chdir(base_dir)
for course in curr_courses: for course in curr_courses:
log.debug('course is ' + curr_courses[course]) log.debug('course is ' + curr_courses[course])

4
src/studip.py
View File

@@ -4,6 +4,7 @@ import logging as log
#from tqdm import tqdm #from tqdm import tqdm
import requests as req import requests as req
from requests.auth import HTTPBasicAuth from requests.auth import HTTPBasicAuth
import json
class Studip: class Studip:
@@ -72,7 +73,7 @@ class Studip:
Returns: Returns:
string: id of the current semester string: id of the current semester
""" """
rsp = self.auth_req('/api.php/user/' + user_id + '/courses') rsp = self.auth_req('/api.php/user/' + user_id + '/courses?limit=1000')
ord_sems = self.get_ordered_semesters() ord_sems = self.get_ordered_semesters()
courses = rsp.json()['collection'] courses = rsp.json()['collection']
course_list = {} course_list = {}
@@ -158,7 +159,6 @@ class Studip:
""" """
rsp = self.auth_req('/api.php/folder/' + folder) rsp = self.auth_req('/api.php/folder/' + folder)
subdirs = rsp.json()['subfolders'] subdirs = rsp.json()['subfolders']
docs = rsp.json()['file_refs']
res_subdirs = {} res_subdirs = {}
for subdir in subdirs: for subdir in subdirs:
sub_id = subdir['id'] sub_id = subdir['id']