Revize 3692d853
Přidáno uživatelem Petr Hlaváč před asi 4 roky(ů)
modules/crawler/DatasetCrawler/WIFI_crawler.py | ||
---|---|---|
30 | 30 |
second_level_links = basic_crawler_functions.get_all_links(link) |
31 | 31 |
filtered_second_level_links = basic_crawler_functions.filter_links(second_level_links, regex) |
32 | 32 |
absolute_second_level_links = basic_crawler_functions.create_absolute_links(filtered_second_level_links, link) |
33 |
final_links = basic_crawler_functions.remove_downloaded_links(absolute_second_level_links, dataset_name) |
|
34 | 33 |
|
35 |
for file_link in final_links:
|
|
34 |
for file_link in absolute_second_level_links:
|
|
36 | 35 |
files.append(file_link) |
37 | 36 |
|
37 |
files = basic_crawler_functions.remove_downloaded_links(files, dataset_name) |
|
38 |
|
|
38 | 39 |
for file in files: |
39 | 40 |
basic_crawler_functions.download_file_from_url(file, dataset_name) |
40 | 41 |
|
Také k dispozici: Unified diff
Úprava komentářů