aswi2020sebela-gitlab/modules/crawler/crone_update_script.py @ afb0cc02
1 |
import pipeline |
---|---|
2 |
import os |
3 |
|
4 |
# Path to configuration files
|
5 |
CONFIG_FILES_PATH = "DatasetConfigs/" |
6 |
|
7 |
|
8 |
def run_pipeline_for_all_datasets(): |
9 |
"""
|
10 |
Runs whole DataScript pipeline for every dataset that has existing configuration file
|
11 |
"""
|
12 |
files_in_dir = os.listdir(CONFIG_FILES_PATH) |
13 |
|
14 |
for file in files_in_dir: |
15 |
name = file.split('.')[0] |
16 |
pipeline.run_full_pipeline_crone(name) |
17 |
|
18 |
|
19 |
run_pipeline_for_all_datasets() |