Muhammad Abdur Rahman Saad
commited on
Commit
·
82a33ed
1
Parent(s):
edb8f2e
add more logging statements
Browse files- main.py +1 -1
- source/csrc.py +3 -1
main.py
CHANGED
@@ -21,7 +21,7 @@ logging.getLogger().setLevel(logging.INFO)
|
|
21 |
delta = int(os.environ.get('DELTA') or '1')
|
22 |
logging.info("DELTA = %s", delta)
|
23 |
|
24 |
-
@flow(name="Data Collection -
|
25 |
def main():
|
26 |
"""
|
27 |
Orchestrates the data collection process by calling the crawl functions of different sources.
|
|
|
21 |
delta = int(os.environ.get('DELTA') or '1')
|
22 |
logging.info("DELTA = %s", delta)
|
23 |
|
24 |
+
@flow(name="Data Collection - Daily", log_prints=True)
|
25 |
def main():
|
26 |
"""
|
27 |
Orchestrates the data collection process by calling the crawl functions of different sources.
|
source/csrc.py
CHANGED
@@ -42,6 +42,7 @@ def crawl(delta):
|
|
42 |
else:
|
43 |
category_url = f"http://www.csrc.gov.cn/csrc/c100039/common_list_{i}.shtml"
|
44 |
i = i + 1
|
|
|
45 |
logger.info(category_url)
|
46 |
req = urllib.request.urlopen(category_url)
|
47 |
text = req.read()
|
@@ -66,7 +67,7 @@ def crawl(delta):
|
|
66 |
article = {}
|
67 |
url = "http://www.csrc.gov.cn" + url
|
68 |
article['category'] = "Policy Interpretation"
|
69 |
-
logger.info(url)
|
70 |
crawl_by_url(url, article)
|
71 |
except Exception as error:
|
72 |
logger.error(error)
|
@@ -130,3 +131,4 @@ def crawl(delta):
|
|
130 |
except Exception as error:
|
131 |
i = -1
|
132 |
logger.error(error)
|
|
|
|
42 |
else:
|
43 |
category_url = f"http://www.csrc.gov.cn/csrc/c100039/common_list_{i}.shtml"
|
44 |
i = i + 1
|
45 |
+
logger.info(f"Fetching from URL: {category_url}")
|
46 |
logger.info(category_url)
|
47 |
req = urllib.request.urlopen(category_url)
|
48 |
text = req.read()
|
|
|
67 |
article = {}
|
68 |
url = "http://www.csrc.gov.cn" + url
|
69 |
article['category'] = "Policy Interpretation"
|
70 |
+
logger.info(f"Processing article URL: {url}")
|
71 |
crawl_by_url(url, article)
|
72 |
except Exception as error:
|
73 |
logger.error(error)
|
|
|
131 |
except Exception as error:
|
132 |
i = -1
|
133 |
logger.error(error)
|
134 |
+
|