Spaces:
Running
Running
Subhajit Chakraborty
commited on
Commit
·
9cdbd5b
1
Parent(s):
91df0bf
update files(2)
Browse files
src/post_extraction_tools/lead_scoring.py
CHANGED
|
@@ -71,7 +71,7 @@ class LeadScoring:
|
|
| 71 |
def score(self, leads: object, conditions: object):
|
| 72 |
# scored_leads = []
|
| 73 |
for lead in leads["companies"]:
|
| 74 |
-
if lead["score"] is None:
|
| 75 |
score = 0
|
| 76 |
if "industry_type" in conditions:
|
| 77 |
try:
|
|
|
|
| 71 |
def score(self, leads: object, conditions: object):
|
| 72 |
# scored_leads = []
|
| 73 |
for lead in leads["companies"]:
|
| 74 |
+
if (lead["score"] is None or lead["score"] == 0):
|
| 75 |
score = 0
|
| 76 |
if "industry_type" in conditions:
|
| 77 |
try:
|
src/services/scrape_worker.py
CHANGED
|
@@ -21,4 +21,10 @@ if __name__ == "__main__":
|
|
| 21 |
|
| 22 |
url = sys.argv[1]
|
| 23 |
result = scrape_website(url)
|
| 24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
url = sys.argv[1]
|
| 23 |
result = scrape_website(url)
|
| 24 |
+
if "error" in result:
|
| 25 |
+
print(result["error"], file=sys.stderr)
|
| 26 |
+
sys.exit(1)
|
| 27 |
+
else:
|
| 28 |
+
print(json.dumps(result))
|
| 29 |
+
sys.exit(0)
|
| 30 |
+
|
src/services/scraper.py
CHANGED
|
@@ -49,6 +49,7 @@ def scrape_website(url: str) -> str:
|
|
| 49 |
text=True,
|
| 50 |
timeout=60
|
| 51 |
)
|
|
|
|
| 52 |
if result.returncode != 0:
|
| 53 |
return f"Error scraping the URL {url}: {result.stderr.strip()}"
|
| 54 |
|
|
|
|
| 49 |
text=True,
|
| 50 |
timeout=60
|
| 51 |
)
|
| 52 |
+
# print("This is the print block: ", result.returncode, result.stderr, result.stdout)
|
| 53 |
if result.returncode != 0:
|
| 54 |
return f"Error scraping the URL {url}: {result.stderr.strip()}"
|
| 55 |
|