Spaces:
Running
Running
NCTCMumbai
commited on
Commit
•
adc862d
1
Parent(s):
62008fe
Update middlewares/search_client.py
Browse files- middlewares/search_client.py +9 -20
middlewares/search_client.py
CHANGED
@@ -2,7 +2,7 @@ import requests
|
|
2 |
from bs4 import BeautifulSoup
|
3 |
import re
|
4 |
import concurrent.futures
|
5 |
-
from
|
6 |
|
7 |
|
8 |
class SearchClient:
|
@@ -43,25 +43,13 @@ class SearchClient:
|
|
43 |
print(f"Error fetching data from {link}: {e}")
|
44 |
return results
|
45 |
|
46 |
-
def _google_search(self,
|
47 |
-
|
48 |
-
for keyword in keywords_lst:
|
49 |
-
googlenews = GoogleNews(start=start_date,end=end_date)
|
50 |
-
googlenews = GoogleNews(lang='en', region='IN')
|
51 |
-
googlenews.get_news(keyword)
|
52 |
-
results=googlenews.results
|
53 |
-
texts = googlenews.get_texts() # List of news texts
|
54 |
-
links = googlenews.get_links() # List of news links
|
55 |
|
56 |
-
|
57 |
-
|
58 |
-
else:
|
59 |
-
data = {'Keyword': [keyword]*(n_crawl), 'Links': links[:n_crawl], 'Text': texts[:n_crawl]}
|
60 |
|
61 |
-
|
62 |
-
corrected_urls = ["https://" + url for url in results]
|
63 |
-
|
64 |
-
text_results = self._fetch_text_from_links(corrected_urls)
|
65 |
return text_results
|
66 |
|
67 |
def _bing_search(self, query, n_crawl):
|
@@ -87,8 +75,9 @@ class SearchClient:
|
|
87 |
else:
|
88 |
return "Invalid vendor"
|
89 |
|
90 |
-
def search_google(self,
|
91 |
if self.vendor == "google":
|
92 |
-
return self._google_search(
|
93 |
else:
|
94 |
return "Invalid vendor"
|
|
|
|
2 |
from bs4 import BeautifulSoup
|
3 |
import re
|
4 |
import concurrent.futures
|
5 |
+
from googlesearch import search
|
6 |
|
7 |
|
8 |
class SearchClient:
|
|
|
43 |
print(f"Error fetching data from {link}: {e}")
|
44 |
return results
|
45 |
|
46 |
+
def _google_search(self,query, n_crawl):
|
47 |
+
search_results = search(query, stop=n_crawl, lang='en', country='IN')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
+
# Convert the search results to a list
|
50 |
+
links = list(search_results)
|
|
|
|
|
51 |
|
52 |
+
text_results = self._fetch_text_from_links(links)
|
|
|
|
|
|
|
53 |
return text_results
|
54 |
|
55 |
def _bing_search(self, query, n_crawl):
|
|
|
75 |
else:
|
76 |
return "Invalid vendor"
|
77 |
|
78 |
+
def search_google(self, query, n_crawl):
|
79 |
if self.vendor == "google":
|
80 |
+
return self._google_search(query, n_crawl)
|
81 |
else:
|
82 |
return "Invalid vendor"
|
83 |
+
|