allowlgroup/parser
5
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

исправил url_ist
All checks were successful
continuous-integration/drone/push Build is passing
Details

Browse Source
This commit is contained in:
admin
2026-04-11 15:39:11 +10:00
parent f39ade5245
commit 5805ab0fe2
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
main.py
View File

@@ -209,7 +209,7 @@ def extract_text_from_url(url, timeout=10, verify=True):
def gpt_response_message(content: str, name_promt: str): def gpt_response_message(content: str, name_promt: str):
contentGPT = wp.get_promt(name_promt).replace('{content}', content) contentGPT = wp.get_promt(name_promt).replace('{content}', content)
print(contentGPT)
url = 'http://45.129.78.228:8484' #10.8.0.14:5500 url = 'http://45.129.78.228:8484' #10.8.0.14:5500
params = {'text': contentGPT} params = {'text': contentGPT}
@@ -336,7 +336,7 @@ def start_pars_one_istochnik(data_init=""):
print(f"Страница {page_number} [{i}/{len(hrefs)}] parsing {link}") print(f"Страница {page_number} [{i}/{len(hrefs)}] parsing {link}")
text = extract_text_from_url_one(link) text = extract_text_from_url_one(link)
if len(text) >= 100: if len(text) >= 100:
response_text = gpt_response_message(text, url_ist = "source1") response_text = gpt_response_message(text, "source1")
print(response_text) print(response_text)
if response_text: if response_text:
update_bd_and_create_document(response_text=response_text, article_date=f"{current_year}/{current_month}/{current_day}", url=link, parsed_at=str(dt.now()), original_text=text, other=url) update_bd_and_create_document(response_text=response_text, article_date=f"{current_year}/{current_month}/{current_day}", url=link, parsed_at=str(dt.now()), original_text=text, other=url)
@@ -364,7 +364,7 @@ def start_pars_two_istochnik():
try: try:
text, time_text = extract_text_from_url(hrefs) text, time_text = extract_text_from_url(hrefs)
if len(text) >= 100: if len(text) >= 100:
response_text = gpt_response_message(text, url_ist = "source2") response_text = gpt_response_message(text, "source2")
print(response_text) print(response_text)
if response_text: if response_text:
update_bd_and_create_document(response_text=response_text, article_date=time_text, url=hrefs, parsed_at=str(dt.now()), original_text=text, other=url) update_bd_and_create_document(response_text=response_text, article_date=time_text, url=hrefs, parsed_at=str(dt.now()), original_text=text, other=url)