Добавление данных LyngSat в базу

This commit is contained in:
2025-11-10 17:59:35 +03:00
parent 0858961410
commit b24ef940ce
4 changed files with 96 additions and 7 deletions

View File

@@ -77,9 +77,11 @@ class LyngSatParser:
}
return polarization_map.get(polarization.upper(), polarization)
def get_region_pages(self) -> list[str]:
def get_region_pages(self, regions: list[str] | None = None) -> list[str]:
html_regions = []
for region in self.regions:
if regions is None:
regions = self.regions
for region in regions:
url = f"{self.BASE_URL}/{region}.html"
payload = {
"cmd": "request.get",
@@ -94,9 +96,34 @@ class LyngSatParser:
print(f"Обработал страницу по {region}")
return html_regions
def get_satellite_urls(self, html_regions: list[str]):
sat_names = []
sat_urls = []
for region_page in html_regions:
soup = BeautifulSoup(region_page, "html.parser")
col_table = soup.find_all("div", class_="desktab")[0]
tables = col_table.find_next_sibling('table').find_all('table')
trs = []
for table in tables:
trs.extend(table.find_all('tr'))
for tr in trs:
sat_name = tr.find('span').text
if self.target_sats is not None:
if sat_name.strip().lower() not in self.target_sats:
continue
try:
sat_url = tr.find_all('a')[2]['href']
except IndexError:
sat_url = tr.find_all('a')[0]['href']
sat_names.append(sat_name)
sat_urls.append(sat_url)
return sat_names, sat_urls
def get_satellites_data(self) -> dict[dict]:
sat_data = {}
for region_page in self.get_region_pages():
for region_page in self.get_region_pages(self.regions):
soup = BeautifulSoup(region_page, "html.parser")
col_table = soup.find_all("div", class_="desktab")[0]
@@ -368,4 +395,9 @@ class KingOfSatParser:
'metadata': data['ini_data']['metadata']
}
return satellite_dict
return satellite_dict
from pprint import pprint
lyngsat = LyngSatParser(regions=['europe'], target_sats=['Türksat 3A', 'Intelsat 22'])
html_regions = lyngsat.get_region_pages()
pprint(lyngsat.get_satellite_urls(html_regions))