dritte spalte
This commit is contained in:
@@ -132,26 +132,29 @@ class DealfrontScraper:
|
||||
|
||||
rows = self.driver.find_elements(*rows_selector)
|
||||
logger.info(f"{len(rows)} Firmen-Datenzeilen zur Verarbeitung gefunden.")
|
||||
self.driver.implicitly_wait(1)
|
||||
|
||||
for i, row in enumerate(rows, 1):
|
||||
try:
|
||||
# Firmennamen holen
|
||||
# Firmennamen holen...
|
||||
name_elem = row.find_element(By.CSS_SELECTOR, ".sticky-column a.t-highlight-text")
|
||||
company_name = (name_elem.get_attribute("title") or name_elem.text).strip()
|
||||
|
||||
# Website aus der zweiten Spalte: erst href, dann Text-Fallback
|
||||
try:
|
||||
website_elem = row.find_element(By.CSS_SELECTOR, "td:nth-of-type(2) a")
|
||||
# HREF bereinigen (ohne https://)
|
||||
website = website_elem.get_attribute("href").split("://")[-1].strip("/")
|
||||
except NoSuchElementException:
|
||||
website = row.find_element(By.CSS_SELECTOR, "td:nth-of-type(2)").text.strip()
|
||||
# Website aus der dritten Spalte: erst href, dann Text-Fallback
|
||||
elems = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3) a")
|
||||
if elems:
|
||||
website = elems[0].get_attribute("href").split("://", 1)[1].rstrip("/")
|
||||
else:
|
||||
website = row.find_element(By.CSS_SELECTOR, "td:nth-of-type(3)").text.strip()
|
||||
|
||||
results.append({'name': company_name, 'website': website})
|
||||
except NoSuchElementException:
|
||||
logger.warning(f"Zeile {i}: Name oder Webseite nicht extrahierbar. Überspringe.")
|
||||
continue
|
||||
|
||||
# Implicit-Wait wiederherstellen (z. B. 10 Sekunden)
|
||||
self.driver.implicitly_wait(10)
|
||||
|
||||
logger.info(f"Extraktion abgeschlossen. {len(results)} Firmen gefunden.")
|
||||
return results
|
||||
|
||||
|
||||
Reference in New Issue
Block a user