Floke/Brancheneinstufung2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

alle Websiten?

Browse Source
This commit is contained in:
Floke
2025-07-10 11:51:42 +00:00
parent f789258638
commit b6be2b183f
1 changed files with 17 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
dealfront_enrichment.py
View File

@@ -140,16 +140,25 @@ class DealfrontScraper:
name_elem = name_elems[0] name_elem = name_elems[0]
company_name = (name_elem.get_attribute("title") or name_elem.text).strip() company_name = (name_elem.get_attribute("title") or name_elem.text).strip()
# Website-Extraktion aus 3. Spalte # Website-Extraktion aus 3. Spalte (externe Links mit target="_blank")
web_elems = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3) a") web_ext = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3) a[target='_blank']")
if web_elems: if web_ext:
# Link-Text ist der Domain-Name href = web_ext[0].get_attribute("href")
website = web_elems[0].text.strip() # ohne Protokoll und Slash am Ende
website = href.split("://")[-1].rstrip("/")
else: else:
# Fallback: reiner Zellen-Text # Fallback 1: Link ohne target (manchmal vorhanden)
web_any = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3) a")
if web_any:
# Domain steht im title-Attribut oder im Text
website = (web_any[0].get_attribute("title") or web_any[0].text).strip()
else:
# Fallback 2: reiner Zellen-Text
cell = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3)") cell = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3)")
website = cell[0].text.strip() if cell else "" website = cell[0].text.strip() if cell else ""
results.append({'name': company_name, 'website': website}) results.append({'name': company_name, 'website': website})
logger.info(f"Extraktion abgeschlossen: {len(results)} Firmen.") logger.info(f"Extraktion abgeschlossen: {len(results)} Firmen.")