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

dealfront_enrichment.py aktualisiert

Browse Source
This commit is contained in:
Floke
2025-07-10 11:24:46 +00:00
parent 286c473a8b
commit c109a9506b
1 changed files with 29 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
dealfront_enrichment.py
View File

@@ -73,44 +73,48 @@ class DealfrontScraper:
time.sleep(1) time.sleep(1)
def extract_current_page_results(self): def extract_current_page_results(self):
# 1) Kurz Implicit-Wait absenken, damit Fehlversuche sofort zurückkehren # 1) Kurz Implicit-Wait absenken
self.driver.implicitly_wait(1) self.driver.implicitly_wait(1)
# 2) Auf erstes Firmen-Element warten (bis zu 15 s), dann kurzen Puffer # 2) Auf erstes Firmen-Element warten und Puffer
first = (By.CSS_SELECTOR, ".sticky-column a.t-highlight-text") first = (By.CSS_SELECTOR, ".sticky-column a.t-highlight-text")
self.wait.until(EC.visibility_of_element_located(first)) self.wait.until(EC.visibility_of_element_located(first))
time.sleep(1) time.sleep(1)
logger.info("Extrahiere Ergebnisse von der aktuellen Seite...") try:
results = [] logger.info("Extrahiere Ergebnisse von der aktuellen Seite...")
results = []
# 3) Auf mindestens eine Tabellenzeile warten # 3) Auf mindestens eine Tabellenzeile warten
rows_sel = (By.CSS_SELECTOR, "table#t-result-table tbody tr[id]") rows_sel = (By.CSS_SELECTOR, "table#t-result-table tbody tr[id]")
self.wait.until(EC.presence_of_all_elements_located(rows_sel)) self.wait.until(EC.presence_of_all_elements_located(rows_sel))
rows = self.driver.find_elements(*rows_sel) rows = self.driver.find_elements(*rows_sel)
logger.info(f"{len(rows)} Firmen-Zeilen gefunden.") logger.info(f"{len(rows)} Firmen-Zeilen gefunden.")
# 4) Schleife ganz ohne Sleeps oder Implicit-Waits # 4) Ohne weitere Sleeps/Implicit-Waits extrahieren
for i, row in enumerate(rows, 1): for i, row in enumerate(rows, 1):
# Name per bewährtem Selector name_elems = row.find_elements(By.CSS_SELECTOR, ".sticky-column a.t-highlight-text")
name_elems = row.find_elements(By.CSS_SELECTOR, ".sticky-column a.t-highlight-text") if not name_elems:
if not name_elems: logger.warning(f"Zeile {i}: Kein Name-Element gefunden. Überspringe.")
logger.warning(f"Zeile {i}: Kein Name-Element gefunden. Überspringe.") continue
continue ne = name_elems[0]
ne = name_elems[0] company_name = (ne.get_attribute("title") or ne.text).strip()
company_name = (ne.get_attribute("title") or ne.text).strip()
# Website per bewährtem Selector web_elems = row.find_elements(By.CSS_SELECTOR, "a.text-gray-400.t-highlight-text")
web_elems = row.find_elements(By.CSS_SELECTOR, "a.text-gray-400.t-highlight-text") website = web_elems[0].text.strip() if web_elems else ""
website = web_elems[0].text.strip() if web_elems 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.")
return results return results
except Exception as e:
logger.error(f"Schwerwiegender Fehler bei der Extraktion: {type(e).__name__}", exc_info=True)
self._save_debug_artifacts()
return []
finally: finally:
# 5) Implicit-Wait zurück auf Standard (z. B. 10 s) # 5) Implicit-Wait wieder auf Standard setzen
self.driver.implicitly_wait(10) self.driver.implicitly_wait(10)
def click_next_page(self): def click_next_page(self):