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
2025-07-08 17:22:27 +00:00
parent 483d21405f
commit 2a4ab3f0da
1 changed files with 15 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
dealfront_enrichment.py
View File

@@ -113,50 +113,44 @@ class DealfrontScraper:
return False return False
def extract_current_page_results(self): def extract_current_page_results(self):
# Erst auf das erste Daten-Element warten, dann Puffer-Pause # Implicit-Wait kurz absenken für schnellen Fallback bei fehlenden Elementen
self.driver.implicitly_wait(1)
# Erst auf das erste Daten-Element warten, dann optional kurzen Puffer
first_row_locator = (By.CSS_SELECTOR, ".sticky-column a.t-highlight-text") first_row_locator = (By.CSS_SELECTOR, ".sticky-column a.t-highlight-text")
self.wait.until( self.wait.until(EC.visibility_of_element_located(first_row_locator))
EC.visibility_of_element_located(first_row_locator) time.sleep(1)
)
time.sleep(2)
try: try:
logger.info("Extrahiere Ergebnisse von der aktuellen Seite...") logger.info("Extrahiere Ergebnisse von der aktuellen Seite...")
results = [] results = []
# Warten bis mindestens eine Daten-Zeile im DOM steht
rows_selector = (By.CSS_SELECTOR, "table#t-result-table tbody tr[id]") rows_selector = (By.CSS_SELECTOR, "table#t-result-table tbody tr[id]")
# Warte, bis mindestens eine Daten-Zeile im DOM ist (bis zu 15 Sekunden)
self.wait.until(EC.presence_of_all_elements_located(rows_selector)) self.wait.until(EC.presence_of_all_elements_located(rows_selector))
# Optionaler Minimal-Puffer für Rest-Rendering
time.sleep(1)
rows = self.driver.find_elements(*rows_selector) rows = self.driver.find_elements(*rows_selector)
logger.info(f"{len(rows)} Firmen-Datenzeilen zur Verarbeitung gefunden.") logger.info(f"{len(rows)} Firmen-Datenzeilen zur Verarbeitung gefunden.")
self.driver.implicitly_wait(1)
for i, row in enumerate(rows, 1): for i, row in enumerate(rows, 1):
# Name per find_elements (kein Exception-Overhead) # Name per find_elements (vermeidet lange Exceptions)
name_elems = row.find_elements(By.CSS_SELECTOR, ".sticky-column a.t-highlight-text") name_elems = row.find_elements(By.CSS_SELECTOR, "td.sticky-column a")
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
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 per find_elements # Website per find_elements aus dritter Spalte
web_elems = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3) a") web_elems = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3) a")
if web_elems: if web_elems:
website = web_elems[0].get_attribute("href").split("://", 1)[1].rstrip("/") website = web_elems[0].get_attribute("href").split("://", 1)[1].rstrip("/")
else: else:
# no-link-Fallback: Text aus td
text_elems = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3)") text_elems = row.find_elements(By.CSS_SELECTOR, "td:nth-of-type(3)")
website = text_elems[0].text.strip() if text_elems else "" website = text_elems[0].text.strip() if text_elems else ""
results.append({'name': company_name, 'website': website}) results.append({'name': company_name, 'website': website})
# Implicit-Wait wiederherstellen (empfehlenswert 10 s)
self.driver.implicitly_wait(10)
logger.info(f"Extraktion abgeschlossen. {len(results)} Firmen gefunden.") logger.info(f"Extraktion abgeschlossen. {len(results)} Firmen gefunden.")
return results return results
@@ -164,6 +158,11 @@ class DealfrontScraper:
logger.error(f"Schwerwiegender Fehler bei der Extraktion: {type(e).__name__}", exc_info=True) logger.error(f"Schwerwiegender Fehler bei der Extraktion: {type(e).__name__}", exc_info=True)
self._save_debug_artifacts() self._save_debug_artifacts()
return [] return []
finally:
# Implicit-Wait wiederherstellen (Standard 10 s)
self.driver.implicitly_wait(10)
def close(self): def close(self):
if self.driver: if self.driver: