Anpassung Verify Wiki
This commit is contained in:
@@ -373,73 +373,81 @@ class DataProcessor:
|
||||
# ======================================================================
|
||||
if run_wiki_step and wiki_processing_needed:
|
||||
any_processing_done = True
|
||||
grund_message_wiki = "Re-Eval" if force_reeval else "Timestamp (Z) leer"
|
||||
self.logger.info(
|
||||
f"Zeile {row_num_in_sheet}: Fuehre WIKI Schritte aus (Grund: {grund_message_wiki})...")
|
||||
|
||||
wiki_data_updated_in_this_run = False
|
||||
self.logger.info(f"Zeile {row_num_in_sheet}: Starte Wikipedia-Pipeline...")
|
||||
|
||||
# --- Stufe 1: Intelligente URL-Suche (Kaskade) ---
|
||||
url_to_process = None
|
||||
crm_vorschlag_url_n = self._get_cell_value_safe(row_data, "CRM Vorschlag Wiki URL").strip()
|
||||
current_wiki_url_r = self._get_cell_value_safe(row_data, "Wiki URL").strip()
|
||||
search_origin = "Nicht festgelegt"
|
||||
parent_name_o = self._get_cell_value_safe(row_data, "System Vorschlag Parent Account").strip()
|
||||
parent_name_for_search = parent_name_o if parent_name_o and parent_name_o.lower() != 'k.a.' else None
|
||||
|
||||
# Priorität 1: URL aus Spalte N (CRM Vorschlag)
|
||||
if crm_vorschlag_url_n and "wikipedia.org" in crm_vorschlag_url_n.lower():
|
||||
self.logger.debug(f" -> Nutze URL aus Spalte N (CRM Vorschlag): {crm_vorschlag_url_n}")
|
||||
url_to_process = crm_vorschlag_url_n
|
||||
# Priorität 2: Bereits existierende URL in Spalte R
|
||||
elif current_wiki_url_r and "wikipedia.org" in current_wiki_url_r.lower():
|
||||
self.logger.debug(f" -> Nutze bestehende URL aus Spalte R: {current_wiki_url_r}")
|
||||
url_to_process = current_wiki_url_r
|
||||
# Priorität 3: Wenn beides leer, suche eine neue URL
|
||||
else:
|
||||
self.logger.debug(f" -> Spalten N und R sind leer/ungültig. Starte neue Wiki-Suche...")
|
||||
# Prio 1: Manuelle URLs
|
||||
for col_name, origin in [("CRM Vorschlag Wiki URL", "Spalte N"), ("Wiki URL", "Spalte R")]:
|
||||
url = self._get_cell_value_safe(row_data, col_name).strip()
|
||||
if url and "wikipedia.org" in url.lower():
|
||||
self.logger.debug(f" -> Stufe 1: Nutze URL aus {origin}: {url}")
|
||||
url_to_process = url
|
||||
search_origin = origin
|
||||
break
|
||||
|
||||
# Prio 2 & 3: Automatische Suche
|
||||
if not url_to_process:
|
||||
self.logger.debug(f" -> Stufe 1: Starte automatische Suche für '{company_name}' (Parent-Kontext: {parent_name_for_search})...")
|
||||
try:
|
||||
search_name = company_name
|
||||
if parent_account_name_d and parent_account_name_d.lower() != 'k.A.':
|
||||
search_name = parent_account_name_d
|
||||
self.logger.debug(f" -> Suche für Parent Account (D): '{search_name}'")
|
||||
|
||||
page_obj = self.wiki_scraper.search_company_article(search_name, website_url)
|
||||
page_obj = self.wiki_scraper.search_company_article(company_name, website_url, parent_name_for_search)
|
||||
if page_obj:
|
||||
url_to_process = page_obj.url
|
||||
self.logger.info(f" -> Neue URL für '{search_name}' gefunden: {url_to_process}")
|
||||
else:
|
||||
self.logger.warning(f" -> Kein passender Artikel für '{search_name}' gefunden.")
|
||||
url_to_process = "Kein Artikel gefunden"
|
||||
except Exception as e_wiki_search:
|
||||
self.logger.error(f" -> FEHLER bei der Wiki-Suche: {e_wiki_search}")
|
||||
url_to_process = "Fehler bei Suche"
|
||||
|
||||
# --- Datenextraktion & Update ---
|
||||
search_origin = "Automatische Suche"
|
||||
self.logger.info(f" -> Stufe 1: URL durch automatische Suche gefunden: {url_to_process}")
|
||||
except Exception as e:
|
||||
self.logger.error(f" -> Stufe 1: Fehler bei der automatischen Suche: {e}")
|
||||
|
||||
# --- Stufe 2: Inhaltsextraktion & Aufbereitung ---
|
||||
if url_to_process and "wikipedia.org" in url_to_process.lower():
|
||||
self.logger.info(f" -> Extrahiere Daten von URL: {url_to_process[:100]}...")
|
||||
self.logger.debug(f" -> Stufe 2: Extrahiere Inhalte von '{url_to_process}'...")
|
||||
try:
|
||||
extracted_data = self.wiki_scraper.extract_company_data(url_to_process)
|
||||
final_wiki_data = extracted_data
|
||||
extracted_content = self.wiki_scraper.extract_company_data(url_to_process)
|
||||
|
||||
# Neue, hochwertige Zusammenfassung erstellen
|
||||
self.logger.debug(" -> Erstelle KI-Zusammenfassung des Artikel-Rohtextes...")
|
||||
new_summary = summarize_wikipedia_article(extracted_content.get('full_text'), company_name)
|
||||
extracted_content['first_paragraph'] = new_summary # Überschreibe den alten Absatz
|
||||
|
||||
final_wiki_data.update(extracted_content)
|
||||
wiki_data_updated_in_this_run = True
|
||||
except Exception as e_extract:
|
||||
self.logger.error(f" -> FEHLER bei Datenextraktion: {e_extract}")
|
||||
final_wiki_data = {key: 'k.A. (FEHLER Extr.)' for key in final_wiki_data}
|
||||
|
||||
except Exception as e:
|
||||
self.logger.error(f" -> Stufe 2: Fehler bei der Inhaltsextraktion: {e}")
|
||||
final_wiki_data.update({key: 'FEHLER (Extraktion)' for key, val in final_wiki_data.items() if key != 'url'})
|
||||
final_wiki_data['url'] = url_to_process
|
||||
wiki_data_updated_in_this_run = True
|
||||
else: # Fall: "Kein Artikel gefunden" oder "Fehler bei Suche"
|
||||
final_wiki_data['url'] = url_to_process if url_to_process else "k.A."
|
||||
else:
|
||||
self.logger.warning(f" -> Stufe 1-2: Keine gültige URL gefunden oder Prozess übersprungen. Setze Wiki-Daten auf 'Kein Artikel gefunden'.")
|
||||
final_wiki_data.update({key: 'Kein Artikel gefunden' for key, val in final_wiki_data.items()})
|
||||
wiki_data_updated_in_this_run = True
|
||||
|
||||
# Schreibe die finalen Wiki-Daten immer, um den Zustand zu aktualisieren
|
||||
# KORREKTE EINRÜCKUNG BEGINNT HIER
|
||||
# --- Stufe 3: Kontextbasierte KI-Verifikation ---
|
||||
verification_needed = "Suche" in search_origin # Nur verifizieren, wenn URL automatisch gesucht wurde
|
||||
if verification_needed and wiki_data_updated_in_this_run and "FEHLER" not in str(final_wiki_data.get('title')):
|
||||
self.logger.debug(" -> Stufe 3: Automatisch gefundene URL erfordert KI-Verifizierung...")
|
||||
verification_result = verify_wiki_article_chatgpt(
|
||||
company_name=company_name,
|
||||
parent_name=parent_name_for_search,
|
||||
website=website_url,
|
||||
wiki_title=final_wiki_data.get('title', 'k.A.'),
|
||||
wiki_summary=final_wiki_data.get('first_paragraph', 'k.A.')
|
||||
)
|
||||
updates.append({'range': f'{self.sheet_handler._get_col_letter(COLUMN_MAP["Chat Wiki Konsistenzpruefung"] + 1)}{row_num_in_sheet}', 'values': [[verification_result.get('consistency', 'X')]]})
|
||||
updates.append({'range': f'{self.sheet_handler._get_col_letter(COLUMN_MAP["Chat Begruendung Wiki Inkonsistenz"] + 1)}{row_num_in_sheet}', 'values': [[verification_result.get('justification', 'Fehler')]]})
|
||||
updates.append({'range': f'{self.sheet_handler._get_col_letter(COLUMN_MAP["Wiki Verif. Timestamp"] + 1)}{row_num_in_sheet}', 'values': [[now_timestamp]]})
|
||||
|
||||
# --- Finales Schreiben ins Sheet ---
|
||||
if wiki_data_updated_in_this_run:
|
||||
# Update für Spalte R (Wiki URL)
|
||||
updates.append({'range': f'{self.sheet_handler._get_col_letter(COLUMN_MAP["Wiki URL"] + 1)}{row_num_in_sheet}', 'values': [[final_wiki_data.get('url', 'k.A.')]]})
|
||||
|
||||
# Update für die restlichen Datenfelder mit korrektem Mapping
|
||||
key_mapping = {
|
||||
'Wiki Sitz Stadt': 'sitz_stadt',
|
||||
'Wiki Sitz Land': 'sitz_land',
|
||||
'Wiki Absatz': 'first_paragraph',
|
||||
'Wiki Branche': 'branche',
|
||||
'Wiki Umsatz': 'umsatz',
|
||||
'Wiki Mitarbeiter': 'mitarbeiter',
|
||||
'Wiki Kategorien': 'categories'
|
||||
'Wiki URL': 'url', 'Wiki Sitz Stadt': 'sitz_stadt', 'Wiki Sitz Land': 'sitz_land',
|
||||
'Wiki Absatz': 'first_paragraph', 'Wiki Branche': 'branche', 'Wiki Umsatz': 'umsatz',
|
||||
'Wiki Mitarbeiter': 'mitarbeiter', 'Wiki Kategorien': 'categories'
|
||||
}
|
||||
for sheet_col_name, data_key in key_mapping.items():
|
||||
updates.append({
|
||||
@@ -447,7 +455,6 @@ class DataProcessor:
|
||||
'values': [[final_wiki_data.get(data_key, 'k.A.')]]
|
||||
})
|
||||
|
||||
# Setze IMMER den Timestamp
|
||||
updates.append({
|
||||
'range': f'{self.sheet_handler._get_col_letter(COLUMN_MAP["Wikipedia Timestamp"] + 1)}{row_num_in_sheet}',
|
||||
'values': [[now_timestamp]]
|
||||
|
||||
Reference in New Issue
Block a user