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

[2ff88f42] feat(GTM-Engine): Add final test and trigger scripts

Browse Source
This commit is contained in:
Floke
2026-02-20 16:41:56 +00:00
parent e824a49450
commit 91ea02820b
5 changed files with 135 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
company-explorer/backend/scripts/inspect_sqlite_native.py Normal file
View File

@@ -0,0 +1,58 @@
import sqlite3
import json
DB_PATH = "/app/companies_v3_fixed_2.db"
def inspect(name_part):
try:
conn = sqlite3.connect(DB_PATH)
cursor = conn.cursor()
print(f"Searching for '{name_part}' in {DB_PATH}...")
cursor.execute("SELECT id, name, website, industry_ai, calculated_metric_value, standardized_metric_value, ai_opener, ai_opener_secondary FROM companies WHERE name LIKE ?", (f'%{name_part}%',))
companies = cursor.fetchall()
if not companies:
print("No hits.")
return
for c in companies:
cid, name, website, industry, metric, std_metric, opener_primary, opener_secondary = c
print("\n" + "="*40)
print(f"🏢 {name} (ID: {cid})")
print(f" Vertical: {industry}")
print(f" Website: {website}")
print(f" Metric: {metric} (Std: {std_metric})")
print(f" Opener (Primary): {opener_primary}")
print(f" Opener (Secondary): {opener_secondary}")
# Fetch Enrichment Data
cursor.execute("SELECT source_type, content FROM enrichment_data WHERE company_id = ?", (cid,))
rows = cursor.fetchall()
print("\n 📚 Enrichment Data:")
for r in rows:
stype, content_raw = r
print(f" - {stype}")
try:
content = json.loads(content_raw)
if stype == "website_scrape":
summary = content.get("summary", "")
raw = content.get("text", "")
print(f" > Summary: {summary[:150]}...")
print(f" > Raw Length: {len(raw)}")
if len(raw) > 500:
print(f" > Raw Snippet: {raw[:300]}...")
elif stype == "wikipedia":
print(f" > URL: {content.get('url')}")
intro = content.get("intro_text", "") or content.get("full_text", "")
print(f" > Intro: {str(intro)[:150]}...")
except:
print(" > (Content not valid JSON)")
except Exception as e:
print(f"Error: {e}")
finally:
if conn: conn.close()
if __name__ == "__main__":
inspect("Therme Erding")

32
company-explorer/backend/services/classification.py
View File

@@ -294,15 +294,18 @@ Gib NUR den finalen Satz aus. Keine Anführungszeichen.
# 1. Load Definitions
industries = self._load_industry_definitions(db)
industry_defs = [{"name": i.name, "description": i.description} for i in industries]
logger.debug(f"Loaded {len(industries)} industry definitions.")
# 2. Get Content (Website)
website_content, _ = self._get_website_content_and_url(company)
if not website_content:
logger.warning(f"No website content for {company.name}. Skipping classification.")
if not website_content or len(website_content) < 100:
logger.warning(f"No or insufficient website content for {company.name} (Length: {len(website_content) if website_content else 0}). Skipping classification.")
return company
logger.debug(f"Website content length for classification: {len(website_content)}")
# 3. Classify Industry
logger.info(f"Running LLM classification prompt for {company.name}...")
suggested_industry_name = self._run_llm_classification_prompt(website_content, company.name, industry_defs)
logger.info(f"AI suggests industry: {suggested_industry_name}")
@@ -311,32 +314,47 @@ Gib NUR den finalen Satz aus. Keine Anführungszeichen.
if matched_industry:
company.industry_ai = matched_industry.name
logger.info(f"Matched company to industry: {matched_industry.name}")
# --- Generate PRIMARY Opener (Infrastructure/Cleaning) ---
logger.info(f"Generating PRIMARY opener for {company.name}...")
op_prim = self._generate_marketing_opener(
company.name, website_content, matched_industry.name, matched_industry.pains, "primary"
)
if op_prim:
company.ai_opener = op_prim
logger.info(f"Opener (Primary): {op_prim}")
logger.info(f"Opener (Primary) generated and set.")
else:
logger.warning(f"Failed to generate PRIMARY opener for {company.name}.")
# --- Generate SECONDARY Opener (Service/Logistics) ---
# Only if relevant (could be optimized, but generating always is safer for "Dual Strategy")
logger.info(f"Generating SECONDARY opener for {company.name}...")
op_sec = self._generate_marketing_opener(
company.name, website_content, matched_industry.name, matched_industry.pains, "secondary"
)
if op_sec:
company.ai_opener_secondary = op_sec
logger.info(f"Opener (Secondary): {op_sec}")
logger.info(f"Opener (Secondary) generated and set.")
else:
logger.warning(f"Failed to generate SECONDARY opener for {company.name}.")
else:
company.industry_ai = "Others"
logger.warning(f"No specific industry matched for {company.name}. Set to 'Others'.")
# 5. Extract Metrics (Cascade)
if matched_industry:
self.extract_metrics_for_industry(company, db, matched_industry)
logger.info(f"Extracting metrics for {company.name} and industry {matched_industry.name}...")
try:
self.extract_metrics_for_industry(company, db, matched_industry)
logger.info(f"Metric extraction completed for {company.name}.")
except Exception as e:
logger.error(f"Error during metric extraction for {company.name}: {e}", exc_info=True)
else:
logger.warning(f"Skipping metric extraction for {company.name} as no specific industry was matched.")
company.last_classification_at = datetime.utcnow()
db.commit()
logger.info(f"Classification and enrichment for {company.name} completed and committed.")
return company