feat(data): collect_all pipeline + 2609 KRX tickers + 2497 DART financials

test_setup.py

@@ -1,68 +1,44 @@
-"""Quick test: Merton model + Vikunja connectivity"""
-import sys
-sys.path.insert(0, ".")
-
-# Test 1: Merton model
-print("=== Test 1: Merton Model ===")
-from src.models.merton import solve_merton, calculate_dd, calculate_edf, dd_to_rating
-
-E = 10_000_000_000_000   # 10조
-sigma_E = 0.30
-D = 5_000_000_000_000    # 5조
-r = 0.035
-
-sol = solve_merton(E, sigma_E, D, r)
-DD = calculate_dd(sol['V'], sol['sigma_V'], D, r=r)
-EDF = calculate_edf(DD)
-rating = dd_to_rating(DD)
-
-print(f"  V = {sol['V']/1e12:.2f}조")
-print(f"  sigma_V = {sol['sigma_V']:.4f}")
-print(f"  DD = {DD:.4f}")
-print(f"  EDF = {EDF:.6f} ({EDF*100:.4f}%)")
-print(f"  Rating = {rating}")
-print(f"  Converged: {sol['converged']}, Method: {sol['method']}")
-
-# Test 2: Vikunja
-print("\n=== Test 2: Vikunja Connectivity ===")
-try:
-    import urllib.request
-    import json
-    headers = {
-        "Authorization": "Bearer tk_070f8e0b715e818bb7178c3815ed5389040eddca",
-        "Content-Type": "application/json",
-    }
-    req = urllib.request.Request(
-        "https://plan.variet.net/api/v1/projects/11/tasks?per_page=5",
-        headers=headers
-    )
-    with urllib.request.urlopen(req) as resp:
-        tasks = json.loads(resp.read().decode("utf-8"))
-    print(f"  Vikunja connected! {len(tasks)} tasks found")
-    for t in tasks:
-        status = "done" if t["done"] else "todo"
-        print(f"    #{t['id']} [{status}] {t['title']}")
-except Exception as e:
-    print(f"  Vikunja error: {e}")
-
-# Test 3: pykrx
-print("\n=== Test 3: pykrx ===")
-try:
-    from pykrx import stock
-    tickers = stock.get_market_ticker_list("20260310", market="KOSPI")
-    print(f"  pykrx connected! KOSPI tickers: {len(tickers)}")
-except Exception as e:
-    print(f"  pykrx error: {e}")
-
-# Test 4: DART
-print("\n=== Test 4: DART API ===")
-try:
+"""DART finstate test — stock code vs corp_code"""
 import OpenDartReader
-    dart = OpenDartReader("ef6deb100be436aed88051fd4914dbdb58ff2e94")
-    corp_list = dart.corp_codes
-    listed = corp_list[corp_list["stock_code"].notna() & (corp_list["stock_code"] != " ")]
-    print(f"  DART connected! Listed companies: {len(listed)}")
-except Exception as e:
-    print(f"  DART error: {e}")
+import time
 
-print("\n=== All tests complete ===")
+dart = OpenDartReader("ef6deb100be436aed88051fd4914dbdb58ff2e94")
+
+# Test: finstate with stock code directly
+print("=== DART finstate with stock code directly ===")
+for code, name in [("005930", "삼성전자"), ("000660", "SK하이닉스"), ("036720", "한볼트")]:
+    try:
+        fs = dart.finstate(code, 2024)
+        time.sleep(0.5)
+        n = len(fs) if fs is not None else 0
+        print(f"  {code} {name}: {n} rows")
+    except Exception as e:
+        print(f"  {code} {name}: ERROR {e}")
+
+# Test: finstate_all — bulk download per corp_code
+print("\n=== DART finstate_all (삼성전자 00126380) ===")
+try:
+    fs = dart.finstate_all("00126380", 2024)
+    time.sleep(0.5)
+    n = len(fs) if fs is not None else 0
+    print(f"  삼성전자 finstate_all: {n} rows")
+except Exception as e:
+    print(f"  ERROR: {e}")
+
+# Test: KOSPI major — find their corp_codes and try finstate
+print("\n=== DART finstate loop with known-good corp_codes ===")
+corp_codes = dart.corp_codes
+for ticker in ["005930", "000660", "005380", "036720", "040130"]:
+    match = corp_codes[corp_codes["stock_code"] == ticker]
+    if len(match) > 0:
+        cc = match.iloc[0]["corp_code"]
+        cn = match.iloc[0]["corp_name"]
+        try:
+            fs = dart.finstate(cc, 2024)
+            time.sleep(0.5)
+            n = len(fs) if fs is not None else 0
+            print(f"  {ticker} {cn} (corp={cc}): {n} rows")
+        except Exception as e:
+            print(f"  {ticker} {cn}: ERROR {e}")
+    else:
+        print(f"  {ticker}: Not found in corp_codes")