init update

2026-04-07 03:07:24 -04:00 · 2025-10-24 00:35:21 +08:00
commit df5c25c98d
205 changed files with 81998 additions and 0 deletions
--- a/data/merge_jsonl.py
+++ b/data/merge_jsonl.py
@@ -0,0 +1,62 @@
+import json
+import os
+import glob
+
+
+all_nasdaq_100_symbols = [
+    "NVDA", "MSFT", "AAPL", "GOOG", "GOOGL", "AMZN", "META", "AVGO", "TSLA",
+    "NFLX", "PLTR", "COST", "ASML", "AMD", "CSCO", "AZN", "TMUS", "MU", "LIN",
+    "PEP", "SHOP", "APP", "INTU", "AMAT", "LRCX", "PDD", "QCOM", "ARM", "INTC",
+    "BKNG", "AMGN", "TXN", "ISRG", "GILD", "KLAC", "PANW", "ADBE", "HON",
+    "CRWD", "CEG", "ADI", "ADP", "DASH", "CMCSA", "VRTX", "MELI", "SBUX",
+    "CDNS", "ORLY", "SNPS", "MSTR", "MDLZ", "ABNB", "MRVL", "CTAS", "TRI",
+    "MAR", "MNST", "CSX", "ADSK", "PYPL", "FTNT", "AEP", "WDAY", "REGN", "ROP",
+    "NXPI", "DDOG", "AXON", "ROST", "IDXX", "EA", "PCAR", "FAST", "EXC", "TTWO",
+    "XEL", "ZS", "PAYX", "WBD", "BKR", "CPRT", "CCEP", "FANG", "TEAM", "CHTR",
+    "KDP", "MCHP", "GEHC", "VRSK", "CTSH", "CSGP", "KHC", "ODFL", "DXCM", "TTD",
+    "ON", "BIIB", "LULU", "CDW", "GFS"
+]
+
+# 合并所有以 daily_price 开头的 json，逐文件一行写入 merged.jsonl
+current_dir = os.path.dirname(__file__)
+pattern = os.path.join(current_dir, 'daily_price*.json')
+files = sorted(glob.glob(pattern))
+
+output_file = os.path.join(current_dir, 'merged.jsonl')
+
+with open(output_file, 'w', encoding='utf-8') as fout:
+    for fp in files:
+        basename = os.path.basename(fp)
+        # 仅当文件名包含任一纳指100成分符号时才写入
+        if not any(symbol in basename for symbol in all_nasdaq_100_symbols):
+            continue
+        with open(fp, 'r', encoding='utf-8') as f:
+            data = json.load(f)
+        # 统一重命名："1. open" -> "1. buy price"；"4. close" -> "4. sell price"
+        # 对于最新的一天，只保留并写入 "1. buy price"
+        try:
+            series = data.get("Time Series (Daily)", {})
+            if isinstance(series, dict) and series:
+                # 先对所有日期做键名重命名
+                for d, bar in list(series.items()):
+                    if not isinstance(bar, dict):
+                        continue
+                    if "1. open" in bar:
+                        bar["1. buy price"] = bar.pop("1. open")
+                    if "4. close" in bar:
+                        bar["4. sell price"] = bar.pop("4. close")
+                # 再处理最新日期，仅保留买入价
+                latest_date = max(series.keys())
+                latest_bar = series.get(latest_date, {})
+                if isinstance(latest_bar, dict):
+                    buy_val = latest_bar.get("1. buy price")
+                    series[latest_date] = {"1. buy price": buy_val} if buy_val is not None else {}
+                # 更新 Meta Data 描述
+                meta = data.get("Meta Data", {})
+                if isinstance(meta, dict):
+                    meta["1. Information"] = "Daily Prices (buy price, high, low, sell price) and Volumes"
+        except Exception:
+            # 若结构异常则原样写入
+            pass
+
+        fout.write(json.dumps(data, ensure_ascii=False) + "\n")