quant1x 0.7.2__tar.gz → 0.7.5__tar.gz

{quant1x-0.7.2 → quant1x-0.7.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: quant1x
-Version: 0.7.2
+Version: 0.7.5
 Summary: Quant1X量化研究
 Home-page: https://gitee.com/quant1x/quant1x
 Author: wangfeng

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/cache/__init__.py

@@ -5,3 +5,5 @@ from .security import securities, stock_name
 from .market import klines, get_period_name, convert_klines_trading, get_minutes_data, get_tick_transaction, date_format
 from .sector import block_list, sector_filename, get_sector_list, get_sector_constituents
 from .f10 import cache_f10, get_f10
+
+trains_begin_date = "2024-10-01"

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/cache/f10.py

@@ -8,12 +8,24 @@ from .. import exchange, config
 
 @lru_cache(maxsize=None)
 def cache_f10(date: str = None) -> DataFrame:
+    """
+    从闪存数据中读取F10因子数据
+    
+    Args:
+        date (str, optional): 交易日期，格式为'YYYYMMDD'。如果未提供，则使用交易所最后一个交易日
+    
+    Returns:
+        DataFrame: 包含F10因子数据的Pandas DataFrame
+    
+    Raises:
+        FileNotFoundError: 如果指定日期的数据文件不存在
+    """
     factor_name = 'f10'
     trade_date = date or exchange.last_trade_date()
     file_extension = exchange.fix_trade_date(trade_date)
     filename = f"{factor_name}.{file_extension}"
     year = trade_date[:4]
-    base_path = os.path.join(config.quant1x_config.data_path, 'flash')
+    base_path = os.path.join(config.data_path, 'flash')
     return pd.read_csv(os.path.join(base_path, year, filename))
 
 def get_f10(code: str, date: str = None) -> Dict[str, Any]:

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/cache/market.py

@@ -14,7 +14,7 @@ def klines(code: str, freq:str = 'D') -> pd.DataFrame | None:
     corrected_symbol = exchange.correct_security_code(code)
     suffix_length = 3  # 修正拼写并明确表示后缀长度
     freq_path = 'day' if freq == 'D' else freq
-    symbol_directory = os.path.join(config.quant1x_config.data_path, freq_path, corrected_symbol[:-suffix_length])  # 更清晰表达目录用途
+    symbol_directory = os.path.join(config.data_path, freq_path, corrected_symbol[:-suffix_length])  # 更清晰表达目录用途
     file_extension = '.csv'
     filename = f"{corrected_symbol}{file_extension}"  # 使用f-string格式化
     full_path = os.path.join(symbol_directory, filename)
@@ -106,7 +106,7 @@ def get_minutes_data(code: str, date: str) -> DataFrame | None:
     # 获取年份
     cache_date = date_format(cache_date, layout='%Y%m%d')
     year = cache_date[:4]
-    base_path = os.path.join(config.quant1x_config.data_path, 'minutes')
+    base_path = os.path.join(config.data_path, 'minutes')
     full_path = os.path.join(base_path, year, cache_date, filename)
 
     if os.path.isfile(full_path):
@@ -126,7 +126,7 @@ def get_tick_transaction(code: str, date: str) -> DataFrame | None:
     # 获取年份
     cache_date = date_format(cache_date, layout='%Y%m%d')
     year = cache_date[:4]
-    base_path = os.path.join(config.quant1x_config.data_path, 'trans')
+    base_path = os.path.join(config.data_path, 'trans')
     full_path = os.path.join(base_path, year, cache_date, filename)
 
     if os.path.isfile(full_path):

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/cache/sector.py

@@ -15,10 +15,10 @@ def sector_filename(date: str = '') -> str:
     cache_date = date.strip()
     if len(cache_date) == 0:
         cache_date = exchange.last_trade_date()
-    filename = os.path.join(config.quant1x_config.meta_path, f'{name}.{cache_date}')
+    filename = os.path.join(config.meta_path, f'{name}.{cache_date}')
     if not os.path.isfile(filename):
         # fallback to blocks.csv
-        filename = os.path.join(config.quant1x_config.meta_path, 'blocks.csv')
+        filename = os.path.join(config.meta_path, 'blocks.csv')
     return filename
 
 @lru_cache(maxsize=None)

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/cache/security.py

@@ -9,7 +9,7 @@ def securities() -> pd.DataFrame:
     """
     证券列表
     """
-    full_path = os.path.join(config.quant1x_config.meta_path, 'securities.csv')
+    full_path = os.path.join(config.meta_path, 'securities.csv')
     if not os.path.isfile(full_path):
         return pd.DataFrame(columns=['code', 'name'])
     # with open(full_path, 'rb') as f:

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/config/__init__.py

@@ -10,10 +10,19 @@
 """
 from .config import (
     get_quant1x_config_filename,
+    get_historical_trade_filename,
     quant1x_config
 )
 
+# 导出常用配置路径
+data_path = quant1x_config.data_path
+meta_path = quant1x_config.meta_path
+kline_path = quant1x_config.kline_path
+
 __all__ = [
-    "quant1x_config",
     "get_quant1x_config_filename",
+    "get_historical_trade_filename",
+    "data_path",
+    "meta_path",
+    "kline_path",
 ]

{quant1x-0.7.2 → quant1x-0.7.5}/quant1x/config/config.py

@@ -117,3 +117,83 @@ class Quant1XConfig:
 
 # 创建配置单例
 quant1x_config = Quant1XConfig()
+
+def get_historical_trade_filename(code: str, date: str) -> str:
+    """
+    获取历史成交记录文件路径
+    目录结构: ${trans}/${YYYY}/${YYYYMMDD}/${SecurityCode}.csv
+    """
+    date_str = date.replace('-', '').replace('/', '')
+    year = date_str[:4]
+    base_path = os.path.join(quant1x_config.data_path, 'trans')
+    return os.path.join(base_path, year, date_str, f"{code}.csv")
+
+def get_xdxr_path() -> str:
+    """获取除权除息文件路径"""
+    return os.path.join(quant1x_config.data_path, 'xdxr')
+
+def get_xdxr_filename(code: str) -> str:
+    """
+    获取除权除息文件路径
+    目录结构: ${xdxr}/${subpath}/${SecurityCode}.csv
+    subpath: code[:-3]
+    """
+    if len(code) <= 3:
+        sub = ""
+    else:
+        sub = code[:-3]
+        
+    base_path = get_xdxr_path()
+    return os.path.join(base_path, sub, f"{code}.csv")
+
+def get_day_path() -> str:
+    """获取日K线文件路径"""
+    return quant1x_config.kline_path
+
+def get_kline_filename(code: str, forward: bool = True) -> str:
+    """
+    获取日K线文件路径
+    目录结构: ${day}/${subpath}/${SecurityCode}.${ext}
+    subpath: code[:-3]
+    ext: csv if forward else raw
+    """
+    if len(code) <= 3:
+        sub = ""
+    else:
+        sub = code[:-3]
+        
+    base_path = get_day_path()
+    ext = "csv" if forward else "raw"
+    return os.path.join(base_path, sub, f"{code}.{ext}")
+
+from quant1x import std
+
+def get_holding_path() -> str:
+    return os.path.join(quant1x_config.data_path, "holding")
+
+def cache_id_path(code: str) -> str:
+    """
+    Generate cache ID path.
+    SH600000 -> SH600/SH600000
+    """
+    if len(code) <= 3:
+        return code
+    
+    prefix = code[:-3]
+    return os.path.join(prefix, code)
+
+def top10_holders_filename(code: str, date_str: str) -> str:
+    quarter, _, _ = std.get_quarter_by_date(date_str)
+    id_path = cache_id_path(code)
+    return os.path.join(get_holding_path(), quarter, f"{id_path}.csv")
+
+def quarterly_cache_path(date: str) -> str:
+    quarter, _, _ = std.get_quarter_by_date(date)
+    return os.path.join(quant1x_config.data_path, "infoq", quarter)
+
+def quarterly_filename(date: str, keyword: str) -> str:
+    return os.path.join(quarterly_cache_path(date), f"{keyword}.csv")
+
+def reports_filename(date: str) -> str:
+    return quarterly_filename(date, "reports")
+

quant1x-0.7.5/quant1x/datasets/__init__.py

@@ -0,0 +1,22 @@
+
+from .trans import (
+    TickTransaction,
+    TurnoverDataSummary,
+    checkout_transaction_data,
+    count_inflow,
+    DataTrans
+)
+
+from .xdxr import (
+    load_xdxr,
+    save_xdxr,
+    DataXdxr
+)
+
+from .kline import (
+    KLine,
+    load_kline,
+    save_kline,
+    fetch_kline,
+    DataKLine
+)

quant1x-0.7.5/quant1x/datasets/kline.py

@@ -0,0 +1,286 @@
+# -*- coding: UTF-8 -*-
+import os
+import pandas as pd
+from dataclasses import dataclass, field
+from typing import List, Optional, Any
+import logging
+
+from quant1x.exchange import Timestamp
+from quant1x.level1 import protocol, client
+from quant1x.level1.security_bars import SecurityBarsRequest, SecurityBarsResponse, KLineType, SecurityBar, SECURITY_BARS_MAX
+from quant1x.factors import base as factors
+from quant1x.level1.xdxr_info import XdxrInfo
+from quant1x.datasets import xdxr
+from quant1x.config import config
+
+logger = logging.getLogger(__name__)
+
+MAX_KLINE_LOOKBACK_DAYS = 1
+
+@dataclass
+class KLine:
+    Date: str = ""
+    Open: float = 0.0
+    Close: float = 0.0
+    High: float = 0.0
+    Low: float = 0.0
+    Volume: float = 0.0
+    Amount: float = 0.0
+    Up: int = 0
+    Down: int = 0
+    Datetime: str = ""
+    AdjustmentCount: int = 0
+
+    def adjust(self, adj: factors.CumulativeAdjustment):
+        self.Open = self.Open * adj.m + adj.a
+        self.Close = self.Close * adj.m + adj.a
+        self.High = self.High * adj.m + adj.a
+        self.Low = self.Low * adj.m + adj.a
+        
+        # 成交量复权
+        if self.Volume != 0:
+            # 1. 计算均价
+            ap = self.Amount / self.Volume
+            # 2. 均价复权
+            ap_adjusted = ap * adj.m + adj.a
+            # 3. 成交量复权
+            self.Volume *= (1 + adj.share_adjustment_ratio)
+            # 4. 以新成交量*均价计算成交额
+            self.Amount = self.Volume * ap_adjusted
+        
+        # 5. 更新除权除息次数
+        self.AdjustmentCount = adj.no
+
+    @staticmethod
+    def headers() -> List[str]:
+        return ["Date", "Open", "Close", "High", "Low", "Volume", "Amount", "Up", "Down", "Datetime", "AdjustmentCount"]
+
+def save_kline(filename: str, values: List[KLine]):
+    if not values:
+        return
+        
+    dirname = os.path.dirname(filename)
+    if dirname:
+        os.makedirs(dirname, exist_ok=True)
+    
+    data = [
+        {
+            "Date": v.Date,
+            "Open": v.Open,
+            "Close": v.Close,
+            "High": v.High,
+            "Low": v.Low,
+            "Volume": v.Volume,
+            "Amount": v.Amount,
+            "Up": v.Up,
+            "Down": v.Down,
+            "Datetime": v.Datetime,
+            "AdjustmentCount": v.AdjustmentCount
+        }
+        for v in values
+    ]
+    
+    df = pd.DataFrame(data, columns=KLine.headers())
+    df.to_csv(filename, index=False)
+
+def read_kline_from_csv(filename: str) -> List[KLine]:
+    klines = []
+    if not os.path.exists(filename):
+        return klines
+        
+    try:
+        df = pd.read_csv(filename)
+        # Ensure columns exist
+        required_cols = KLine.headers()
+        if not all(col in df.columns for col in required_cols):
+            return klines
+            
+        for _, row in df.iterrows():
+            kline = KLine(
+                Date=str(row['Date']),
+                Open=float(row['Open']),
+                Close=float(row['Close']),
+                High=float(row['High']),
+                Low=float(row['Low']),
+                Volume=float(row['Volume']),
+                Amount=float(row['Amount']),
+                Up=int(row['Up']),
+                Down=int(row['Down']),
+                Datetime=str(row['Datetime']),
+                AdjustmentCount=int(row['AdjustmentCount'])
+            )
+            klines.append(kline)
+    except Exception as e:
+        logger.error(f"Failed to read kline csv {filename}: {e}")
+        
+    return klines
+
+def load_kline(code: str) -> List[KLine]:
+    filename = config.get_kline_filename(code)
+    logger.debug(f"[dataset::KLine] kline file: {filename}")
+    return read_kline_from_csv(filename)
+
+def fetch_kline(code: str, start: int, count: int, kline_type: KLineType = KLineType.DAILY) -> List[SecurityBar]:
+    try:
+        with client.client() as conn:
+            req = SecurityBarsRequest(code, kline_type.value, start, count)
+            # Determine is_index from request logic or pass it explicitly if needed
+            # SecurityBarsRequest constructor determines is_index
+            
+            resp = SecurityBarsResponse(req.is_index, kline_type.value)
+            protocol.process(conn.socket, req, resp)
+            return resp.list
+    except Exception as e:
+        logger.error(f"[dataset::KLine] fetch_kline error: {e}")
+        return []
+
+def apply_forward_adjustment_for_event(klines: List[KLine], 
+                                       current_start_date: Timestamp, 
+                                       dividends: List[XdxrInfo]):
+    if not klines:
+        return
+        
+    # 最后一根K线的日期
+    last_day = klines[-1].Date
+    # 转成时间戳且对齐时间
+    ts_last_day = Timestamp.parse(last_day).get_pre_market_time()
+    # 计算最后一根K线的下一个交易日的日期
+    # TODO: Implement next_trading_day properly. For now, use +1 day approximation or just rely on date comparison
+    last_day_next = ts_last_day.offset(hour=24).only_date() # Approximation
+    start_date_str = current_start_date.only_date()
+    
+    # Filter dividends
+    xdxr_infos = [
+        x for x in dividends 
+        if x.Date <= last_day_next and x.Category == 1
+    ]
+    
+    # Sort by date? C++ uses std::views::filter, order depends on input.
+    # Assuming dividends are sorted by date.
+    
+    times = len(xdxr_infos)
+    
+    for info in xdxr_infos:
+        if info.Date <= start_date_str:
+            # IPO check logic in C++ is commented out or specific?
+            # "除权除息数据在日线第一条数据之前... continue" is commented out in C++.
+            pass
+        else:
+            m, a = info.adjust_factor()
+            share_ratio = info.compute_share_adjustment_ratio()
+            
+            for kline in klines:
+                if kline.Date >= info.Date:
+                    break
+                
+                if kline.Date < info.Date:
+                    kline.Open = kline.Open * m + a
+                    kline.Close = kline.Close * m + a
+                    kline.High = kline.High * m + a
+                    kline.Low = kline.Low * m + a
+                    
+                    if kline.Volume != 0:
+                        ap = kline.Amount / kline.Volume
+                        ap_adjusted = ap * m + a
+                        kline.Volume *= (1 + share_ratio)
+                        kline.Amount = kline.Volume * ap_adjusted
+                    
+                    kline.AdjustmentCount += 1
+        
+        times -= 1
+
+class DataKLine:
+    def update(self, code: str, date: Timestamp):
+        # 1. Determine start date from local cache
+        current_start_date = Timestamp.parse("1990-12-19") # market_first_date
+        cache_filename = config.get_kline_filename(code)
+        cache_klines = read_kline_from_csv(cache_filename)
+        
+        klines_length = len(cache_klines)
+        klines_offset_days = MAX_KLINE_LOOKBACK_DAYS
+        adjust_times = 0
+        
+        if klines_length > 0:
+            if klines_offset_days > klines_length:
+                klines_offset_days = klines_length
+            
+            kline = cache_klines[klines_length - klines_offset_days]
+            current_start_date = Timestamp.parse(kline.Date)
+            adjust_times = kline.AdjustmentCount
+            
+        # 2. Determine end date
+        current_end_date = Timestamp.now().get_pre_market_time()
+        logger.debug(f"[dataset::KLine] [{code}]: from {current_start_date.only_date()} to {current_end_date.only_date()}")
+        
+        step = SECURITY_BARS_MAX
+        start = 0
+        hs: List[List[SecurityBar]] = []
+        element_count = 0
+        
+        while True:
+            count = step
+            reply = fetch_kline(code, start, count)
+            if not reply:
+                break
+                
+            element_count += len(reply)
+            hs.append(reply)
+            
+            last_bar = reply[-1]
+            last_bar_date = Timestamp.parse(f"{last_bar.Year}-{last_bar.Month:02d}-{last_bar.Day:02d}").get_pre_market_time()
+            
+            if last_bar_date < current_start_date:
+                break
+                
+            if len(reply) < count:
+                break
+                
+            start += count
+            
+        hs.reverse()
+        
+        incremental_klines: List[KLine] = []
+        
+        for vec in hs:
+            for row in vec:
+                date_time = Timestamp.parse(f"{row.Year}-{row.Month:02d}-{row.Day:02d}").get_pre_market_time()
+                
+                if date_time < current_start_date or date_time > current_end_date:
+                    continue
+                    
+                kx = KLine(
+                    Date=date_time.only_date(),
+                    Open=row.Open,
+                    Close=row.Close,
+                    High=row.High,
+                    Low=row.Low,
+                    Volume=row.Vol * 100, # Convert to shares
+                    Amount=row.Amount,
+                    Up=row.UpCount,
+                    Down=row.DownCount,
+                    Datetime=row.DateTime,
+                    AdjustmentCount=0
+                )
+                incremental_klines.append(kx)
+                
+        # 6. Adjustment logic
+        is_fresh_fetch_require_adjustment = (adjust_times == 1)
+        dividends = xdxr.load_xdxr(code)
+        
+        if is_fresh_fetch_require_adjustment:
+            apply_forward_adjustment_for_event(incremental_klines, current_start_date, dividends)
+            
+        # 7. Merge
+        klines = []
+        if klines_length > klines_offset_days:
+            klines.extend(cache_klines[:klines_length - klines_offset_days])
+            
+        klines.extend(incremental_klines)
+        
+        # 8. Forward adjust
+        if not is_fresh_fetch_require_adjustment:
+            apply_forward_adjustment_for_event(klines, current_start_date, dividends)
+            
+        # 9. Save
+        save_kline(cache_filename, klines)
+