vizflow 0.4.1__py3-none-any.whl → 0.4.3__py3-none-any.whl

vizflow/__init__.py

@@ -5,10 +5,18 @@ Usage:
     import vizflow as vf
 """
 
-__version__ = "0.4.1"
+__version__ = "0.4.3"
 
 from .config import ColumnSchema, Config, get_config, set_config
-from .io import load_alpha, load_calendar, load_trade, scan_trade, scan_trades
+from .io import (
+    load_alpha,
+    load_calendar,
+    load_trade,
+    scan_alpha,
+    scan_alphas,
+    scan_trade,
+    scan_trades,
+)
 from .market import CN, CRYPTO, Market, Session
 from .ops import aggregate, bin, parse_time
-from .presets import YLIN
+from .presets import JYAO_V20251114, PRESETS, YLIN_V20251204

vizflow/config.py

@@ -70,9 +70,9 @@ class Config:
     alpha_schema: dict[str, ColumnSchema] = field(default_factory=dict)
     trade_schema: dict[str, ColumnSchema] = field(default_factory=dict)
 
-    # === Column Mapping ===
-    column_preset: str | None = None  # "ylin" or None
-    column_rename: dict[str, str] = field(default_factory=dict)  # Custom rename map
+    # === Column Mapping Presets ===
+    trade_preset: str | None = None  # "ylin" or None
+    alpha_preset: str | None = None  # "jyao_v20251114" or None
 
     # === Aggregation ===
     binwidths: dict[str, float] = field(default_factory=dict)

vizflow/io.py

@@ -141,7 +141,7 @@ def scan_trade(date: str, config: Config | None = None) -> pl.LazyFrame:
         >>> config = vf.Config(
         ...     trade_dir=Path("/data/yuanzhao/"),
         ...     trade_pattern="{date}.meords",
-        ...     column_preset="ylin",
+        ...     trade_preset="ylin_v20251204",
         ... )
         >>> vf.set_config(config)
         >>> df = vf.scan_trade("20241001")
@@ -149,7 +149,7 @@ def scan_trade(date: str, config: Config | None = None) -> pl.LazyFrame:
     config = config or get_config()
     path = config.get_trade_path(date)
     df = _scan_file(path)
-    return _apply_mapping(df, config)
+    return _apply_trade_mapping(df, config)
 
 
 def scan_trades(config: Config | None = None) -> pl.LazyFrame:
@@ -168,7 +168,7 @@ def scan_trades(config: Config | None = None) -> pl.LazyFrame:
         >>> config = vf.Config(
         ...     trade_dir=Path("/data/yuanzhao/"),
         ...     trade_pattern="{date}.feather",
-        ...     column_preset="ylin",
+        ...     trade_preset="ylin_v20251204",
         ... )
         >>> vf.set_config(config)
         >>> df = vf.scan_trades()
@@ -185,18 +185,31 @@ def scan_trades(config: Config | None = None) -> pl.LazyFrame:
     # Concatenate all files using lazy scanning
     dfs = [_scan_file(f) for f in files]
     df = pl.concat(dfs)
-    return _apply_mapping(df, config)
+    return _apply_trade_mapping(df, config)
 
 
-def _apply_mapping(df: pl.LazyFrame, config: Config) -> pl.LazyFrame:
-    """Apply column rename + schema evolution.
+def _apply_trade_mapping(df: pl.LazyFrame, config: Config) -> pl.LazyFrame:
+    """Apply column rename + schema evolution for trade data."""
+    df = _apply_rename(df, config.trade_preset)
+    for col_name, schema in config.trade_schema.items():
+        df = df.with_columns(pl.col(col_name).cast(schema.cast_to))
+    return df
 
-    Args:
-        df: Input LazyFrame
-        config: Config with mapping settings
 
-    Returns:
-        LazyFrame with mapping applied
+def _apply_alpha_mapping(df: pl.LazyFrame, config: Config) -> pl.LazyFrame:
+    """Apply column rename + schema evolution for alpha data."""
+    df = _apply_rename(df, config.alpha_preset)
+    for col_name, schema in config.alpha_schema.items():
+        df = df.with_columns(pl.col(col_name).cast(schema.cast_to))
+    return df
+
+
+def _apply_rename(df: pl.LazyFrame, preset: str | None) -> pl.LazyFrame:
+    """Apply column rename from preset name.
+
+    Args:
+        df: LazyFrame to rename columns
+        preset: Preset name (e.g., "ylin", "jyao_v20251114") or None
     """
     # Drop record type prefix column if present (from CSV files)
     existing = set(df.collect_schema().names())
@@ -204,8 +217,8 @@ def _apply_mapping(df: pl.LazyFrame, config: Config) -> pl.LazyFrame:
         df = df.drop("#HFTORD")
         existing.remove("#HFTORD")
 
-    # Get rename map from preset or custom
-    rename_map = _get_rename_map(config)
+    # Get rename map from preset
+    rename_map = _get_rename_map(preset)
 
     if rename_map:
         existing = set(df.collect_schema().names())
@@ -213,27 +226,71 @@ def _apply_mapping(df: pl.LazyFrame, config: Config) -> pl.LazyFrame:
         if to_rename:
             df = df.rename(to_rename)
 
-    # Schema evolution (type casting) - use renamed column names
-    for col_name, schema in config.trade_schema.items():
-        df = df.with_columns(pl.col(col_name).cast(schema.cast_to))
-
     return df
 
 
-def _get_rename_map(config: Config) -> dict[str, str]:
-    """Get rename map from preset name or custom dict.
+def scan_alpha(date: str, config: Config | None = None) -> pl.LazyFrame:
+    """Scan single date alpha file with column mapping.
 
     Args:
-        config: Config with column_preset or column_rename
+        date: Date string, e.g. "20241001"
+        config: Config to use, or get_config() if None
+
+    Returns:
+        LazyFrame with column mapping and schema evolution applied
+
+    Example:
+        >>> config = vf.Config(
+        ...     alpha_dir=Path("/data/jyao/alpha"),
+        ...     alpha_pattern="alpha_{date}.feather",
+        ...     alpha_preset="jyao_v20251114",
+        ... )
+        >>> vf.set_config(config)
+        >>> df = vf.scan_alpha("20251114")
+    """
+    config = config or get_config()
+    path = config.get_alpha_path(date)
+    df = _scan_file(path)
+    return _apply_alpha_mapping(df, config)
+
+
+def scan_alphas(config: Config | None = None) -> pl.LazyFrame:
+    """Scan all alpha files with column mapping.
+
+    Args:
+        config: Config to use, or get_config() if None
+
+    Returns:
+        LazyFrame with column mapping and schema evolution applied
+
+    Raises:
+        ValueError: If alpha_dir is not set or no files found
+    """
+    config = config or get_config()
+    if config.alpha_dir is None:
+        raise ValueError("alpha_dir is not set in Config")
+
+    pattern = config.alpha_pattern.replace("{date}", "*")
+    files = sorted(config.alpha_dir.glob(pattern))
+    if not files:
+        raise ValueError(f"No files found matching {pattern} in {config.alpha_dir}")
+
+    dfs = [_scan_file(f) for f in files]
+    df = pl.concat(dfs)
+    return _apply_alpha_mapping(df, config)
+
+
+def _get_rename_map(preset: str | None) -> dict[str, str]:
+    """Get rename map from preset name.
+
+    Args:
+        preset: Preset name (e.g., "ylin_v20251204", "jyao_v20251114") or None
 
     Returns:
         Dict mapping old column names to new names
     """
-    if config.column_rename:
-        return config.column_rename
-    if config.column_preset:
-        from .presets import YLIN
-
-        presets = {"ylin": YLIN}
-        return presets.get(config.column_preset.lower(), {})
-    return {}
+    if not preset:
+        return {}
+    from .presets import PRESETS
+
+    return PRESETS.get(preset.lower(), {})

vizflow/presets.py

@@ -1,7 +1,7 @@
 """Column mapping presets for VizFlow."""
 
-# Yuanzhao's log format -> VizFlow standard
-YUANZHAO = {
+# ylin's trade format (v2025-12-04)
+YLIN_V20251204 = {
     # Order columns (18)
     "symbol": "ukey",
     "orderId": "order_id",
@@ -59,5 +59,29 @@ YUANZHAO = {
     "globalCumSellNotional": "cum_sell_filled_notional",
 }
 
-# Alias: ylin (Yuanzhao's username)
-YLIN = YUANZHAO
+# jyao's alpha format (v2025-11-14)
+JYAO_V20251114 = {
+    # Quote columns
+    "BidPrice1": "bid_px0",
+    "AskPrice1": "ask_px0",
+    "BidVolume1": "bid_size0",
+    "AskVolume1": "ask_size0",
+    # Time columns
+    "TimeStamp": "timestamp",
+    "GlobalExTime": "global_exchange_ts",
+    "DataDate": "data_date",
+    # Volume
+    "Volume": "volume",
+    # Alpha columns (Ŷ - predictions)
+    # Rule: ≤60s → s, >60s → m
+    "x10s": "alpha_10s",
+    "x60s": "alpha_60s",
+    "alpha1": "alpha_3m",
+    "alpha2": "alpha_30m",
+}
+
+# Preset registry for dynamic lookup
+PRESETS: dict[str, dict[str, str]] = {
+    "ylin_v20251204": YLIN_V20251204,
+    "jyao_v20251114": JYAO_V20251114,
+}

{vizflow-0.4.1.dist-info → vizflow-0.4.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vizflow
-Version: 0.4.1
+Version: 0.4.3
 Requires-Python: >=3.9
 Requires-Dist: polars>=0.20.0
 Provides-Extra: dev

vizflow-0.4.3.dist-info/RECORD

@@ -0,0 +1,10 @@
+vizflow/__init__.py,sha256=yZZ76XlpFGgaBxlMINGLysOHdjCosqZ5I6tE-1P9KJI,480
+vizflow/config.py,sha256=zSZnhdHzgXSqhDenHcHKm4CDGrMpKAdkNNRoUYYF1uc,6530
+vizflow/io.py,sha256=zmN0fFQOTmSBEBKangMExr0Q5mC2gajZM6GgdAyWkw4,8824
+vizflow/market.py,sha256=MtNz_nnZxC66Aq-i2PXEwaFCTknijFWYZUUv6798k2s,2493
+vizflow/ops.py,sha256=NL-Gtv-m_O1hv-0RUb9Wt43916HsQ5tYK_0e_uKR90w,4062
+vizflow/presets.py,sha256=brvYVJyc6fEgFoEz8JIJv8NqZVvaGK27_vq_GiBe82s,2457
+vizflow/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+vizflow-0.4.3.dist-info/METADATA,sha256=P2kaZ25OcLDtnUceLTecMskv9wmuY_sdfR_MK4iKZjk,388
+vizflow-0.4.3.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+vizflow-0.4.3.dist-info/RECORD,,

vizflow-0.4.1.dist-info/RECORD

@@ -1,10 +0,0 @@
-vizflow/__init__.py,sha256=s03IltxCkl4odJo8ywVERpOD8Dl0GqOaKhFyELjGwM4,387
-vizflow/config.py,sha256=IiVau-4WyO_7NOWlR7Tw58RpDeGFYBbgkNzH7xuQIUg,6544
-vizflow/io.py,sha256=Z2W-jU8nrX52DjzzXCxfHkKSWl3AGnIUC3cFDg4dCTk,6906
-vizflow/market.py,sha256=MtNz_nnZxC66Aq-i2PXEwaFCTknijFWYZUUv6798k2s,2493
-vizflow/ops.py,sha256=NL-Gtv-m_O1hv-0RUb9Wt43916HsQ5tYK_0e_uKR90w,4062
-vizflow/presets.py,sha256=RdxlFouysHBH8IftQx3v5e9Mq82DznxZFbMxv4w1vnA,1825
-vizflow/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-vizflow-0.4.1.dist-info/METADATA,sha256=qRa3pRup8YXzTvR9kxBFmhNZR_BBqsR1CkYrOa3t-As,388
-vizflow-0.4.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-vizflow-0.4.1.dist-info/RECORD,,