vizflow 0.5.6__py3-none-any.whl → 0.5.8__py3-none-any.whl

vizflow/__init__.py

@@ -5,7 +5,7 @@ Usage:
     import vizflow as vf
 """
 
-__version__ = "0.5.5"
+__version__ = "0.5.8"
 
 from .config import Config, get_config, set_config
 from .io import (
@@ -19,6 +19,7 @@ from .io import (
 )
 from .market import CN, CRYPTO, Market, Session
 from .ops import aggregate, bin, forward_return, mark_to_close, parse_time, sign_by_side
+from .viz import add_tod
 from .schema_evolution import (
     JYAO_V20251114,
     SCHEMAS,

vizflow/io.py

@@ -167,6 +167,8 @@ def scan_trade(date: str, config: Config | None = None) -> pl.LazyFrame:
     if schema:
         df = _apply_schema_evolution(df, schema)
 
+    df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").cast(pl.Date).alias("data_date"))
+
     return df
 
 
@@ -213,7 +215,7 @@ def scan_trades(config: Config | None = None) -> pl.LazyFrame:
             df = _apply_schema_evolution(df, schema)
         date = _extract_date_from_path(f, config.trade_pattern)
         if date:
-            df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").alias("data_date"))
+            df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").cast(pl.Date).alias("data_date"))
         dfs.append(df)
 
     return pl.concat(dfs)
@@ -246,6 +248,8 @@ def scan_alpha(date: str, config: Config | None = None) -> pl.LazyFrame:
     if schema:
         df = _apply_schema_evolution(df, schema)
 
+    df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").cast(pl.Date).alias("data_date"))
+
     return df
 
 
@@ -281,7 +285,7 @@ def scan_alphas(config: Config | None = None) -> pl.LazyFrame:
             df = _apply_schema_evolution(df, schema)
         # Convert data_date to Date type (may be Int64 from feather or String from CSV)
         df = df.with_columns(
-            pl.col("data_date").cast(pl.String).str.to_date("%Y%m%d")
+            pl.col("data_date").cast(pl.String).str.to_date("%Y%m%d").cast(pl.Date)
         )
         dfs.append(df)
 
@@ -314,6 +318,8 @@ def scan_univ(date: str, config: Config | None = None) -> pl.LazyFrame:
     if schema:
         df = _apply_schema_evolution(df, schema)
 
+    df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").cast(pl.Date).alias("data_date"))
+
     return df
 
 
@@ -351,7 +357,7 @@ def scan_univs(config: Config | None = None) -> pl.LazyFrame:
             df = _apply_schema_evolution(df, schema)
         date = _extract_date_from_path(f, config.univ_pattern)
         if date:
-            df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").alias("data_date"))
+            df = df.with_columns(pl.lit(date).str.to_date("%Y%m%d").cast(pl.Date).alias("data_date"))
         dfs.append(df)
 
     return pl.concat(dfs)

vizflow/ops.py

@@ -11,19 +11,20 @@ def parse_time(
     df: pl.LazyFrame,
     timestamp_col: str = "ticktime",
 ) -> pl.LazyFrame:
-    """Parse HHMMSSMMM timestamp to time-of-day and elapsed milliseconds.
+    """Parse HHMMSSMMM timestamp to elapsed milliseconds.
 
-    Adds two columns:
-    - tod_{timestamp_col}: pl.Time (time-of-day HH:MM:SS.mmm) - good for plotting
+    Adds one column:
     - elapsed_{timestamp_col}: pl.Int64 (milliseconds since market open)
 
+    For time-of-day column (pl.Time) for plotting, use vf.add_tod() separately.
+
     Args:
         df: Input LazyFrame
         timestamp_col: Column with integer HHMMSSMMM format timestamps
             e.g., 93012145 = 09:30:12.145, 142058425 = 14:20:58.425
 
     Returns:
-        LazyFrame with tod and elapsed columns added
+        LazyFrame with elapsed column added
 
     Raises:
         RuntimeError: If config not set via set_config()
@@ -33,7 +34,7 @@ def parse_time(
         >>> config = vf.Config(market="CN", input_dir=".", output_dir=".")
         >>> vf.set_config(config)
         >>> df = vf.parse_time(df, "ticktime")
-        >>> # Creates: tod_ticktime (pl.Time), elapsed_ticktime (pl.Int64)
+        >>> # Creates: elapsed_ticktime (pl.Int64)
     """
     config = get_config()
 
@@ -48,19 +49,8 @@ def parse_time(
         (pl.col(timestamp_col) % 1000).alias("_ms"),
     ])
 
-    # Add time-of-day column (pl.Time)
-    # Convert to nanoseconds since midnight
-    tod_ns = (
-        pl.col("_hour") * 3_600_000_000_000
-        + pl.col("_minute") * 60_000_000_000
-        + pl.col("_second") * 1_000_000_000
-        + pl.col("_ms") * 1_000_000
-    )
-    df = df.with_columns(tod_ns.cast(pl.Time).alias(f"tod_{timestamp_col}"))
-
     # Add elapsed milliseconds (int)
     # CN market: 09:30-11:30 (morning), 13:00-15:00 (afternoon)
-    # Using user's hardcoded logic
     elapsed_ms = (
         pl.when(pl.col("_hour") < 12)
         .then(

vizflow/schema_evolution.py

@@ -353,7 +353,7 @@ JYAO_V20251114 = SchemaEvolution(
         # Time columns
         "TimeStamp": ColumnSpec(rename_to="timestamp", parse_dtype=pl.Int64),
         "GlobalExTime": ColumnSpec(rename_to="global_exchange_ts", parse_dtype=pl.Int64),
-        "DataDate": ColumnSpec(rename_to="data_date", parse_dtype=pl.String),
+        "DataDate": ColumnSpec(rename_to="data_date", parse_dtype=pl.String, cast_dtype=pl.Date),
         # Volume
         "Volume": ColumnSpec(
             rename_to="volume",
@@ -371,6 +371,52 @@ JYAO_V20251114 = SchemaEvolution(
 )
 
 
+# =============================================================================
+# JYAO Univ Format (v2025-12-30)
+# =============================================================================
+
+JYAO_UNIV_V20251230 = SchemaEvolution(
+    columns={
+        # ID
+        "ukey": ColumnSpec(parse_dtype=pl.Int64),
+        # Price columns (Float64)
+        "ydclose": ColumnSpec(parse_dtype=pl.Float64),
+        "preclose": ColumnSpec(parse_dtype=pl.Float64),
+        "open": ColumnSpec(parse_dtype=pl.Float64),
+        "close": ColumnSpec(parse_dtype=pl.Float64),
+        "upper_limit_price": ColumnSpec(parse_dtype=pl.Float64),
+        "lower_limit_price": ColumnSpec(parse_dtype=pl.Float64),
+        "tick_size": ColumnSpec(parse_dtype=pl.Float64),
+        # Lot size columns (parse Float64 → cast Int64)
+        "trade_min_size": ColumnSpec(parse_dtype=pl.Float64, cast_dtype=pl.Int64),
+        "trade_unit_size": ColumnSpec(parse_dtype=pl.Float64, cast_dtype=pl.Int64),
+        "qty_unit": ColumnSpec(parse_dtype=pl.Float64, cast_dtype=pl.Int64),
+        # Average/aggregated columns (Float64)
+        "trade_max_size": ColumnSpec(parse_dtype=pl.Float64),
+        "adv": ColumnSpec(parse_dtype=pl.Float64),
+        "roll_spread": ColumnSpec(parse_dtype=pl.Float64),
+        "buy_avg_volume": ColumnSpec(parse_dtype=pl.Float64),
+        "sell_avg_volume": ColumnSpec(parse_dtype=pl.Float64),
+        "avg_touch_size_mean": ColumnSpec(parse_dtype=pl.Float64),
+        "avg_touch_order_size": ColumnSpec(parse_dtype=pl.Float64),
+        # Risk columns (Float64)
+        "TotalRisk": ColumnSpec(parse_dtype=pl.Float64),
+        "SpecRisk": ColumnSpec(parse_dtype=pl.Float64),
+        # Boolean columns (TRUE/FALSE strings)
+        "is_price_limited": ColumnSpec(parse_dtype=pl.Boolean),
+        "is_t0": ColumnSpec(parse_dtype=pl.Boolean),
+        # Integer columns
+        "category": ColumnSpec(parse_dtype=pl.Float64, cast_dtype=pl.Int64),
+        "is_ST": ColumnSpec(parse_dtype=pl.Float64, cast_dtype=pl.Int64),  # 0/1 numeric
+        # String columns
+        "UNIVERSE": ColumnSpec(parse_dtype=pl.String),
+        "INDUSTRY": ColumnSpec(parse_dtype=pl.String),
+        "INDEX": ColumnSpec(parse_dtype=pl.String),
+    },
+    null_values=["", "NA"],
+)
+
+
 # =============================================================================
 # Schema Registry
 # =============================================================================
@@ -378,6 +424,7 @@ JYAO_V20251114 = SchemaEvolution(
 SCHEMAS: dict[str, SchemaEvolution] = {
     "ylin_v20251204": YLIN_V20251204,
     "jyao_v20251114": JYAO_V20251114,
+    "jyao_univ_v20251230": JYAO_UNIV_V20251230,
 }
 
 

vizflow/viz.py

@@ -0,0 +1,35 @@
+"""Visualization utilities for VizFlow."""
+
+import polars as pl
+
+
+def add_tod(
+    df: pl.LazyFrame,
+    timestamp_col: str = "ticktime",
+) -> pl.LazyFrame:
+    """Add time-of-day column for plotting.
+
+    Converts HHMMSSMMM integer timestamp to pl.Time for visualization.
+    Note: pl.Time type is not supported by Delta Lake - use this only
+    for plotting, not for data that will be written to Delta Lake.
+
+    Args:
+        df: Input LazyFrame with HHMMSSMMM timestamp column
+        timestamp_col: Column with integer HHMMSSMMM format timestamps
+            e.g., 93012145 = 09:30:12.145, 142058425 = 14:20:58.425
+
+    Returns:
+        LazyFrame with tod_{timestamp_col} (pl.Time) column added
+
+    Example:
+        >>> df = vf.add_tod(df, "ticktime")
+        >>> # Creates: tod_ticktime (pl.Time)
+    """
+    # Parse HHMMSSMMM to nanoseconds since midnight
+    tod_ns = (
+        (pl.col(timestamp_col) // 10000000) * 3_600_000_000_000
+        + (pl.col(timestamp_col) // 100000 % 100) * 60_000_000_000
+        + (pl.col(timestamp_col) // 1000 % 100) * 1_000_000_000
+        + (pl.col(timestamp_col) % 1000) * 1_000_000
+    )
+    return df.with_columns(tod_ns.cast(pl.Time).alias(f"tod_{timestamp_col}"))

{vizflow-0.5.6.dist-info → vizflow-0.5.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vizflow
-Version: 0.5.6
+Version: 0.5.8
 Requires-Python: >=3.9
 Requires-Dist: polars>=0.20.0
 Provides-Extra: dev

vizflow-0.5.8.dist-info/RECORD

@@ -0,0 +1,11 @@
+vizflow/__init__.py,sha256=KGksMU9wGsdGSbd2OgSHM2ofMZbTwgaCN92ZDEgASwU,614
+vizflow/config.py,sha256=nPZPXlqQbaY8u_FAdtPShvb0mdx3e2TRaQ2CILzliAU,7192
+vizflow/io.py,sha256=1T7t-L1ijrfEkE-gr4f45yiupJKA4-DxbJhsyN_tLnI,11939
+vizflow/market.py,sha256=MtNz_nnZxC66Aq-i2PXEwaFCTknijFWYZUUv6798k2s,2493
+vizflow/ops.py,sha256=QO8airzHFfKw8nGsStsecic2Z1dmq1rciSyAfSElh7M,10421
+vizflow/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+vizflow/schema_evolution.py,sha256=3_qFIQJAgXUK0vQzanb355YVjytmfOACkelZlykIO8w,16349
+vizflow/viz.py,sha256=dzcY72hWMVbxWIyjwfUW3Ot3CunaP7O4GLVUzzOjkbY,1212
+vizflow-0.5.8.dist-info/METADATA,sha256=uapEZDP5ntZ1_a8uzexB_K45cIpK9rdOiqq87fSv_4M,388
+vizflow-0.5.8.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+vizflow-0.5.8.dist-info/RECORD,,

vizflow-0.5.6.dist-info/RECORD

@@ -1,10 +0,0 @@
-vizflow/__init__.py,sha256=_jP6bp5U2iBFkPGTSLNwvOQay9XU_opNdrylF22iR9s,589
-vizflow/config.py,sha256=nPZPXlqQbaY8u_FAdtPShvb0mdx3e2TRaQ2CILzliAU,7192
-vizflow/io.py,sha256=5gucUD68zBhfnVOmFRbDyxqVM5ikzPWRTeOY20gXQOM,11612
-vizflow/market.py,sha256=MtNz_nnZxC66Aq-i2PXEwaFCTknijFWYZUUv6798k2s,2493
-vizflow/ops.py,sha256=oR44HYKrfaXLh0SmbfXefl714UESSIC5lTNJBrR1kto,10858
-vizflow/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-vizflow/schema_evolution.py,sha256=puwuuJ8LAzelHq1JTwQD968J7GYLvgAsCaHJnDHzu4U,14025
-vizflow-0.5.6.dist-info/METADATA,sha256=oadPQYbrOT8G2aPZ-QZ1_VZ0uqRbde62ddsZnnmDKHE,388
-vizflow-0.5.6.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-vizflow-0.5.6.dist-info/RECORD,,