vizflow 0.5.7__py3-none-any.whl → 0.5.9__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/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(
@@ -226,11 +216,12 @@ def forward_return(
         )
 
         # Add forward price and calculate return (guard against zero price)
+        # Explicitly cast to Float64 to ensure consistent schema across dates
         trade = joined.with_columns([
-            pl.col("_forward_price").alias(forward_col),
+            pl.col("_forward_price").cast(pl.Float64).alias(forward_col),
             pl.when(pl.col(price_col) != 0)
             .then((pl.col("_forward_price") - pl.col(price_col)) / pl.col(price_col))
-            .otherwise(pl.lit(None))
+            .otherwise(pl.lit(None, dtype=pl.Float64))
             .alias(return_col),
         ]).drop(["_forward_time", "_alpha_time", "_forward_price"])
 

vizflow/schema_evolution.py

@@ -337,6 +337,8 @@ JYAO_V20251114 = SchemaEvolution(
         # Symbol column - parse_dtype for CSV, cast_dtype for feather/IPC
         # (feather files have embedded types, so cast is needed post-load)
         "ukey": ColumnSpec(parse_dtype=pl.Int64, cast_dtype=pl.Int64),
+        # Time columns
+        "ticktime": ColumnSpec(parse_dtype=pl.Int64, cast_dtype=pl.Int64),
         # Quote columns
         "BidPrice1": ColumnSpec(rename_to="bid_px0", parse_dtype=pl.Float64),
         "AskPrice1": ColumnSpec(rename_to="ask_px0", parse_dtype=pl.Float64),

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-0.5.7.dist-info → vizflow-0.5.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vizflow
-Version: 0.5.7
+Version: 0.5.9
 Requires-Python: >=3.9
 Requires-Dist: polars>=0.20.0
 Provides-Extra: dev

vizflow-0.5.9.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=cUFeDdWfMk2hG8hpevTz_Dfnzk3fnhTI79UwMGzUkHw,10534
+vizflow/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+vizflow/schema_evolution.py,sha256=HuyfHNGXUS82_oat1aSixd3msvOnBC9l5u0ec01blck,16447
+vizflow/viz.py,sha256=dzcY72hWMVbxWIyjwfUW3Ot3CunaP7O4GLVUzzOjkbY,1212
+vizflow-0.5.9.dist-info/METADATA,sha256=wrZtGCshKtCZDr96T8QfmYvdhBi7x2y0Bc_Fe5eAiC4,388
+vizflow-0.5.9.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+vizflow-0.5.9.dist-info/RECORD,,

vizflow-0.5.7.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=1T7t-L1ijrfEkE-gr4f45yiupJKA4-DxbJhsyN_tLnI,11939
-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=3_qFIQJAgXUK0vQzanb355YVjytmfOACkelZlykIO8w,16349
-vizflow-0.5.7.dist-info/METADATA,sha256=nkd-hejnIVFMaJzYnQP6EIHRkrJfBDsohONlj4zFVu4,388
-vizflow-0.5.7.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-vizflow-0.5.7.dist-info/RECORD,,