vizflow 0.5.5__py3-none-any.whl → 0.5.7__py3-none-any.whl

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).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
 
 
@@ -273,12 +277,16 @@ def scan_alphas(config: Config | None = None) -> pl.LazyFrame:
     schema = _resolve_schema(config.alpha_schema)
 
     # Apply schema evolution per-file BEFORE concat to ensure matching schemas
-    # Note: Alpha files already have data_date column, no need to add it
+    # Note: Alpha files already have data_date column, convert to Date type for consistency
     dfs = []
     for f in files:
         df = _scan_file(f, schema=schema)
         if schema:
             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").cast(pl.Date)
+        )
         dfs.append(df)
 
     return pl.concat(dfs)
@@ -310,6 +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
 
 
@@ -347,7 +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).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/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-0.5.5.dist-info → vizflow-0.5.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vizflow
-Version: 0.5.5
+Version: 0.5.7
 Requires-Python: >=3.9
 Requires-Dist: polars>=0.20.0
 Provides-Extra: dev

{vizflow-0.5.5.dist-info → vizflow-0.5.7.dist-info}/RECORD

@@ -1,10 +1,10 @@
 vizflow/__init__.py,sha256=_jP6bp5U2iBFkPGTSLNwvOQay9XU_opNdrylF22iR9s,589
 vizflow/config.py,sha256=nPZPXlqQbaY8u_FAdtPShvb0mdx3e2TRaQ2CILzliAU,7192
-vizflow/io.py,sha256=qaILo7JYHRILZv7RgPdwltOkCew5ac3nmwOTn1p2ha4,11351
+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=puwuuJ8LAzelHq1JTwQD968J7GYLvgAsCaHJnDHzu4U,14025
-vizflow-0.5.5.dist-info/METADATA,sha256=m6Zj_2AjsyBr1Tvq7uf20dzH9X9fNrFwpgM6wrDpcVg,388
-vizflow-0.5.5.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-vizflow-0.5.5.dist-info/RECORD,,
+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,,