Updates
2025 → 2026

# ── 1. Acquire token (Fabric notebook — no app registration needed) ──────
import notebookutils

PBI_RESOURCE = "https://analysis.windows.net/powerbi/api"
access_token = notebookutils.credentials.getToken(PBI_RESOURCE)

# ── 2. Helper: POST DAX → return pandas DataFrame via Arrow IPC ──────────
import io, requests, pandas as pd, pyarrow as pa

def execute_dax(workspace_id: str, dataset_id: str, query: str) -> pd.DataFrame:
    url = (
        f"https://api.powerbi.com/v1.0/myorg/groups/{workspace_id}"
        f"/datasets/{dataset_id}/executeDaxQueries"
    )
    resp = requests.post(url, headers={
        "Authorization": f"Bearer {access_token}",
        "Content-Type": "application/json"
    }, json={
        "query": query,
        "resultsetRowcountLimit": 500000
    }, timeout=180)
    resp.raise_for_status()
    reader = pa.ipc.open_stream(io.BytesIO(resp.content))
    return reader.read_all().to_pandas()

# ── 3. Run DAX query against your semantic model ─────────────────────────
WORKSPACE_ID  = "your-workspace-guid"
DATASET_ID    = "your-semantic-model-guid"

df = execute_dax(WORKSPACE_ID, DATASET_ID, """
EVALUATE
SUMMARIZECOLUMNS(
    'Date'[Year],
    'Sales Territory'[Region],
    "Total Sales", [Total Sales Amount],
    "YoY Growth",  [YoY Sales Growth %]
)
ORDER BY 'Date'[Year] DESC
""")

# ── 4. Persist as V-Ordered Delta table for Direct Lake consumption ───────
import re
from delta.tables import DeltaTable

spark_df = spark.createDataFrame(df)
spark_df = spark_df.toDF(*[re.sub(r"[ ,;{}()\n\t=]", "_", c) for c in spark_df.columns])

spark_df.write.format("delta") \
    .mode("overwrite") \
    .option("overwriteSchema", "true") \
    .option("delta.parquet.vorder.enabled", "true") \
    .saveAsTable("semantic_extract_sales_delta")

print(f"✓ Exported {len(df):,} rows to Delta table")
spark.sql("SELECT * FROM semantic_extract_sales_delta LIMIT 5").show()