Rediger

Retrieve data with mssql-python

The mssql-python driver offers several fetch methods, row access patterns, and cursor navigation features for retrieving query results.

Fetch methods

After executing a SELECT query, use fetch methods to retrieve results.

fetchone()

Returns a single row or None if no more rows are available:

cursor.execute("SELECT ProductID, Name, ListPrice FROM Production.Product")

row = cursor.fetchone()
while row:
    print(f"{row.ProductID}: {row.Name} - ${row.ListPrice}")
    row = cursor.fetchone()

fetchmany()

Returns a list of rows. cursor.arraysize controls the default batch size (default: 1):

cursor.execute("SELECT * FROM Production.Product")
cursor.arraysize = 100  # Fetch 100 rows at a time

while True:
    rows = cursor.fetchmany()
    if not rows:
        break
    for row in rows:
        print(row.Name)

You can also specify the size directly:

rows = cursor.fetchmany(50)  # Fetch up to 50 rows

fetchall()

Returns all remaining rows as a list:

cursor.execute("SELECT * FROM Production.Product WHERE Color = 'Black'")
rows = cursor.fetchall()

print(f"Found {len(rows)} products")
for row in rows:
    print(row.Name)

fetchval()

Returns the first column of the first row, which is useful for scalar queries.

count = cursor.execute("SELECT COUNT(*) FROM Production.Product").fetchval()
print(f"Total products: {count}")

max_price = cursor.execute("SELECT MAX(ListPrice) FROM Production.Product").fetchval()
print(f"Highest price: ${max_price}")

Row access patterns

The Row class supports multiple access patterns.

Index access

Access columns by position (zero-based):

cursor.execute("SELECT ProductID, Name, ListPrice FROM Production.Product WHERE ProductID < 5")
row = cursor.fetchone()

product_id = row[0]
name = row[1]
price = row[2]

Attribute access

Access columns by name:

cursor.execute("SELECT ProductID, Name, ListPrice FROM Production.Product WHERE ProductID < 5")
row = cursor.fetchone()

product_id = row.ProductID
name = row.Name
price = row.ListPrice

Lowercase column names

Enable lowercase attribute names globally:

import mssql_python

settings = mssql_python.get_settings()
settings.lowercase = True

cursor.execute("SELECT ProductID, Name FROM Production.Product WHERE ProductID < 5")
row = cursor.fetchone()
print(row.productid, row.name)  # Lowercase access
settings.lowercase = False  # Restore default

Iteration

Rows support iteration over values:

cursor.execute("SELECT ProductID, Name, ListPrice FROM Production.Product WHERE ProductID < 5")
row = cursor.fetchone()

for value in row:
    print(value)

Cursor iteration

Iterate directly over the cursor to process rows:

cursor.execute("SELECT * FROM Production.Product")

for row in cursor:
    print(row.Name)

This pattern is equivalent to calling fetchone() repeatedly.

Column metadata

Access column information through cursor.description:

cursor.execute("SELECT ProductID, Name, ListPrice FROM Production.Product WHERE ProductID < 5")

for col in cursor.description:
    name, type_code, display_size, internal_size, precision, scale, null_ok = col
    print(f"Column: {name}, Type: {type_code}, Nullable: {null_ok}")

Row count

The cursor.rowcount attribute indicates:

  • For SELECT: Returns -1 after execute() until fetching begins. Once you start fetching, it reflects the cumulative number of rows fetched so far.
  • For INSERT/UPDATE/DELETE: Number of rows affected.
cursor.execute("SELECT * FROM Production.Product")
print(f"Rows returned: {cursor.rowcount}")

cursor.execute("CREATE TABLE #PriceUpd (Name NVARCHAR(50), Price DECIMAL(10,2), CategoryID INT)")
cursor.execute("INSERT INTO #PriceUpd VALUES ('A',10,1),('B',20,1),('C',30,2)")
cursor.execute("UPDATE #PriceUpd SET Price = Price * 1.1 WHERE CategoryID = 1")
print(f"Rows updated: {cursor.rowcount}")

Cursor navigation

skip()

Skip rows without fetching them:

cursor.execute("SELECT * FROM Production.Product ORDER BY ProductID")
cursor.skip(10)  # Skip first 10 rows
row = cursor.fetchone()  # Returns 11th row

scroll()

Move the cursor position forward:

cursor.execute("SELECT * FROM Production.Product ORDER BY ProductID")

# Move forward 5 rows from current position
cursor.scroll(5, mode='relative')

row = cursor.fetchone()

Note

The driver supports only mode='relative' with positive values. Absolute positioning and backward scrolling raise NotSupportedError because the driver uses forward-only cursors.

rownumber

Track the current position:

cursor.execute("SELECT * FROM Production.Product")

print(f"Initial position: {cursor.rownumber}")  # -1 (before first fetch)

row = cursor.fetchone()
print(f"After fetchone: {cursor.rownumber}")    # 0 (first row fetched)

Multiple result sets

Use nextset() to process multiple result sets:

cursor.execute("""
    SELECT * FROM Production.Product WHERE Color = 'Black';
    SELECT * FROM Production.ProductCategory;
    SELECT COUNT(*) FROM Production.Product;
""")

# First result set
products = cursor.fetchall()
print(f"Products: {len(products)}")

# Move to second result set
if cursor.nextset():
    categories = cursor.fetchall()
    print(f"Categories: {len(categories)}")

# Move to third result set
if cursor.nextset():
    count = cursor.fetchval()
    print(f"Total count: {count}")

Large result sets

For large result sets, process rows in batches to manage memory:

def process_batch(rows):
    # Example: print each row. Replace with your own logic.
    for row in rows:
        print(row)

cursor.execute("SELECT * FROM LargeTable")
cursor.arraysize = 1000

while True:
    rows = cursor.fetchmany()
    if not rows:
        break

    process_batch(rows)
    print(f"Processed {cursor.rownumber} rows so far")

Context managers

Use context managers for automatic resource cleanup:

with mssql_python.connect(connection_string) as conn:
    with conn.cursor() as cursor:
        cursor.execute("SELECT * FROM Production.Product")
        for row in cursor:
            print(row.Name)
# Cursor and connection closed automatically

Best practices

  • Use fetchmany() for large results to avoid loading everything into memory.
  • Close cursors when done to release server resources.
  • Use column names (attribute access) for more readable code.
  • Check rowcount after data modification statements.
  • Handle None values explicitly when columns are nullable.