snowflake-connector-python

Version 3.1.0a2 is built on top of 3.1.0a1, including all the latest changes in the released GA version 3.1.0.

Version 3.1.0a1 is our first efforts to build snowflake-connector-python based on apache nanoarrow project, which reduces the package size as well as removes a hard dependency on a specific version of pyarrow. This version also includes the following features and bug fixes from the unreleased v3.0.5:

New Features and Improvements

• Added a feature that lets you add connection definitions to the config.toml configuration file. A connection definition refers to a collection of connection parameters. The connection configuration name must begin with connections, similar to the following that defines the parameters for the prod connection:

    [connections.prod]
    account = "my_account"
    user = "my_user"
    password = "my_password"
  

  By default, we look for the config.toml file in the location specified in the SNOWFLAKE_HOME environment variable (default: ~/.snowflake). If this folder does not exist, the Python connector looks for the file in the platformdirs location, as follows:

  • On Linux: ~/.config/snowflake/, but follows XDG settings
  • On Mac: ~/Library/Application Support/snowflake/
  • On Windows: %USERPROFILE%\AppData\Local\snowflake\

  You can determine which file is used by running the following command:

  python -c "from snowflake.connector.constants import CONFIG_FILE; print(str(CONFIG_FILE))"
  

• Bumped cryptography dependency from <41.0.0,>=3.1.0 to >=3.1.0,<42.0.0.

• Allowed to pass type_mapper to fetch_pandas_batches() and fetch_pandas_all().

• Added a parameter server_session_keep_alive in SnowflakeConnection that skips session deletion when client connection closes.

• Added retry reason for queries that are retried by the client.

• Improved error handling of connection reset error.

• Improved OCSP response caching to remove tmp cache files on Windows.

• Improved OCSP response caching to reduce the times of disk writing.

• Tightened our pinning of platformdirs, to prevent their new releases breaking us.

• Remove Python 3.7 support.

Bug Fixes

• Fixed a bug where SFPlatformDirs would incorrectly append application_name/version to its path.
• Fixed a bug where write_pandas fails when user does not have the privilege to create stage or file format in the target schema, but has the right privilege for the current schema.
• Fixed a bug about deleting the temporary files happened when running PUT command.
• Fixed a bug where pickle.dump segfaults during cache serialization in multi-threaded scenarios.
• Worked around a segfault which sometimes occurred during cache serialization in multi-threaded scenarios.

New Features and Improvements

• Added the json_result_force_utf8_decoding connection parameter to force decoding JSON content in utf-8 when the result format is JSON.
• Bumped vendored library urllib3 to 1.26.15
• Bumped vendored library requests to 2.29.0
• Bumped pandas dependency from <1.6.0,>=1.0.0 to >=1.0.0,<2.1.0
• Add support for Geometry types.

Bug Fixes

• Fixed a bug in which cursor.execute() could modify the argument statement_params dictionary object when executing a multistatement query.
• Fixed a bug in which we cannot call SnowflakeCursor.nextset before fetching the result of the first query if the cursor runs an async multistatement query.
• Fixed a bug when _prefetch_hook() was not called before yielding results of execute_async().
• Fixed a bug where some ResultMetadata fields were marked as required when they were optional.
• Fixed a bug where bulk insert converts date incorrectly.

Please check our community page for release notes.

New Features and Improvements

• Added a parameter that allows users to skip file uploads to stage if file exists on stage and contents of the file match.
• Improved type hint of SnowflakeCursor.execute method.
• Improved GET logging to warn when downloading multiple files with the same name.

Bug Fixes

• Fixed a bug that prints error in logs for GET command on GCS.
• Fixed a bug that occurred when writing a Pandas DataFrame with non-default index in snowflake.connector.pandas_tool.write_pandas.
• Fixed a bug that occurred when writing a Pandas DataFrame with column names containing double quotes in snowflake.connector.pandas_tool.write_pandas.
• Fixed a bug that occurred when writing a Pandas DataFrame with binary data in snowflake.connector.pandas_tool.write_pandas.

Please check our community page for release notes.

New Features

• Improved logging to mask tokens in case of errors.
• Validate SSO URL before opening it in the browser for External browser authenticator.

Bug Fixes

• Fixed a memory leak in the logging module of the Cython extension.
• Fixed a bug where the put command on AWS raised AttributeError when uploading file composed of multiple parts.
• Fixed a bug of incorrect type hints of SnowflakeCursor.fetch_arrow_all and SnowflakeCursor.fetchall.
• Fixed a bug where snowflake.connector.util_text.split_statements swallows the final line break in the case when there are no space between lines.

Please check our community page for release notes.

New Features

• Improved the robustness of OCSP response caching to handle errors in cases of serialization and deserialization.
• Replaced the dependency on setuptools in favor of packaging.
• Updated async_executes method's doc-string.
• Errors raised now have a query field that contains the SQL query that caused them when available.

Bug Fixes

• Fixed a bug where AuthByKeyPair.handle_timeout should pass keyword arguments instead of positional arguments when calling AuthByKeyPair.prepare.
• Fixed a bug where MFA token caching would refuse to work until restarted instead of reauthenticating. Please check our community page for release notes.

Behavior Changes

• Fixed a bug where write_pandas did not use user-specified schemas and databases to create intermediate objects

New Features and Updates

• Bumped pyarrow dependency from >=8.0.0,<8.1.0 to >=10.0.1,<10.1.0
• Bumped pyOpenSSL dependency from <23.0.0 to <24.0.0
• During browser-based authentication, the SSO url is now printed before opening it in the browser
• Increased the level of a log for when ArrowResult cannot be imported
• Added a minimum MacOS version check when compiling C-extensions

Bug Fixes

• Fixed a bug where the HTTP 429 response code was not retried
• Fixed a bug where MFA token caching was not working

Please check our community page for release notes.

New Features

• Reworked authentication internals to allow users to plug custom key-pair authenticators.
• Multi-statement query execution is now supported through cursor.execute and cursor.executemany.
  • The Snowflake parameter MULTI_STATEMENT_COUNT can be altered at the account, session, or statement level. An additional argument, num_statements, can be provided to execute to use this parameter at the statement level. It must be provided to executemany to submit a multi-statement query through the method. Note that bulk insert optimizations available through executemany are not available when submitting multi-statement queries.
    • By default the parameter is 1, meaning only a single query can be submitted at a time.
    • Set to 0 to submit any number of statements in a multi-statement query.
    • Set to >1 to submit the specified exact number of statements in a multi-statement query.
  • Bindings are accepted in the same way for multi-statements as they are for single statement queries.
  • Asynchronous multi-statement query execution is supported. Users should still use get_results_from_sfqid to retrieve results.
  • To access the results of each query, users can call SnowflakeCursor.nextset() as specified in the DB 2.0 API (PEP-249), to iterate through each statements results.
    • The first statement's results are accessible immediately after calling execute (or get_results_from_sfqid if asynchronous) through the existing fetch*() methods.

Bug Fixes

• Fixed a bug where the permission of the file downloaded via GET command is changed.

Released to PyPI on December 14, 2022

• Bumped cryptography dependency from <39.0.0 to <41.0.0
• Fixed a bug where expired OCSP response cache caused infinite recursion during cache loading

• Improved performance of OCSP response caching
• During the execution of GET commands we no longer resolve target location on the local machine
• Improved performance of regexes used for PUT/GET SQL statement detection. CVE-2022-42965

• Bumped cryptography dependency from <37.0.0 to <39.0.0
• Bumped pandas dependency from <1.5.0 to <1.6.0
• Fixed a bug where write_pandas wouldn't write an empty DataFrame to Snowflake
• When closing connection async query status checking is now parallelized
• Fixed a bug where test logging would be enabled on Jenkins workers in non-Snowflake Jenkins machines
• Enhanced the atomicity of write_pandas when overwrite is set to True

• Fixed a bug where rowcount was deleted when the cursor was closed
  • Fixed a bug where extTypeName was used even when it was empty
  • Updated how telemetry entries are constructed
  • Added telemetry for imported root packages during run-time
  • Added telemetry for using write_pandas
  • Fixed missing dtypes when calling fetch_pandas_all() on empty result
  • The write_pandas function now supports providing additional arguments to be used by DataFrame.to_parquet
  • All optional parameters of write_pandas can now be provided to pd_writer and make_pd_writer to be used with DataFrame.to_sql

• Fixed a bug where timestamps fetched as pandas.DataFrame or pyarrow.Table would overflow for the sake of unnecessary precision. In the case where an overflow cannot be prevented a clear error will be raised now.
• Added in-file caching for OCSP response caching
• The write_pandas function now supports transient tables through the new table_type argument which supersedes create_temp_table argument
• Fixed a bug where calling fetch_pandas_batches incorrectly raised NotSupportedError after an async query was executed

• Added minimum version pin to typing_extensions

• Release wheels are now built on manylinux2014
• Bumped supported pyarrow version to >=8.0.0,<8.1.0
• Updated vendored library versions requests to 2.28.1 and urllib3 to 1.26.10
• Added in-memory cache to OCSP requests
• Added overwrite option to write_pandas
• Added attribute lastrowid to SnowflakeCursor in compliance with PEP249.
• Fixed a bug where gzip compressed http requests might be garbled by an unflushed buffer
• Added new connection diagnostics capabilities to snowflake-connector-python
• Bumped numpy dependency from <1.23.0 to <1.24.0

• Fixed a bug where errors raised during get_results_from_sfqid() were missing errno
  • Fixed a bug where empty results containing GEOGRAPHY type raised IndexError