You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
DataFrameReader.documents() (Private Preview) reads with a schema or a prompt now return the AI_COMPLETE token usage of each document in INPUT_TOKENS and OUTPUT_TOKENS columns. A schema field can no longer be named like one of these columns.
Improvements
to_timestamp, to_char/to_varchar, to_binary and try_to_binary now accept the format as a Column or a literal str. to_binary and try_to_binary now pass an empty-string format to Snowflake instead of silently ignoring it.
Bug Fixes
Fixed a bug where to_binary and try_to_binary raised TypeError when the format was a Column.
Fixed a bug where DataFrameReader.documents() (Private Preview) reads with a schema rejected .jpg, .jpeg, .png, .gif and .webp images with a "file type is not supported" error.
Snowpark Local Testing Updates
Bug Fixes
Fixed a bug where to_binary in local testing treated an empty-string format as HEX and failed with an unclear error for an unknown format, instead of raising an invalid binary format error as Snowflake does.