Add test to do map->flatmap and then passthrough that same flatmap #89
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 5 commits
September 26, 2024 11:04
Summary: Allows for already encoded dictionaries to passthrough the writer as opposed to being decoded + deduped again. Specifically, we want to allow for a "passthrough" without decoding when the schema indicates that we are ingesting an ArrayVector, but the array we find is a dictionaryVector with ArrayVector values. When this is the case, we will check to see if the dictionary is a valid run-length encoding (meaning, it likely has been deduped by upstream). The way in which the vector is written to storage does not change, we will still write the offsets and the ArrayVector elements. A caveat is that right now, `isDictionaryValidRunLengthEncoded` function doesn't actually verify that there is a duplicate. It's possible that a scenario exists in which a DictionaryVector was passed to the ArrayVector field writer, but the DictionaryVector has not been deduped, but its offsets still technically fulfill a run-length encoding (each offset is greater than or equal to the offset at index - 1). Reviewed By: helfman Differential Revision: D59863671
…tLocalDecodedVector Summary: As title, in order to verify passthrough code is being correctly called and the local decoded vector is NOT being accessed, we need a way to track visits to the decode function. On suggestion/idea by helfman, we've created a visit callback passed in via the writer options which allows us to track if the decode visit has been called from the writer code. I did verify that the callback is being called through some tests that I know _do_ use the decode functionality. I plan to add decode visit counters for those tests later but that is out of scope for this diff for now This decode visit functionality will be used in the flatmap passthrough functionality as well Reviewed By: helfman Differential Revision: D60241883
Summary: Allows for caching of the last element when a dictionary encoded array is given for passthrough. This will bring us up to parity of local file encoding with the original arrayWithOffsets, as two consecutive datasets, either passed as unencoded array or already encoded dictionaryArray will now have the same output into the file. Reviewed By: helfman Differential Revision: D60413733
Summary: As title, allows for ingestion of RowVector in the FlatMapFieldWriter. Takes each item in the row and writes its child vector based on the ranges provided. Currently not handled: - Encoded row vectors with flatmap schema - What happens when there are subsequent writes that have a different set of keys from the original write Reviewed By: helfman Differential Revision: D61456412
Summary: Test that creates a fuzz map, writes it to a nimble file with flatmap encoding turned on, then reads it back and writes the same vector again. Test ensures that the two paths map input ->flatmap vector to write and flatmap input->flatmap vector to write have the same right side output Reviewed By: helfman Differential Revision: D62110004
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D62110004 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
Test that creates a fuzz map, writes it to a nimble file with flatmap encoding turned on, then reads it back and writes the same vector again. Test ensures that the two paths
map input ->flatmap vector to write
and
flatmap input->flatmap vector to write
have the same right side output
Reviewed By: helfman
Differential Revision: D62110004