BOKJUNSOO opened a new pull request, #4505:
URL: https://github.com/apache/arrow-adbc/pull/4505
### Summary
PostgreSQL binary COPY represents a `jsonb` field as a version byte followed
by
the JSON text. The PostgreSQL driver currently sends the physical string
storage
of `extension<arrow.json>` unchanged, so PostgreSQL interprets the first
JSON byte
as the version (for example, `{` becomes version 123) and rejects the row.
This adds a target-aware JSONB COPY field writer. When the Arrow source is
`extension<arrow.json>` and the existing PostgreSQL target column is
`jsonb`, the
writer prefixes the JSON bytes with version `0x01` and includes that byte in
the
field length. Other string inputs and PostgreSQL `json` targets continue to
use
the existing writer.
The existing JSONB ingest regression test now verifies a successful
round-trip,
including a SQL NULL value.
### Testing
- `pre-commit run --files c/driver/postgresql/copy/writer.h
c/driver/postgresql/postgresql_test.cc` (all applicable C/C++ checks passed)
- `adbc-driver-postgresql-test` (242 passed, 6 skipped)
- `adbc-driver-postgresql-copy-test` (73 passed)
- Loaded 100 `extension<arrow.json>` values from the reported sample Parquet
file through the locally built driver into a PostgreSQL `jsonb` column without
a compatibility shim (100 inserted and validated)
Depends on #4499 for resolving existing PostgreSQL target column types during
COPY writer selection.
Closes #3293
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
