as per the ft.search docs ( https://redis.io/docs/latest/commands/ft.search ):
If a relevant key expires while a query is running,
an attempt to load the key's value will return a null array.
However, the key is still counted in the total number of results.
So, instead of crashing when seeing a null as a value, we
return empty object.
fixes#2772
see https://github.com/redis/node-redis/pull/2814
* release-it/bumper
* remove git:false
* fix package ordering
* adjust git add
* fix git config
* adjust git config for all packages
* add noop release script for test-utils
* no need to try to release root
* better way to handle skipping
* pass parameters down
* better version hint
* update node version
* return git arguments from before
* rename release workflow
* rename workflow
* set git.tagMatch
* add link to docs
* update description
* update workspace order in package-lock
* fix secondary releases
release-it/bumper was removing the ^ before the peerDep to client
npm is not happy with that. one potential fix would be to bump all
packages together as a prestep and then proceed without bupming again.
for now, this fix should bring us to the previous state ( what was used
in the manual process )
* require clean working dir in root
* remove root release-it config
not needed
* (docs) bloom: add jsdocs for all commands
* (docs) json: add jsdocs
* (docs) search: add jsdocs for all commands
* (docs) jsdocs for std commands
* (docs) jsdoc comments to time series commands
> FT.AGGREGATE returns an array reply where each row is an array reply and represents a single aggregate result.
The integer reply at position 1 does not represent a valid value.
https://redis.io/docs/latest/commands/ft.aggregate/#return
> FT.AGGREGATE returns an array reply where each row is an array reply
and represents a single aggregate result.
> The integer reply at position 1 does not represent a valid value.
We now calculate the result length bazed on the number of results
instead of the integer reply at pos 1
* churn(test): use redislabs/client-libs-test for testing
This switches our testing infrastructure from redis/redis-stack to
redislabs/client-libs-test Docker image across all packages. This change
also updates the default Docker version from 7.4.0-v1 to 8.0-M04-pre.
* churn(test): verify CONFIG SET / GET compatibility with Redis 8
- Add tests for Redis 8 search configuration settings
- Deprecate Redis Search CONFIG commands in favor of standard CONFIG
- Test read-only config restrictions for Redis 8
* churn(test): handle Redis 8 coordinate precision in GEOPOS
- Update GEOPOS tests to handle increased precision in Redis 8 (17 decimal places vs 14)
- Add precision-aware coordinate comparison helper
- Add comprehensive test suite for coordinate comparison function
* test(search): adapt SUGGET tests for Redis 8 empty results
- Update tests to expect empty array ([]) instead of null for SUGGET variants
- Affects sugGet, sugGetWithPayloads, sugGetWithScores, and sugGetWithScoresWithPayloads
* test(search): support Redis 8 INFO indexes_all field
- Add indexes_all field introduced in Redis 8 to index definition test
* refactor!(search): simplify PROFILE commands to return raw response
- BREAKING CHANGE: FT.PROFILE now returns raw response, letting users implement their own parsing
* test: improve version-specific test coverage
- Add `testWithClientIfVersionWithinRange` method to run tests for specific Redis versions
- Refactor TestUtils to handle version comparisons more accurately
- Update test utilities across Redis modules to run tests against multiple versions, and not against latest only
* Parser support with all commands
* remove "dist" from all imports for consistency
* address most of my review comments
* small tweak to multi type mapping handling
* tweak multi commands / fix addScript cases
* nits
* addressed all in person review comments
* revert addCommand/addScript changes to multi-commands
addCommand needs to be there for sendCommand like ability within a multi.
If its there, it might as well be used by createCommand() et al, to avoid repeating code.
addScript is there (even though only used once), but now made private to keep the logic for bookkeeping near each other.
RESP3 Support
- Some commands responses in RESP3 aren't stable yet and therefore return an "untyped" ReplyUnion.
Sentinel
TypeMapping
Correctly types Multi commands
Note: some API changes to be further documented in v4-to-v5.md
* small refactor per discussion with leibele
* move true type to boolean type
* fix geoshape to support NOINDEX & SORTABLE, clean code
* fix for last commit
---------
Co-authored-by: Leibale Eidelman <me@leibale.com>
* Add support for NOCONTENT in FT.SEARCH
* Move support for NOCONTENT search option from client.search to client.searchNoContent
* Add test for SEARCH_NOCONTENT#transformReply
* Fix typo
* Enable test
* Update test field type
---------
Co-authored-by: Leibale <me@leibale.com>
* #2486: add timeout as optional param in FT.Search
* return timeout from aggregate
* add test case for TIMEOUT in aggregate
* add TIMEOUT option in search file
* add test cases for TIMEOUT option in search file
* uodate search/aggregates to add timeout when it is not undefuned
* update search to add timeout when it is not undefuned
* update test case for AGGREGATE
