-
Notifications
You must be signed in to change notification settings - Fork 682
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
bleve v2.0.0 proposal PR #1494
bleve v2.0.0 proposal PR #1494
Commits on Nov 10, 2020
-
Configuration menu - View commit details
-
Copy full SHA for bb883f6 - Browse repository at this point
Copy the full SHA bb883f6View commit details
Commits on Nov 11, 2020
-
Configuration menu - View commit details
-
Copy full SHA for 81209a8 - Browse repository at this point
Copy the full SHA 81209a8View commit details
Commits on Nov 16, 2020
-
Configuration menu - View commit details
-
Copy full SHA for f25d522 - Browse repository at this point
Copy the full SHA f25d522View commit details
Commits on Nov 23, 2020
-
reduce index and segment API surface area (#1498)
Logic inside segment implementations has been moved up the stack into scorch, so that fewer methods are required in the API. - Iterator() becomes AutomatonIterator(nil, nil, nil) - RangeIterator(start, end) now computes an end key compatible with traditional inclusive start, exclusive end keys. Then becomes: AutomatonIterator(nil, start, endExclusive) - PrefixIterator(prefix) now computes an end key and becomes: AutomatonIterator(nil, prefix, prefixEnd) NumericRangeSearcher updated to no longer require the IndexReaderOnly interface. This interface was only used for non-scorch indexes, but only scorch actually implemented it. Removing this also meant we can remove the OnlyIterator method from the segment dictionary API.
Configuration menu - View commit details
-
Copy full SHA for c32ba08 - Browse repository at this point
Copy the full SHA c32ba08View commit details
Commits on Nov 24, 2020
-
move regexp compilation into scorch (#1500)
this code was originally placed in the segment package, however it is only used by scorch, and makes more sense to be placed at this level
Configuration menu - View commit details
-
Copy full SHA for b85c082 - Browse repository at this point
Copy the full SHA b85c082View commit details -
move in more things from the segment package (#1501)
the empty postings iterator is now only used by scorch's optimize.go, so it is relocated here the uvarint ascending encode/decode is only used by scorch, so it too is moved inside this package
Configuration menu - View commit details
-
Copy full SHA for 01f884d - Browse repository at this point
Copy the full SHA 01f884dView commit details -
move unadorned into scorch (#1502)
the unadorned postings lists are only used by scorch so they are relocated to this package
Configuration menu - View commit details
-
Copy full SHA for 8cd7ab1 - Browse repository at this point
Copy the full SHA 8cd7ab1View commit details
Commits on Nov 27, 2020
-
switch to tagged api versions (#1503)
note: while doing this it was observed that 'go mod tidy' was failing with an error. this relates to the kvstores in the blevex package which will needed to be updated separately. this change addresses this by removing the config_* and upsidedown/benchmark_* files which caused the problem. this does not remove support for these kvstores, just the build tag support.
Configuration menu - View commit details
-
Copy full SHA for 9d86800 - Browse repository at this point
Copy the full SHA 9d86800View commit details
Commits on Nov 30, 2020
-
update to latest versions (#1506)
also fixup some places we missed during refactor
Configuration menu - View commit details
-
Copy full SHA for 1460d32 - Browse repository at this point
Copy the full SHA 1460d32View commit details
Commits on Dec 2, 2020
-
move the plugin interface into scorch (#1507)
* move the plugin interface into scorch * rename scorch Plugin to SegmentPlugin and related functions * update to make all analysis take place in func() (#1508) * update to make all analysis take place in func() by making analysis take place inside a func() a closure can be used to encapsulate all the index specific details. this allows the index API to not know about scorch vs upsidedown analysis structures * rewrite variable declaration for clarity also update comment, per review feedback * update to latest versions selects latest versions compatible with these changes
Configuration menu - View commit details
-
Copy full SHA for cf6396d - Browse repository at this point
Copy the full SHA cf6396dView commit details
Commits on Dec 3, 2020
-
update Advanced() method (#1509)
* update Advanced() method at the top-level of bleve, Advanced no longer knows about K/V stores, instead it only returns reference to the internal index implementation. this will allow us to remove the Advanced() method from the internal index implementation. NOTE: users who previously needed to access the underlying K/V store, will still be able to do so, but must first type assert that the internal index impl is *UpsideDownCouch and if it is, they can use that Advanced() method directly * relocate store pkg inside upsidedown (#1510) * relocate store pkg inside upsidedown since kvstores are a concept unique to the upsidedown index implementation, the store folder should be relocated inside of that package * remove Advanced() method from scorch (#1511) * remove Advanced() method from scorch after the API is updated, scorch no longer requires this useless method to satisfy the interface * update to latest apis
Configuration menu - View commit details
-
Copy full SHA for 9548dc5 - Browse repository at this point
Copy the full SHA 9548dc5View commit details
Commits on Dec 4, 2020
-
make upsidedown store API ext module (#1512)
also, place shared kvstore test code in this module, not in upsidedown (avoids circular deps)
Configuration menu - View commit details
-
Copy full SHA for 9c73f9d - Browse repository at this point
Copy the full SHA 9c73f9dView commit details -
move FieldCache into upsidedown (#1513)
previous the FieldCache type was in the index package, but it is only used by upsidedown, so we relocate it there.
Configuration menu - View commit details
-
Copy full SHA for 1fcbc4f - Browse repository at this point
Copy the full SHA 1fcbc4fView commit details -
make ErrUnknownIndexType upsidedown specific (#1514)
this removes the top-level bleve err ErrUnknownIndexType instead, now the upsidedown specific error is returned. users who actually care and are checking for this error type still can, only now it is defined in the upsidedown package.
Configuration menu - View commit details
-
Copy full SHA for 010cf8f - Browse repository at this point
Copy the full SHA 010cf8fView commit details -
move regexp interface to searcher (#1515)
this interface is only used by the searcher package, thus it makes sense to define it there
Configuration menu - View commit details
-
Copy full SHA for bcacd48 - Browse repository at this point
Copy the full SHA bcacd48View commit details
Commits on Dec 6, 2020
-
remove use of dump methods from index reader (#1516)
the dump methods were only ever supposed by upsidedown now, users of this method must type-assert that their reader is from the upsidedown package, where these methods remain, unchanged.
Configuration menu - View commit details
-
Copy full SHA for 66ef9cf - Browse repository at this point
Copy the full SHA 66ef9cfView commit details
Commits on Dec 16, 2020
-
remove remaining blevex imports (#1520)
these cause circular dependencies and are easy for applications to do themselves
Configuration menu - View commit details
-
Copy full SHA for 0050aac - Browse repository at this point
Copy the full SHA 0050aacView commit details -
use the moved indexing options (#1521)
* use the moved indexing options * update analysis.TokenFrequency to use options (#1522) * update analysis.TokenFrequency to use options previously it used boolean, and future changes would require additional booleans. using options directly keeps the function signatures cleaner. * update to latest apis
Configuration menu - View commit details
-
Copy full SHA for 4da47c3 - Browse repository at this point
Copy the full SHA 4da47c3View commit details
Commits on Dec 18, 2020
-
update to work with removed items (#1524)
* update to work with removed items * update to latest api
Configuration menu - View commit details
-
Copy full SHA for 1af4ca2 - Browse repository at this point
Copy the full SHA 1af4ca2View commit details -
do not type assert on document or field (#1525)
instead use the interface exposed by index api
Configuration menu - View commit details
-
Copy full SHA for 3f65216 - Browse repository at this point
Copy the full SHA 3f65216View commit details
Commits on Dec 22, 2020
-
update for latest renaming (#1527)
* update for latest renaming * update to latest apis
Configuration menu - View commit details
-
Copy full SHA for a6e5e07 - Browse repository at this point
Copy the full SHA a6e5e07View commit details -
switch default index to scorch with latest zap (#1528)
one example required updating as it's custom sort contained a tie (both fields with age 0). upsidedown always returned the results in a particular order, which was not guaranteed by scorch. to address this, the exmaple has been updated to include the doc id as an additional sort criteria (ensuring the original order expected is kept by scorch)
Configuration menu - View commit details
-
Copy full SHA for 44887af - Browse repository at this point
Copy the full SHA 44887afView commit details
Commits on Dec 31, 2020
-
Configuration menu - View commit details
-
Copy full SHA for 3f670ba - Browse repository at this point
Copy the full SHA 3f670baView commit details -
Adding configurable freq/norm processing (#1526)
* Adding configurable freq/norm processing Useful for use cases where score relevancy isn't needed for the search requests. A special frequency value of zero/0 is used for skipping the freq/norm. * update to work with removed items (#1524) * update to work with removed items * update to latest api * do not type assert on document or field (#1525) instead use the interface exposed by index api * update for latest renaming (#1527) * update for latest renaming * update to latest apis * switch default index to scorch with latest zap (#1528) one example required updating as it's custom sort contained a tie (both fields with age 0). upsidedown always returned the results in a particular order, which was not guaranteed by scorch. to address this, the exmaple has been updated to include the doc id as an additional sort criteria (ensuring the original order expected is kept by scorch) * update to release candidate apis (#1531) Co-authored-by: Marty Schoch <marty.schoch@gmail.com>
Configuration menu - View commit details
-
Copy full SHA for d464b26 - Browse repository at this point
Copy the full SHA d464b26View commit details
Commits on Jan 5, 2021
-
restore correct copyright header (#1532)
it appears that while moving files around I mistakenly changed a copyright header, which was not my intention
Configuration menu - View commit details
-
Copy full SHA for 4f3ee06 - Browse repository at this point
Copy the full SHA 4f3ee06View commit details
Commits on Jan 6, 2021
-
MatchOperator should be that type instead of int (#1410)
Right now `MatchOperator` constants are ints and when you're trying to assign them to a variable you get an int. If you try to use those variables as operator you get the following error: ``` cannot use operator (type int) as type MatchQueryOperator in argument to q.SetOperator ``` So, to fix this problem we just need to create `MatchOperator` constants
Configuration menu - View commit details
-
Copy full SHA for 8e14fc9 - Browse repository at this point
Copy the full SHA 8e14fc9View commit details -
Configuration menu - View commit details
-
Copy full SHA for 987d561 - Browse repository at this point
Copy the full SHA 987d561View commit details
Commits on Jan 12, 2021
-
update to v1 apis and zapx versions (#1537)
* update to v1 apis and zapx versions * update to v1.0.1 upsidedown_store_api * make sort integration tests more precise the upsidedown implementation would break tie sort order in a predictable way, and some tests inadvertely relied on this behavior. these tests have been adjusted to use a more specific sort order, breaking all ties, such that the expected results no longer depend on implementation details.
Configuration menu - View commit details
-
Copy full SHA for eedb767 - Browse repository at this point
Copy the full SHA eedb767View commit details -
Configuration menu - View commit details
-
Copy full SHA for e843f84 - Browse repository at this point
Copy the full SHA e843f84View commit details