1312 Commits

Author SHA1 Message Date
Kishore Nallan
b18e2a41b3 Early exit for exact match calculation based on distance. 2021-08-29 14:36:05 +05:30
Kishore Nallan
adc816e662 Use token separators whule parsing search query as well. 2021-08-28 20:59:05 +05:30
Kishore Nallan
9659d60047 Exhaustive search should ignore typo and drop token thresholds. 2021-08-28 19:33:40 +05:30
Kishore Nallan
b6f1885aec Stricter bounding of typo correction threshold. 2021-08-28 16:38:07 +05:30
Kishore Nallan
ce7b6e12e9 Prioritize record with a field containing all tokens in the query. 2021-08-27 20:52:51 +05:30
Kishore Nallan
07d838e385 Make symbols for indexing and segmentation configurable. 2021-08-26 10:27:18 +05:30
Kishore Nallan
a931bb4b2a Handle highlighting on a field with empty array value. 2021-08-25 17:05:06 +05:30
Kishore Nallan
67e67b7d06 Add array index check for highlight of string array. 2021-08-25 07:06:43 +05:30
Kishore Nallan
d4bd6e67e5 Further tweak exact match logic. 2021-08-22 15:47:21 +05:30
Kishore Nallan
2df55e7991 Fix exact value matching. 2021-08-22 13:45:26 +05:30
Kishore Nallan
ced4163062 Handle stack trace on batched indexing. 2021-08-22 08:43:32 +05:30
Kishore Nallan
76915eed20 Speed up exact matching on single tokens. 2021-08-19 16:43:53 +05:30
Kishore Nallan
793e21a1c2 Ensure that search does not fetch existing tokens. 2021-08-18 18:51:39 +05:30
Kishore Nallan
f00f756535 Fix queue for collection creation write. 2021-08-18 07:24:35 +05:30
Kishore Nallan
0e2adb4242 Copy-free intersect + score. 2021-08-17 18:37:42 +05:30
Kishore Nallan
22670b1342 Skip offset compute for non-exact single token. 2021-08-13 16:22:57 +05:30
Kishore Nallan
b24df3c26a Allow read/write lags to be updated via config API. 2021-08-11 20:28:47 +05:30
Kishore Nallan
a890d036ed Account for queued writes when computing read+write lags. 2021-08-11 18:18:49 +05:30
Kishore Nallan
26351a6984 Change default value of typo/drop tokens threshold to 1. 2021-08-11 14:20:28 +05:30
Kishore Nallan
8c46fed1b4 Return error if import is attempted on unsupported curl version. 2021-08-10 20:55:37 +05:30
Kishore Nallan
812ed37496 Bump h2o dep. 2021-08-10 20:55:08 +05:30
Kishore Nallan
1e1e4c8e07 Close connection after follower is done proxying import. 2021-08-09 19:10:47 +05:30
Kishore Nallan
203230e051 Upgrade CURL. 2021-08-09 18:13:05 +05:30
Kishore Nallan
044be85006 Change log level to warning when nodes file is empty. 2021-08-08 08:13:39 +05:30
Kishore Nallan
7703939430 Parameterize combination limit for more exhaustive searching. 2021-08-07 18:13:29 +05:30
Kishore Nallan
55535198a4 Prefix search to be used only for last token. 2021-08-07 13:12:06 +05:30
Kishore Nallan
27ebae1f88 Address valgrind warnings. 2021-08-06 17:09:28 +05:30
Kishore Nallan
b5e3a28ace More fixes for highlighting. 2021-08-05 21:31:04 +05:30
Kishore Nallan
3abb3896fb Bug: use resize instead of reserve. 2021-08-05 21:03:21 +05:30
Kishore Nallan
eedf791a8d Reduce logging. 2021-08-01 19:20:26 +05:30
Kishore Nallan
35409f8493 Batched background indexer.
Enables write parallelism across collections.
2021-07-31 21:44:06 +05:30
Kishore Nallan
dedb6669c8 Upgrade RocksDB. 2021-07-31 21:38:45 +05:30
Kishore Nallan
cf96d9fee3 Disable TLS for glog as workaround for a bug.
See: https://github.com/google/glog/issues/409
2021-07-31 21:38:28 +05:30
Kishore Nallan
261536d0f4 Merge branch '0.22.0-rc' into postings-refactor-integration
# Conflicts:
#	src/collection.cpp
#	src/index.cpp
#	test/collection_specific_test.cpp
2021-07-31 21:35:30 +05:30
Kishore Nallan
b2c12a9b2c Fix more edge cases in highlighting. 2021-07-31 08:59:49 +05:30
Kishore Nallan
331db4f27e Add precision option to geo field sorting. 2021-07-27 19:57:56 +05:30
Kishore Nallan
13cb7b9364 Revert "Highlight field value that is a prefix of the query."
This reverts commit 545027a59bc55b24c2fece112b4fa6a655a1f79e.

# Conflicts:
#	test/collection_specific_test.cpp
2021-07-27 17:57:49 +05:30
Kishore Nallan
b4c222064c Handle bad data in ingestion text gracefully. 2021-07-26 19:44:38 +05:30
Kishore Nallan
e45f18785f Ignore id field present in schema. 2021-07-26 19:44:10 +05:30
Kishore Nallan
38d44a7c8a Highlight field value that is a prefix of the query. 2021-07-26 15:33:03 +05:30
Kishore Nallan
41c16fb7a7 Merge branch '0.22.0-rc' into postings-refactor-integration
# Conflicts:
#	include/index.h
#	include/posting.h
#	include/posting_list.h
#	src/art.cpp
#	src/collection.cpp
#	src/index.cpp
#	src/posting.cpp
#	src/posting_list.cpp
#	test/art_test.cpp
#	test/collection_specific_test.cpp
#	test/collection_test.cpp
#	test/posting_list_test.cpp
2021-07-24 17:10:54 +05:30
Kishore Nallan
89a509513a Ensure that weights can fully control cross-field matching. 2021-07-24 15:08:08 +05:30
Kishore Nallan
e42f78a695 Fix single character full field value highlight. 2021-07-21 19:26:09 +05:30
Kishore Nallan
672c895805 Typo and drop tokens thresholds must be applied independently. 2021-07-16 13:39:52 +05:30
Kishore Nallan
0ae718d067 Use all candidates of a given num_typo value.
Typo tokens threshold should not trigger when we have explored only some of the candidates of a given num_typo value.
2021-07-16 12:07:44 +05:30
Kishore Nallan
bfb122bfec Repeating tokens in an array: fix relevancy. 2021-07-15 15:58:55 +05:30
Kishore Nallan
21fd7f8e33 Update README 2021-07-14 11:45:26 +05:30
Kishore Nallan
a062dbbb3f Fix peering address environment variable assignment. 2021-07-14 11:44:01 +05:30
Kishore Nallan
56247ce6ac Prefix match must be differentiated from single typo. 2021-07-14 11:44:01 +05:30
Kishore Nallan
53ad55ae18 Fix cpu usage metric. 2021-07-14 11:44:01 +05:30