Viktor Lofgren
fa36689597
(index-reverse) Simplify priority index
...
* Do not emit a documents file
* Do not interlace metadata or offsets with doc ids
2024-07-06 18:04:08 +02:00
Viktor Lofgren
85c99ae808
(index-reverse) Split index construction into separate packages for full and priority index
2024-07-06 15:44:47 +02:00
Viktor Lofgren
a4ecd5f4ce
(minor) Fix non-compiling test due to previous refactor
2024-07-06 15:11:43 +02:00
Viktor Lofgren
a6b03a66dc
(crawl) Reduce Charset.forName() object churn
...
Cache the Charset object returned from Charset.forName() for future use, since we're likely to see the same charset again and Charset.forName(...) can be surprisingly expensive and its built-in caching strategy, which just caches the 2 last values seen doesn't cope well with how we're hitting it with a wide array of random charsets
2024-07-04 20:49:07 +02:00
Viktor Lofgren
d023e399d2
(index) Remove unnecessary allocations in journal reader
...
The term data iterator is quite hot and was performing buffer slice operations that were not necessary.
Replacing with a fixed pointer alias that can be repositioned to the relevant data.
The positions data was also being wrapped in a GammaCodedSequence only to be immediately un-wrapped.
Removed this unnecessary step and move to copying the buffer directly instead.
2024-07-04 15:38:22 +02:00
Viktor Lofgren
e8ab1e14e0
(keyword-extraction) Update upper limit to number of positions per word
...
After real-world testing, it was determined that 256 was still a bit too low, but 512 seems like it will only truncate outlier cases like assembly code and certain tabulations.
2024-07-02 20:52:32 +02:00
Viktor Lofgren
a6e15cb338
(keyword-extraction) Update upper limit to number of positions per word
...
100 was a bit too low, let's try 256.
2024-06-30 22:46:56 +02:00
Viktor Lofgren
4fbb863a10
(keyword-extraction) Add upper limit to number of positions per word
...
Also adding some logging for this event to get a feel for how big these lists get with realistic data. To be cleaned up later.
2024-06-30 22:41:38 +02:00
Viktor Lofgren
6ee4d1eb90
(keyword) Increase the work area for position encoding
...
The change also moves the allocation outside of the build()-method to allow re-use of this rather large temporary buffer.
2024-06-28 16:42:39 +02:00
Viktor Lofgren
738e0e5fed
(process) Add option for automatic profiling
...
The change adds a new system property 'system.profile' that makes ProcessService automatically trigger JFR profiling of the processes it spawns. By default, these are put in the log directory.
The change also adds a JVM parameter that makes it shut up about native access.
2024-06-27 13:58:36 +02:00
Viktor Lofgren
0e4dd3d76d
(minor) Remove accidentally committed debug printf
2024-06-27 13:40:53 +02:00
Viktor Lofgren
10fe5a78cb
(log) Prevent tests from trying to log to file
...
They would never have succeeded, but it adds an annoying preamble of error spam in the console window.
2024-06-27 13:19:48 +02:00
Viktor Lofgren
975b8ae2e9
(minor) Tidy code
2024-06-27 13:15:31 +02:00
Viktor Lofgren
935234939c
(test) Add query parsing to IntegrationTest
2024-06-27 13:15:20 +02:00
Viktor Lofgren
87e38e6181
(search-query) refac: Move query factory
2024-06-27 13:14:47 +02:00
Viktor Lofgren
f73fc8dd57
(search-query) Fix end-inclusion bug in QWordGraphIterator
2024-06-27 13:13:42 +02:00
Viktor Lofgren
3faa5bf521
(search-query) Tidy up QueryGRPCService and IndexClient
2024-06-26 14:03:30 +02:00
Viktor Lofgren
6973712480
(query) Tidy up code
2024-06-26 13:40:06 +02:00
Viktor Lofgren
02df421c94
(*) Trim the stopwords list
...
Having an overlong stopwords list leads to quoted terms not performing well. For now we'll slash it to just "a" and "the".
2024-06-26 12:22:57 +02:00
Viktor Lofgren
95b9af92a0
(index) Implement working optional TermCoherences
2024-06-26 12:22:06 +02:00
Viktor Lofgren
8ee64c0771
(index) Correct TermCoherence requirements
2024-06-25 22:18:10 +02:00
Viktor Lofgren
b805f6daa8
(gamma) Fix readCount() behavior in EGC
2024-06-25 22:17:54 +02:00
Viktor Lofgren
dae22ccbe0
(test) Integration test from crawl->query
2024-06-25 22:17:26 +02:00
Viktor Lofgren
9d00243d7f
(index) Partial re-implementation of position constraints
2024-06-24 15:55:54 +02:00
Viktor Lofgren
5461634616
(doc) Add readme.md for coded-sequence library
...
This commit introduces a readme.md file to document the functionality and usage of the coded-sequence library. It covers the Elias Gamma code support, how sequences are encoded, and methods the library offers to query sequences, iterate over values, access data, and decode sequences.
2024-06-24 14:28:51 +02:00
Viktor Lofgren
40bca93884
(gamma) Minor clean-up
2024-06-24 13:56:43 +02:00
Viktor Lofgren
b798f28443
(journal) Fixing journal encoding
...
Adjusting some bit widths for entry and record sizes to ensure these don't overflow, as this would corrupt the written journal.
2024-06-24 13:56:27 +02:00
Viktor Lofgren
fff2ce5721
(gamma) Correctly decode zero-length sequences
2024-06-24 13:11:41 +02:00
Viktor Lofgren
23759a7243
(loader) Correctly clamp document size
2024-06-10 18:29:14 +02:00
Viktor Lofgren
55b2b7636b
(loader) Correctly load the positions column in the keyword projection
2024-06-10 18:27:15 +02:00
Viktor Lofgren
36160988e2
(index) Integrate positions data with indexes WIP
...
This change integrates the new positions data with the forward and reverse indexes.
The ranking code is still only partially re-written.
2024-06-10 15:09:06 +02:00
Viktor Lofgren
9f982a0c3d
(index) Integrate positions file properly
2024-06-06 16:45:42 +02:00
Viktor Lofgren
dcbec9414f
(index) Fix non-compiling tests
2024-06-06 16:35:09 +02:00
Viktor Lofgren
4a8afa6b9f
(index, WIP) Position data partially integrated with forward and reverse indexes.
...
There's no graceful way of doing this in small commits, pushing to avoid the risk of data loss.
2024-06-06 12:54:52 +02:00
Viktor Lofgren
9b922af075
(converter) Amend existing modifications to use gamma coded positions lists
...
... instead of serialized RoaringBitmaps as was the initial take on the problem.
2024-05-30 14:20:36 +02:00
Viktor Lofgren
0112ae725c
(gamma) Implement a small library for Elias gamma coding an integer sequence
2024-05-30 14:19:13 +02:00
Viktor Lofgren
619392edf9
(keywords) Add position information to keywords
2024-05-28 16:54:53 +02:00
Viktor Lofgren
0894822b68
(converter) Add position information to serialized document data
...
This is not hooked in yet, and the term metadata is still left intact. It should probably shrink to a smaller representation (byte?) with the upcoming removal of the position mask.
2024-05-28 14:18:03 +02:00
Viktor Lofgren
a69ab311c7
(qword) Fix tests that broke due to stopword removal
2024-05-28 14:15:45 +02:00
Viktor Lofgren
6985ab762a
(query) Improve handling of stopwords in queries
2024-05-23 20:50:55 +02:00
Viktor Lofgren
0e8300979b
(search) Update the no result text to request bug reports.
2024-05-23 20:18:16 +02:00
Viktor Lofgren
0b60411e5f
(query) Bugfix stopword issue
...
Add a new rule that crates an alternative path that omits a word if it's a stopword.
In queries where a stopword is present, and no query ngram expansion is possible, the query should not require the stopword to be present in the index, as this results in no search results being found.
2024-05-23 20:15:14 +02:00
Viktor Lofgren
f83f777fff
(converter) Experimental support for searching by URL
...
Add up to synthetic 128 keywords per document, corresponding to links to other websites.
2024-05-23 17:10:57 +02:00
Viktor Lofgren
89aae93e60
(*) Lift jetty and guava-dependencies
2024-05-23 14:20:01 +02:00
Viktor Lofgren
65b74f9cab
(registry) Fix broken test
2024-05-23 14:15:01 +02:00
Viktor Lofgren
59ec70eb73
(*) Clean up code related to crawl parquet inspection
2024-05-22 12:55:08 +02:00
Viktor Lofgren
365229991b
(control) Improve pagination for crawl data inspector
2024-05-21 19:44:48 +02:00
Viktor Lofgren
959a8e29ee
(control) Improve pagination for crawl data inspector
2024-05-21 19:27:25 +02:00
Viktor Lofgren
197c82acd4
(control) Add filter functionality for crawl data inspector
2024-05-21 19:05:44 +02:00
Viktor Lofgren
9539fdb53c
(control) Clean up UX for crawl data inspector
2024-05-21 18:27:24 +02:00
