Commit Graph

187 Commits

Author SHA1 Message Date
Loïc Lecrenier
264a04922d Add prefix_word_pair_proximity database
Similar to the word_prefix_pair_proximity one but instead the keys are:
(proximity, prefix, word2)
2022-10-18 10:37:34 +02:00
Kerollmops
6603437cb1
Introduce an indexation abortion function when indexing documents 2022-10-17 17:28:03 +02:00
Ewan Higgs
beb987d3d1 Fixing piles of clippy errors.
Most of these are calling clone when the struct supports Copy.

Many are using & and &mut on `self` when the function they are called
from already has an immutable or mutable borrow so this isn't needed.

I tried to stay away from actual changes or places where I'd have to
name fresh variables.
2022-10-13 22:02:54 +02:00
bors[bot]
15d478cf4d
Merge #635
635: Use an unstable algorithm for `grenad::Sorter` when possible r=Kerollmops a=loiclec

# Pull Request
## What does this PR do?

Use an unstable algorithm to sort the internal vector used by `grenad::Sorter` whenever possible to speed up indexing.

In practice, every time the merge function creates a `RoaringBitmap`, we use an unstable sort. For every other merge function, such as `keep_first`, `keep_last`, etc., a stable sort is used.


Co-authored-by: Loïc Lecrenier <loic@meilisearch.com>
2022-09-14 12:00:52 +00:00
Loïc Lecrenier
3794962330 Use an unstable algorithm for grenad::Sorter when possible 2022-09-13 14:49:53 +02:00
Kerollmops
d4d7c9d577
We avoid skipping errors in the indexing pipeline 2022-09-13 14:03:00 +02:00
Kerollmops
c83c3cd796
Add a test to make sure that long words are correctly skipped 2022-09-07 14:12:36 +02:00
ManyTheFish
5391e3842c replace optional_words by term_matching_strategy 2022-08-22 17:47:19 +02:00
ManyTheFish
9640976c79 Rename TermMatchingPolicies 2022-08-18 17:36:08 +02:00
Irevoire
e96b852107
bump heed 2022-08-17 17:05:50 +02:00
ManyTheFish
e9e2349ce6 Fix typo in comment 2022-08-17 15:09:48 +02:00
ManyTheFish
2668f841d1 Fix update indexing 2022-08-17 15:03:37 +02:00
ManyTheFish
7384650d85 Update test to showcase the bug 2022-08-17 15:03:08 +02:00
Loïc Lecrenier
58cb1c1bda Simplify unit tests in facet/filter.rs 2022-08-04 12:03:44 +02:00
Loïc Lecrenier
acff17fb88 Simplify indexing tests 2022-08-04 12:03:13 +02:00
bors[bot]
21284cf235
Merge #556
556: Add EXISTS filter r=loiclec a=loiclec

## What does this PR do?

Fixes issue [#2484](https://github.com/meilisearch/meilisearch/issues/2484) in the meilisearch repo.

It creates a `field EXISTS` filter which selects all documents containing the `field` key. 
For example, with the following documents:
```json
[{
	"id": 0,
	"colour": []
},
{
	"id": 1,
	"colour": ["blue", "green"]
},
{
	"id": 2,
	"colour": 145238
},
{
	"id": 3,
	"colour": null
},
{
	"id": 4,
	"colour": {
		"green": []
	}
},
{
	"id": 5,
	"colour": {}
},
{
	"id": 6
}]
```
Then the filter `colour EXISTS` selects the ids `[0, 1, 2, 3, 4, 5]`. The filter `colour NOT EXISTS` selects `[6]`.

## Details
There is a new database named `facet-id-exists-docids`. Its keys are field ids and its values are bitmaps of all the document ids where the corresponding field exists.

To create this database, the indexing part of milli had to be adapted. The implementation there is basically copy/pasted from the code handling the `facet-id-f64-docids` database, with appropriate modifications in place.

There was an issue involving the flattening of documents during (re)indexing. Previously, the following JSON:
```json
{
    "id": 0,
    "colour": [],
    "size": {}
}
```
would be flattened to:
```json
{
    "id": 0
}
```
prior to being given to the extraction pipeline.

This transformation would lose the information that is needed to populate the `facet-id-exists-docids` database. Therefore, I have also changed the implementation of the `flatten-serde-json` crate. Now, as it traverses the Json, it keeps track of which key was encountered. Then, at the end, if a previously encountered key is not present in the flattened object, it adds that key to the object with an empty array as value. For example:
```json
{
    "id": 0,
    "colour": {
        "green": [],
        "blue": 1
    },
    "size": {}
} 
```
becomes
```json
{
    "id": 0,
    "colour": [],
    "colour.green": [],
    "colour.blue": 1,
    "size": []
} 
```


Co-authored-by: Kerollmops <clement@meilisearch.com>
2022-08-04 09:46:06 +00:00
ManyTheFish
d6f9a60a32 fix: Remove whitespace trimming during document id validation
fix #592
2022-08-03 11:38:40 +02:00
Loïc Lecrenier
07003704a8 Merge branch 'filter/field-exist' 2022-07-21 14:51:41 +02:00
Loïc Lecrenier
1eb1e73bb3 Add integration tests for the EXISTS filter 2022-07-19 10:07:33 +02:00
Loïc Lecrenier
c17d616250 Refactor index_documents_check_exists_database tests 2022-07-19 10:07:33 +02:00
Loïc Lecrenier
453d593ce8 Add a database containing the docids where each field exists 2022-07-19 10:07:33 +02:00
Kerollmops
192793ee38
Add some tests to check for the nested documents ids 2022-07-12 15:14:07 +02:00
Kerollmops
dc61105554
Fix the nested document id fetching function 2022-07-12 15:14:06 +02:00
Kerollmops
2eec290424
Check the validity of the latitute and longitude numbers 2022-07-12 15:14:06 +02:00
Kerollmops
0bbcc7b180
Expose the DocumentId struct to be sure to inject the generated ids 2022-07-12 15:14:06 +02:00
Kerollmops
c8ebf0de47
Rename the validate function as an enriching function 2022-07-12 15:14:06 +02:00
Kerollmops
6a0a0ae94f
Make the Transform read from an EnrichedDocumentsBatchReader 2022-07-12 14:55:52 +02:00
Kerollmops
8ebf5eed0d
Make the nested primary key work 2022-07-12 14:55:52 +02:00
Kerollmops
19eb3b4708
Make sur that we do not accept floats as documents ids 2022-07-12 14:55:52 +02:00
Kerollmops
2ceeb51c37
Support the auto-generated ids when validating documents 2022-07-12 14:55:51 +02:00
Kerollmops
399eec5c01
Fix the indexation tests 2022-07-12 14:55:51 +02:00
Kerollmops
0146175fe6
Introduce the validate_documents_batch function 2022-07-12 14:55:51 +02:00
Kerollmops
e8297ad27e
Fix the tests for the new DocumentsBatchBuilder/Reader 2022-07-12 14:52:56 +02:00
Tamo
eaf28b0628
Apply review suggestions
Co-authored-by: Clément Renault <clement@meilisearch.com>
2022-07-05 15:30:33 +02:00
Tamo
3b309f654a
Fasten the document deletion
When a document deletion occurs, instead of deleting the document we mark it as deleted
in the new “soft deleted” bitmap. It is then removed from the search, and all the other
endpoints.
2022-07-05 15:30:33 +02:00
ad hoc
31776fdc3f
add failing test 2022-06-07 15:49:33 +02:00
bors[bot]
08c6d50cd1
Merge #531
531: fix the mixed dataset geosearch indexing bug r=Kerollmops a=irevoire

port #529 to main

Co-authored-by: Tamo <tamo@meilisearch.com>
2022-05-16 16:06:36 +00:00
Tamo
0af399a6d7
fix the mixed dataset geosearch indexing bug 2022-05-16 17:37:45 +02:00
Tamo
f586028f9a
fix the searchable fields bug when a field is nested
Update milli/src/index.rs

Co-authored-by: Clément Renault <clement@meilisearch.com>
2022-05-16 17:24:36 +02:00
bors[bot]
65e6aa0de2
Merge #523
523: Improve geosearch error messages r=irevoire a=irevoire

Improve the geosearch error messages (#488).
And try to parse the string as specified in https://github.com/meilisearch/meilisearch/issues/2354

Co-authored-by: Tamo <tamo@meilisearch.com>
2022-05-04 13:36:11 +00:00
Kerollmops
211c8763b9
Make sure that we do not generate too long keys 2022-05-03 10:03:15 +02:00
Kerollmops
7e47031bdc
Add a test for long keys in LMDB 2022-05-03 10:03:13 +02:00
Tamo
3cb1f6d0a1
improve geosearch error messages 2022-05-02 19:20:47 +02:00
Tamo
f19d2dc548
Only flatten the required fields
apply review comments

Co-authored-by: Kerollmops <kero@meilisearch.com>
2022-04-26 12:33:46 +02:00
Clément Renault
eb5830aa40
Add a test to make sure that long words are handled 2022-04-21 13:45:28 +02:00
Tamo
ee64f4a936
Use smartstring to store the external id in our hashmap
We need to store all the external id (primary key) in a hashmap
associated to their internal id during.
The smartstring remove heap allocation / memory usage and should
improve the cache locality.
2022-04-13 21:22:07 +02:00
Irevoire
4f3ce6d9cd
nested fields 2022-04-07 16:58:46 +02:00
ad hoc
e8f06f6c06
extract exact_word_prefix_docids 2022-04-04 20:54:03 +02:00
ad hoc
ba0bb29cd8
refactor WordPrefixDocids to take dbs instead of indexes 2022-04-04 20:54:02 +02:00
ad hoc
c4c6e35352
query exact_word_docids in resolve_query_tree 2022-04-04 20:54:02 +02:00
ad hoc
8d46a5b0b5
extract exact word docids 2022-04-04 20:54:02 +02:00
ad hoc
0a77be4ec0
introduce exact_word_docids db 2022-04-04 20:54:02 +02:00
Kerollmops
d5b8b5a2f8
Replace the ugly unwraps by clean if let Somes 2022-02-28 16:31:33 +01:00
Kerollmops
8d26f3040c
Remove a useless grenad file merging 2022-02-28 16:31:33 +01:00
bors[bot]
25123af3b8
Merge #436
436: Speed up the word prefix databases computation time r=Kerollmops a=Kerollmops

This PR depends on the fixes done in #431 and must be merged after it.

In this PR we will bring the `WordPrefixPairProximityDocids`, `WordPrefixDocids` and, `WordPrefixPositionDocids` update structures to a new era, a better era, where computing the word prefix pair proximities costs much fewer CPU cycles, an era where this update structure can use the, previously computed, set of new word docids from the newly indexed batch of documents.

---

The `WordPrefixPairProximityDocids` is an update structure, which means that it is an object that we feed with some parameters and which modifies the LMDB database of an index when asked for. This structure specifically computes the list of word prefix pair proximities, which correspond to a list of pairs of words associated with a proximity (the distance between both words) where the second word is not a word but a prefix e.g. `s`, `se`, `a`. This word prefix pair proximity is associated with the list of documents ids which contains the pair of words and prefix at the given proximity.

The origin of the performances issue that this struct brings is related to the fact that it starts its job from the beginning, it clears the LMDB database before rewriting everything from scratch, using the other LMDB databases to achieve that. I hope you understand that this is absolutely not an optimized way of doing things.

Co-authored-by: Clément Renault <clement@meilisearch.com>
Co-authored-by: Kerollmops <clement@meilisearch.com>
2022-02-16 15:41:14 +00:00
Clément Renault
ff8d7a810d
Change the behavior of the as_cloneable_grenad by taking a ref 2022-02-16 15:40:08 +01:00
Many
d59bcea749 Revert "Revert "Change chunk size to 4MiB to fit more the end user usage"" 2022-02-02 17:01:13 +01:00
Kerollmops
fb79c32430
Compute the new, common and, deleted prefix words fst once 2022-01-27 11:00:18 +01:00
Clément Renault
51d1e64b23
Remove, now useless, the WriteMethod enum 2022-01-27 10:08:35 +01:00
Clément Renault
e9c02173cf
Rework the WordsPrefixPositionDocids update to compute a subset of the database 2022-01-27 10:08:35 +01:00
Clément Renault
d59e559317
Fix the computation of the newly added and common prefix words 2022-01-27 10:08:34 +01:00
Clément Renault
28692f65be
Rework the WordPrefixDocids update to compute a subset of the database 2022-01-27 10:08:34 +01:00
Clément Renault
5404bc02dd
Move the fst_stream_into_hashset method in the helper methods 2022-01-27 10:06:00 +01:00
Clément Renault
822f67e9ad
Bring the newly created word pair proximity docids 2022-01-27 10:06:00 +01:00
Clément Renault
d28f18658e
Retrieve the previous version of the words prefixes FST 2022-01-27 10:05:59 +01:00
Marin Postma
0c84a40298 document batch support
reusable transform

rework update api

add indexer config

fix tests

review changes

Co-authored-by: Clément Renault <clement@meilisearch.com>

fmt
2022-01-19 12:40:20 +01:00
Marin Postma
6eb47ab792 remove update_id in UpdateBuilder 2021-11-16 13:07:04 +01:00
Marin Postma
09b4281cff improve document addition returned metaimprove document addition
returned metaimprove document addition returned metaimprove document
addition returned metaimprove document addition returned metaimprove
document addition returned metaimprove document addition returned
metaimprove document addition returned meta
2021-11-10 14:08:36 +01:00
marin postma
baddd80069
implement review suggestions 2021-10-25 18:29:12 +02:00
marin postma
2e62925a6e
fix tests 2021-10-25 10:26:42 +02:00
marin postma
8d70b01714
optimize document deserialization 2021-10-25 10:26:42 +02:00
many
c5a6075484
Make max_position_per_attributes changable 2021-10-12 10:10:50 +02:00
many
360c5ff3df
Remove limit of 1000 position per attribute
Instead of using an arbitrary limit we encode the absolute position in a u32
using one strong u16 for the field id and a weak u16 for the relative position in the attribute.
2021-10-12 10:10:50 +02:00
many
3296bb243c
Simplify word level position DB into a word position DB 2021-10-05 12:15:02 +02:00
Many
26b5dad042
Revert "Change chunk size to 4MiB to fit more the end user usage" 2021-09-29 15:08:39 +02:00
Tamo
f65153ad64
stop casting integer docids to string 2021-09-28 18:35:54 +02:00
many
1988416295
Add failing test related to Meilisearch#1714 2021-09-28 12:05:11 +02:00
many
b188063869
Change chunk size to 4MiB to fit more the end user usage 2021-09-27 14:26:21 +02:00
many
551df0cb77
Add test checking the bug reported in meilisearch issue 1716 2021-09-23 15:55:39 +02:00
mpostma
aa6c5df0bc Implement documents format
document reader transform

remove update format

support document sequences

fix document transform

clean transform

improve error handling

add documents! macro

fix transform bug

fix tests

remove csv dependency

Add comments on the transform process

replace search cli

fmt

review edits

fix http ui

fix clippy warnings

Revert "fix clippy warnings"

This reverts commit a1ce3cd96e603633dbf43e9e0b12b2453c9c5620.

fix review comments

remove smallvec in transform loop

review edits
2021-09-21 16:58:33 +02:00
bors[bot]
31c8de1cca
Merge #322
322: Geosearch r=ManyTheFish a=irevoire

This PR introduces [basic geo-search functionalities](https://github.com/meilisearch/specifications/pull/59), it makes the engine able to index, filter and, sort by geo-point. We decided to use [the rstar library](https://docs.rs/rstar) and to save the points in [an RTree](https://docs.rs/rstar/0.9.1/rstar/struct.RTree.html) that we de/serialize in the index database [by using serde](https://serde.rs/) with [bincode](https://docs.rs/bincode). This is not an efficient way to query this tree as it will consume a lot of CPU and memory when a search is made, but at least it is an easy first way to do so.

### What we will have to do on the indexing part:
 - [x] Index the `_geo` fields from the documents.
   - [x] Create a new module with an extractor in the `extract` module that takes the `obkv_documents` and retrieves the latitude and longitude coordinates, outputting them in a `grenad::Reader` for further process.
   - [x] Call the extractor in the `extract::extract_documents_data` function and send the result to the `TypedChunk` module.
   - [x] Get the `grenad::Reader` in the `typed_chunk::write_typed_chunk_into_index` function and store all the points in the `rtree`
- [x] Delete the documents from the `RTree` when deleting documents from the database. All this can be done in the `delete_documents.rs` file by getting the data structure and removing the points from it, inserting it back after the modification.
- [x] Clearing the `RTree` entirely when we clear the documents from the database, everything happens in the `clear_documents.rs` file.
- [x] save a Roaring bitmap of all documents containing the `_geo` field

### What we will have to do on the query part:
- [x] Filter the documents at a certain distance around a point, this is done by [collecting the documents from the searched point](https://docs.rs/rstar/0.9.1/rstar/struct.RTree.html#method.nearest_neighbor_iter) while they are in range.
  - [x] We must introduce new `geoLowerThan` and `geoGreaterThan` variants to the `Operator` filter enum.
  - [x] Implement the `negative` method on both variants where the `geoGreaterThan` variant is implemented by executing the `geoLowerThan` and removing the results found from the whole list of geo faceted documents.
  - [x] Add the `_geoRadius` function in the pest parser.
- [x] Introduce a `_geo` ascending ranking function that takes a point in parameter, ~~this function must keep the iterator on the `RTree` and make it peekable~~ This was not possible for now, we had to collect the whole iterator. Only the documents that are part of the candidates must be sent too!
  - [x] This ascending ranking rule will only be active if the search is set up with the `_geoPoint` parameter that indicates the center point of the ascending ranking rule.

-----------

- On Meilisearch part: We must introduce a new concept, returning the documents with a new `_geoDistance` field when it passed by the `_geo` ranking rule, this has never been done before. We could maybe just do it afterward when the documents have been retrieved from the database, computing the distance from the `_geoPoint` and all of the documents to be returned.

Co-authored-by: Irevoire <tamo@meilisearch.com>
Co-authored-by: cvermand <33010418+bidoubiwa@users.noreply.github.com>
Co-authored-by: Tamo <tamo@meilisearch.com>
2021-09-20 19:04:57 +00:00
many
26deeb45a3
Add lacking parameter to word level position builder 2021-09-09 17:49:04 +02:00
Irevoire
a84f3a8b31
Apply suggestions from code review
Co-authored-by: Clément Renault <clement@meilisearch.com>
2021-09-09 15:09:35 +02:00
Tamo
bad8ea47d5
edit the two lasts TODO comments 2021-09-08 18:24:09 +02:00
Tamo
bd4c248292
improve the error handling in general and introduce the concept of reserved keywords 2021-09-08 18:24:09 +02:00
Tamo
5bb175fc90
only index _geo if it's set as sortable OR filterable
and only allow the filters if geo was set to filterable
2021-09-08 17:51:08 +02:00
Irevoire
216a8aa3b2
add a tests for the indexation of the geosearch 2021-09-08 17:51:07 +02:00
Irevoire
44d6b6ae9e
Index the geo points 2021-09-08 17:51:07 +02:00
many
741a4444a9
Remove log in chunk generator 2021-09-02 16:57:46 +02:00
many
7f7fafb857
Make document_chunk_size settable from update builder 2021-09-02 15:25:39 +02:00
many
db0c681bae
Fix Pr comments 2021-09-02 15:17:52 +02:00
many
8f702828ca
Ignore errors comming from crossbeam channel senders 2021-09-01 16:48:40 +02:00
many
fc7cc770d4
Add logging timers 2021-09-01 16:48:40 +02:00
many
a2f59a28f7
Remove unwrap sending errors in channel 2021-09-01 16:48:40 +02:00
many
5c962c03dd
Fix and optimize word_prefix_pair_proximity_docids database 2021-09-01 16:48:40 +02:00
many
2d1727697d
Take stop word in account 2021-09-01 16:48:40 +02:00
many
823da19745
Fix test and use progress callback 2021-09-01 16:48:39 +02:00
many
1d314328f0
Plug new indexer 2021-09-01 16:48:36 +02:00
bors[bot]
200e98c211
Merge #293
293: Make sure that the relevancy is not impacted by other settings r=Kerollmops a=Kerollmops

Fix https://github.com/meilisearch/meilisearch/issues/1505.

fix https://github.com/meilisearch/MeiliSearch/issues/1529

Co-authored-by: Kerollmops <clement@meilisearch.com>
2021-07-27 16:04:52 +00:00
Kerollmops
dc2b63abdf
Introduce an empty FilterCondition variant to support unknown fields 2021-07-27 16:34:04 +02:00