2022-10-12 03:21:25 +02:00
|
|
|
use std::cmp::min;
|
|
|
|
use std::collections::{BTreeMap, BTreeSet, HashSet};
|
2023-01-11 12:33:56 +01:00
|
|
|
use std::str::FromStr;
|
2022-10-12 03:21:25 +02:00
|
|
|
use std::time::Instant;
|
|
|
|
|
2023-02-13 18:45:13 +01:00
|
|
|
use deserr::Deserr;
|
2022-10-12 03:21:25 +02:00
|
|
|
use either::Either;
|
2023-06-22 23:12:29 +02:00
|
|
|
use index_scheduler::RoFeatures;
|
2023-06-27 12:31:52 +02:00
|
|
|
use log::warn;
|
2023-01-24 15:18:00 +01:00
|
|
|
use meilisearch_auth::IndexSearchRules;
|
2023-01-16 16:59:26 +01:00
|
|
|
use meilisearch_types::deserr::DeserrJsonError;
|
2023-01-11 17:10:32 +01:00
|
|
|
use meilisearch_types::error::deserr_codes::*;
|
2023-04-13 18:16:33 +02:00
|
|
|
use meilisearch_types::heed::RoTxn;
|
2023-01-24 15:18:00 +01:00
|
|
|
use meilisearch_types::index_uid::IndexUid;
|
2023-06-06 18:26:33 +02:00
|
|
|
use meilisearch_types::milli::score_details::{ScoreDetails, ScoringStrategy};
|
2023-04-26 11:55:43 +02:00
|
|
|
use meilisearch_types::milli::{
|
2023-04-27 10:43:02 +02:00
|
|
|
dot_product_similarity, FacetValueHit, InternalError, SearchForFacetValues,
|
2023-04-26 11:55:43 +02:00
|
|
|
};
|
2022-10-13 15:02:59 +02:00
|
|
|
use meilisearch_types::settings::DEFAULT_PAGINATION_MAX_TOTAL_HITS;
|
2022-10-12 03:21:25 +02:00
|
|
|
use meilisearch_types::{milli, Document};
|
|
|
|
use milli::tokenizer::TokenizerBuilder;
|
|
|
|
use milli::{
|
|
|
|
AscDesc, FieldId, FieldsIdsMap, Filter, FormatOptions, Index, MatchBounds, MatcherBuilder,
|
2023-06-20 15:54:28 +02:00
|
|
|
SortError, TermsMatchingStrategy, VectorOrArrayOfVectors, DEFAULT_VALUES_PER_FACET,
|
2022-10-12 03:21:25 +02:00
|
|
|
};
|
2023-06-20 14:38:58 +02:00
|
|
|
use ordered_float::OrderedFloat;
|
2022-10-12 03:21:25 +02:00
|
|
|
use regex::Regex;
|
2023-01-16 16:59:26 +01:00
|
|
|
use serde::Serialize;
|
2022-10-12 03:21:25 +02:00
|
|
|
use serde_json::{json, Value};
|
|
|
|
|
|
|
|
use crate::error::MeilisearchHttpError;
|
|
|
|
|
|
|
|
type MatchesPosition = BTreeMap<String, Vec<MatchBounds>>;
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
pub const DEFAULT_SEARCH_OFFSET: fn() -> usize = || 0;
|
2022-10-12 03:21:25 +02:00
|
|
|
pub const DEFAULT_SEARCH_LIMIT: fn() -> usize = || 20;
|
|
|
|
pub const DEFAULT_CROP_LENGTH: fn() -> usize = || 10;
|
|
|
|
pub const DEFAULT_CROP_MARKER: fn() -> String = || "…".to_string();
|
|
|
|
pub const DEFAULT_HIGHLIGHT_PRE_TAG: fn() -> String = || "<em>".to_string();
|
|
|
|
pub const DEFAULT_HIGHLIGHT_POST_TAG: fn() -> String = || "</em>".to_string();
|
|
|
|
|
2023-06-08 15:44:03 +02:00
|
|
|
#[derive(Debug, Clone, Default, PartialEq, Deserr)]
|
2023-01-12 13:55:53 +01:00
|
|
|
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub struct SearchQuery {
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchQ>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub q: Option<String>,
|
2023-06-26 18:07:56 +02:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchVector>)]
|
2023-06-08 15:44:03 +02:00
|
|
|
pub vector: Option<Vec<f32>>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default = DEFAULT_SEARCH_OFFSET(), error = DeserrJsonError<InvalidSearchOffset>)]
|
2022-10-26 19:25:59 +02:00
|
|
|
pub offset: usize,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default = DEFAULT_SEARCH_LIMIT(), error = DeserrJsonError<InvalidSearchLimit>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub limit: usize,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchPage>)]
|
2022-10-26 19:25:59 +02:00
|
|
|
pub page: Option<usize>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchHitsPerPage>)]
|
2022-10-26 19:25:59 +02:00
|
|
|
pub hits_per_page: Option<usize>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchAttributesToRetrieve>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub attributes_to_retrieve: Option<BTreeSet<String>>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchAttributesToCrop>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub attributes_to_crop: Option<Vec<String>>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchCropLength>, default = DEFAULT_CROP_LENGTH())]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub crop_length: usize,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchAttributesToHighlight>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub attributes_to_highlight: Option<HashSet<String>>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchShowMatchesPosition>, default)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub show_matches_position: bool,
|
2023-06-06 18:26:33 +02:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchShowRankingScore>, default)]
|
|
|
|
pub show_ranking_score: bool,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchShowRankingScoreDetails>, default)]
|
|
|
|
pub show_ranking_score_details: bool,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchFilter>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub filter: Option<Value>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchSort>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub sort: Option<Vec<String>>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchFacets>)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub facets: Option<Vec<String>>,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchHighlightPreTag>, default = DEFAULT_HIGHLIGHT_PRE_TAG())]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub highlight_pre_tag: String,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchHighlightPostTag>, default = DEFAULT_HIGHLIGHT_POST_TAG())]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub highlight_post_tag: String,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchCropMarker>, default = DEFAULT_CROP_MARKER())]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub crop_marker: String,
|
2023-01-12 15:35:03 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchMatchingStrategy>, default)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub matching_strategy: MatchingStrategy,
|
2023-06-20 10:00:35 +02:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidAttributesToSearchOn>, default)]
|
|
|
|
pub attributes_to_search_on: Option<Vec<String>>,
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
impl SearchQuery {
|
|
|
|
pub fn is_finite_pagination(&self) -> bool {
|
|
|
|
self.page.or(self.hits_per_page).is_some()
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-01-24 15:18:00 +01:00
|
|
|
/// A `SearchQuery` + an index UID.
|
|
|
|
// This struct contains the fields of `SearchQuery` inline.
|
|
|
|
// This is because neither deserr nor serde support `flatten` when using `deny_unknown_fields.
|
|
|
|
// The `From<SearchQueryWithIndex>` implementation ensures both structs remain up to date.
|
2023-06-08 15:44:03 +02:00
|
|
|
#[derive(Debug, Clone, PartialEq, Deserr)]
|
2023-01-24 15:18:00 +01:00
|
|
|
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
|
|
|
|
pub struct SearchQueryWithIndex {
|
|
|
|
#[deserr(error = DeserrJsonError<InvalidIndexUid>, missing_field_error = DeserrJsonError::missing_index_uid)]
|
|
|
|
pub index_uid: IndexUid,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchQ>)]
|
|
|
|
pub q: Option<String>,
|
2023-06-08 15:44:03 +02:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchQ>)]
|
|
|
|
pub vector: Option<Vec<f32>>,
|
2023-01-24 15:18:00 +01:00
|
|
|
#[deserr(default = DEFAULT_SEARCH_OFFSET(), error = DeserrJsonError<InvalidSearchOffset>)]
|
|
|
|
pub offset: usize,
|
|
|
|
#[deserr(default = DEFAULT_SEARCH_LIMIT(), error = DeserrJsonError<InvalidSearchLimit>)]
|
|
|
|
pub limit: usize,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchPage>)]
|
|
|
|
pub page: Option<usize>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchHitsPerPage>)]
|
|
|
|
pub hits_per_page: Option<usize>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchAttributesToRetrieve>)]
|
|
|
|
pub attributes_to_retrieve: Option<BTreeSet<String>>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchAttributesToCrop>)]
|
|
|
|
pub attributes_to_crop: Option<Vec<String>>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchCropLength>, default = DEFAULT_CROP_LENGTH())]
|
|
|
|
pub crop_length: usize,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchAttributesToHighlight>)]
|
|
|
|
pub attributes_to_highlight: Option<HashSet<String>>,
|
2023-06-06 18:26:33 +02:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchShowRankingScore>, default)]
|
|
|
|
pub show_ranking_score: bool,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchShowRankingScoreDetails>, default)]
|
|
|
|
pub show_ranking_score_details: bool,
|
2023-01-24 15:18:00 +01:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchShowMatchesPosition>, default)]
|
|
|
|
pub show_matches_position: bool,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchFilter>)]
|
|
|
|
pub filter: Option<Value>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchSort>)]
|
|
|
|
pub sort: Option<Vec<String>>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchFacets>)]
|
|
|
|
pub facets: Option<Vec<String>>,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchHighlightPreTag>, default = DEFAULT_HIGHLIGHT_PRE_TAG())]
|
|
|
|
pub highlight_pre_tag: String,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchHighlightPostTag>, default = DEFAULT_HIGHLIGHT_POST_TAG())]
|
|
|
|
pub highlight_post_tag: String,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchCropMarker>, default = DEFAULT_CROP_MARKER())]
|
|
|
|
pub crop_marker: String,
|
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidSearchMatchingStrategy>, default)]
|
|
|
|
pub matching_strategy: MatchingStrategy,
|
2023-06-20 10:00:35 +02:00
|
|
|
#[deserr(default, error = DeserrJsonError<InvalidAttributesToSearchOn>, default)]
|
|
|
|
pub attributes_to_search_on: Option<Vec<String>>,
|
2023-01-24 15:18:00 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
impl SearchQueryWithIndex {
|
|
|
|
pub fn into_index_query(self) -> (IndexUid, SearchQuery) {
|
|
|
|
let SearchQueryWithIndex {
|
|
|
|
index_uid,
|
|
|
|
q,
|
2023-06-08 15:44:03 +02:00
|
|
|
vector,
|
2023-01-24 15:18:00 +01:00
|
|
|
offset,
|
|
|
|
limit,
|
|
|
|
page,
|
|
|
|
hits_per_page,
|
|
|
|
attributes_to_retrieve,
|
|
|
|
attributes_to_crop,
|
|
|
|
crop_length,
|
|
|
|
attributes_to_highlight,
|
2023-06-06 18:26:33 +02:00
|
|
|
show_ranking_score,
|
|
|
|
show_ranking_score_details,
|
2023-01-24 15:18:00 +01:00
|
|
|
show_matches_position,
|
|
|
|
filter,
|
|
|
|
sort,
|
|
|
|
facets,
|
|
|
|
highlight_pre_tag,
|
|
|
|
highlight_post_tag,
|
|
|
|
crop_marker,
|
|
|
|
matching_strategy,
|
2023-06-20 10:00:35 +02:00
|
|
|
attributes_to_search_on,
|
2023-01-24 15:18:00 +01:00
|
|
|
} = self;
|
|
|
|
(
|
|
|
|
index_uid,
|
|
|
|
SearchQuery {
|
|
|
|
q,
|
2023-06-08 15:44:03 +02:00
|
|
|
vector,
|
2023-01-24 15:18:00 +01:00
|
|
|
offset,
|
|
|
|
limit,
|
|
|
|
page,
|
|
|
|
hits_per_page,
|
|
|
|
attributes_to_retrieve,
|
|
|
|
attributes_to_crop,
|
|
|
|
crop_length,
|
|
|
|
attributes_to_highlight,
|
2023-06-06 18:26:33 +02:00
|
|
|
show_ranking_score,
|
|
|
|
show_ranking_score_details,
|
2023-01-24 15:18:00 +01:00
|
|
|
show_matches_position,
|
|
|
|
filter,
|
|
|
|
sort,
|
|
|
|
facets,
|
|
|
|
highlight_pre_tag,
|
|
|
|
highlight_post_tag,
|
|
|
|
crop_marker,
|
|
|
|
matching_strategy,
|
2023-06-20 10:00:35 +02:00
|
|
|
attributes_to_search_on,
|
2023-01-24 15:18:00 +01:00
|
|
|
// do not use ..Default::default() here,
|
|
|
|
// rather add any missing field from `SearchQuery` to `SearchQueryWithIndex`
|
|
|
|
},
|
|
|
|
)
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-04-13 18:16:33 +02:00
|
|
|
#[derive(Debug, Copy, Clone, PartialEq, Eq, Deserr)]
|
2022-12-14 13:00:43 +01:00
|
|
|
#[deserr(rename_all = camelCase)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub enum MatchingStrategy {
|
|
|
|
/// Remove query words from last to first
|
|
|
|
Last,
|
|
|
|
/// All query words are mandatory
|
|
|
|
All,
|
|
|
|
}
|
|
|
|
|
|
|
|
impl Default for MatchingStrategy {
|
|
|
|
fn default() -> Self {
|
|
|
|
Self::Last
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
impl From<MatchingStrategy> for TermsMatchingStrategy {
|
|
|
|
fn from(other: MatchingStrategy) -> Self {
|
|
|
|
match other {
|
|
|
|
MatchingStrategy::Last => Self::Last,
|
|
|
|
MatchingStrategy::All => Self::All,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-06 18:26:33 +02:00
|
|
|
#[derive(Debug, Clone, Serialize, PartialEq)]
|
2022-10-12 03:21:25 +02:00
|
|
|
pub struct SearchHit {
|
|
|
|
#[serde(flatten)]
|
|
|
|
pub document: Document,
|
|
|
|
#[serde(rename = "_formatted", skip_serializing_if = "Document::is_empty")]
|
|
|
|
pub formatted: Document,
|
|
|
|
#[serde(rename = "_matchesPosition", skip_serializing_if = "Option::is_none")]
|
|
|
|
pub matches_position: Option<MatchesPosition>,
|
2023-06-06 18:26:33 +02:00
|
|
|
#[serde(rename = "_rankingScore", skip_serializing_if = "Option::is_none")]
|
|
|
|
pub ranking_score: Option<f64>,
|
|
|
|
#[serde(rename = "_rankingScoreDetails", skip_serializing_if = "Option::is_none")]
|
|
|
|
pub ranking_score_details: Option<serde_json::Map<String, serde_json::Value>>,
|
2023-06-27 12:31:23 +02:00
|
|
|
#[serde(rename = "_semanticScore", skip_serializing_if = "Option::is_none")]
|
|
|
|
pub semantic_score: Option<f32>,
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
|
2023-01-25 15:04:08 +01:00
|
|
|
#[derive(Serialize, Debug, Clone, PartialEq)]
|
2022-10-12 03:21:25 +02:00
|
|
|
#[serde(rename_all = "camelCase")]
|
|
|
|
pub struct SearchResult {
|
|
|
|
pub hits: Vec<SearchHit>,
|
|
|
|
pub query: String,
|
2023-06-20 10:14:25 +02:00
|
|
|
#[serde(skip_serializing_if = "Option::is_none")]
|
|
|
|
pub vector: Option<Vec<f32>>,
|
2022-10-12 03:21:25 +02:00
|
|
|
pub processing_time_ms: u128,
|
2022-10-26 19:25:59 +02:00
|
|
|
#[serde(flatten)]
|
|
|
|
pub hits_info: HitsInfo,
|
2022-10-12 03:21:25 +02:00
|
|
|
#[serde(skip_serializing_if = "Option::is_none")]
|
|
|
|
pub facet_distribution: Option<BTreeMap<String, BTreeMap<String, u64>>>,
|
2023-01-25 15:04:08 +01:00
|
|
|
#[serde(skip_serializing_if = "Option::is_none")]
|
|
|
|
pub facet_stats: Option<BTreeMap<String, FacetStats>>,
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
|
2023-01-24 15:18:00 +01:00
|
|
|
#[derive(Serialize, Debug, Clone, PartialEq)]
|
|
|
|
#[serde(rename_all = "camelCase")]
|
|
|
|
pub struct SearchResultWithIndex {
|
|
|
|
pub index_uid: String,
|
|
|
|
#[serde(flatten)]
|
|
|
|
pub result: SearchResult,
|
|
|
|
}
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
#[derive(Serialize, Debug, Clone, PartialEq, Eq)]
|
|
|
|
#[serde(untagged)]
|
|
|
|
pub enum HitsInfo {
|
|
|
|
#[serde(rename_all = "camelCase")]
|
|
|
|
Pagination { hits_per_page: usize, page: usize, total_pages: usize, total_hits: usize },
|
|
|
|
#[serde(rename_all = "camelCase")]
|
|
|
|
OffsetLimit { limit: usize, offset: usize, estimated_total_hits: usize },
|
|
|
|
}
|
|
|
|
|
2023-01-25 15:04:08 +01:00
|
|
|
#[derive(Serialize, Debug, Clone, PartialEq)]
|
|
|
|
pub struct FacetStats {
|
|
|
|
pub min: f64,
|
|
|
|
pub max: f64,
|
|
|
|
}
|
|
|
|
|
2023-04-26 17:09:59 +02:00
|
|
|
#[derive(Serialize, Debug, Clone, PartialEq)]
|
|
|
|
#[serde(rename_all = "camelCase")]
|
|
|
|
pub struct FacetSearchResult {
|
|
|
|
pub hits: Vec<FacetValueHit>,
|
|
|
|
pub query: Option<String>,
|
|
|
|
pub processing_time_ms: u128,
|
|
|
|
}
|
|
|
|
|
2023-01-24 15:18:00 +01:00
|
|
|
/// Incorporate search rules in search query
|
|
|
|
pub fn add_search_rules(query: &mut SearchQuery, rules: IndexSearchRules) {
|
|
|
|
query.filter = match (query.filter.take(), rules.filter) {
|
|
|
|
(None, rules_filter) => rules_filter,
|
|
|
|
(filter, None) => filter,
|
|
|
|
(Some(filter), Some(rules_filter)) => {
|
|
|
|
let filter = match filter {
|
|
|
|
Value::Array(filter) => filter,
|
|
|
|
filter => vec![filter],
|
|
|
|
};
|
|
|
|
let rules_filter = match rules_filter {
|
|
|
|
Value::Array(rules_filter) => rules_filter,
|
|
|
|
rules_filter => vec![rules_filter],
|
|
|
|
};
|
|
|
|
|
|
|
|
Some(Value::Array([filter, rules_filter].concat()))
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-04-13 18:16:33 +02:00
|
|
|
fn prepare_search<'t>(
|
|
|
|
index: &'t Index,
|
|
|
|
rtxn: &'t RoTxn,
|
|
|
|
query: &'t SearchQuery,
|
2023-06-22 23:12:29 +02:00
|
|
|
features: RoFeatures,
|
2023-04-13 18:16:33 +02:00
|
|
|
) -> Result<(milli::Search<'t>, bool, usize, usize), MeilisearchHttpError> {
|
2023-04-26 18:35:06 +02:00
|
|
|
let mut search = index.search(rtxn);
|
2022-10-12 03:21:25 +02:00
|
|
|
|
2023-06-27 12:31:52 +02:00
|
|
|
if query.vector.is_some() && query.q.is_some() {
|
|
|
|
warn!("Ignoring the query string `q` when used with the `vector` parameter.");
|
|
|
|
}
|
|
|
|
|
2023-06-08 18:47:06 +02:00
|
|
|
if let Some(ref vector) = query.vector {
|
|
|
|
search.vector(vector.clone());
|
|
|
|
}
|
|
|
|
|
2022-10-12 03:21:25 +02:00
|
|
|
if let Some(ref query) = query.q {
|
|
|
|
search.query(query);
|
|
|
|
}
|
|
|
|
|
2023-06-20 10:00:35 +02:00
|
|
|
if let Some(ref searchable) = query.attributes_to_search_on {
|
2023-06-07 11:24:40 +02:00
|
|
|
search.searchable_attributes(searchable);
|
|
|
|
}
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
let is_finite_pagination = query.is_finite_pagination();
|
2022-10-12 03:21:25 +02:00
|
|
|
search.terms_matching_strategy(query.matching_strategy.into());
|
|
|
|
|
|
|
|
let max_total_hits = index
|
2023-04-26 18:35:06 +02:00
|
|
|
.pagination_max_total_hits(rtxn)
|
2022-10-12 03:21:25 +02:00
|
|
|
.map_err(milli::Error::from)?
|
|
|
|
.unwrap_or(DEFAULT_PAGINATION_MAX_TOTAL_HITS);
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
search.exhaustive_number_hits(is_finite_pagination);
|
2023-06-06 18:26:33 +02:00
|
|
|
search.scoring_strategy(if query.show_ranking_score || query.show_ranking_score_details {
|
|
|
|
ScoringStrategy::Detailed
|
|
|
|
} else {
|
|
|
|
ScoringStrategy::Skip
|
|
|
|
});
|
2022-10-26 19:25:59 +02:00
|
|
|
|
2023-06-22 23:12:29 +02:00
|
|
|
if query.show_ranking_score_details {
|
|
|
|
features.check_score_details()?;
|
|
|
|
}
|
|
|
|
|
2023-06-26 16:36:01 +02:00
|
|
|
if query.vector.is_some() {
|
|
|
|
features.check_vector()?;
|
|
|
|
}
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
// compute the offset on the limit depending on the pagination mode.
|
|
|
|
let (offset, limit) = if is_finite_pagination {
|
|
|
|
let limit = query.hits_per_page.unwrap_or_else(DEFAULT_SEARCH_LIMIT);
|
|
|
|
let page = query.page.unwrap_or(1);
|
|
|
|
|
|
|
|
// page 0 gives a limit of 0 forcing Meilisearch to return no document.
|
|
|
|
page.checked_sub(1).map_or((0, 0), |p| (limit * p, limit))
|
|
|
|
} else {
|
|
|
|
(query.offset, query.limit)
|
|
|
|
};
|
|
|
|
|
2022-10-12 03:21:25 +02:00
|
|
|
// Make sure that a user can't get more documents than the hard limit,
|
|
|
|
// we align that on the offset too.
|
2022-10-26 19:25:59 +02:00
|
|
|
let offset = min(offset, max_total_hits);
|
|
|
|
let limit = min(limit, max_total_hits.saturating_sub(offset));
|
2022-10-12 03:21:25 +02:00
|
|
|
|
|
|
|
search.offset(offset);
|
|
|
|
search.limit(limit);
|
|
|
|
|
|
|
|
if let Some(ref filter) = query.filter {
|
|
|
|
if let Some(facets) = parse_filter(filter)? {
|
|
|
|
search.filter(facets);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if let Some(ref sort) = query.sort {
|
|
|
|
let sort = match sort.iter().map(|s| AscDesc::from_str(s)).collect() {
|
|
|
|
Ok(sorts) => sorts,
|
|
|
|
Err(asc_desc_error) => {
|
|
|
|
return Err(milli::Error::from(SortError::from(asc_desc_error)).into())
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
search.sort_criteria(sort);
|
|
|
|
}
|
|
|
|
|
2023-04-13 18:16:33 +02:00
|
|
|
Ok((search, is_finite_pagination, max_total_hits, offset))
|
|
|
|
}
|
|
|
|
|
|
|
|
pub fn perform_search(
|
|
|
|
index: &Index,
|
|
|
|
query: SearchQuery,
|
|
|
|
features: RoFeatures,
|
|
|
|
) -> Result<SearchResult, MeilisearchHttpError> {
|
|
|
|
let before_search = Instant::now();
|
|
|
|
let rtxn = index.read_txn()?;
|
|
|
|
|
|
|
|
let (search, is_finite_pagination, max_total_hits, offset) =
|
|
|
|
prepare_search(index, &rtxn, &query, features)?;
|
|
|
|
|
2023-06-06 18:26:33 +02:00
|
|
|
let milli::SearchResult { documents_ids, matching_words, candidates, document_scores, .. } =
|
|
|
|
search.execute()?;
|
2022-10-12 03:21:25 +02:00
|
|
|
|
|
|
|
let fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
|
|
|
|
|
|
|
|
let displayed_ids = index
|
|
|
|
.displayed_fields_ids(&rtxn)?
|
|
|
|
.map(|fields| fields.into_iter().collect::<BTreeSet<_>>())
|
|
|
|
.unwrap_or_else(|| fields_ids_map.iter().map(|(id, _)| id).collect());
|
|
|
|
|
|
|
|
let fids = |attrs: &BTreeSet<String>| {
|
|
|
|
let mut ids = BTreeSet::new();
|
|
|
|
for attr in attrs {
|
|
|
|
if attr == "*" {
|
|
|
|
ids = displayed_ids.clone();
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if let Some(id) = fields_ids_map.id(attr) {
|
|
|
|
ids.insert(id);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
ids
|
|
|
|
};
|
|
|
|
|
|
|
|
// The attributes to retrieve are the ones explicitly marked as to retrieve (all by default),
|
|
|
|
// but these attributes must be also be present
|
|
|
|
// - in the fields_ids_map
|
|
|
|
// - in the the displayed attributes
|
|
|
|
let to_retrieve_ids: BTreeSet<_> = query
|
|
|
|
.attributes_to_retrieve
|
|
|
|
.as_ref()
|
|
|
|
.map(fids)
|
|
|
|
.unwrap_or_else(|| displayed_ids.clone())
|
|
|
|
.intersection(&displayed_ids)
|
|
|
|
.cloned()
|
|
|
|
.collect();
|
|
|
|
|
|
|
|
let attr_to_highlight = query.attributes_to_highlight.unwrap_or_default();
|
|
|
|
|
|
|
|
let attr_to_crop = query.attributes_to_crop.unwrap_or_default();
|
|
|
|
|
|
|
|
// Attributes in `formatted_options` correspond to the attributes that will be in `_formatted`
|
|
|
|
// These attributes are:
|
|
|
|
// - the attributes asked to be highlighted or cropped (with `attributesToCrop` or `attributesToHighlight`)
|
|
|
|
// - the attributes asked to be retrieved: these attributes will not be highlighted/cropped
|
|
|
|
// But these attributes must be also present in displayed attributes
|
|
|
|
let formatted_options = compute_formatted_options(
|
|
|
|
&attr_to_highlight,
|
|
|
|
&attr_to_crop,
|
|
|
|
query.crop_length,
|
|
|
|
&to_retrieve_ids,
|
|
|
|
&fields_ids_map,
|
|
|
|
&displayed_ids,
|
|
|
|
);
|
|
|
|
|
2023-03-09 10:56:17 +01:00
|
|
|
let mut tokenizer_builder = TokenizerBuilder::default();
|
|
|
|
tokenizer_builder.create_char_map(true);
|
2022-10-12 03:21:25 +02:00
|
|
|
|
2023-03-08 12:44:16 +01:00
|
|
|
let script_lang_map = index.script_language(&rtxn)?;
|
|
|
|
if !script_lang_map.is_empty() {
|
2023-03-09 10:56:17 +01:00
|
|
|
tokenizer_builder.allow_list(&script_lang_map);
|
2023-03-08 12:44:16 +01:00
|
|
|
}
|
|
|
|
|
2023-03-09 10:56:17 +01:00
|
|
|
let mut formatter_builder = MatcherBuilder::new(matching_words, tokenizer_builder.build());
|
2022-10-12 03:21:25 +02:00
|
|
|
formatter_builder.crop_marker(query.crop_marker);
|
|
|
|
formatter_builder.highlight_prefix(query.highlight_pre_tag);
|
|
|
|
formatter_builder.highlight_suffix(query.highlight_post_tag);
|
|
|
|
|
|
|
|
let mut documents = Vec::new();
|
|
|
|
let documents_iter = index.documents(&rtxn, documents_ids)?;
|
|
|
|
|
2023-06-06 18:26:33 +02:00
|
|
|
for ((_id, obkv), score) in documents_iter.into_iter().zip(document_scores.into_iter()) {
|
2022-10-12 03:21:25 +02:00
|
|
|
// First generate a document with all the displayed fields
|
|
|
|
let displayed_document = make_document(&displayed_ids, &fields_ids_map, obkv)?;
|
|
|
|
|
|
|
|
// select the attributes to retrieve
|
|
|
|
let attributes_to_retrieve = to_retrieve_ids
|
|
|
|
.iter()
|
|
|
|
.map(|&fid| fields_ids_map.name(fid).expect("Missing field name"));
|
|
|
|
let mut document =
|
|
|
|
permissive_json_pointer::select_values(&displayed_document, attributes_to_retrieve);
|
|
|
|
|
|
|
|
let (matches_position, formatted) = format_fields(
|
|
|
|
&displayed_document,
|
|
|
|
&fields_ids_map,
|
|
|
|
&formatter_builder,
|
|
|
|
&formatted_options,
|
|
|
|
query.show_matches_position,
|
|
|
|
&displayed_ids,
|
|
|
|
)?;
|
|
|
|
|
|
|
|
if let Some(sort) = query.sort.as_ref() {
|
|
|
|
insert_geo_distance(sort, &mut document);
|
|
|
|
}
|
|
|
|
|
2023-06-27 12:31:23 +02:00
|
|
|
let semantic_score = match query.vector.as_ref() {
|
|
|
|
Some(vector) => match extract_field("_vectors", &fields_ids_map, obkv)? {
|
|
|
|
Some(vectors) => compute_semantic_score(vector, vectors)?,
|
|
|
|
None => None,
|
|
|
|
},
|
|
|
|
None => None,
|
|
|
|
};
|
2023-06-20 14:38:58 +02:00
|
|
|
|
2023-06-06 18:26:33 +02:00
|
|
|
let ranking_score =
|
|
|
|
query.show_ranking_score.then(|| ScoreDetails::global_score(score.iter()));
|
|
|
|
let ranking_score_details =
|
|
|
|
query.show_ranking_score_details.then(|| ScoreDetails::to_json_map(score.iter()));
|
|
|
|
|
|
|
|
let hit = SearchHit {
|
|
|
|
document,
|
|
|
|
formatted,
|
|
|
|
matches_position,
|
|
|
|
ranking_score_details,
|
|
|
|
ranking_score,
|
2023-06-27 12:31:23 +02:00
|
|
|
semantic_score,
|
2023-06-06 18:26:33 +02:00
|
|
|
};
|
2022-10-12 03:21:25 +02:00
|
|
|
documents.push(hit);
|
|
|
|
}
|
|
|
|
|
2022-10-26 19:25:59 +02:00
|
|
|
let number_of_hits = min(candidates.len() as usize, max_total_hits);
|
|
|
|
let hits_info = if is_finite_pagination {
|
|
|
|
let hits_per_page = query.hits_per_page.unwrap_or_else(DEFAULT_SEARCH_LIMIT);
|
|
|
|
// If hit_per_page is 0, then pages can't be computed and so we respond 0.
|
|
|
|
let total_pages = (number_of_hits + hits_per_page.saturating_sub(1))
|
|
|
|
.checked_div(hits_per_page)
|
|
|
|
.unwrap_or(0);
|
|
|
|
|
|
|
|
HitsInfo::Pagination {
|
|
|
|
hits_per_page,
|
|
|
|
page: query.page.unwrap_or(1),
|
|
|
|
total_pages,
|
|
|
|
total_hits: number_of_hits,
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
HitsInfo::OffsetLimit { limit: query.limit, offset, estimated_total_hits: number_of_hits }
|
|
|
|
};
|
2022-10-12 03:21:25 +02:00
|
|
|
|
2023-01-25 15:04:08 +01:00
|
|
|
let (facet_distribution, facet_stats) = match query.facets {
|
2022-10-12 03:21:25 +02:00
|
|
|
Some(ref fields) => {
|
|
|
|
let mut facet_distribution = index.facets_distribution(&rtxn);
|
|
|
|
|
|
|
|
let max_values_by_facet = index
|
|
|
|
.max_values_per_facet(&rtxn)
|
|
|
|
.map_err(milli::Error::from)?
|
|
|
|
.unwrap_or(DEFAULT_VALUES_PER_FACET);
|
|
|
|
facet_distribution.max_values_per_facet(max_values_by_facet);
|
|
|
|
|
|
|
|
if fields.iter().all(|f| f != "*") {
|
|
|
|
facet_distribution.facets(fields);
|
|
|
|
}
|
|
|
|
let distribution = facet_distribution.candidates(candidates).execute()?;
|
2023-01-25 15:04:08 +01:00
|
|
|
let stats = facet_distribution.compute_stats()?;
|
|
|
|
(Some(distribution), Some(stats))
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
2023-01-25 15:04:08 +01:00
|
|
|
None => (None, None),
|
2022-10-12 03:21:25 +02:00
|
|
|
};
|
|
|
|
|
2023-01-25 15:04:08 +01:00
|
|
|
let facet_stats = facet_stats.map(|stats| {
|
|
|
|
stats.into_iter().map(|(k, (min, max))| (k, FacetStats { min, max })).collect()
|
|
|
|
});
|
|
|
|
|
2022-10-12 03:21:25 +02:00
|
|
|
let result = SearchResult {
|
|
|
|
hits: documents,
|
2022-10-26 19:25:59 +02:00
|
|
|
hits_info,
|
2023-06-20 10:14:25 +02:00
|
|
|
query: query.q.unwrap_or_default(),
|
|
|
|
vector: query.vector,
|
2022-10-12 03:21:25 +02:00
|
|
|
processing_time_ms: before_search.elapsed().as_millis(),
|
|
|
|
facet_distribution,
|
2023-01-25 15:04:08 +01:00
|
|
|
facet_stats,
|
2022-10-12 03:21:25 +02:00
|
|
|
};
|
|
|
|
Ok(result)
|
|
|
|
}
|
|
|
|
|
2023-04-13 18:16:33 +02:00
|
|
|
pub fn perform_facet_search(
|
|
|
|
index: &Index,
|
|
|
|
search_query: SearchQuery,
|
2023-04-26 17:09:59 +02:00
|
|
|
facet_query: Option<String>,
|
2023-04-13 18:16:33 +02:00
|
|
|
facet_name: String,
|
2023-05-04 12:27:19 +02:00
|
|
|
features: RoFeatures,
|
2023-04-26 17:09:59 +02:00
|
|
|
) -> Result<FacetSearchResult, MeilisearchHttpError> {
|
2023-04-13 18:16:33 +02:00
|
|
|
let before_search = Instant::now();
|
|
|
|
let rtxn = index.read_txn()?;
|
|
|
|
|
2023-05-04 12:27:19 +02:00
|
|
|
let (search, _, _, _) = prepare_search(index, &rtxn, &search_query, features)?;
|
2023-04-27 10:43:02 +02:00
|
|
|
let mut facet_search = SearchForFacetValues::new(facet_name, search);
|
2023-04-26 17:09:59 +02:00
|
|
|
if let Some(facet_query) = &facet_query {
|
2023-04-26 11:55:43 +02:00
|
|
|
facet_search.query(facet_query);
|
|
|
|
}
|
2023-04-13 18:16:33 +02:00
|
|
|
|
2023-04-26 17:09:59 +02:00
|
|
|
let hits = facet_search.execute()?;
|
|
|
|
|
|
|
|
Ok(FacetSearchResult {
|
|
|
|
hits,
|
|
|
|
query: facet_query,
|
|
|
|
processing_time_ms: before_search.elapsed().as_millis(),
|
|
|
|
})
|
2023-04-13 18:16:33 +02:00
|
|
|
}
|
|
|
|
|
2022-10-12 03:21:25 +02:00
|
|
|
fn insert_geo_distance(sorts: &[String], document: &mut Document) {
|
|
|
|
lazy_static::lazy_static! {
|
|
|
|
static ref GEO_REGEX: Regex =
|
|
|
|
Regex::new(r"_geoPoint\(\s*([[:digit:].\-]+)\s*,\s*([[:digit:].\-]+)\s*\)").unwrap();
|
|
|
|
};
|
|
|
|
if let Some(capture_group) = sorts.iter().find_map(|sort| GEO_REGEX.captures(sort)) {
|
|
|
|
// TODO: TAMO: milli encountered an internal error, what do we want to do?
|
2022-10-20 18:00:07 +02:00
|
|
|
let base = [capture_group[1].parse().unwrap(), capture_group[2].parse().unwrap()];
|
2022-10-12 03:21:25 +02:00
|
|
|
let geo_point = &document.get("_geo").unwrap_or(&json!(null));
|
|
|
|
if let Some((lat, lng)) = geo_point["lat"].as_f64().zip(geo_point["lng"].as_f64()) {
|
|
|
|
let distance = milli::distance_between_two_points(&base, &[lat, lng]);
|
|
|
|
document.insert("_geoDistance".to_string(), json!(distance.round() as usize));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-27 12:31:23 +02:00
|
|
|
fn compute_semantic_score(query: &[f32], vectors: Value) -> milli::Result<Option<f32>> {
|
|
|
|
let vectors = serde_json::from_value(vectors)
|
|
|
|
.map(VectorOrArrayOfVectors::into_array_of_vectors)
|
|
|
|
.map_err(InternalError::SerdeJson)?;
|
|
|
|
Ok(vectors
|
2023-06-20 15:54:28 +02:00
|
|
|
.into_iter()
|
|
|
|
.map(|v| OrderedFloat(dot_product_similarity(query, &v)))
|
|
|
|
.max()
|
2023-06-27 12:31:23 +02:00
|
|
|
.map(OrderedFloat::into_inner))
|
2023-06-20 14:38:58 +02:00
|
|
|
}
|
|
|
|
|
2022-10-12 03:21:25 +02:00
|
|
|
fn compute_formatted_options(
|
|
|
|
attr_to_highlight: &HashSet<String>,
|
|
|
|
attr_to_crop: &[String],
|
|
|
|
query_crop_length: usize,
|
|
|
|
to_retrieve_ids: &BTreeSet<FieldId>,
|
|
|
|
fields_ids_map: &FieldsIdsMap,
|
|
|
|
displayed_ids: &BTreeSet<FieldId>,
|
|
|
|
) -> BTreeMap<FieldId, FormatOptions> {
|
|
|
|
let mut formatted_options = BTreeMap::new();
|
|
|
|
|
|
|
|
add_highlight_to_formatted_options(
|
|
|
|
&mut formatted_options,
|
|
|
|
attr_to_highlight,
|
|
|
|
fields_ids_map,
|
|
|
|
displayed_ids,
|
|
|
|
);
|
|
|
|
|
|
|
|
add_crop_to_formatted_options(
|
|
|
|
&mut formatted_options,
|
|
|
|
attr_to_crop,
|
|
|
|
query_crop_length,
|
|
|
|
fields_ids_map,
|
|
|
|
displayed_ids,
|
|
|
|
);
|
|
|
|
|
|
|
|
// Should not return `_formatted` if no valid attributes to highlight/crop
|
|
|
|
if !formatted_options.is_empty() {
|
|
|
|
add_non_formatted_ids_to_formatted_options(&mut formatted_options, to_retrieve_ids);
|
|
|
|
}
|
|
|
|
|
|
|
|
formatted_options
|
|
|
|
}
|
|
|
|
|
|
|
|
fn add_highlight_to_formatted_options(
|
|
|
|
formatted_options: &mut BTreeMap<FieldId, FormatOptions>,
|
|
|
|
attr_to_highlight: &HashSet<String>,
|
|
|
|
fields_ids_map: &FieldsIdsMap,
|
|
|
|
displayed_ids: &BTreeSet<FieldId>,
|
|
|
|
) {
|
|
|
|
for attr in attr_to_highlight {
|
2022-10-20 18:00:07 +02:00
|
|
|
let new_format = FormatOptions { highlight: true, crop: None };
|
2022-10-12 03:21:25 +02:00
|
|
|
|
|
|
|
if attr == "*" {
|
|
|
|
for id in displayed_ids {
|
|
|
|
formatted_options.insert(*id, new_format);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if let Some(id) = fields_ids_map.id(attr) {
|
|
|
|
if displayed_ids.contains(&id) {
|
|
|
|
formatted_options.insert(id, new_format);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
fn add_crop_to_formatted_options(
|
|
|
|
formatted_options: &mut BTreeMap<FieldId, FormatOptions>,
|
|
|
|
attr_to_crop: &[String],
|
|
|
|
crop_length: usize,
|
|
|
|
fields_ids_map: &FieldsIdsMap,
|
|
|
|
displayed_ids: &BTreeSet<FieldId>,
|
|
|
|
) {
|
|
|
|
for attr in attr_to_crop {
|
|
|
|
let mut split = attr.rsplitn(2, ':');
|
|
|
|
let (attr_name, attr_len) = match split.next().zip(split.next()) {
|
|
|
|
Some((len, name)) => {
|
|
|
|
let crop_len = len.parse::<usize>().unwrap_or(crop_length);
|
|
|
|
(name, crop_len)
|
|
|
|
}
|
|
|
|
None => (attr.as_str(), crop_length),
|
|
|
|
};
|
|
|
|
|
|
|
|
if attr_name == "*" {
|
|
|
|
for id in displayed_ids {
|
|
|
|
formatted_options
|
|
|
|
.entry(*id)
|
|
|
|
.and_modify(|f| f.crop = Some(attr_len))
|
2022-10-20 18:00:07 +02:00
|
|
|
.or_insert(FormatOptions { highlight: false, crop: Some(attr_len) });
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if let Some(id) = fields_ids_map.id(attr_name) {
|
|
|
|
if displayed_ids.contains(&id) {
|
|
|
|
formatted_options
|
|
|
|
.entry(id)
|
|
|
|
.and_modify(|f| f.crop = Some(attr_len))
|
2022-10-20 18:00:07 +02:00
|
|
|
.or_insert(FormatOptions { highlight: false, crop: Some(attr_len) });
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
fn add_non_formatted_ids_to_formatted_options(
|
|
|
|
formatted_options: &mut BTreeMap<FieldId, FormatOptions>,
|
|
|
|
to_retrieve_ids: &BTreeSet<FieldId>,
|
|
|
|
) {
|
|
|
|
for id in to_retrieve_ids {
|
2022-10-20 18:00:07 +02:00
|
|
|
formatted_options.entry(*id).or_insert(FormatOptions { highlight: false, crop: None });
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
fn make_document(
|
|
|
|
displayed_attributes: &BTreeSet<FieldId>,
|
|
|
|
field_ids_map: &FieldsIdsMap,
|
|
|
|
obkv: obkv::KvReaderU16,
|
|
|
|
) -> Result<Document, MeilisearchHttpError> {
|
|
|
|
let mut document = serde_json::Map::new();
|
|
|
|
|
|
|
|
// recreate the original json
|
|
|
|
for (key, value) in obkv.iter() {
|
|
|
|
let value = serde_json::from_slice(value)?;
|
2022-10-20 18:00:07 +02:00
|
|
|
let key = field_ids_map.name(key).expect("Missing field name").to_string();
|
2022-10-12 03:21:25 +02:00
|
|
|
|
|
|
|
document.insert(key, value);
|
|
|
|
}
|
|
|
|
|
|
|
|
// select the attributes to retrieve
|
|
|
|
let displayed_attributes = displayed_attributes
|
|
|
|
.iter()
|
|
|
|
.map(|&fid| field_ids_map.name(fid).expect("Missing field name"));
|
|
|
|
|
|
|
|
let document = permissive_json_pointer::select_values(&document, displayed_attributes);
|
|
|
|
Ok(document)
|
|
|
|
}
|
|
|
|
|
2023-06-20 15:54:28 +02:00
|
|
|
/// Extract the JSON value under the field name specified
|
|
|
|
/// but doesn't support nested objects.
|
|
|
|
fn extract_field(
|
|
|
|
field_name: &str,
|
|
|
|
field_ids_map: &FieldsIdsMap,
|
|
|
|
obkv: obkv::KvReaderU16,
|
|
|
|
) -> Result<Option<serde_json::Value>, MeilisearchHttpError> {
|
|
|
|
match field_ids_map.id(field_name) {
|
|
|
|
Some(fid) => match obkv.get(fid) {
|
|
|
|
Some(value) => Ok(serde_json::from_slice(value).map(Some)?),
|
|
|
|
None => Ok(None),
|
|
|
|
},
|
|
|
|
None => Ok(None),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-01-30 17:18:02 +01:00
|
|
|
fn format_fields<A: AsRef<[u8]>>(
|
2022-10-12 03:21:25 +02:00
|
|
|
document: &Document,
|
|
|
|
field_ids_map: &FieldsIdsMap,
|
2023-01-30 17:18:02 +01:00
|
|
|
builder: &MatcherBuilder<'_, A>,
|
2022-10-12 03:21:25 +02:00
|
|
|
formatted_options: &BTreeMap<FieldId, FormatOptions>,
|
|
|
|
compute_matches: bool,
|
|
|
|
displayable_ids: &BTreeSet<FieldId>,
|
|
|
|
) -> Result<(Option<MatchesPosition>, Document), MeilisearchHttpError> {
|
|
|
|
let mut matches_position = compute_matches.then(BTreeMap::new);
|
|
|
|
let mut document = document.clone();
|
|
|
|
|
|
|
|
// select the attributes to retrieve
|
2022-10-20 18:00:07 +02:00
|
|
|
let displayable_names =
|
|
|
|
displayable_ids.iter().map(|&fid| field_ids_map.name(fid).expect("Missing field name"));
|
2022-10-12 03:21:25 +02:00
|
|
|
permissive_json_pointer::map_leaf_values(&mut document, displayable_names, |key, value| {
|
|
|
|
// To get the formatting option of each key we need to see all the rules that applies
|
|
|
|
// to the value and merge them together. eg. If a user said he wanted to highlight `doggo`
|
|
|
|
// and crop `doggo.name`. `doggo.name` needs to be highlighted + cropped while `doggo.age` is only
|
|
|
|
// highlighted.
|
|
|
|
let format = formatted_options
|
|
|
|
.iter()
|
|
|
|
.filter(|(field, _option)| {
|
|
|
|
let name = field_ids_map.name(**field).unwrap();
|
|
|
|
milli::is_faceted_by(name, key) || milli::is_faceted_by(key, name)
|
|
|
|
})
|
|
|
|
.map(|(_, option)| *option)
|
|
|
|
.reduce(|acc, option| acc.merge(option));
|
|
|
|
let mut infos = Vec::new();
|
|
|
|
|
2022-10-20 18:00:07 +02:00
|
|
|
*value = format_value(std::mem::take(value), builder, format, &mut infos, compute_matches);
|
2022-10-12 03:21:25 +02:00
|
|
|
|
|
|
|
if let Some(matches) = matches_position.as_mut() {
|
|
|
|
if !infos.is_empty() {
|
|
|
|
matches.insert(key.to_owned(), infos);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
});
|
|
|
|
|
|
|
|
let selectors = formatted_options
|
|
|
|
.keys()
|
|
|
|
// This unwrap must be safe since we got the ids from the fields_ids_map just
|
|
|
|
// before.
|
|
|
|
.map(|&fid| field_ids_map.name(fid).unwrap());
|
|
|
|
let document = permissive_json_pointer::select_values(&document, selectors);
|
|
|
|
|
|
|
|
Ok((matches_position, document))
|
|
|
|
}
|
|
|
|
|
2023-01-30 17:18:02 +01:00
|
|
|
fn format_value<A: AsRef<[u8]>>(
|
2022-10-12 03:21:25 +02:00
|
|
|
value: Value,
|
2023-01-30 17:18:02 +01:00
|
|
|
builder: &MatcherBuilder<'_, A>,
|
2022-10-12 03:21:25 +02:00
|
|
|
format_options: Option<FormatOptions>,
|
|
|
|
infos: &mut Vec<MatchBounds>,
|
|
|
|
compute_matches: bool,
|
|
|
|
) -> Value {
|
|
|
|
match value {
|
|
|
|
Value::String(old_string) => {
|
|
|
|
let mut matcher = builder.build(&old_string);
|
|
|
|
if compute_matches {
|
|
|
|
let matches = matcher.matches();
|
|
|
|
infos.extend_from_slice(&matches[..]);
|
|
|
|
}
|
|
|
|
|
|
|
|
match format_options {
|
|
|
|
Some(format_options) => {
|
|
|
|
let value = matcher.format(format_options);
|
|
|
|
Value::String(value.into_owned())
|
|
|
|
}
|
|
|
|
None => Value::String(old_string),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
Value::Array(values) => Value::Array(
|
|
|
|
values
|
|
|
|
.into_iter()
|
|
|
|
.map(|v| {
|
|
|
|
format_value(
|
|
|
|
v,
|
|
|
|
builder,
|
|
|
|
format_options.map(|format_options| FormatOptions {
|
|
|
|
highlight: format_options.highlight,
|
|
|
|
crop: None,
|
|
|
|
}),
|
|
|
|
infos,
|
|
|
|
compute_matches,
|
|
|
|
)
|
|
|
|
})
|
|
|
|
.collect(),
|
|
|
|
),
|
|
|
|
Value::Object(object) => Value::Object(
|
|
|
|
object
|
|
|
|
.into_iter()
|
|
|
|
.map(|(k, v)| {
|
|
|
|
(
|
|
|
|
k,
|
|
|
|
format_value(
|
|
|
|
v,
|
|
|
|
builder,
|
|
|
|
format_options.map(|format_options| FormatOptions {
|
|
|
|
highlight: format_options.highlight,
|
|
|
|
crop: None,
|
|
|
|
}),
|
|
|
|
infos,
|
|
|
|
compute_matches,
|
|
|
|
),
|
|
|
|
)
|
|
|
|
})
|
|
|
|
.collect(),
|
|
|
|
),
|
|
|
|
Value::Number(number) => {
|
|
|
|
let s = number.to_string();
|
|
|
|
|
|
|
|
let mut matcher = builder.build(&s);
|
|
|
|
if compute_matches {
|
|
|
|
let matches = matcher.matches();
|
|
|
|
infos.extend_from_slice(&matches[..]);
|
|
|
|
}
|
|
|
|
|
|
|
|
match format_options {
|
|
|
|
Some(format_options) => {
|
|
|
|
let value = matcher.format(format_options);
|
|
|
|
Value::String(value.into_owned())
|
|
|
|
}
|
|
|
|
None => Value::Number(number),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
value => value,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-03-08 09:43:40 +01:00
|
|
|
pub(crate) fn parse_filter(facets: &Value) -> Result<Option<Filter>, MeilisearchHttpError> {
|
2022-10-12 03:21:25 +02:00
|
|
|
match facets {
|
|
|
|
Value::String(expr) => {
|
|
|
|
let condition = Filter::from_str(expr)?;
|
|
|
|
Ok(condition)
|
|
|
|
}
|
|
|
|
Value::Array(arr) => parse_filter_array(arr),
|
2023-01-11 11:37:12 +01:00
|
|
|
v => Err(MeilisearchHttpError::InvalidExpression(&["String", "Array"], v.clone())),
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
fn parse_filter_array(arr: &[Value]) -> Result<Option<Filter>, MeilisearchHttpError> {
|
|
|
|
let mut ands = Vec::new();
|
|
|
|
for value in arr {
|
|
|
|
match value {
|
|
|
|
Value::String(s) => ands.push(Either::Right(s.as_str())),
|
|
|
|
Value::Array(arr) => {
|
|
|
|
let mut ors = Vec::new();
|
|
|
|
for value in arr {
|
|
|
|
match value {
|
|
|
|
Value::String(s) => ors.push(s.as_str()),
|
|
|
|
v => {
|
|
|
|
return Err(MeilisearchHttpError::InvalidExpression(
|
|
|
|
&["String"],
|
|
|
|
v.clone(),
|
2022-10-22 16:35:42 +02:00
|
|
|
))
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
ands.push(Either::Left(ors));
|
|
|
|
}
|
|
|
|
v => {
|
|
|
|
return Err(MeilisearchHttpError::InvalidExpression(
|
|
|
|
&["String", "[String]"],
|
|
|
|
v.clone(),
|
2022-10-22 16:35:42 +02:00
|
|
|
))
|
2022-10-12 03:21:25 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
Ok(Filter::from_array(ands)?)
|
|
|
|
}
|
|
|
|
|
|
|
|
#[cfg(test)]
|
|
|
|
mod test {
|
|
|
|
use super::*;
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn test_insert_geo_distance() {
|
|
|
|
let value: Document = serde_json::from_str(
|
|
|
|
r#"{
|
|
|
|
"_geo": {
|
|
|
|
"lat": 50.629973371633746,
|
|
|
|
"lng": 3.0569447399419567
|
|
|
|
},
|
|
|
|
"city": "Lille",
|
|
|
|
"id": "1"
|
|
|
|
}"#,
|
|
|
|
)
|
|
|
|
.unwrap();
|
|
|
|
|
|
|
|
let sorters = &["_geoPoint(50.629973371633746,3.0569447399419567):desc".to_string()];
|
|
|
|
let mut document = value.clone();
|
|
|
|
insert_geo_distance(sorters, &mut document);
|
|
|
|
assert_eq!(document.get("_geoDistance"), Some(&json!(0)));
|
|
|
|
|
|
|
|
let sorters = &["_geoPoint(50.629973371633746, 3.0569447399419567):asc".to_string()];
|
|
|
|
let mut document = value.clone();
|
|
|
|
insert_geo_distance(sorters, &mut document);
|
|
|
|
assert_eq!(document.get("_geoDistance"), Some(&json!(0)));
|
|
|
|
|
|
|
|
let sorters =
|
|
|
|
&["_geoPoint( 50.629973371633746 , 3.0569447399419567 ):desc".to_string()];
|
|
|
|
let mut document = value.clone();
|
|
|
|
insert_geo_distance(sorters, &mut document);
|
|
|
|
assert_eq!(document.get("_geoDistance"), Some(&json!(0)));
|
|
|
|
|
|
|
|
let sorters = &[
|
|
|
|
"prix:asc",
|
|
|
|
"villeneuve:desc",
|
|
|
|
"_geoPoint(50.629973371633746, 3.0569447399419567):asc",
|
|
|
|
"ubu:asc",
|
|
|
|
]
|
|
|
|
.map(|s| s.to_string());
|
|
|
|
let mut document = value.clone();
|
|
|
|
insert_geo_distance(sorters, &mut document);
|
|
|
|
assert_eq!(document.get("_geoDistance"), Some(&json!(0)));
|
|
|
|
|
|
|
|
// only the first geoPoint is used to compute the distance
|
|
|
|
let sorters = &[
|
|
|
|
"chien:desc",
|
|
|
|
"_geoPoint(50.629973371633746, 3.0569447399419567):asc",
|
|
|
|
"pangolin:desc",
|
|
|
|
"_geoPoint(100.0, -80.0):asc",
|
|
|
|
"chat:asc",
|
|
|
|
]
|
|
|
|
.map(|s| s.to_string());
|
|
|
|
let mut document = value.clone();
|
|
|
|
insert_geo_distance(sorters, &mut document);
|
|
|
|
assert_eq!(document.get("_geoDistance"), Some(&json!(0)));
|
|
|
|
|
|
|
|
// there was no _geoPoint so nothing is inserted in the document
|
|
|
|
let sorters = &["chien:asc".to_string()];
|
|
|
|
let mut document = value;
|
|
|
|
insert_geo_distance(sorters, &mut document);
|
|
|
|
assert_eq!(document.get("_geoDistance"), None);
|
|
|
|
}
|
|
|
|
}
|