Compare commits

...

1 Commits

Author SHA1 Message Date
bc0e219158 feat: improve search experience 2024-09-09 19:23:21 +02:00
20 changed files with 230 additions and 158 deletions

View File

@@ -8,8 +8,7 @@ const config = {
database: { database: {
name: 'elasticsearch', name: 'elasticsearch',
query: { query: {
minMatch: '60%', fields: ["name"]
queryType: 'query_string',
}, },
}, },
}, },

View File

@@ -39,7 +39,7 @@ const boostings = {
type: SCThingType.AcademicEvent, type: SCThingType.AcademicEvent,
}, },
{ {
factor: 1.6, factor: 2,
type: SCThingType.Building, type: SCThingType.Building,
}, },
{ {
@@ -85,7 +85,7 @@ const boostings = {
], ],
place: [ place: [
{ {
factor: 2, factor: 3,
type: SCThingType.Building, type: SCThingType.Building,
}, },
{ {

View File

@@ -17,12 +17,21 @@ const config = {
name: 'elasticsearch', name: 'elasticsearch',
version: '8.4.2', version: '8.4.2',
query: { query: {
minMatch: '75%', type: 'best_fields',
queryType: 'dis_max', fields: [
matchBoosting: 1.3, 'identifiers^20',
fuzziness: 'AUTO', 'name^10',
cutoffFrequency: 0, 'translations.*.name^10',
tieBreaker: 0, 'alternateNames^10',
'translations.*.alternateNames^10',
'description^2',
'translations.*.description^2',
'categories^5',
],
},
searchAsYouTypeQuery: {
type: 'phrase_prefix',
fields: ['name.completion', 'name.completion._2gram', 'name.completion._3gram'],
}, },
}, },
}, },

View File

@@ -20,6 +20,8 @@ import {
IndicesGetAliasResponse, IndicesGetAliasResponse,
SearchHit, SearchHit,
SearchResponse, SearchResponse,
SearchTermSuggest,
SearchTermSuggestOption,
} from '@elastic/elasticsearch/lib/api/types.js'; } from '@elastic/elasticsearch/lib/api/types.js';
import {SCConfigFile, SCSearchQuery, SCSearchResponse, SCThings, SCUuid} from '@openstapps/core'; import {SCConfigFile, SCSearchQuery, SCSearchResponse, SCThings, SCUuid} from '@openstapps/core';
import {Logger} from '@openstapps/logger'; import {Logger} from '@openstapps/logger';
@@ -47,6 +49,9 @@ import {
import {noUndefined} from './util/no-undefined.js'; import {noUndefined} from './util/no-undefined.js';
import {retryCatch, RetryOptions} from './util/retry.js'; import {retryCatch, RetryOptions} from './util/retry.js';
import {Feature, Point, Polygon} from 'geojson'; import {Feature, Point, Polygon} from 'geojson';
import {parseSuggestions} from './util/parse-suggestions.js';
import {buildScoringFunctions} from './query/boost/scoring-functions.js';
import {buildFilter} from './query/filter.js';
/** /**
* A database interface for elasticsearch * A database interface for elasticsearch
@@ -355,6 +360,39 @@ export class Elasticsearch implements Database {
throw new Error('You tried to PUT an non-existing object. PUT is only supported on existing objects.'); throw new Error('You tried to PUT an non-existing object. PUT is only supported on existing objects.');
} }
public async searchAsYouType(parameters: SCSearchQuery): Promise<SCSearchResponse> {
const result = await this.client.search({
_source: 'name',
query: {
function_score: {
functions: buildScoringFunctions(this.config.internal.boostings, parameters.context),
query: {
bool: {
must: {
multi_match: {
query: parameters.query,
type: 'bool_prefix',
fields: ['name.completion', 'name.completion._2gram', 'name.completion._3gram'],
},
},
should: [],
filter: parameters.filter === undefined ? undefined : buildFilter(parameters.filter),
},
},
score_mode: 'max',
boost_mode: 'multiply',
},
},
index: ACTIVE_INDICES_ALIAS,
allow_no_indices: true,
size: 5,
});
const suggestions = result.hits.hits.map(it => (it._source as any).name);
console.log(suggestions);
console.log(result.took);
}
/** /**
* Search all indexed data * Search all indexed data
* @param parameters search query * @param parameters search query
@@ -364,18 +402,23 @@ export class Elasticsearch implements Database {
throw new TypeError('Database is undefined. You have to configure the query build'); throw new TypeError('Database is undefined. You have to configure the query build');
} }
const esConfig: ElasticsearchConfig = { const esConfig = this.config.internal.database as object as ElasticsearchConfig;
name: this.config.internal.database.name as 'elasticsearch',
version: this.config.internal.database.version as string,
query: this.config.internal.database.query as
| ElasticsearchQueryDisMaxConfig
| ElasticsearchQueryQueryStringConfig
| undefined,
};
const response: SearchResponse<SCThings> = await this.client.search({ const response: SearchResponse<SCThings> = await this.client.search({
aggs: aggregations, aggs: aggregations,
query: buildQuery(parameters, this.config, esConfig), query: buildQuery(parameters, this.config, esConfig),
suggest:
parameters.query === undefined
? undefined
: {
text: parameters.query,
terms: {
term: {
field: 'name',
suggest_mode: 'missing',
},
},
},
from: parameters.from, from: parameters.from,
index: ACTIVE_INDICES_ALIAS, index: ACTIVE_INDICES_ALIAS,
allow_no_indices: true, allow_no_indices: true,
@@ -395,6 +438,7 @@ export class Elasticsearch implements Database {
response.aggregations === undefined response.aggregations === undefined
? [] ? []
: parseAggregations(response.aggregations as Record<AggregateName, AggregationsMultiTermsBucket>), : parseAggregations(response.aggregations as Record<AggregateName, AggregationsMultiTermsBucket>),
suggestions: response.suggest === undefined ? undefined : parseSuggestions(response.suggest),
pagination: { pagination: {
count: response.hits.hits.length, count: response.hits.hits.length,
offset: typeof parameters.from === 'number' ? parameters.from : 0, offset: typeof parameters.from === 'number' ? parameters.from : 0,

View File

@@ -30,84 +30,21 @@ export const buildQuery = function buildQuery(
defaultConfig: SCConfigFile, defaultConfig: SCConfigFile,
elasticsearchConfig: ElasticsearchConfig, elasticsearchConfig: ElasticsearchConfig,
): QueryDslQueryContainer { ): QueryDslQueryContainer {
// if config provides a minMatch parameter, we use query_string instead of a match query return {
let query;
if (elasticsearchConfig.query === undefined) {
query = {
query_string: {
analyzer: 'search_german',
default_field: 'name',
minimum_should_match: '90%',
query: typeof parameters.query === 'string' ? parameters.query : '*',
},
};
} else if (elasticsearchConfig.query.queryType === 'query_string') {
query = {
query_string: {
analyzer: 'search_german',
default_field: 'name',
minimum_should_match: elasticsearchConfig.query.minMatch,
query: typeof parameters.query === 'string' ? parameters.query : '*',
},
};
} else if (elasticsearchConfig.query.queryType === 'dis_max') {
if (typeof parameters.query === 'string' && parameters.query !== '*') {
query = {
dis_max: {
boost: 1.2,
queries: [
{
match: {
name: {
boost: elasticsearchConfig.query.matchBoosting,
fuzziness: elasticsearchConfig.query.fuzziness,
query: parameters.query,
},
},
},
{
query_string: {
default_field: 'name',
minimum_should_match: elasticsearchConfig.query.minMatch,
query: parameters.query,
},
},
],
tie_breaker: elasticsearchConfig.query.tieBreaker,
},
};
}
} else {
throw new Error(
'Unsupported query type. Check your config file and reconfigure your elasticsearch query',
);
}
const functionScoreQuery: QueryDslQueryContainer = {
function_score: { function_score: {
functions: buildScoringFunctions(defaultConfig.internal.boostings, parameters.context), functions: buildScoringFunctions(defaultConfig.internal.boostings, parameters.context),
query: { query: {
bool: { bool: {
minimum_should_match: 0, // if we have no should, nothing can match must:
must: [], parameters.query === undefined || parameters.query === '' || parameters.query === '*'
? {match_all: {}}
: {multi_match: {...elasticsearchConfig.query, query: parameters.query}},
should: [], should: [],
filter: parameters.filter === undefined ? undefined : buildFilter(parameters.filter),
}, },
}, },
score_mode: 'multiply', score_mode: 'max',
boost_mode: 'multiply',
}, },
}; };
const mustMatch = functionScoreQuery.function_score?.query?.bool?.must;
if (Array.isArray(mustMatch)) {
if (query !== undefined) {
mustMatch.push(query);
}
if (parameters.filter !== undefined) {
mustMatch.push(buildFilter(parameters.filter));
}
}
return functionScoreQuery;
}; };

View File

@@ -13,68 +13,7 @@
* this program. If not, see <https://www.gnu.org/licenses/>. * this program. If not, see <https://www.gnu.org/licenses/>.
*/ */
/** import {QueryDslMultiMatchQuery} from '@elastic/elasticsearch/lib/api/types.js';
* A configuration for using the Dis Max Query
*
* See https://www.elastic.co/guide/en/elasticsearch/reference/5.5/query-dsl-dis-max-query.html for further
* explanation of what the parameters mean
*/
export interface ElasticsearchQueryDisMaxConfig {
/**
* Relative (to a total number of documents) or absolute number to exclude meaningless matches that frequently appear
*/
cutoffFrequency: number;
/**
* The maximum allowed Levenshtein Edit Distance (or number of edits)
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/common-options.html#fuzziness
*/
fuzziness: number | string;
/**
* Increase the importance (relevance score) of a field
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/mapping-boost.html
*/
matchBoosting: number;
/**
* Minimal number (or percentage) of words that should match in a query
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/query-dsl-minimum-should-match.html
*/
minMatch: string;
/**
* Type of the query - in this case 'dis_max' which is a union of its subqueries
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/query-dsl-dis-max-query.html
*/
queryType: 'dis_max';
/**
* Changes behavior of default calculation of the score when multiple results match
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/query-dsl-multi-match-query.html#tie-breaker
*/
tieBreaker: number;
}
/**
* A configuration for using Query String Query
*
* See https://www.elastic.co/guide/en/elasticsearch/reference/5.5/query-dsl-query-string-query.html for further
* explanation of what the parameters mean
*/
export interface ElasticsearchQueryQueryStringConfig {
/**
* Minimal number (or percentage) of words that should match in a query
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/query-dsl-minimum-should-match.html
*/
minMatch: string;
/**
* Type of the query - in this case 'query_string' which uses a query parser in order to parse content
* @see https://www.elastic.co/guide/en/elasticsearch/reference/5.6/query-dsl-query-string-query.html
*/
queryType: 'query_string';
}
/** /**
* An config file for the elasticsearch database interface * An config file for the elasticsearch database interface
@@ -105,7 +44,12 @@ export interface ElasticsearchConfig {
/** /**
* Configuration for using queries * Configuration for using queries
*/ */
query?: ElasticsearchQueryDisMaxConfig | ElasticsearchQueryQueryStringConfig; query: Omit<QueryDslMultiMatchQuery, 'query'>;
/**
*
*/
searchAsYouTypeQuery: Omit<QueryDslMultiMatchQuery, 'query'>;
/** /**
* Version of the used elasticsearch * Version of the used elasticsearch

View File

@@ -0,0 +1,28 @@
import {
SearchSuggest,
SearchTermSuggest,
SearchTermSuggestOption,
SuggestionName,
} from '@elastic/elasticsearch/lib/api/types.js';
import {SCSearchSuggestions} from '@openstapps/core';
/**
* Parse ES Suggestions to SC Search Suggestions
*/
export function parseSuggestions(suggest: Record<SuggestionName, SearchSuggest[]>): SCSearchSuggestions {
const termsSuggestions =
suggest.terms === undefined
? []
: (suggest.terms as SearchTermSuggest[])
?.map(
({text, options}) =>
[
text,
(options as SearchTermSuggestOption[] | undefined)?.map(({text}) => text) ?? [],
] as const,
)
.filter(([, suggestions]) => suggestions.length > 0) ?? [];
return {
terms: termsSuggestions.length === 0 ? undefined : Object.fromEntries(termsSuggestions),
};
}

View File

@@ -24,6 +24,7 @@ import {
SCSearchQuery, SCSearchQuery,
SCSearchSort, SCSearchSort,
SCThings, SCThings,
SCSearchSuggestions,
} from '@openstapps/core'; } from '@openstapps/core';
import {NGXLogger} from 'ngx-logger'; import {NGXLogger} from 'ngx-logger';
import {combineLatest, Subject} from 'rxjs'; import {combineLatest, Subject} from 'rxjs';
@@ -110,6 +111,8 @@ export class SearchPageComponent implements OnInit {
*/ */
items: Promise<SCThings[]>; items: Promise<SCThings[]>;
suggestions: SCSearchSuggestions | undefined;
/** /**
* Page size of queries * Page size of queries
*/ */
@@ -219,6 +222,7 @@ export class SearchPageComponent implements OnInit {
try { try {
const result = await this.dataProvider.search(searchOptions); const result = await this.dataProvider.search(searchOptions);
this.suggestions = result.suggestions;
this.singleTypeResponse = result.facets.find(facet => facet.field === 'type')?.buckets.length === 1; this.singleTypeResponse = result.facets.find(facet => facet.field === 'type')?.buckets.length === 1;
if (append) { if (append) {
// append results // append results
@@ -283,6 +287,12 @@ export class SearchPageComponent implements OnInit {
this.contextMenuService.updateContextFilter(facets); this.contextMenuService.updateContextFilter(facets);
} }
applySuggestion(target: string, suggestion: string) {
this.queryText = this.queryText.replaceAll(target, suggestion);
this.suggestions = undefined;
this.searchStringChanged(this.queryText);
}
ngOnInit(defaultListeners = true) { ngOnInit(defaultListeners = true) {
this.initialize(); this.initialize();
this.contextMenuService.setContextSort({ this.contextMenuService.setContextSort({

View File

@@ -69,7 +69,23 @@
</ion-header> </ion-header>
<ion-content class="content"> <ion-content class="content">
<div class="suggestions">
@if (suggestions?.terms; as terms) {
<span>{{ 'search.SUGGESTIONS' | translate }}: </span>
@for (suggestion of terms | keyvalue; track suggestion) {
@for (term of suggestion.value; track term) {
@if ($index == 0) {
<b (click)="applySuggestion(suggestion.key, term)" class="suggestion">{{ term }}</b>
} @else {
<span (click)="applySuggestion(suggestion.key, term)" class="suggestion">{{ term }}</span>
}
}
}
}
</div>
<div <div
class="hint"
[class.no-results]="!showDefaultData && !items && !loading" [class.no-results]="!showDefaultData && !items && !loading"
[style.display]="!showDefaultData && !items && !loading ? 'block' : 'none'" [style.display]="!showDefaultData && !items && !loading ? 'block' : 'none'"
> >

View File

@@ -46,7 +46,7 @@ ion-content {
--background: var(--ion-background-color); --background: var(--ion-background-color);
} }
.content > div { .content > .hint {
height: 100%; height: 100%;
ion-label.centered-message-container { ion-label.centered-message-container {
@@ -60,3 +60,19 @@ ion-content {
ion-header { ion-header {
background: var(--ion-color-primary); background: var(--ion-color-primary);
} }
.suggestions {
padding: var(--spacing-md);
padding-block-end: 0;
}
.suggestion {
cursor: pointer;
}
.suggestion + .suggestion::before {
cursor: text;
content: ',';
padding-inline-end: 0.25ch;
font-weight: normal;
}

View File

@@ -422,7 +422,8 @@
"placeholder": "Veranstaltungen, Personen, Orte und mehr" "placeholder": "Veranstaltungen, Personen, Orte und mehr"
}, },
"instruction": "Finde alle Informationen rund ums Studium und den Campus", "instruction": "Finde alle Informationen rund ums Studium und den Campus",
"nothing_found": "Keine Ergebnisse" "nothing_found": "Keine Ergebnisse",
"SUGGESTIONS": "Meintest du"
}, },
"hebisSearch": { "hebisSearch": {
"title": "Bibliothekssuche", "title": "Bibliothekssuche",

View File

@@ -422,7 +422,8 @@
"placeholder": "Events, places, persons and more" "placeholder": "Events, places, persons and more"
}, },
"instruction": "Find all information related to your studies and campus", "instruction": "Find all information related to your studies and campus",
"nothing_found": "No results" "nothing_found": "No results",
"SUGGESTIONS": "Did you mean"
}, },
"hebisSearch": { "hebisSearch": {
"title": "Library Search", "title": "Library Search",

View File

@@ -18,7 +18,7 @@
// The list of which env maps to which file can be found in `.angular-cli.json`. // The list of which env maps to which file can be found in `.angular-cli.json`.
export const environment = { export const environment = {
backend_url: 'https://mobile.server.uni-frankfurt.de', backend_url: 'http://localhost:3000',
app_host: 'mobile.app.uni-frankfurt.de', app_host: 'mobile.app.uni-frankfurt.de',
custom_url_scheme: 'de.anyschool.app', custom_url_scheme: 'de.anyschool.app',
backend_version: '999.0.0', backend_version: '999.0.0',

View File

@@ -23,7 +23,7 @@ node ./lib/cli.js e2e http://localhost:3000
Example to clone the full database Example to clone the full database
```shell ```shell
node app.js copy "*" https://mobile.app.uni-frankfurt.de http://localhost:3000 100 node app.js copy "*" -a "999.0.0" https://mobile.server.uni-frankfurt.de http://localhost:3000 100
``` ```
### Program arguments ### Program arguments

View File

@@ -97,6 +97,8 @@
"date", "date",
"validatable", "validatable",
"filterable", "filterable",
"suggestable",
"completable",
"inheritTags", "inheritTags",
"minLength", "minLength",
"pattern", "pattern",

View File

@@ -24,6 +24,11 @@ export interface SCSearchResult {
*/ */
data: SCThings[]; data: SCThings[];
/**
* Suggestions for query corrections
*/
suggestions?: SCSearchSuggestions;
/** /**
* Facets (aggregations over all matching data) * Facets (aggregations over all matching data)
*/ */
@@ -40,6 +45,18 @@ export interface SCSearchResult {
stats: SCSearchResultSearchEngineStats; stats: SCSearchResultSearchEngineStats;
} }
/**
* Seach suggestions
*
* Not to be confused with search-as-you-type suggestions
*/
export interface SCSearchSuggestions {
/**
* Suggestions for query terms that might have been misspelled
*/
terms?: Record<string, string[]>;
}
/** /**
* Stores information about Pagination * Stores information about Pagination
*/ */

View File

@@ -63,7 +63,7 @@ export interface SCThingWithoutReferences {
/** /**
* Alternate names of the thing * Alternate names of the thing
* @filterable * @filterable
* @keyword * @text
*/ */
alternateNames?: string[]; alternateNames?: string[];
@@ -92,6 +92,8 @@ export interface SCThingWithoutReferences {
* @filterable * @filterable
* @minLength 1 * @minLength 1
* @sortable ducet * @sortable ducet
* @completable
* @suggestable
* @text * @text
*/ */
name: string; name: string;
@@ -240,6 +242,8 @@ export interface SCThingTranslatableProperties {
* Translation of the name of the thing * Translation of the name of the thing
* @sortable ducet * @sortable ducet
* @text * @text
* @suggestable
* @completable
*/ */
name?: string; name?: string;
/** /**

View File

@@ -43,6 +43,27 @@ export const fieldmap: ElasticsearchFieldmap = {
}, },
ignore: ['price'], ignore: ['price'],
}, },
suggestable: {
default: {
trigram: {
type: 'text',
analyzer: 'trigram',
},
reverse: {
type: 'text',
analyzer: 'reverse',
},
},
ignore: [],
},
completable: {
default: {
completion: {
type: 'search_as_you_type',
},
},
ignore: [],
},
}; };
export const filterableTagName = 'filterable'; export const filterableTagName = 'filterable';

View File

@@ -19,4 +19,27 @@ export const settings: IndicesPutTemplateRequest['settings'] = {
'max_result_window': 30_000, 'max_result_window': 30_000,
'number_of_replicas': 0, 'number_of_replicas': 0,
'number_of_shards': 1, 'number_of_shards': 1,
'index': {
analysis: {
analyzer: {
trigram: {
type: 'custom',
tokenizer: 'standard',
filter: ['lowercase', 'shingle'],
},
reverse: {
type: 'custom',
tokenizer: 'standard',
filter: ['lowercase', 'reverse'],
},
},
filter: {
shingle: {
type: 'shingle',
min_shingle_size: 2,
max_shingle_size: 3,
},
},
},
},
}; };