Enable wildcard search in terms list

This commit is contained in:
2026-06-20 06:33:16 +02:00
parent cff72f167a
commit 18fd3950c9
7 changed files with 99 additions and 17 deletions
@@ -0,0 +1,9 @@
namespace Rsdo.Concordancer.Core.Extensions;
public static class StringExtensions
{
public static bool IsWildcardSearch(this string value)
{
return !string.IsNullOrEmpty(value) && (value.Contains('*') || value.Contains('?'));
}
}
@@ -1,5 +1,7 @@
using System.Collections.Generic;
using System.Linq;
using OpenSearch.Client;
using Rsdo.Concordancer.Core.Extensions;
namespace Rsdo.Concordancer.Infrastructure.Extensions;
@@ -30,4 +32,36 @@ public static class ElasticQueryExtensions
},
};
}
public static QueryContainer ToQuery(this string value, string field)
{
if (value.IsWildcardSearch())
{
return new WildcardQuery()
{
Field = field,
Value = value,
};
}
return new TermQuery()
{
Field = field,
Value = value,
};
}
public static QueryContainer ToQuery(this List<string> values, string field)
{
if (values.Any(v => v.IsWildcardSearch()))
{
return values.Select(value => value.ToQuery(field)).ToList().ToBooleanOrQuery();
}
return new TermsQuery()
{
Field = field,
Terms = values,
};
}
}
@@ -35,22 +35,12 @@ public class TermListQueryBuilder : IQueryBuilder<TermListQuery>
if (!string.IsNullOrEmpty(wordQuery.Form))
{
queries.Add(
new TermQuery()
{
Field = $"{tokenField}.formLower",
Value = wordQuery.Form.ToLower(),
});
queries.Add(wordQuery.Form.ToLower().ToQuery($"{tokenField}.formLower"));
}
if (!wordQuery.Lemmas.IsNullOrEmpty())
{
queries.Add(
new TermsQuery()
{
Field = $"{tokenField}.lemma",
Terms = wordQuery.Lemmas,
});
queries.Add(wordQuery.Lemmas.ToQuery($"{tokenField}.lemma"));
}
return queries.Count == 0 ? new MatchNoneQuery() : queries.ToBooleanAndQuery();
@@ -1,10 +1,14 @@
using System;
using System.Reflection;
using Autofac;
using Autofac.Core;
using FluentValidation;
using Microsoft.Extensions.Caching.Memory;
using Rsdo.Concordancer.Core.Interfaces;
using Rsdo.Concordancer.Core.Search.Queries.Concordances;
using Rsdo.Concordancer.Core.Search.Queries.TermLists;
using Rsdo.Concordancer.ServiceModel.Requests.Concordances;
using Rsdo.Concordancer.ServiceModel.Requests.TermLists;
using Rsdo.Concordancer.ServiceModel.Types;
using Rsdo.Concordancer.Services.Framework;
using Rsdo.Concordancer.Services.Framework.BulkLoaders;
@@ -15,6 +19,8 @@ using Rsdo.Concordancer.Services.Framework.Decorators;
using Rsdo.Concordancer.Services.Search.Aggregations;
using Rsdo.Concordancer.Services.Search.AlternateSearches;
using Rsdo.Concordancer.Services.Search.QueryFactories;
using Rsdo.Concordancer.Services.Search.QueryFactories.Concordances;
using Rsdo.Concordancer.Services.Search.QueryFactories.TermLists;
using Rsdo.Concordancer.Services.Services.InputQueryParser;
using Rsdo.Concordancer.Services.Services.LemmatizationService;
using Rsdo.Concordancer.Services.Services.ParagraphService;
@@ -83,7 +89,8 @@ public class ServicesModule : Module
builder.RegisterType<DefaultTokenizerService>().Keyed<ITokenizerService>(TokenizerType.Default).SingleInstance();
builder.RegisterType<InputQueryParser>().As<IInputQueryParser>().SingleInstance();
builder.RegisterType<ParagraphService>().As<IParagraphService>().InstancePerLifetimeScope();
builder.RegisterType<LemmatizationService>().As<ILemmatizationService>().InstancePerLifetimeScope();
builder.RegisterType<SimpleLemmatizationService>().Keyed<ILemmatizationService>(LemmatizationType.Simple).InstancePerLifetimeScope();
builder.RegisterType<WildcardLemmatizationService>().Keyed<ILemmatizationService>(LemmatizationType.Wildcard).InstancePerLifetimeScope();
builder.RegisterType<PartOfSpeechService>().As<IPartOfSpeechService>().InstancePerLifetimeScope();
}
@@ -103,7 +110,33 @@ public class ServicesModule : Module
private void RegisterSearch(ContainerBuilder builder)
{
// Query factories
builder.RegisterAssemblyTypes(ServicesAssembly).AsClosedTypesOf(typeof(IQueryFactory<,>)).AsImplementedInterfaces().InstancePerLifetimeScope();
var wildcardLemmatizationServiceParameter = new ResolvedParameter(
(p, ctx) => p.ParameterType == typeof(ILemmatizationService),
(p, ctx) => ctx.ResolveKeyed<ILemmatizationService>(LemmatizationType.Wildcard));
var simpleLemmatizationServiceParameter = new ResolvedParameter(
(p, ctx) => p.ParameterType == typeof(ILemmatizationService),
(p, ctx) => ctx.ResolveKeyed<ILemmatizationService>(LemmatizationType.Simple));
builder.RegisterType<ExportConcordancesQueryFactory>()
.As<IQueryFactory<ExportConcordances, ConcordancesQuery>>()
.WithParameter(simpleLemmatizationServiceParameter)
.InstancePerLifetimeScope();
builder.RegisterType<SearchConcordancesQueryFactory>()
.As<IQueryFactory<SearchConcordances, ConcordancesQuery>>()
.WithParameter(simpleLemmatizationServiceParameter)
.InstancePerLifetimeScope();
builder.RegisterType<ConcordanceDetailsQueryFactory>()
.As<IQueryFactory<ConcordanceDetails, ConcordancesQuery>>()
.WithParameter(simpleLemmatizationServiceParameter)
.InstancePerLifetimeScope();
builder.RegisterType<ExportTermListQueryFactory>()
.As<IQueryFactory<ExportTermList, TermListQuery>>()
.WithParameter(wildcardLemmatizationServiceParameter)
.InstancePerLifetimeScope();
builder.RegisterType<SearchTermListQueryFactory>()
.As<IQueryFactory<SearchTermList, TermListQuery>>()
.WithParameter(wildcardLemmatizationServiceParameter)
.InstancePerLifetimeScope();
// Aggregations
builder.RegisterType<AggregationProviderFactory>().As<IAggregationProviderFactory>().InstancePerLifetimeScope();
@@ -112,6 +145,7 @@ public class ServicesModule : Module
// Alternate searches
builder.RegisterType<LemmasAlternateSearchProvider>()
.As<IAlternateSearchProvider<SearchConcordances, SearchConcordancesResponse>>()
.WithParameter(simpleLemmatizationServiceParameter)
.InstancePerLifetimeScope();
}
}
@@ -4,6 +4,7 @@ using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Autofac.Features.Indexed;
using Rsdo.Concordancer.Core.Extensions;
using Rsdo.Concordancer.ServiceModel.Requests.Concordances;
using Rsdo.Concordancer.ServiceModel.Types;
using Rsdo.Concordancer.Services.Services.TokenizerService;
@@ -21,6 +22,12 @@ public class InputQueryParser : IInputQueryParser
public async Task<IEnumerable<(TokenType type, string form)>> Tokenize(string query)
{
if (query.IsWildcardSearch())
{
// Classla tokenizer doesn't support wildcards, so use default tokenizer
return await tokenizers[TokenizerType.Default].Tokenize(query);
}
try
{
return await tokenizers[TokenizerType.Classla].Tokenize(query);
@@ -0,0 +1,7 @@
namespace Rsdo.Concordancer.Services.Services.LemmatizationService;
public enum LemmatizationType
{
Simple,
Wildcard,
}
@@ -7,17 +7,18 @@ using Rsdo.Concordancer.Services.Framework.DbContext;
namespace Rsdo.Concordancer.Services.Services.LemmatizationService;
public class LemmatizationService : ILemmatizationService
public class SimpleLemmatizationService : ILemmatizationService
{
private readonly MasterDbContext dbContext;
public LemmatizationService(MasterDbContext dbContext)
public SimpleLemmatizationService(MasterDbContext dbContext)
{
this.dbContext = dbContext;
}
public async Task<List<string>> GetLemmas(string form)
public virtual async Task<List<string>> GetLemmas(string form)
{
// Retrieving lemmas without support for wildcard search
var lemmas = await dbContext.LemmaFormPair.Where(f => f.Form.ToLower() == form.ToLower()).Select(f => f.Lemma).Distinct().ToListAsync();
if (lemmas.IsNullOrEmpty())
{