searcherside/core/services/bleve_searcher.go

package services

import (
	"context"
	"slices"

	"github.com/blevesearch/bleve/v2"
	"github.com/blevesearch/bleve/v2/search/query"
	"github.com/mitchellh/mapstructure"

	"code.icb4dc0.de/prskr/searcherside/core/dto"
	"code.icb4dc0.de/prskr/searcherside/core/ports"
)

var (
	_            ports.Searcher = (*BleveSearcher)(nil)
	searchFields                = []string{
		"mainTitle",
		"pageTitle",
		"url",
		"headings",
		"content",
		"metaDescription",
		"breadcrumbs",
	}
)

func NewBleveSearcher(indexPath string) (*BleveSearcher, error) {
	idx, err := bleve.Open(indexPath)
	if err != nil {
		return nil, err
	}
	return &BleveSearcher{idx: idx}, nil
}

type BleveSearcher struct {
	idx bleve.Index
}

func (b BleveSearcher) Search(ctx context.Context, req ports.IndexSearchRequest) (resp dto.SearchResponse, err error) {
	var searchQuery query.Query
	if req.ExactSearch {
		searchQuery = bleve.NewTermQuery(req.Query)
	} else {
		searchQuery = bleve.NewFuzzyQuery(req.Query)
	}

	maxHits := req.MaxHits
	if maxHits == 0 {
		maxHits = 10
	}

	searchRequest := bleve.NewSearchRequestOptions(searchQuery, maxHits, 0, false)
	searchRequest.Fields = slices.Clone(searchFields)
	searchRequest.Highlight = bleve.NewHighlight()

	result, err := b.idx.Search(searchRequest)
	if err != nil {
		return resp, err
	}

	for _, hit := range result.Hits {
		var page pageToIndex
		if err := mapstructure.Decode(hit.Fields, &page); err != nil {
			return resp, err
		}

		searchHit := dto.SearchHit{
			ObjectID:    hit.ID,
			MainTitle:   page.MainTitle,
			PageTitle:   page.PageTitle,
			Url:         page.Url,
			Breadcrumbs: page.Breadcrumbs,
			Snippet: dto.SnippetResult{
				Content: dto.SnippetContent{
					Value:      page.Content,
					MatchLevel: dto.MatchLevelFull,
				},
			},
			Highlight: make(map[string]dto.Match),
		}

		if contentFragment, ok := hit.Fragments["content"]; ok && len(contentFragment) > 0 {
			searchHit.Snippet.Content.Value = contentFragment[0]
		}

		for _, field := range searchFields {
			fieldValue := hit.Fields[field].(string)
			if locationMap, ok := hit.Locations[field]; ok {
				fieldMatch := dto.Match{
					MatchLevel:   dto.MatchLevelFull,
					Value:        hit.Fragments[field][0],
					MatchedWords: make([]string, 0),
				}
				for _, locations := range locationMap {
					for _, location := range locations {
						matchedWord := fieldValue[location.Start:location.End]
						if pos, found := slices.BinarySearch(fieldMatch.MatchedWords, matchedWord); found {
							continue
						} else if pos == len(fieldMatch.MatchedWords) {
							fieldMatch.MatchedWords = append(fieldMatch.MatchedWords, matchedWord)
						} else {
							fieldMatch.MatchedWords = append(fieldMatch.MatchedWords[:pos], append([]string{matchedWord}, fieldMatch.MatchedWords[pos:]...)...)
						}
					}
				}
				searchHit.Highlight[field] = fieldMatch
			} else {
				searchHit.Highlight[field] = dto.Match{
					Value:        fieldValue,
					MatchLevel:   dto.MatchLevelNone,
					MatchedWords: make([]string, 0),
				}
			}
		}

		resp.Hits = append(resp.Hits, searchHit)
	}

	return resp, nil
}

func (b BleveSearcher) Close() error {
	return b.idx.Close()
}