gitea-search/internal/meili/client.go

package meili

import (
	"crypto/sha256"
	"fmt"
	"log"
	"time"

	"github.com/meilisearch/meilisearch-go"
)

// Document represents an indexed file in MeiliSearch.
type Document struct {
	ID        string `json:"id"`
	Repo      string `json:"repo"`
	Branch    string `json:"branch"`
	Path      string `json:"path"`
	Filename  string `json:"filename"`
	Extension string `json:"extension"`
	Content   string `json:"content"`
	Language  string `json:"language"`
	UpdatedAt int64  `json:"updated_at"`
}

// SearchResult holds a single search hit.
type SearchResult struct {
	Repo      string `json:"repo"`
	Branch    string `json:"branch"`
	Path      string `json:"path"`
	Filename  string `json:"filename"`
	Extension string `json:"extension"`
	Snippet   string `json:"snippet"`
}

// Client wraps the MeiliSearch SDK.
type Client struct {
	client    meilisearch.ServiceManager
	indexName string
}

// NewClient creates a new MeiliSearch client.
func NewClient(url, apiKey, indexName string) (*Client, error) {
	client := meilisearch.New(url, meilisearch.WithAPIKey(apiKey))

	c := &Client{
		client:    client,
		indexName: indexName,
	}

	if err := c.ensureIndex(); err != nil {
		return nil, fmt.Errorf("ensuring index: %w", err)
	}

	return c, nil
}

// ensureIndex creates the index if it doesn't exist and configures settings.
func (c *Client) ensureIndex() error {
	_, err := c.client.GetIndex(c.indexName)
	if err != nil {
		log.Printf("Creating index %q", c.indexName)
		task, err := c.client.CreateIndex(&meilisearch.IndexConfig{
			Uid:        c.indexName,
			PrimaryKey: "id",
		})
		if err != nil {
			return fmt.Errorf("creating index: %w", err)
		}
		if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
			return fmt.Errorf("waiting for index creation: %w", err)
		}
	}

	index := c.client.Index(c.indexName)

	// Configure searchable attributes
	task, err := index.UpdateSearchableAttributes(&[]string{
		"content", "path", "filename", "repo",
	})
	if err != nil {
		return fmt.Errorf("updating searchable attributes: %w", err)
	}
	if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
		return fmt.Errorf("waiting for searchable attributes: %w", err)
	}

	// Configure filterable attributes
	task, err = index.UpdateFilterableAttributes(&[]string{
		"repo", "extension", "branch",
	})
	if err != nil {
		return fmt.Errorf("updating filterable attributes: %w", err)
	}
	if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
		return fmt.Errorf("waiting for filterable attributes: %w", err)
	}

	// Configure displayed attributes (exclude full content)
	task, err = index.UpdateDisplayedAttributes(&[]string{
		"id", "repo", "branch", "path", "filename", "extension", "updated_at",
	})
	if err != nil {
		return fmt.Errorf("updating displayed attributes: %w", err)
	}
	if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
		return fmt.Errorf("waiting for displayed attributes: %w", err)
	}

	log.Printf("Index %q configured", c.indexName)
	return nil
}

// DocumentID generates a deterministic ID from repo, branch, and path.
func DocumentID(repo, branch, path string) string {
	h := sha256.Sum256([]byte(repo + ":" + branch + ":" + path))
	return fmt.Sprintf("%x", h[:16])
}

// IndexDocuments adds or updates documents in MeiliSearch.
// It batches documents in chunks to avoid overwhelming MeiliSearch.
func (c *Client) IndexDocuments(docs []Document) error {
	if len(docs) == 0 {
		return nil
	}

	const batchSize = 100
	index := c.client.Index(c.indexName)

	for i := 0; i < len(docs); i += batchSize {
		end := i + batchSize
		if end > len(docs) {
			end = len(docs)
		}
		batch := docs[i:end]

		task, err := index.AddDocuments(batch, "id")
		if err != nil {
			return fmt.Errorf("adding documents batch %d-%d: %w", i, end, err)
		}

		if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
			return fmt.Errorf("waiting for batch %d-%d: %w", i, end, err)
		}

		log.Printf("Indexed documents %d-%d of %d", i+1, end, len(docs))
	}

	return nil
}

// Search queries MeiliSearch and returns formatted results.
func (c *Client) Search(query string, repo string, filetype string, limit int64) ([]SearchResult, error) {
	if limit <= 0 {
		limit = 10
	}

	index := c.client.Index(c.indexName)

	// Build filter
	var filters []string
	if repo != "" {
		filters = append(filters, fmt.Sprintf("repo = %q", repo))
	}
	if filetype != "" {
		filters = append(filters, fmt.Sprintf("extension = %q", filetype))
	}

	filterStr := ""
	if len(filters) > 0 {
		filterStr = filters[0]
		for _, f := range filters[1:] {
			filterStr += " AND " + f
		}
	}

	searchReq := &meilisearch.SearchRequest{
		Limit:                limit,
		AttributesToRetrieve: []string{"repo", "branch", "path", "filename", "extension"},
		AttributesToCrop:     []string{"content:40"},
		CropLength:           40,
		AttributesToHighlight: []string{"content"},
		ShowMatchesPosition:  true,
	}
	if filterStr != "" {
		searchReq.Filter = filterStr
	}

	resp, err := index.Search(query, searchReq)
	if err != nil {
		return nil, fmt.Errorf("searching: %w", err)
	}

	var results []SearchResult
	for _, hit := range resp.Hits {
		m, ok := hit.(map[string]interface{})
		if !ok {
			continue
		}

		result := SearchResult{
			Repo:      strVal(m, "repo"),
			Branch:    strVal(m, "branch"),
			Path:      strVal(m, "path"),
			Filename:  strVal(m, "filename"),
			Extension: strVal(m, "extension"),
		}

		// Extract highlighted snippet from _formatted
		if formatted, ok := m["_formatted"].(map[string]interface{}); ok {
			if content, ok := formatted["content"].(string); ok {
				result.Snippet = content
			}
		}

		// Fall back to cropped content
		if result.Snippet == "" {
			if cropped, ok := m["_croppped"].(map[string]interface{}); ok {
				if content, ok := cropped["content"].(string); ok {
					result.Snippet = content
				}
			}
		}

		results = append(results, result)
	}

	return results, nil
}

// DeleteByRepo removes all documents for a given repo.
func (c *Client) DeleteByRepo(repo string) error {
	index := c.client.Index(c.indexName)

	task, err := index.DeleteDocumentsByFilter(fmt.Sprintf("repo = %q", repo))
	if err != nil {
		return fmt.Errorf("deleting documents for repo %s: %w", repo, err)
	}

	if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
		return fmt.Errorf("waiting for deletion: %w", err)
	}

	return nil
}

// DeleteAll removes all documents from the index.
func (c *Client) DeleteAll() error {
	index := c.client.Index(c.indexName)

	task, err := index.DeleteAllDocuments()
	if err != nil {
		return fmt.Errorf("deleting all documents: %w", err)
	}

	if _, err := c.client.WaitForTask(task.TaskUID, 500 * time.Millisecond); err != nil {
		return fmt.Errorf("waiting for deletion: %w", err)
	}

	return nil
}

func strVal(m map[string]interface{}, key string) string {
	if v, ok := m[key].(string); ok {
		return v
	}
	return ""
}