gitea/services/wiki/wiki_index.go

// Copyright 2026 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT

package wiki

import (
	"context"
	"crypto/sha256"
	"encoding/hex"
	"regexp"
	"slices"
	"strings"

	repo_model "code.gitea.io/gitea/models/repo"
	"code.gitea.io/gitea/modules/git"
	"code.gitea.io/gitea/modules/gitrepo"
	"code.gitea.io/gitea/modules/json"
	"code.gitea.io/gitea/modules/log"
)

// IndexWikiPage indexes a single wiki page for search
func IndexWikiPage(ctx context.Context, repo *repo_model.Repository, pageName string) error {
	wikiRepo, commit, err := findWikiRepoCommit(ctx, repo)
	if err != nil {
		return err
	}
	if wikiRepo != nil {
		defer wikiRepo.Close()
	}
	if commit == nil {
		return nil
	}

	// Get the page content
	pagePath := WebPathToGitPath(WebPath(pageName))
	entry, err := commit.GetTreeEntryByPath(pagePath)
	if err != nil {
		return err
	}

	blob := entry.Blob()
	content, err := blob.GetBlobContent(1024 * 1024) // 1MB max
	if err != nil {
		return err
	}

	// Calculate hash
	hash := sha256.Sum256([]byte(content))
	contentHash := hex.EncodeToString(hash[:])

	// Check if already indexed with same hash
	existing, err := repo_model.GetWikiIndex(ctx, repo.ID, pageName)
	if err != nil {
		return err
	}
	if existing != nil && existing.ContentHash == contentHash {
		return nil // Already up to date
	}

	// Extract links from content
	links := extractWikiLinks(content)
	linksJSON, _ := json.Marshal(links)

	// Count words
	wordCount := countWords(content)

	// Get title from first heading or page name
	title := extractTitle(content, pageName)

	// Create/update index
	idx := &repo_model.WikiIndex{
		RepoID:      repo.ID,
		PageName:    pageName,
		PagePath:    pagePath,
		Title:       title,
		Content:     content,
		ContentHash: contentHash,
		CommitSHA:   commit.ID.String(),
		WordCount:   wordCount,
		LinksOut:    string(linksJSON),
	}

	return repo_model.CreateOrUpdateWikiIndex(ctx, idx)
}

// IndexAllWikiPages indexes all pages in a wiki
func IndexAllWikiPages(ctx context.Context, repo *repo_model.Repository) error {
	wikiRepo, commit, err := findWikiRepoCommit(ctx, repo)
	if err != nil {
		return err
	}
	if wikiRepo != nil {
		defer wikiRepo.Close()
	}
	if commit == nil {
		return nil
	}

	// Get all entries
	entries, err := commit.ListEntries()
	if err != nil {
		return err
	}

	indexedPages := make(map[string]bool)

	for _, entry := range entries {
		if entry.IsDir() {
			continue
		}
		if !strings.HasSuffix(entry.Name(), ".md") {
			continue
		}

		webPath, err := GitPathToWebPath(entry.Name())
		if err != nil {
			continue
		}
		pageName := string(webPath)
		if pageName == "" {
			continue
		}

		if err := IndexWikiPage(ctx, repo, pageName); err != nil {
			log.Warn("Failed to index wiki page %s: %v", pageName, err)
			continue
		}
		indexedPages[pageName] = true
	}

	// Remove deleted pages from index
	existingIndexes, err := repo_model.GetWikiIndexByRepo(ctx, repo.ID)
	if err != nil {
		return err
	}

	for _, idx := range existingIndexes {
		if !indexedPages[idx.PageName] {
			if err := repo_model.DeleteWikiIndex(ctx, repo.ID, idx.PageName); err != nil {
				log.Warn("Failed to remove deleted wiki page from index %s: %v", idx.PageName, err)
			}
		}
	}

	return nil
}

// RemoveWikiPageFromIndex removes a page from the search index
func RemoveWikiPageFromIndex(ctx context.Context, repoID int64, pageName string) error {
	return repo_model.DeleteWikiIndex(ctx, repoID, pageName)
}

// ClearWikiIndex removes all indexed pages for a repository
func ClearWikiIndex(ctx context.Context, repoID int64) error {
	return repo_model.DeleteWikiIndexByRepo(ctx, repoID)
}

// GetWikiGraph returns the link graph for a wiki
func GetWikiGraph(ctx context.Context, repoID int64) (
	nodes []map[string]any,
	edges []map[string]any,
	err error,
) {
	indexes, err := repo_model.GetWikiIndexByRepo(ctx, repoID)
	if err != nil {
		return nil, nil, err
	}

	nodes = make([]map[string]any, 0, len(indexes))
	edges = make([]map[string]any, 0)
	pageSet := make(map[string]bool)

	// Build nodes
	for _, idx := range indexes {
		pageSet[idx.PageName] = true
		nodes = append(nodes, map[string]any{
			"name":       idx.PageName,
			"title":      idx.Title,
			"word_count": idx.WordCount,
		})
	}

	// Build edges from links
	for _, idx := range indexes {
		var links []string
		if idx.LinksOut != "" {
			_ = json.Unmarshal([]byte(idx.LinksOut), &links)
		}
		for _, link := range links {
			if pageSet[link] { // Only include links to existing pages
				edges = append(edges, map[string]any{
					"source": idx.PageName,
					"target": link,
				})
			}
		}
	}

	return nodes, edges, nil
}

// GetWikiIncomingLinks returns pages that link to the given page
func GetWikiIncomingLinks(ctx context.Context, repoID int64, pageName string) ([]string, error) {
	indexes, err := repo_model.GetWikiIndexByRepo(ctx, repoID)
	if err != nil {
		return nil, err
	}

	incoming := make([]string, 0)
	for _, idx := range indexes {
		var links []string
		if idx.LinksOut != "" {
			_ = json.Unmarshal([]byte(idx.LinksOut), &links)
		}
		if slices.Contains(links, pageName) {
			incoming = append(incoming, idx.PageName)
		}
	}

	return incoming, nil
}

// GetOrphanedPages returns pages with no incoming links
func GetOrphanedPages(ctx context.Context, repoID int64) ([]*repo_model.WikiIndex, error) {
	indexes, err := repo_model.GetWikiIndexByRepo(ctx, repoID)
	if err != nil {
		return nil, err
	}

	// Build set of pages that are linked to
	linkedPages := make(map[string]bool)
	for _, idx := range indexes {
		var links []string
		if idx.LinksOut != "" {
			_ = json.Unmarshal([]byte(idx.LinksOut), &links)
		}
		for _, link := range links {
			linkedPages[link] = true
		}
	}

	// Find orphaned pages (excluding Home which is always accessible)
	orphaned := make([]*repo_model.WikiIndex, 0)
	for _, idx := range indexes {
		if idx.PageName != "Home" && !linkedPages[idx.PageName] {
			orphaned = append(orphaned, idx)
		}
	}

	return orphaned, nil
}

// GetDeadLinks returns links to non-existent pages
func GetDeadLinks(ctx context.Context, repoID int64) ([]map[string]string, error) {
	indexes, err := repo_model.GetWikiIndexByRepo(ctx, repoID)
	if err != nil {
		return nil, err
	}

	// Build set of existing pages
	existingPages := make(map[string]bool)
	for _, idx := range indexes {
		existingPages[idx.PageName] = true
	}

	// Find dead links
	deadLinks := make([]map[string]string, 0)
	for _, idx := range indexes {
		var links []string
		if idx.LinksOut != "" {
			_ = json.Unmarshal([]byte(idx.LinksOut), &links)
		}
		for _, link := range links {
			if !existingPages[link] {
				deadLinks = append(deadLinks, map[string]string{
					"page":        idx.PageName,
					"broken_link": link,
				})
			}
		}
	}

	return deadLinks, nil
}

// findWikiRepoCommit opens the wiki repo and gets the latest commit
func findWikiRepoCommit(ctx context.Context, repo *repo_model.Repository) (*git.Repository, *git.Commit, error) {
	wikiRepo, err := gitrepo.OpenRepository(ctx, repo.WikiStorageRepo())
	if err != nil {
		if git.IsErrNotExist(err) {
			return nil, nil, nil
		}
		return nil, nil, err
	}

	branch := repo.DefaultWikiBranch
	if branch == "" {
		branch = "master"
	}

	commit, err := wikiRepo.GetBranchCommit(branch)
	if err != nil {
		wikiRepo.Close()
		if git.IsErrNotExist(err) {
			return nil, nil, nil
		}
		return nil, nil, err
	}

	return wikiRepo, commit, nil
}

// extractWikiLinks extracts wiki page links from markdown content
func extractWikiLinks(content string) []string {
	links := make([]string, 0)
	seen := make(map[string]bool)

	// Match [[Page Name]] style wiki links
	wikiLinkRe := regexp.MustCompile(`\[\[([^\]|]+)(?:\|[^\]]+)?\]\]`)
	matches := wikiLinkRe.FindAllStringSubmatch(content, -1)
	for _, match := range matches {
		if len(match) > 1 {
			link := strings.TrimSpace(match[1])
			// Convert to page name format
			link = strings.ReplaceAll(link, " ", "-")
			if !seen[link] {
				links = append(links, link)
				seen[link] = true
			}
		}
	}

	// Match [text](wiki/Page-Name) style links
	mdLinkRe := regexp.MustCompile(`\[([^\]]+)\]\((?:\.\.?/)?(?:wiki/)?([^)]+)\)`)
	matches = mdLinkRe.FindAllStringSubmatch(content, -1)
	for _, match := range matches {
		if len(match) > 2 {
			link := match[2]
			// Skip external links
			if strings.HasPrefix(link, "http://") || strings.HasPrefix(link, "https://") {
				continue
			}
			// Clean up the link
			link = strings.TrimPrefix(link, "./")
			link = strings.TrimSuffix(link, ".md")
			if !seen[link] && link != "" {
				links = append(links, link)
				seen[link] = true
			}
		}
	}

	return links
}

// extractTitle extracts the title from markdown content
func extractTitle(content, defaultTitle string) string {
	// Look for first H1 heading
	for line := range strings.SplitSeq(content, "\n") {
		line = strings.TrimSpace(line)
		if title, ok := strings.CutPrefix(line, "# "); ok {
			return title
		}
	}
	return defaultTitle
}

// countWords counts the number of words in content
func countWords(content string) int {
	// Remove markdown formatting
	content = regexp.MustCompile(`[#*_\[\](){}]`).ReplaceAllString(content, " ")
	// Split on whitespace
	words := strings.Fields(content)
	return len(words)
}