Method processContent

internal/adapter/github.go:120–176

processContent processes a GitHub content item recursively

(ctx context.Context, owner, repo string, content *github.RepositoryContent, path string, knowledgeID string)

// processContent processes a GitHub content item recursively
func (g *GitHubAdapter) processContent(ctx context.Context, owner, repo string, content *github.RepositoryContent, path string, knowledgeID string) ([]*File, error) {
	if content == nil {
		return nil, nil
	}

	currentPath := filepath.Join(path, content.GetName())

	// Skip binary files and non-text files
	if content.GetType() == "file" {
		// Check if it's a text file
		if !isTextFile(content.GetName()) {
			return nil, nil
		}

		// Get file content
		fileContent, err := g.getFileContent(ctx, owner, repo, content)
		if err != nil {
			return nil, fmt.Errorf("failed to get file content: %w", err)
		}

		// Calculate hash
		hash := fmt.Sprintf("%x", sha256.Sum256(fileContent))

		return []*File{{
			Path:        currentPath,
			Content:     fileContent,
			Hash:        hash,
			Modified:    time.Now(), // GitHub API doesn't provide modification time for content
			Size:        int64(len(fileContent)),
			Source:      fmt.Sprintf("%s/%s", owner, repo),
			KnowledgeID: knowledgeID,
		}}, nil
	}

	// If it's a directory, recurse
	if content.GetType() == "dir" {
		_, contents, _, err := g.client.Repositories.GetContents(ctx, owner, repo, content.GetPath(), nil)
		if err != nil {
			return nil, fmt.Errorf("failed to get directory contents: %w", err)
		}

		var allFiles []*File
		for _, subContent := range contents {
			files, err := g.processContent(ctx, owner, repo, subContent, currentPath, knowledgeID)
			if err != nil {
				continue
			}
			if files != nil {
				allFiles = append(allFiles, files...)
			}
		}

		return allFiles, nil
	}

	return nil, nil
}