collect/collector/collector.go

package collector

import (
	"fmt"
	"io/fs"
	"os"
	"path/filepath"
	"strings"
)

// FileEntry represents a file to be archived
type FileEntry struct {
	Path     string // Relative path from sourceDir
	FullPath string // Absolute path for reading
}

// Collector handles file collection based on a matcher and optional directory filter
type Collector struct {
	matcher   Matcher
	dirFilter DirectoryFilter
	verbose   bool
}

// New creates a new collector with the specified matcher
func New(matcher Matcher, verbose bool) *Collector {
	return &Collector{
		matcher:   matcher,
		dirFilter: nil,
		verbose:   verbose,
	}
}

// NewWithDirectoryFilter creates a new collector with the specified matcher and directory filter
func NewWithDirectoryFilter(matcher Matcher, dirFilter DirectoryFilter, verbose bool) *Collector {
	return &Collector{
		matcher:   matcher,
		dirFilter: dirFilter,
		verbose:   verbose,
	}
}

// Collect walks the source directory and collects matching files
func (c *Collector) Collect(sourceDir string) ([]FileEntry, error) {
	// Clean and convert to absolute path
	absSourceDir, err := filepath.Abs(sourceDir)
	if err != nil {
		return nil, fmt.Errorf("failed to get absolute path: %w", err)
	}

	// Check if source directory exists
	info, err := os.Stat(absSourceDir)
	if err != nil {
		return nil, fmt.Errorf("source directory error: %w", err)
	}
	if !info.IsDir() {
		return nil, fmt.Errorf("source path is not a directory: %s", sourceDir)
	}

	var files []FileEntry

	err = filepath.WalkDir(absSourceDir, func(path string, d fs.DirEntry, err error) error {
		if err != nil {
			// Log permission errors but continue walking
			if os.IsPermission(err) {
				fmt.Fprintf(os.Stderr, "Warning: Permission denied: %s\n", path)
				return nil
			}
			return err
		}

		// Get file info
		info, err := d.Info()
		if err != nil {
			if os.IsPermission(err) {
				fmt.Fprintf(os.Stderr, "Warning: Cannot stat file: %s\n", path)
				return nil
			}
			return err
		}

		// Skip symlinks
		if info.Mode()&os.ModeSymlink != 0 {
			return nil
		}

		// Check directory filter for directories
		if info.IsDir() && c.dirFilter != nil {
			if !c.dirFilter.ShouldTraverse(path, absSourceDir) {
				return filepath.SkipDir
			}
		}

		// Check if this file should be included
		if c.matcher.ShouldInclude(path, info) {
			// Calculate relative path from source directory
			relPath, err := filepath.Rel(absSourceDir, path)
			if err != nil {
				return fmt.Errorf("failed to get relative path: %w", err)
			}

			// Clean the relative path to ensure consistent formatting
			relPath = filepath.ToSlash(relPath)

			// Show verbose output when file is found
			if c.verbose {
				fmt.Printf("found: %s\n", relPath)
			}

			files = append(files, FileEntry{
				Path:     relPath,
				FullPath: path,
			})
		}

		return nil
	})

	if err != nil {
		return nil, fmt.Errorf("walk error: %w", err)
	}

	// Check if any files were found
	if len(files) == 0 {
		return nil, fmt.Errorf("no files found matching criteria")
	}

	return files, nil
}

// GetArchiveFormat determines the archive format from the filename
func GetArchiveFormat(filename string) string {
	lower := strings.ToLower(filename)

	if strings.HasSuffix(lower, ".tar.gz") || strings.HasSuffix(lower, ".tgz") {
		return "tar.gz"
	}

	if strings.HasSuffix(lower, ".zip") {
		return "zip"
	}

	return ""
}