sync file-type support across upload/convert and fix collected_at timezone handling

internal/parser/archive.go

@@ -9,23 +9,38 @@ import (
 	"io"
 	"os"
 	"path/filepath"
+	"sort"
 	"strings"
+	"time"
 )
 
 const maxSingleFileSize = 10 * 1024 * 1024
 const maxZipArchiveSize = 50 * 1024 * 1024
 const maxGzipDecompressedSize = 50 * 1024 * 1024
 
+var supportedArchiveExt = map[string]struct{}{
+	".gz":  {},
+	".tgz": {},
+	".tar": {},
+	".zip": {},
+	".txt": {},
+	".log": {},
+}
+
 // ExtractedFile represents a file extracted from archive
 type ExtractedFile struct {
 	Path             string
 	Content          []byte
+	ModTime          time.Time
 	Truncated        bool
 	TruncatedMessage string
 }
 
 // ExtractArchive extracts tar.gz or zip archive and returns file contents
 func ExtractArchive(archivePath string) ([]ExtractedFile, error) {
+	if !IsSupportedArchiveFilename(archivePath) {
+		return nil, fmt.Errorf("unsupported archive format: %s", strings.ToLower(filepath.Ext(archivePath)))
+	}
 	ext := strings.ToLower(filepath.Ext(archivePath))
 
 	switch ext {
@@ -44,6 +59,9 @@ func ExtractArchive(archivePath string) ([]ExtractedFile, error) {
 
 // ExtractArchiveFromReader extracts archive from reader
 func ExtractArchiveFromReader(r io.Reader, filename string) ([]ExtractedFile, error) {
+	if !IsSupportedArchiveFilename(filename) {
+		return nil, fmt.Errorf("unsupported archive format: %s", strings.ToLower(filepath.Ext(filename)))
+	}
 	ext := strings.ToLower(filepath.Ext(filename))
 
 	switch ext {
@@ -60,6 +78,27 @@ func ExtractArchiveFromReader(r io.Reader, filename string) ([]ExtractedFile, er
 	}
 }
 
+// IsSupportedArchiveFilename reports whether filename extension is supported by archive extractor.
+func IsSupportedArchiveFilename(filename string) bool {
+	ext := strings.ToLower(strings.TrimSpace(filepath.Ext(filename)))
+	if ext == "" {
+		return false
+	}
+	_, ok := supportedArchiveExt[ext]
+	return ok
+}
+
+// SupportedArchiveExtensions returns sorted list of archive/file extensions
+// accepted by archive extractor.
+func SupportedArchiveExtensions() []string {
+	out := make([]string, 0, len(supportedArchiveExt))
+	for ext := range supportedArchiveExt {
+		out = append(out, ext)
+	}
+	sort.Strings(out)
+	return out
+}
+
 func extractTarGz(archivePath string) ([]ExtractedFile, error) {
 	f, err := os.Open(archivePath)
 	if err != nil {
@@ -111,6 +150,7 @@ func extractTarFromReader(r io.Reader) ([]ExtractedFile, error) {
 		files = append(files, ExtractedFile{
 			Path:    header.Name,
 			Content: content,
+			ModTime: header.ModTime,
 		})
 	}
 
@@ -152,6 +192,7 @@ func extractTarGzFromReader(r io.Reader, filename string) ([]ExtractedFile, erro
 			file := ExtractedFile{
 				Path:    baseName,
 				Content: decompressed,
+				ModTime: gzr.ModTime,
 			}
 			if gzipTruncated {
 				file.Truncated = true
@@ -180,6 +221,7 @@ func extractTarGzFromReader(r io.Reader, filename string) ([]ExtractedFile, erro
 				files = append(files, ExtractedFile{
 					Path:    header.Name,
 					Content: content,
+					ModTime: header.ModTime,
 				})
 			}
 		}
@@ -230,6 +272,7 @@ func extractZip(archivePath string) ([]ExtractedFile, error) {
 		files = append(files, ExtractedFile{
 			Path:    f.Name,
 			Content: content,
+			ModTime: f.Modified,
 		})
 	}
 
@@ -281,6 +324,7 @@ func extractZipFromReader(r io.Reader) ([]ExtractedFile, error) {
 		files = append(files, ExtractedFile{
 			Path:    f.Name,
 			Content: content,
+			ModTime: f.Modified,
 		})
 	}
 
@@ -288,13 +332,24 @@ func extractZipFromReader(r io.Reader) ([]ExtractedFile, error) {
 }
 
 func extractSingleFile(path string) ([]ExtractedFile, error) {
+	info, err := os.Stat(path)
+	if err != nil {
+		return nil, fmt.Errorf("stat file: %w", err)
+	}
 	f, err := os.Open(path)
 	if err != nil {
 		return nil, fmt.Errorf("open file: %w", err)
 	}
 	defer f.Close()
 
-	return extractSingleFileFromReader(f, filepath.Base(path))
+	files, err := extractSingleFileFromReader(f, filepath.Base(path))
+	if err != nil {
+		return nil, err
+	}
+	if len(files) > 0 {
+		files[0].ModTime = info.ModTime()
+	}
+	return files, nil
 }
 
 func extractSingleFileFromReader(r io.Reader, filename string) ([]ExtractedFile, error) {