Inspect exact file sizes, totals and extension breakdowns without reading file contents.
No file selected.
Up to 2000 files; metadata only, no content reading. Choose whether the next selection appends or replaces the queue. File input values are reset after capture; retained selections remain visible below.
Choose files to begin. Nothing is uploaded.
Choose a source or try the sample.
Up to 2000 files, metadata only: contents are not read. Exact byte totals use BigInt, with separate decimal SI and binary IEC displays. Type and last-modified values are browser-supplied metadata, not file-format certification or measured storage allocation. Reports include original names and individual sizes.
Operations use a local worker with a 60-second deadline. Bounded processing can still exceed a small device’s memory; errors leave no enabled partial result. ZIPs are data containers, not encryption. Keep original files; all actions generate downloads or reports.
fflate 0.8.3, MIT license. No CDN, server processing or input history. Older tools retain their existing library.
No files selected.
Selections and queue order affect naming and processing. Remove changes only this in-memory queue; source files on disk remain untouched. Filter does not change selection. Lists paginate 50 rows; exports include the full applicable result.
No generated file. Choose a source and settings.
Your files are never uploaded to any server. Archive compression, extraction, naming, metadata inspection and duplicate-content comparison run locally with self-hosted code. No account, external processing service or application input history is used.
Up to 2000 files, metadata only: contents are not read. Exact byte totals use BigInt, with separate decimal SI and binary IEC displays. Type and last-modified values are browser-supplied metadata, not file-format certification or measured storage allocation. Reports include original names and individual sizes.
No. Files are read from the user-selected browser File objects. Renaming/extraction/compression creates downloaded copies. Removing from the queue only releases the in-memory selection. Duplicate Finder produces reports and never deletes source files.
No. ZIP reader support is explicit: classic unencrypted STORE/DEFLATE with validated safe relative paths. ZIP64, multi-part, encrypted, symlink, self-extracting and other compression methods reject. RAR/7z/TAR support is not claimed. GZIP restores a single data stream; it does not unpack TAR contents.
No. Already-compressed formats such as JPG, PNG, MP4 or ZIP can become slightly larger. The actual generated size is shown. ZIP is a container and does not add password protection. Compression settings never change original content bytes.
ZIP structure/path checks and declared size budgets precede decoding. Selected extraction and generated ZIP files check actual expanded byte counts and CRC32. GZIP uses the native browser decoder plus CRC/length checks. Duplicate groups use SHA-256 candidates with final byte-for-byte confirmation. Integrity checks are not a malware scan or proof that a file is safe to execute.
