Preserving subfolder hierarchy when you create a ZIP file online
Bundling project deliverables, reports, or photo sets into one archive keeps attachments organized and prevents email clients from stripping folder structure. When you create a ZIP file online by dropping items onto the workspace or clicking Choose files, every file is queued locally with its exact byte size displayed in the list.
Preserving subfolder trees with Add a folder
Clicking Add a folder opens your browser’s directory picker powered by the MDN HTMLInputElement.webkitdirectory Web API. The browser reads each file’s relative path locally, preserving nested subdirectories such as docs/2026/ or assets/icons/ inside the archive.
If you combine files from multiple tools in one deliverable, you can turn Markdown notes into Word documents first and drop the resulting .docx drafts alongside spreadsheets and images. When every queued item shares a single top-level directory, the ZIP file name field automatically adopts that folder’s name.
Automatic archive naming and filename sanitization
Before building the archive, the tool cleans the value in ZIP file name so the downloaded file works across Windows, macOS, and Linux filesystems. Any trailing .zip extension is normalized, and reserved characters (\, /, :, *, ?, ", <, >, |) are replaced with hyphens.
If you leave ZIP file name blank, the archive defaults to files.zip. Once Create ZIP finishes, the result screen displays the original total size, the compressed output size, and the percentage saved.
Why do JPG and MP4 files barely shrink inside a ZIP archive?
Photos, videos, audio tracks, and modern office documents are already compressed internally, so running Deflate over them yields almost zero additional byte savings. Because high-entropy compressed streams lack repeating byte patterns, re-compressing them wastes CPU cycles without shrinking the archive.
Under the IETF RFC 1951 DEFLATE Compressed Data Format, compression relies on LZ77 sliding-window matching and Huffman coding to eliminate repeated sequences. Plain text, source code, HTML, XML, and raw tabular exports contain heavy repetition and shrink by roughly 60% to 85% under Deflate level 6 versus uncompressed Store mode (level 0).
When Compress text, CSV and documents (smaller ZIP) is checked, the packer inspects each file extension against 37 already-compressed formats and routes them to Store mode (level 0) automatically. This automatic Deflate vs Store ZIP routing lets you package outputs from the CSV to Excel workbook converter alongside raw CSVs in the same archive without slowing down the build.
| File category | Representative extensions | Compression level used | Expected size reduction |
|---|---|---|---|
| Plain text and markup | .txt, .csv, .json, .xml, .md |
Deflate level 6 | High (often 60% to 85%) |
| Source code and logs | .js, .ts, .css, .html, .log |
Deflate level 6 | High (often 55% to 80%) |
| Raster and vector photos | .jpg, .png, .webp, .avif, .heic |
Store level 0 (bypassed) | Minimal (0% to 3%) |
| Audio and video streams | .mp4, .mov, .mkv, .mp3, .flac |
Store level 0 (bypassed) | Minimal (0% to 1%) |
| Modern archives and Office | .zip, .7z, .docx, .xlsx, .pptx |
Store level 0 (bypassed) | Minimal (0% to 2%) |
How does path sanitization keep ZIP files safe across platforms?
Path sanitization strips drive letters, relative traversal tokens, and reserved characters so every archive entry extracts cleanly on Windows, macOS, and Linux. Before packing begins, each queued path is normalized and checked case-insensitively to prevent silent file overwrites or prototype collisions.
Case-insensitive deduplication and path normalization
The path sanitizer converts Windows backslashes to forward slashes, strips leading drive letters or ./ prefixes, and drops . and .. segments. Special property names such as __proto__ are rewritten to _proto_ so archive maps cannot collide with JavaScript object prototypes.
Duplicate entries are checked case-insensitively so archives unpack cleanly on both case-sensitive Linux filesystems and case-insensitive Windows or macOS volumes. For example, if you flatten XML feeds into clean CSV tables twice with the same filename, the second copy is renamed automatically rather than overwriting the first.
Off-thread Web Worker execution without server uploads
Packaging runs inside a dedicated Web Worker using fflate zipSync, keeping the browser interface responsive while bytes are transferred with zero-copy ArrayBuffer handoffs. Because no remote server enforces a 50 MB or 100 MB upload ceiling, desktop tabs can bundle hundreds of files up to available RAM under the IANA application/zip Media Type Registration.
Tip: Click Add more files or Add a folder multiple times before clicking Create ZIP to merge items from different local directories into a single unified archive structure.
Checking and unpacking finished ZIP archives
After downloading an archive, verifying its contents ensures every expected file and subfolder arrived intact before you send it to a client or upload it to a portal. Checking the archive locally also catches accidental inclusions such as hidden system files or temporary exports.
Verifying archive contents before sharing
On the result screen, you can hand the new archive directly to the companion extractor to inspect and extract files from a ZIP archive without leaving your browser. The extractor lists every entry path and uncompressed size so you can confirm the folder hierarchy matches your intent.
Handling encrypted or multi-part archive needs
This builder creates single-volume, unencrypted ZIP archives using standard Store (method 0) and Deflate (method 8) streams. If you require 256-bit AES archive encryption or split multi-part volumes for strict IT compliance gateways, package the unencrypted files here or use a desktop archiver such as 7-Zip or macOS Archive Utility for encrypted containers.