ZIP Viewer

Open one ZIP archive of up to 50 MB and read its directory — every entry with its path, size and compression — without unpacking anything. Tick the files you actually want and only those are extracted, in your browser.

Drop files here or choose files
Your ZIP stays on your device · the directory is read in the browser · list first, extract after · one archive up to 50 MB

🔒 Reading the directory and extracting the files both happened inside your browser. The archive was not uploaded to any server.

How to use

  1. Choose a ZIP archive: Drop a .zip file onto the box or click to choose one. One archive at a time, up to 50 MB. The directory is read as soon as you pick the file, so the table appears without you clicking anything.
  2. Read the list and tick what you need: Every entry shows its full path, whether it is a file or a folder, its unzipped and packed size, the saving and the compression method. Every extractable file starts ticked — untick the rest, or search by name and sort By name, By size or By type to find one row.
  3. Click Extract selected: Only the ticked entries are unpacked. Each extracted file appears in the result area with a numbered preview button, a Download button and Download all, and the files come back byte for byte.

How does this ZIP viewer show contents without extracting?

Archive size: up to 50 MB
Entries: up to 10,000, including folders
Extraction: selected unencrypted Stored or Deflate entries only

This ZIP viewer lists the entries in one archive up to 50 MB without decompressing their contents. A ZIP archive carries an index of its own contents, called the central directory, at the end of the file. Each entry is recorded there with its name, its uncompressed size, its packed size and the compression method used — so listing an archive means reading that index, not unpacking anything. This page does exactly that: the archive is handed to a small JavaScript ZIP decoder (fflate 0.8.3), but every entry is refused as the decoder walks the list, so no entry's data is ever expanded. What you get back is a table of what is inside the archive, built without opening a single file in it.

One consequence is worth knowing: the sizes in the table are the archive's own declarations, not bytes that were counted. Listing skips decompression, but reading the archive and processing more directory entries still takes time and memory. Declared sizes are also why a rare archive can mislead you: hand-edited archives and some streaming packers declare sizes that do not match the data they actually hold. When you extract a file, the size reported for it is the real number of bytes that came out.

Encrypted entries are marked when their flags can be read. They cannot be selected for extraction.

Nothing is uploaded. The archive is read from your disk straight into the tab's memory, both the listing and the extraction run on your own processor, and the preview and download links point at copies held in that same memory. (The code for the two steps is itself fetched the first time you use it, so this is not a page that makes no network requests at all — your archive is simply not part of any of them.) Nothing is stored either: there is no cookie and no local storage, and a refresh or a closed tab clears the archive, the listing and anything you extracted.

Reading the table: path, type, sizes, saving, method

Each row is one entry in the archive, shown under its full path — docs/notes.md rather than notes.md — so the original folder structure is visible at a glance. Folder entries get their own rows and sort to the top, marked Folder in the type column with dashes instead of sizes, because a folder has nothing to extract. Above the table, the summary line adds up the unzipped total, the packed total and the overall saving, and the file row counts files and folders separately.

The Saving column is worked out from the two declared sizes, and it can be negative. Every entry in an archive carries its own headers, and data that was already compressed — a JPEG, an MP4, another ZIP — has little left to squeeze, so small or already-packed files often come out larger than they went in. The 652-byte sample archive used to smoke-test this page is a clean example: its five entries declare 119 bytes unzipped against 126 bytes packed, so its overall saving reads as a negative percentage. Folders and empty files show a dash, since there is no ratio to report.

Two controls help with archives that have more rows than you want to scroll. The search box filters by file name as you type, and the header count next to Name follows what is currently listed. Sort offers By name, By size (largest first) and By type (by extension, then by name), with folders kept at the top either way. Neither affects what you get: extraction follows the archive's own order, so the numbered results line up with the order the files were packed in, not with however you happen to have sorted the table.

The Method column names what the entry was packed with: Stored for method 0 (packed without compression), deflate for method 8, and Method followed by the raw number for anything else. Listing is indifferent to the method — the number comes from the directory — but extraction is not: only Stored and deflate entries can actually be unpacked here, and an archive that uses BZip2, LZMA or PPMd will list and then report the same message as a damaged file when you try to extract from it.

Extracting only the files you tick

Extraction is selective, and that is the point of listing first. Every extractable file starts ticked so the tool works without touching the table, but the usual move is to untick the rest and leave the one or two files you came for. Click Extract selected and only those entries are decompressed; everything else is refused on the way past, exactly as during the listing. Because only the ticked entries are expanded, pulling one document out of a large archive costs you that one document — but tick everything and you are doing the same work as unzipping the whole archive, so the saving only exists when you really do want a subset.

What comes back is the files themselves, byte for byte. Decompression restores the original bytes and nothing re-encodes or re-packs them afterwards — the ticked entries are handed straight over rather than being bundled into a new archive first — so a photo, a spreadsheet or an installer matches what was packed, bit for bit. Each file arrives in the result area with a numbered preview button. Images and PDFs can be previewed, and supported MP4 videos can be played. Other formats remain available to download. Download saves one file; Download all saves them one after another.

Folders are not rebuilt when you save, because a browser writes single files to your downloads. Folder separators become underscores, so docs/notes.md downloads as docs_notes.md. Different paths can produce the same download name; check names when saving multiple files. On a phone, saving is handed to the system share sheet where the browser supports it, which is how you get the file into Files or pass it on to another app; if you dismiss the share sheet nothing is saved, so tap the button again. Where sharing is not available, the file downloads the ordinary way.

The table does not lie about what you will get. Selection and extraction both work by name, so an archive that records two entries under one identical full path — or an entry named __proto__ at its root — is refused before any table is drawn, with the offending path quoted in the message. Those are the two cases where ticking one row could hand you a different file, and refusing them is what makes the rest of the promise hold: every row is a real entry, and the entries you tick are the files you get.

Need to pass a few of those files on as one attachment? Pack them into a fresh archive with Create ZIP.

Limits, and what the list will not tell you

One archive at a time, up to 50 MB, and only .zip. Larger files are stopped at the drop box with a size notice and other file types are filtered out before anything is read; to swap archives, clear the current one with the ✕ on the file row. RAR, 7Z, TAR and GZ are different formats and are not handled here whatever the file name says, and split or multi-volume archives (.z01, .zip.001, .part1.rar) are out of reach because a browser only receives the one file you picked — handing it one volume on its own is reported as damage. A half-finished download, something that was never a ZIP, and an archive whose central directory cannot be read all land on that same message — in the last case it is reported as damaged rather than listed as an empty archive, which is the more useful answer.

Four guardrails are checked while the directory is read, which means an oversized archive is stopped before anything is unpacked rather than after. An archive is refused if it declares more than 10,000 entries (folder entries counted, so this is an entry limit rather than a file limit), any single entry over 256 MB unzipped, more than 512 MB unzipped in total, or an entry that declares at least 10 MB and expands more than 200 times — the signature of a ZIP bomb. Entry names that point outside the extraction folder are refused too: paths starting with a slash or a drive letter, UNC paths, or any name with a path segment equal to .. (the segment has to be exactly that, so an ordinary name such as report..txt is fine).

Encrypted entries can be read about but not read. They appear in the table with their name, sizes and an Encrypted badge, and the archive shows a notice at the top; their checkboxes are disabled and they are left out of the select-all, so normal use never reaches the password error. The reason is in the extractor rather than in your browser: fflate has no decryption step at all, and for the common ZipCrypto case it fails quietly instead of loudly. A test archive makes the point: a file holding 15 bytes of plain text came back out of the decoder as 27 bytes of nonsense — the 12-byte ZipCrypto header with the still-encrypted content behind it — and no error was raised at all. Catching the flag and stopping is the only honest outcome available. To get at those files, open the archive on your computer with 7-Zip, WinRAR or Archive Utility and type the password there.

Finally, the list is names and numbers, nothing more. Modification dates, created dates and permission bits are not shown and do not survive extraction, so files you save carry the time you saved them; the archive comment and any per-entry comments are skipped rather than displayed; and the tool only reads — it cannot rename an entry, delete one, add one or repack the archive, and your original ZIP is not modified by a single byte. File names recorded in UTF-8, which is what current archivers write, display correctly including accented and Vietnamese or Spanish names; names stored by older software in a local code page such as GBK or Windows-1252 come out garbled, and that affects the names only — the contents you extract are still exact. Peak memory runs at roughly the size of the archive plus whatever you ticked, and extraction runs on the main thread, so the page stops responding while it works.

Want the whole archive rather than a few files from it? Unzip Files unpacks everything in one pass.

Checking an archive you were not expecting

Reading the directory first is also how you look at an archive you do not trust. The table shows every entry's full name and extension before anything is unpacked, which is how people spot the things that are easy to miss in a file manager: an .exe, .js, .vbs, .bat or .lnk inside what was described as a folder of photos, a double extension such as invoice.pdf.exe, or another .zip nested inside this one. The declared sizes are visible in the same row, so an archive of a few hundred kilobytes that claims to unpack into gigabytes shows what it is, and the compression-ratio guardrail stops that kind of archive outright.

Be clear about what this is not, though. A ZIP viewer reads the index; it does not scan anything, and a harmless-looking name is not evidence about a file's contents. Extracting a file here does not run it either — the bytes are saved or previewed, not executed — but once a file is on your device, the usual care applies: run your own security software over anything executable, and when the sender or the delivery channel does not match something you asked for, the safest reading of the list is to not extract at all.

Frequently asked questions

Can I see what is inside a ZIP without extracting it?
Yes — that is what this tool is for. A ZIP records an index of its contents (the central directory) at the end of the file, so the list of entries, their paths, their sizes and their compression methods can be read without unpacking any of them. Extraction is a separate step you trigger yourself, and it only covers the entries you tick.
Is my ZIP file uploaded to a server?
No. The archive is read from your disk into the browser tab, both the listing and the extraction run on your device, and the download links point at copies in the tab's memory. No request carries your archive anywhere. Closing or refreshing the tab clears it along with anything you extracted; files you saved yourself stay on your device.
Can I extract just one file from a ZIP?
Yes. Untick everything else and leave that one row ticked, then click Extract selected — only that entry is decompressed. This is the main reason to list an archive first: pulling one document out of a large ZIP costs you that one document, while ticking every row does the same work as unzipping the whole archive.
Can it open a password-protected ZIP?
It can list a password-protected ZIP when its directory is readable, but it cannot extract encrypted entries. Unencrypted entries in the same archive can still be selected and extracted. Encrypted entries appear with their name, size and an Encrypted badge, and their checkboxes are disabled. The ZIP decoder behind this page has no decryption step for either ZipCrypto or AES, so there is nothing to type a password into. Open the archive on your computer with 7-Zip, WinRAR or Archive Utility and enter the password there.
Why is the saving column negative for some files?
Because those entries came out of compression larger than they went in. Every entry carries its own headers, and data that was already compressed — a JPEG, an MP4, a nested ZIP — has almost nothing left to squeeze. The 652-byte sample archive used to test this page shows the same thing overall: 119 bytes declared unzipped against 126 bytes packed.
Are the sizes in the list exact?
They are the sizes the archive declares about itself, read from its directory rather than measured, which is what makes listing possible without unpacking. For ordinary archives they match. Hand-edited archives and some streaming packers can declare sizes that differ from the data they hold; the size shown after you extract a file is the real byte count.
Why was my archive refused instead of listed?
One of the checks that run while the directory is read stopped it. The size guardrails are 10,000 entries (folders included), 256 MB for a single unzipped entry, 512 MB unzipped in total, and an entry declaring at least 10 MB that expands more than 200 times. Separately, entry names that point outside the extraction folder are refused, and so is an archive that stores two entries under one identical full path or an entry named __proto__ at its root — those would make the table point at the wrong file. The message names the path that caused it.
Can I view RAR or 7Z files here?
No. This page reads ZIP archives only; RAR, 7Z, TAR and GZ are different container formats and need their own tools. Renaming a .rar to .zip does not help, because the format is detected from the file's contents, not its name — it is reported as damaged or invalid.
Does it work on iPhone and Android?
Yes, in any modern mobile browser. Pick the archive from Files on an iPhone or Files by Google on Android and the list appears. When you save an extracted file, the browser is asked to hand it to the system share sheet where that is supported, so you can store it or pass it on; where it is not, the file downloads as usual. Keep archives modest on a phone: peak memory is roughly the archive plus whatever you ticked.