On a remote machine I am trying to use wget to download extensions for VS Code. However, the downloaded files are always corrupt and cannot be used. It turns out that using wget or curl will retrieve files which are a bit smaller than using a browser, so it appears reasonable that they cannot be unzipped. The URL I have used for downloading is https://marketplace.visualstudio.com/_apis/public/gallery/publishers/stkb/vsextensions/rewrap/1.9.1/vspackage
Downloads/tmp >ls -la
total 168
drwxrwxrwx 1 enno enno 4096 Feb 6 08:45 .
drwxrwxrwx 1 enno enno 4096 Feb 6 08:45 ..
-rwxrwxrwx 1 enno enno 85317 Feb 6 08:45 curl.vsix
-rwxrwxrwx 1 enno enno 85317 Feb 6 08:45 wget.vsix
BWP-181WD4:Downloads/tmp >unzip curl.vsix
Archive: curl.vsix
End-of-central-directory signature not found. Either this file is not
a zipfile, or it constitutes one disk of a multi-part archive. In the
latter case the central directory and zipfile comment will be found on
the last disk(s) of this archive.
note: curl.vsix may be a plain executable, not an archive
unzip: cannot find zipfile directory in one of curl.vsix or
curl.vsix.zip, and cannot find curl.vsix.ZIP, period.
Downloads/tmp >unzip wget.vsix
Archive: wget.vsix
End-of-central-directory signature not found. Either this file is not
a zipfile, or it constitutes one disk of a multi-part archive. In the
latter case the central directory and zipfile comment will be found on
the last disk(s) of this archive.
note: wget.vsix may be a plain executable, not an archive
unzip: cannot find zipfile directory in one of wget.vsix or
wget.vsix.zip, and cannot find wget.vsix.ZIP, period.
However, using a browser-downloaded file works just fine:
BWP-181WD4:Downloads/tmp >unzip ./browser.vsix
Archive: ./browser.vsix
inflating: extension.vsixmanifest
inflating: [Content_Types].xml
inflating: extension/CHANGELOG.md
inflating: extension/images/logo.png
inflating: extension/images/logo.svg
inflating: extension/package.json
inflating: extension/README.md
...
How come? The file sizes are clearly different:
Downloads/tmp >ls -la
total 256
drwxrwxrwx 1 enno enno 4096 Feb 6 08:52 .
drwxrwxrwx 1 enno enno 4096 Feb 6 08:50 ..
-rwxrwxrwx 1 enno enno 90035 Feb 6 08:49 browser.vsix
-rwxrwxrwx 1 enno enno 85317 Feb 6 08:45 curl.vsix
-rwxrwxrwx 1 enno enno 85317 Feb 6 08:45 wget.vsix
Is this related to how wget and curl retrieve files or how Microsoft serves these files?
The files downloaded with wget and curl are gzip files.
Change the filename to vspackage.gz and use gunzip to extract the vsix file
If you look at the response header in the browser you see they are gzip compressed.
Find a way for curl/wget to use the content-disposition and content-encoding of the response header.