curl/wget does not resolve a URL same as a browser when redirected

Question

How do I get the correct redirected URL by curl/wget?

Target URL:

• https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm-3.18.git;a=commit;h=430f3805c82634a3cb969d83acc4fc4c0ee6af27
• You need to escape ; (semi-colon) to %3B when passing curl/wget

A browser, like Chrome/FireFox, resolves the above target URL to the following URL which includes commit and this is expected result.

• https://source.codeaurora.org/quic/la//kernel/msm-3.18/commit/?id=430f3805c82634a3cb969d83acc4fc4c0ee6af27

However, both curl and wget resolve the target URL to the following URL which includes tree and this is NOT expected behavior.

• https://source.codeaurora.org/quic/la//kernel/msm-3.18/tree/?id=430f3805c82634a3cb969d83acc4fc4c0ee6af27

The deference of redirected URL is including commit or tree, and it might be a gitweb issue instead of curl/wget. How do I get the correct redirected URL by curl/wget? To change a User-Agent is helpless. Any ideas? My curl command is as follows:

$ curl -v -L -o output.html \
https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm-3.18.git%3Ba=commit%3Bh=430f3805c82634a3cb969d83acc4fc4c0ee6af27

Answer 1

I found that the issue is that you need to url encode the / that is part of the project name. Encoding the ; with %3B broke things for me.

Also, because the URL contains ;, you need to quote it since this is a command separator.

In the query string, the p=kernel/msm-3.18 should be p=kernel%2fmsm-3.18

I was able to make the following two cURL commands work:

curl -v -L -o output.html \
'https://www.codeaurora.org/gitweb/quic/la/?p=kernel%2fmsm-3.18.git;a=commit;h=430f3805c82634a3cb969d83acc4fc4c0ee6af27'

Or simply,

curl -v -L -o output.html \
'https://www.codeaurora.org/gitweb/quic/la/?p=kernel/msm-3.18.git;a=commit;h=430f3805c82634a3cb969d83acc4fc4c0ee6af27'

This might be because the software reads the raw parameters and does not decode %3B and expects a semi-colon in the query string.

Hope that helps!