The function lws_parse_uri has a defect when parsing the URL wss://xcx-login.yscq-wy.yscq.com?ip=s325-xcx-zfb.yscq-wy.yscq.com&port=8061.

[adamscm]

The function lws_parse_uri has a defect when parsing the URL wss://xcx-login.yscq-wy.yscq.com?ip=s325-xcx-zfb.yscq-wy.yscq.com&port=8061.

[adamscm]

The function lws_parse_uri parses the URL wss://xcx-login.yscq-wy.yscq.com?ip=s325-xcx-zfb.yscq-wy.yscq.com&port=8061 and returns an ads field value of cx-login.yscq-wy.yscq.com?ip=s325-xcx-zfb.yscq-wy.yscq.com&port=8061, which does not match expectations (the expected host should be xcx-login.yscq-wy.yscq.com).

[lws-team]

Basically it expects a / between the address and path parts. To my surprise, the official BNF for the URL syntax doesn't require it, it's legal to terminate the address with the ? as in your example.

https://www.w3.org/Addressing/URL/url-spec.html

That's a bit of a problem for lws_parse_uri() (which is about 15y old) because it takes the approach to insert NULs and return pointers to sections using the original string... this is good in the sense that it can handle huge URLs without allocation, but if there is no / separating the DNS address part and the "path" (actually the urlargs in this case), using the ? as the delimiter and NUL site will truncate the ? off the "path".

What do you feel about this situation?

[adamscm]

I have modified the lws_parse_uri function in my project to extend compatibility and support parsing URLs like wss://xcx-login.yscq-wy.yscq.com?ip=s325-xcx-zfb.yscq-wy.yscq.com&port=8061, ensuring the correct extraction of the ads (authority) and path fields.

[adamscm]

Basically it expects a / between the address and path parts. To my surprise, the official BNF for the URL syntax doesn't require it, it's legal to terminate the address with the ? as in your example.

https://www.w3.org/Addressing/URL/url-spec.html

That's a bit of a problem for lws_parse_uri() (which is about 15y old) because it takes the approach to insert NULs and return pointers to sections using the original string... this is good in the sense that it can handle huge URLs without allocation, but if there is no / separating the DNS address part and the "path" (actually the urlargs in this case), using the ? as the delimiter and NUL site will truncate the ? off the "path".

What do you feel about this situation?

ok