Re: [Bug-wget] wget IDN support

[Giuseppe Scrivano]

Thanks to have reported these problems.  I'll take a look at them in the
next few days.

Cheers,
Giuseppe



Merinov Nikolay <address@hidden> writes:

> Current realisation of IDN support in wget not worked when system uses
> UTF-8 locale.
>
> Current realisation of function `url_parse' from src/url.c call
> `remote_to_utf8' from src/iri.c and set `iri->utf8_encode' to returned
> value.
>
> Function `remote_to_utf8' can return false in two cases:
> 1. They can not convert string to UTF-8
> 2. Source text is same as result text
>
> Second case appear when system use UTF-8 encoding.
>
> This can be fixed in several places:
> In src/url.c (url_parse) with adding comparing iri->orig_url with "UTF-8"
> In src/iri.c (remote_to_utf8) with removing "if (!strcmp (str, *new))"
>   test at the end of function.
> Or in src/iri.c (remote_to_utf8) with replacing return status when
>   result is same as input string.
>
> Last variant can be written like this:
>
> === modified file 'src/iri.c'
> --- src/iri.c 2011-01-01 12:19:37 +0000
> +++ src/iri.c 2011-06-23 16:34:10 +0000
> @@ -277,7 +277,7 @@
>    if (!strcmp (str, *new))
>      {
>        xfree ((char *) *new);
> -      return false;
> +      *new = NULL;
>      }
>  
>    return ret;
>
>
> Also it can be a good idea to fix src/host.c(lookup_host) with replacing
> usage `gethostbyname_with_timeout' by `getaddrinfo_with_timeout' and
> using AI_IDN flag, if wget compiled with glibc version 2.3.4 or
> newer. It can be helpful when wget compiled without iri support.