URL encodes values that are already encoded

[mtdowling]

uri_template will url encode values that are already encoded and are not inside of a template expansion.

php -a
php > echo uri_template('http://foo.com/baz?bar=bam_%21', array());
http://foo.com/baz?bar=bam_%%2

And strangely, the parser does not URL encode the value when it is unencoded:

php > echo uri_template('http://foo.com/baz?bar=bam_!', array());
http://foo.com/baz?bar=bam_!

I think the correct behavior would be to not touch things outside of URI template expansion blocks.

[ioseb]

@mtdowling

Thanks for feedback!

php -a
php > echo uri_template('http://foo.com/baz?bar=bam_%21', array());
http://foo.com/baz?bar=bam_%%2

Example above looks like a bug i'm sure it needs to be copied as "bam_%21" and not "%%2" it doesn't look like doubly encoded value it looks like incorrectly copied triplet, will investigate this.

This one:

php > echo uri_template('http://foo.com/baz?bar=bam_!', array());
http://foo.com/baz?bar=bam_!

Is correct i think. Spec says that:

2.1.  Literals
   The characters outside of expressions in a URI Template string are
   intended to be copied literally to the URI reference if the character
   is allowed in a URI (reserved / unreserved / pct-encoded) or, if not
   allowed, copied to the URI reference as the sequence of pct-encoded
   triplets corresponding to that character's encoding in UTF-8
   [RFC3629].

     literals      =  %x21 / %x23-24 / %x26 / %x28-3B / %x3D / %x3F-5B
                   /  %x5D / %x5F / %x61-7A / %x7E / ucschar / iprivate
                   /  pct-encoded
                        ; any Unicode character except: CTL, SP,
                        ;  DQUOTE, "'", "%" (aside from pct-encoded),
                        ;  "<", ">", "\", "^", "`", "{", "|", "}"

Maybe i'm misinterpreting something? but "!" is one of the literals which needs to be copied directly(additionally see Reserved characters section of URI spec http://tools.ietf.org/html/rfc3986#section-2.2)

[mtdowling]

Thanks. Yeah, The second example does the right thing and just an example was just me showing that you're not url encoding stuff, so there must be another issue in the parser.

[ioseb]

Okay, will try to resolve the issue as quickly as possible. Thanks again for the bug report.

[ioseb]

@mtdowling

Fixed the bug with URI encoded triplets, additionally fixed copy issue with utf8 characters. Release is published on pecl: http://pecl.php.net/package/uri_template

[mtdowling]

Nice!