LOAD does not handle correctly URLs containing encoded delimiters

[Siskin-Bot]

Submitted by: Ladislav

Encoded delimiter characters are not delimiters in URLs. LOAD "disrespects" this property "unencoding" such characters and transforming them to delimiters, in fact.

mold load "http://a.b.c/d?e=f%26&g=h"
; == "http://a.b.c/d?e=f&&g=h"
; "http://a.b.c/d?e=f%26&g=h" expected!

---

^{Imported from: CureCode [ Version: r3 master Type: Bug Platform: All Category: Datatype Reproduce: Always Fixed-in:none ]
Imported from: metaeducation#2011}

Comments:

Rebolbot commented on Apr 4, 2013:

Submitted by: BrianH

This is related to #1327, #1333 and #1644.

---

Rebolbot commented on Sep 19, 2013:

Submitted by: Ladislav

In the core-tests suite.

---

Rebolbot mentioned this issue on Jan 12, 2016:
Internal representation of URLs

---

Rebolbot added Type.bug, Status.important, Test.written on Jan 12, 2016

---

[Oldes]

Related to: #2012

[Oldes]

It should be noted, that there are 2 escaping:

1. Rebol url in source code escaping (required to escape Rebol delimiter chars which are allowed in url as ;[](){}")
2. RFC's url escaping

Unfortunately both of them are using % as an escape char.

So it may be strange, but it is correct. One must be careful when using url-encoded data in urls inside a Rebol source code.

>> #[url! "http://a.b.c/d?e=f%26&g=h"]
== http://a.b.c/d?e=f%2526&g=h

>> join http://a.b.c/d? "e=f%26&g=h"
== http://a.b.c/d?e=f%2526&g=h

>> length? probe #[url! "a:%26"]
a:%2526
== 5 ;<--- because in construction syntax the string is not decoded

>> length? probe a:%26
a:&
== 3 ;<--- because Rebol lexer decoded the & char

To have load respecting the escaping, we would first need to allow lexer's delimiter chars be part of the url syntax
(for example: http://example.com/?data=a;b)