Remove _HttpUtils from dart:io.

Remove _HttpUtils from dart:io.

The Uri rewrite had all but one of these functions public available, so
they are used insted. The decodeHttpEntityString was only used by
HttpMultipartFormData, so moved to that class.

The function parseCookieDate was moved to HttpDate as a private
function. Renamed the file http_utils.dart to http_date.dart to match the content.

BUG=
R=lrn@google.com, whesse@google.com

Committed: https://code.google.com/p/dart/source/detail?r=24783

[Lasse Reichstein Nielsen, 2013-07-04 14:56:33]

lgtm

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart
File sdk/lib/core/uri.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode774
sdk/lib/core/uri.dart:774: {String decodeString(List<int> bytes): decodeUtf8}) {
Document the method.
Mention that "decodeString" is a temporary name/parameter. because we don't have
a type for general decoders.

"decodeString" sounds like it is a string, not that it decodes a string (and it
doesn't, it decodes bytes into a string). Just call it "decode".

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode817
sdk/lib/core/uri.dart:817: * If [decodeString] is specified, that function will
be used to decode the
again "decodeString" -> decode.
It will be used to decode the bytes into a String.

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode914
sdk/lib/core/uri.dart:914: * A JavaScript-like decodeURI function. It unescapes
the string [text] and
This methods needs a new documentation (a standalone one, that doesn't try to
piggyback on the JavaScript method).

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_body_impl.dart
File sdk/lib/io/http_body_impl.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_body_impl.dar...
sdk/lib/io/http_body_impl.dart:123: var map = Uri.splitQueryString(body.body,
contentType doesn't have a "CharSet"?

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_date.dart
File sdk/lib/io/http_date.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_date.dart#new...
sdk/lib/io/http_date.dart:207: static DateTime _parseCookieDate(String date) {
Yey, less code!

https://codereview.chromium.org/18576006/diff/1/tests/standalone/io/http_body...
File tests/standalone/io/http_body_test.dart (right):

https://codereview.chromium.org/18576006/diff/1/tests/standalone/io/http_body...
tests/standalone/io/http_body_test.dart:258: { 'a' : '\u{FFFD} &#548;' },
This is doing UTF-8 decoding, right?
The \u{FFFD} (just \uFFFD would do too) comes from failing to UTF-8 decode
0xF8,0x2B. It only removes the first byte, assuming it's a short UTF-8 sequence
followed by a normal "+".

I think that is the correct behavior.

The original result must come from a decoder that considers "+" part of the
malformed sequence, and eats that too. Does that decoder still exist? If so, are
there tests of it that would catch this.

How about strings like
 "%C0%A0" (well-formed but invalid UTF-8)
 "x%A0x" (Another malformed UTF-8)
 "x%C0x" (Another malformed UTF-8)

[Bill Hesse, 2013-07-05 08:52:13]

lgtm

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart
File sdk/lib/core/uri.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode914
sdk/lib/core/uri.dart:914: * A JavaScript-like decodeURI function. It unescapes
the string [text] and
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> This methods needs a new documentation (a standalone one, that doesn't try to
> piggyback on the JavaScript method).

I would make clear that it decodes URI encoding to a byte stream, and then
decodes the byte stream to a String using, by default, UTF-8.

[Anders Johnsen, 2013-07-05 09:51:28]

Thanks, landing.

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart
File sdk/lib/core/uri.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode774
sdk/lib/core/uri.dart:774: {String decodeString(List<int> bytes): decodeUtf8}) {
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> Document the method.
> Mention that "decodeString" is a temporary name/parameter. because we don't
have
> a type for general decoders.
> 
> "decodeString" sounds like it is a string, not that it decodes a string (and
it
> doesn't, it decodes bytes into a string). Just call it "decode".

Done.

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode817
sdk/lib/core/uri.dart:817: * If [decodeString] is specified, that function will
be used to decode the
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> again "decodeString" -> decode.
> It will be used to decode the bytes into a String.

Done.

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode914
sdk/lib/core/uri.dart:914: * A JavaScript-like decodeURI function. It unescapes
the string [text] and
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> This methods needs a new documentation (a standalone one, that doesn't try to
> piggyback on the JavaScript method).

Done.

https://codereview.chromium.org/18576006/diff/1/sdk/lib/core/uri.dart#newcode914
sdk/lib/core/uri.dart:914: * A JavaScript-like decodeURI function. It unescapes
the string [text] and
On 2013/07/05 08:52:13, Bill Hesse wrote:
> On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> > This methods needs a new documentation (a standalone one, that doesn't try
to
> > piggyback on the JavaScript method).
> 
> I would make clear that it decodes URI encoding to a byte stream, and then
> decodes the byte stream to a String using, by default, UTF-8.

Done.

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_body_impl.dart
File sdk/lib/io/http_body_impl.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_body_impl.dar...
sdk/lib/io/http_body_impl.dart:123: var map = Uri.splitQueryString(body.body,
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> contentType doesn't have a "CharSet"?

No, even though people would like to have it. That's why we use defaultEncoding
(should be the encoding that the form page was served with).

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_date.dart
File sdk/lib/io/http_date.dart (right):

https://codereview.chromium.org/18576006/diff/1/sdk/lib/io/http_date.dart#new...
sdk/lib/io/http_date.dart:207: static DateTime _parseCookieDate(String date) {
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> Yey, less code!

:)

https://codereview.chromium.org/18576006/diff/1/tests/standalone/io/http_body...
File tests/standalone/io/http_body_test.dart (right):

https://codereview.chromium.org/18576006/diff/1/tests/standalone/io/http_body...
tests/standalone/io/http_body_test.dart:258: { 'a' : '\u{FFFD} &#548;' },
On 2013/07/04 14:56:33, Lasse Reichstein Nielsen wrote:
> This is doing UTF-8 decoding, right?
> The \u{FFFD} (just \uFFFD would do too) comes from failing to UTF-8 decode
> 0xF8,0x2B. It only removes the first byte, assuming it's a short UTF-8
sequence
> followed by a normal "+".
> 
> I think that is the correct behavior.
> 
> The original result must come from a decoder that considers "+" part of the
> malformed sequence, and eats that too. Does that decoder still exist? If so,
are
> there tests of it that would catch this.
> 
> How about strings like
>  "%C0%A0" (well-formed but invalid UTF-8)
>  "x%A0x" (Another malformed UTF-8)
>  "x%C0x" (Another malformed UTF-8)
> 

Yes, I also think this is correct. Added those tests.

[Anders Johnsen, 2013-07-05 09:51:51]

Committed patchset #2 manually as r24783 (presubmit successful).