weird that UTF8ToWide is in utf_string_conversions, but ASCIIToWide is in
string_util.h. This should help some dependencies since string_util changes
much more frequently than utf_string_conversions and fewer files will now need
string_utils.
Since this requires a lot of changes, this keeps a forward-declaration in
string_util so I can update the entire project incrementally. This change
updates base and net only.
I removed some includes of string_util from header files in net. In particular,
url_request_context which involved creating a new .cc file to implement a
function (already virtual so there's no speed penalty). It turns out a lot of
files were getting string_util from this include, so I had to update a bunch of
random files to now explicitly include string_util.h
TEST=it compiles
BUG=none
Review URL: http://codereview.chromium.org/3076013
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@54456 0039d316-1c4b-4281-b951-d872f2087c98
This does not split out the "adjust" versions of the UTF conversions into their own header/implementation; that's coming in the next patch.
BUG=4010
TEST=none
Review URL: http://codereview.chromium.org/380007
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@31533 0039d316-1c4b-4281-b951-d872f2087c98
* URLs with http auth info, which gets stripped
* URLs with IDN hosts
* URLs with escaped values that get unescaped
In cases like these, we'd inline autocomplete from the wrong locations, highlight the wrong portions of the URL as matches, and sometimes DCHECK() in debug mode.
The fix is to track how fixup affects the offsets into the URL we care about. Plumbing this required an enormous number of additions :(
There is also a fix here to the URL Fixer Upper, which was obviously modified at some point in the past to use the Parsed components, but without updating the comments or some of the functionality to match. Since this isn't supposed to "fix up" things that aren't simple typos, I removed some code to "fix" bogus ports, which was causing bizarre effects when typing HTTP auth URLs ("http://foo:bar" would be fixed to "http://foo" and then matched for inline autocompletion, which was clearly wrong). This is tested incidentally by one of the new History URL Provider tests (which is how I discovered it).
BUG=4010
TEST=Covered by unittests
Review URL: http://codereview.chromium.org/372017
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@31352 0039d316-1c4b-4281-b951-d872f2087c98
so we can do basic UTF8/16/32 conversions without linking all of ICU.
Change callers who used to call SysUTF8ToWide/SysWideToUTF8 in base to using
these new functions. I will remove the Sys versions of these functions in a
later patch.
TEST=none
BUG=none
Review URL: http://codereview.chromium.org/243102
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@28219 0039d316-1c4b-4281-b951-d872f2087c98
