Magic-Mirror/gecko-dev
1
0
Fork 0
mirror of https://github.com/mozilla/gecko-dev.git synced 2024-12-11 16:32:59 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Backed out changeset 6fc2f69df1ee (bug 969928) for mochitest-1 bustage on a CLOSED TREE

Browse Source
This commit is contained in:
Wes Kocher 2014-02-18 14:31:57 -08:00
parent 2eebe74598
commit 281a4f7ec2
2 changed files with 8 additions and 46 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
content/html/content/test/forms/test_input_number_data.js
View File

@ -16,15 +16,6 @@ var tests = [
langTag: "de", inputWithGrouping: "123.456,78",
inputWithoutGrouping: "123456,78", value: 123456.78
},
// Extra german test to check that a locale that uses '.' as its grouping
// separator doesn't result in it being invalid (due to step mismatch) due
// to the de-localization code mishandling numbers that look like other
// numbers formatted for English speakers (i.e. treating this as 123.456
// instead of 123456):
{ desc: "German (test 2)",
langTag: "de", inputWithGrouping: "123.456",
inputWithoutGrouping: "123456", value: 123456
},
{ desc: "Hebrew",
langTag: "he", inputWithGrouping: "123,456.78",
inputWithoutGrouping: "123456.78", value: 123456.78

45
layout/forms/nsNumberControlFrame.cpp
View File

@ -627,52 +627,23 @@ nsNumberControlFrame::GetValueOfAnonTextControl(nsAString& aValue)
HTMLInputElement::FromContent(mTextField)->GetValue(aValue);
#ifdef ENABLE_INTL_API
// Here we need to de-localize any number typed in by the user. That is, we
// need to convert it from the number format of the user's language, region,
// etc. to the format that the HTML 5 spec defines to be a "valid
// floating-point number":
// Here we check if the text field's value is a localized serialization of a
// number. If it is we set aValue to the de-localize value, but only if the
// localized value isn't also a valid floating-point number according to the
// HTML 5 spec:
//
// http://www.whatwg.org/specs/web-apps/current-work/multipage/common-microsyntaxes.html#floating-point-numbers
//
// so that it can be parsed by functions like HTMLInputElement::
// StringToDecimal (the HTML-5-conforming parsing function) which don't know
// how to handle numbers that are formatted differently (for example, with
// non-ASCII digits, with grouping separator characters or with a decimal
// separator character other than '.').
//
// We need to be careful to avoid normalizing numbers that are already
// formatted for a locale that matches the format of HTML 5's "valid
// floating-point number" and have no grouping separator characters. (In
// other words we want to return the number as specified by the user, not the
// de-localized serialization, since the latter will normalize the value.)
// For example, if the user's locale is English and the user types in "2e2"
// then inputElement.value should be "2e2" and not "100". This is because
// content (and tests) expect us to avoid "normalizing" the number that the
// user types in if it's not necessary in order to make sure it conforms to
// HTML 5's "valid floating-point number" format.
//
// Note that we also need to be careful when trying to avoid normalization.
// For example, just because "1.234" _looks_ like a valid floating-point
// number according to the spec does not mean that it should be returned
// as-is. If the user's locale is German, then this represents the value
// 1234, not 1.234, so it still needs to be de-localized. Alternatively, if
// the user's locale is English and they type in "1,234" we _do_ need to
// normalize the number to "1234" because HTML 5's valid floating-point
// number format does not allow the ',' grouping separator. We can detect all
// the cases where we need to convert by seeing if the locale-specific
// parsing function understands the user input to mean the same thing as the
// HTML-5-conforming parsing function. If so, then we should return the value
// as-is to avoid normalization. Otherwise, we return the de-localized
// serialization.
// This is because content (and tests) expect us to avoid "normalizing" the
// number that the user types in if it's not necessary. (E.g. if the user
// types "2e2" then inputElement.value should be "2e2" and not "100".
ICUUtils::LanguageTagIterForContent langTagIter(mContent);
double value = ICUUtils::ParseNumber(aValue, langTagIter);
if (NS_finite(value) &&
value != HTMLInputElement::StringToDecimal(aValue).toDouble()) {
!HTMLInputElement::StringToDecimal(aValue).isFinite()) {
aValue.Truncate();
aValue.AppendFloat(value);
}
// else, we return whatever FromContent put into aValue (the number as typed
// in by the user)
#endif
}