<i18n dev> RFR: 8333755: NumberFormat integer only parsing breaks when format has suffix [v3]

Naoto Sato naoto at openjdk.org
Thu Jun 13 18:52:13 UTC 2024


On Wed, 12 Jun 2024 21:58:25 GMT, Justin Lu <jlu at openjdk.org> wrote:

>> Please review this PR and [CSR](https://bugs.openjdk.org/browse/JDK-8333920) which corrects a bug where NumberFormat cannot successfully parse values in integer only mode, when the format expects a suffix.
>> 
>> For example,
>> 
>> // a format that expects a currency suffix
>> var fmt = NumberFormat.getCurrencyInstance(Locale.FRANCE);
>> fmt.setParseIntegerOnly(true);
>> failFmt.parse("5,00 €"); // throws ParseException when you would have expected 5 returned
>> 
>> 
>> When parsing in integer only mode, instead of breaking upon a decimal symbol encounter, we should store the index but continue to fully parse so that we can verify the entire string and increment our position to search for and match the suffix. Upon a successful suffix match, we can then set `ParsePosition.index` back to the stored decimal index.
>> 
>> It should be noted that CompactNumberFormat did previously have code that would traverse the rest of the String, to allow matching of the suffix. The difference is that NumberFormat returns the index upon decimal symbol encounter, while CompactNumberFormat was implemented to return the index reflected by the entire string traversal. Such differences cannot be standardized due to behavioral compatibility concerns.
>> 
>> Thus while parsing in integer only, CNF sets index to the `Position.fullPos()`, while DF sets index to the `Position.intPos()`.
>
> Justin Lu has updated the pull request incrementally with one additional commit since the last revision:
> 
>   reflect review comments

LGTM

-------------

Marked as reviewed by naoto (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/19664#pullrequestreview-2116669878


More information about the i18n-dev mailing list