Updating number format data generation and NodeJS number_fmt #468
Conversation
|
Also clean up regex expressions in collation data generator. |
|
The C++ number formatter is now working, classifying some problems as "unsupported". |
|
ICU4J now has no errors or test failures. |
executors/cpp/number_fmt.cpp
Outdated
| std::regex check1("(0+0.#+E)"); | ||
| std::regex check2("(^.0#*E)"); |
| Pattern check1 = Pattern.compile("0+0.#+E"); | ||
| Matcher matcher1 = check1.matcher(input.pattern); | ||
| Pattern check2 = Pattern.compile("^.0#*E"); |
|
|
||
| // Check unsupport options in skeleton | ||
| if (input.skeleton != null) { | ||
| Pattern check_unnessary = Pattern.compile("rounding-mode-(unnecessary|half-odd)"); |
There was a problem hiding this comment.
Check if rounding-mode-unnecessary is unsupported or not. It appears to exist in Java: NumberFormatterSettings
However, in that enum, there is no HALF_ODD, only HALF_EVEN
| "error_detail", | ||
| json_object_new_string(pattern.c_str())); | ||
|
|
||
| // Nothing more to do here. |
There was a problem hiding this comment.
See if the options UNNECESSARY and HALF_ODD should be used, as they currently are here in C++, or not.
| const unsupported_rounding_modes = [ | ||
| "halfOdd", |
There was a problem hiding this comment.
After checking whether HALF_ODD and UNNECESSARY are supported options or not in ICU4C, double check whether that accords with here in JS where they are currently listed as unsupported.
There was a problem hiding this comment.
HALF_ODD and UNNECESSARY are definitely defined here: https://unicode-org.github.io/icu-docs/apidoc/dev/icu4c/unumberoptions_8h.html#a2f46653ee48f5d8d5b8e67d4a1b8230a
The question about "unnecessary" and "inexact" are if we can create and use a test that is meaningful?
executors/node/numberformat.js
Outdated
|
|
||
| testDecimalFormat: function(json, doLogInput) { | ||
| const node_version = process.version; | ||
| const label = json['label']; | ||
| const skeleton = json['skeleton']; | ||
|
|
||
| const pattern = json['pattern']; | ||
| const rounding = json['roundingMode']; | ||
| // Leave input as string unless it is adjusted. |
There was a problem hiding this comment.
This comment line is confusing. Remove. It's written about input, which is a float, and this PR is adding input_as_string, which seems straightforward when looking at the code.
There was a problem hiding this comment.
Restructured this a bit to only use input as floating value in the special cases of percent and scale options.
executors/node/numberformat.js
Outdated
| const scale_value = parseFloat(match_scale[1]); | ||
| input = input * scale_value; | ||
| } | ||
| // |
There was a problem hiding this comment.
What happened to the common?
executors/node/numberformat.js
Outdated
| } | ||
| } | ||
| } | ||
| // Check for skelection terms that are not supported |
There was a problem hiding this comment.
| // Check for skelection terms that are not supported | |
| // Check for skeleton terms that are not supported |
| @@ -159,7 +159,8 @@ | |||
| "description": "rounding mode to be used", | |||
| "enum": ["ceil", "floor", "expand", "trunc", | |||
| "halfCeil", "halfFloor", "halfExpand", | |||
| "halfTrunc", "halfEven", | |||
| "halfTrunc", "halfEven", "halfOdd", | |||
| "floor", "ceil", | |||
There was a problem hiding this comment.
Why are floor and ceil being duplicated? See line 160
Fixing problems with number format data generation and executors.
Starting with NodeJS.