summaryrefslogtreecommitdiff
path: root/resources/mediawiki.libs/CLDRPluralRuleParser.js
diff options
context:
space:
mode:
Diffstat (limited to 'resources/mediawiki.libs/CLDRPluralRuleParser.js')
-rw-r--r--resources/mediawiki.libs/CLDRPluralRuleParser.js356
1 files changed, 258 insertions, 98 deletions
diff --git a/resources/mediawiki.libs/CLDRPluralRuleParser.js b/resources/mediawiki.libs/CLDRPluralRuleParser.js
index 441bc91f..3def37c5 100644
--- a/resources/mediawiki.libs/CLDRPluralRuleParser.js
+++ b/resources/mediawiki.libs/CLDRPluralRuleParser.js
@@ -1,7 +1,7 @@
-/* This is cldrpluralparser 1.0, ported to MediaWiki ResourceLoader */
+/* This is CLDRPluralRuleParser v1.1, ported to MediaWiki ResourceLoader */
/**
-* cldrpluralparser.js
+* CLDRPluralRuleParser.js
* A parser engine for CLDR plural rules.
*
* Copyright 2012 GPLV3+, Santhosh Thottingal
@@ -13,59 +13,83 @@
* @author Amir Aharoni
*/
+( function ( mw ) {
/**
* Evaluates a plural rule in CLDR syntax for a number
- * @param rule
- * @param number
- * @return true|false|null
+ * @param {string} rule
+ * @param {integer} number
+ * @return {boolean} true if evaluation passed, false if evaluation failed.
*/
-( function( mw ) {
function pluralRuleParser(rule, number) {
/*
Syntax: see http://unicode.org/reports/tr35/#Language_Plural_Rules
-----------------------------------------------------------------
-
condition = and_condition ('or' and_condition)*
+ ('@integer' samples)?
+ ('@decimal' samples)?
and_condition = relation ('and' relation)*
- relation = is_relation | in_relation | within_relation | 'n' <EOL>
+ relation = is_relation | in_relation | within_relation
is_relation = expr 'is' ('not')? value
- in_relation = expr ('not')? 'in' range_list
+ in_relation = expr (('not')? 'in' | '=' | '!=') range_list
within_relation = expr ('not')? 'within' range_list
- expr = 'n' ('mod' value)?
+ expr = operand (('mod' | '%') value)?
+ operand = 'n' | 'i' | 'f' | 't' | 'v' | 'w'
range_list = (range | value) (',' range_list)*
value = digit+
digit = 0|1|2|3|4|5|6|7|8|9
range = value'..'value
-
+ samples = sampleRange (',' sampleRange)* (',' ('…'|'...'))?
+ sampleRange = decimalValue '~' decimalValue
+ decimalValue = value ('.' value)?
*/
+
+ // we don't evaluate the samples section of the rule. Ignore it.
+ rule = rule.split('@')[0].trim();
+
+ if (!rule.length) {
+ // empty rule or 'other' rule.
+ return true;
+ }
// Indicates current position in the rule as we parse through it.
// Shared among all parsing functions below.
- var pos = 0;
-
- var whitespace = makeRegexParser(/^\s+/);
- var digits = makeRegexParser(/^\d+/);
-
- var _n_ = makeStringParser('n');
- var _is_ = makeStringParser('is');
- var _mod_ = makeStringParser('mod');
- var _not_ = makeStringParser('not');
- var _in_ = makeStringParser('in');
- var _within_ = makeStringParser('within');
- var _range_ = makeStringParser('..');
- var _comma_ = makeStringParser(',');
- var _or_ = makeStringParser('or');
- var _and_ = makeStringParser('and');
+ var pos = 0,
+ operand,
+ expression,
+ relation,
+ result,
+ whitespace = makeRegexParser(/^\s+/),
+ value = makeRegexParser(/^\d+/),
+ _n_ = makeStringParser('n'),
+ _i_ = makeStringParser('i'),
+ _f_ = makeStringParser('f'),
+ _t_ = makeStringParser('t'),
+ _v_ = makeStringParser('v'),
+ _w_ = makeStringParser('w'),
+ _is_ = makeStringParser('is'),
+ _isnot_ = makeStringParser('is not'),
+ _isnot_sign_ = makeStringParser('!='),
+ _equal_ = makeStringParser('='),
+ _mod_ = makeStringParser('mod'),
+ _percent_ = makeStringParser('%'),
+ _not_ = makeStringParser('not'),
+ _in_ = makeStringParser('in'),
+ _within_ = makeStringParser('within'),
+ _range_ = makeStringParser('..'),
+ _comma_ = makeStringParser(','),
+ _or_ = makeStringParser('or'),
+ _and_ = makeStringParser('and');
function debug() {
- /* console.log.apply(console, arguments);*/
+ // console.log.apply(console, arguments);
}
debug('pluralRuleParser', rule, number);
// Try parsers until one works, if none work return null
+
function choice(parserSyntax) {
- return function () {
+ return function() {
for (var i = 0; i < parserSyntax.length; i++) {
var result = parserSyntax[i]();
if (result !== null) {
@@ -79,6 +103,7 @@ function pluralRuleParser(rule, number) {
// Try several parserSyntax-es in a row.
// All must succeed; otherwise, return null.
// This is the only eager one.
+
function sequence(parserSyntax) {
var originalPos = pos;
var result = [];
@@ -95,8 +120,9 @@ function pluralRuleParser(rule, number) {
// Run the same parser over and over until it fails.
// Must succeed a minimum of n times; otherwise, return null.
+
function nOrMore(n, p) {
- return function () {
+ return function() {
var originalPos = pos;
var result = [];
var parsed = p();
@@ -113,21 +139,21 @@ function pluralRuleParser(rule, number) {
}
// Helpers -- just make parserSyntax out of simpler JS builtin types
-
function makeStringParser(s) {
var len = s.length;
- return function () {
+ return function() {
var result = null;
if (rule.substr(pos, len) === s) {
result = s;
pos += len;
}
+
return result;
};
}
function makeRegexParser(regex) {
- return function () {
+ return function() {
var matches = rule.substr(pos).match(regex);
if (matches === null) {
return null;
@@ -137,62 +163,166 @@ function pluralRuleParser(rule, number) {
};
}
+ /*
+ * integer digits of n.
+ */
+ function i() {
+ var result = _i_();
+ if (result === null) {
+ debug(' -- failed i', parseInt(number, 10));
+ return result;
+ }
+ result = parseInt(number, 10);
+ debug(' -- passed i ', result);
+ return result;
+ }
+
+ /*
+ * absolute value of the source number (integer and decimals).
+ */
function n() {
var result = _n_();
if (result === null) {
- debug(" -- failed n");
+ debug(' -- failed n ', number);
return result;
}
- result = parseInt(number, 10);
- debug(" -- passed n ", result);
+ result = parseFloat(number, 10);
+ debug(' -- passed n ', result);
+ return result;
+ }
+
+ /*
+ * visible fractional digits in n, with trailing zeros.
+ */
+ function f() {
+ var result = _f_();
+ if (result === null) {
+ debug(' -- failed f ', number);
+ return result;
+ }
+ result = (number + '.').split('.')[1] || 0;
+ debug(' -- passed f ', result);
+ return result;
+ }
+
+ /*
+ * visible fractional digits in n, without trailing zeros.
+ */
+ function t() {
+ var result = _t_();
+ if (result === null) {
+ debug(' -- failed t ', number);
+ return result;
+ }
+ result = (number + '.').split('.')[1].replace(/0$/, '') || 0;
+ debug(' -- passed t ', result);
+ return result;
+ }
+
+ /*
+ * number of visible fraction digits in n, with trailing zeros.
+ */
+ function v() {
+ var result = _v_();
+ if (result === null) {
+ debug(' -- failed v ', number);
+ return result;
+ }
+ result = (number + '.').split('.')[1].length || 0;
+ debug(' -- passed v ', result);
+ return result;
+ }
+
+ /*
+ * number of visible fraction digits in n, without trailing zeros.
+ */
+ function w() {
+ var result = _w_();
+ if (result === null) {
+ debug(' -- failed w ', number);
+ return result;
+ }
+ result = (number + '.').split('.')[1].replace(/0$/, '').length || 0;
+ debug(' -- passed w ', result);
return result;
}
- var expression = choice([mod, n]);
+ // operand = 'n' | 'i' | 'f' | 't' | 'v' | 'w'
+ operand = choice([n, i, f, t, v, w]);
+
+ // expr = operand (('mod' | '%') value)?
+ expression = choice([mod, operand]);
function mod() {
- var result = sequence([n, whitespace, _mod_, whitespace, digits]);
+ var result = sequence([operand, whitespace, choice([_mod_, _percent_]), whitespace, value]);
if (result === null) {
- debug(" -- failed mod");
+ debug(' -- failed mod');
return null;
}
- debug(" -- passed mod");
+ debug(' -- passed ' + parseInt(result[0], 10) + ' ' + result[2] + ' ' + parseInt(result[4], 10));
return parseInt(result[0], 10) % parseInt(result[4], 10);
}
function not() {
var result = sequence([whitespace, _not_]);
if (result === null) {
- debug(" -- failed not");
+ debug(' -- failed not');
return null;
- } else {
- return result[1];
}
+
+ return result[1];
}
+ // is_relation = expr 'is' ('not')? value
function is() {
- var result = sequence([expression, whitespace, _is_, nOrMore(0, not), whitespace, digits]);
+ var result = sequence([expression, whitespace, choice([_is_]), whitespace, value]);
if (result !== null) {
- debug(" -- passed is");
- if (result[3][0] === 'not') {
- return result[0] !== parseInt(result[5], 10);
- } else {
- return result[0] === parseInt(result[5], 10);
+ debug(' -- passed is : ' + result[0] + ' == ' + parseInt(result[4], 10));
+ return result[0] === parseInt(result[4], 10);
+ }
+ debug(' -- failed is');
+ return null;
+ }
+
+ // is_relation = expr 'is' ('not')? value
+ function isnot() {
+ var result = sequence([expression, whitespace, choice([_isnot_, _isnot_sign_]), whitespace, value]);
+ if (result !== null) {
+ debug(' -- passed isnot: ' + result[0] + ' != ' + parseInt(result[4], 10));
+ return result[0] !== parseInt(result[4], 10);
+ }
+ debug(' -- failed isnot');
+ return null;
+ }
+
+ function not_in() {
+ var result = sequence([expression, whitespace, _isnot_sign_, whitespace, rangeList]);
+ if (result !== null) {
+ debug(' -- passed not_in: ' + result[0] + ' != ' + result[4]);
+ var range_list = result[4];
+ for (var i = 0; i < range_list.length; i++) {
+ if (parseInt(range_list[i], 10) === parseInt(result[0], 10)) {
+ return false;
+ }
}
+ return true;
}
- debug(" -- failed is");
+ debug(' -- failed not_in');
return null;
}
+ // range_list = (range | value) (',' range_list)*
function rangeList() {
- // range_list = (range | value) (',' range_list)*
- var result = sequence([choice([range, digits]), nOrMore(0, rangeTail)]);
+ var result = sequence([choice([range, value]), nOrMore(0, rangeTail)]);
var resultList = [];
if (result !== null) {
- resultList = resultList.concat(result[0], result[1][0]);
+ resultList = resultList.concat(result[0]);
+ if (result[1][0]) {
+ resultList = resultList.concat(result[1][0]);
+ }
return resultList;
}
- debug(" -- failed rangeList");
+ debug(' -- failed rangeList');
return null;
}
@@ -202,111 +332,141 @@ function pluralRuleParser(rule, number) {
if (result !== null) {
return result[1];
}
- debug(" -- failed rangeTail");
+ debug(' -- failed rangeTail');
return null;
}
+ // range = value'..'value
+
function range() {
var i;
- var result = sequence([digits, _range_, digits]);
+ var result = sequence([value, _range_, value]);
if (result !== null) {
- debug(" -- passed range");
+ debug(' -- passed range');
var array = [];
var left = parseInt(result[0], 10);
var right = parseInt(result[2], 10);
- for ( i = left; i <= right; i++) {
+ for (i = left; i <= right; i++) {
array.push(i);
}
return array;
}
- debug(" -- failed range");
+ debug(' -- failed range');
return null;
}
function _in() {
// in_relation = expr ('not')? 'in' range_list
- var result = sequence([expression, nOrMore(0, not), whitespace, _in_, whitespace, rangeList]);
+ var result = sequence([expression, nOrMore(0, not), whitespace, choice([_in_, _equal_]), whitespace, rangeList]);
if (result !== null) {
- debug(" -- passed _in");
+ debug(' -- passed _in:' + result);
var range_list = result[5];
for (var i = 0; i < range_list.length; i++) {
- if (parseInt(range_list[i], 10) === result[0]) {
+ if (parseInt(range_list[i], 10) === parseInt(result[0], 10)) {
return (result[1][0] !== 'not');
}
}
return (result[1][0] === 'not');
}
- debug(" -- failed _in ");
+ debug(' -- failed _in ');
return null;
}
+ /*
+ * The difference between in and within is that in only includes integers in the specified range,
+ * while within includes all values.
+ */
+
function within() {
- var result = sequence([expression, whitespace, _within_, whitespace, rangeList]);
+ // within_relation = expr ('not')? 'within' range_list
+ var result = sequence([expression, nOrMore(0, not), whitespace, _within_, whitespace, rangeList]);
if (result !== null) {
- debug(" -- passed within ");
- var range_list = result[4];
- return (parseInt( range_list[0],10 )<= result[0] && result[0] <= parseInt( range_list[1], 10));
+ debug(' -- passed within');
+ var range_list = result[5];
+ if ((result[0] >= parseInt(range_list[0], 10)) &&
+ (result[0] < parseInt(range_list[range_list.length - 1], 10))) {
+ return (result[1][0] !== 'not');
+ }
+ return (result[1][0] === 'not');
}
- debug(" -- failed within ");
+ debug(' -- failed within ');
return null;
}
+ // relation = is_relation | in_relation | within_relation
+ relation = choice([is, not_in, isnot, _in, within]);
- var relation = choice([is, _in, within]);
-
+ // and_condition = relation ('and' relation)*
function and() {
- var result = sequence([relation, whitespace, _and_, whitespace, condition]);
+ var result = sequence([relation, nOrMore(0, andTail)]);
if (result) {
- debug(" -- passed and");
- return result[0] && result[4];
+ if (!result[0]) {
+ return false;
+ }
+ for (var i = 0; i < result[1].length; i++) {
+ if (!result[1][i]) {
+ return false;
+ }
+ }
+ return true;
}
- debug(" -- failed and");
+ debug(' -- failed and');
return null;
}
- function or() {
- var result = sequence([relation, whitespace, _or_, whitespace, condition]);
- if (result) {
- debug(" -- passed or");
- return result[0] || result[4];
+ // ('and' relation)*
+ function andTail() {
+ var result = sequence([whitespace, _and_, whitespace, relation]);
+ if (result !== null) {
+ debug(' -- passed andTail' + result);
+ return result[3];
}
- debug(" -- failed or");
+ debug(' -- failed andTail');
return null;
- }
- var condition = choice([and, or, relation]);
+ }
+ // ('or' and_condition)*
+ function orTail() {
+ var result = sequence([whitespace, _or_, whitespace, and]);
+ if (result !== null) {
+ debug(' -- passed orTail: ' + result[3]);
+ return result[3];
+ }
+ debug(' -- failed orTail');
+ return null;
- function isInt(n) {
- return parseFloat(n) % 1 === 0;
}
+ // condition = and_condition ('or' and_condition)*
+ function condition() {
+ var result = sequence([and, nOrMore(0, orTail)]);
+ if (result) {
+ for (var i = 0; i < result[1].length; i++) {
+ if (result[1][i]) {
+ return true;
+ }
+ }
+ return result[0];
- function start() {
- if (!isInt(number)) {
- return false;
}
- var result = condition();
- return result;
+ return false;
}
-
- var result = start();
-
+ result = condition();
/*
* For success, the pos must have gotten to the end of the rule
* and returned a non-null.
* n.b. This is part of language infrastructure, so we do not throw an internationalizable message.
*/
- if (result === null || pos !== rule.length) {
- // throw new Error("Parse error at position " + pos.toString() + " in input: " + rule + " result is " + result);
+ if (result === null) {
+ throw new Error('Parse error at position ' + pos.toString() + ' for rule: ' + rule);
}
- return result;
-}
+ if (pos !== rule.length) {
+ debug('Warning: Rule not parsed completely. Parser stopped at ' + rule.substr(0, pos) + ' for rule: ' + rule);
+ }
-/* For module loaders, e.g. NodeJS, NPM */
-if (typeof module !== 'undefined' && module.exports) {
- module.exports = pluralRuleParser;
+ return result;
}
/* pluralRuleParser ends here */