@echogarden/text-segmentation 0.2.1 → 0.3.0

package/README.md

@@ -30,7 +30,7 @@ npm install @echogarden/icu-segmentation-wasm
 import { splitToWords } from '@echogarden/text-segmentation'
 
 const wordSequence: WordSequence =
-  await splitToWords('Hello, world! How are you doing today?', { language: 'en' })
+	await splitToWords('Hello, world! How are you doing today?', { language: 'en' })
 
 console.log(wordSequence.words)
 ```
@@ -64,22 +64,22 @@ prints a list of objects, including metadata on each word:
 
 ```ts
 [
-  { text: 'Hello', startOffset: 0, endOffset: 5, isPunctuation: false },
-  { text: ',', startOffset: 5, endOffset: 6, isPunctuation: true },
-  { text: ' ', startOffset: 6, endOffset: 7, isPunctuation: true },
-  { text: 'world', startOffset: 7, endOffset: 12, isPunctuation: false },
-  { text: '!', startOffset: 12, endOffset: 13, isPunctuation: true },
-  { text: ' ', startOffset: 13, endOffset: 14, isPunctuation: true },
-  { text: 'How', startOffset: 14, endOffset: 17, isPunctuation: false },
-  { text: ' ', startOffset: 17, endOffset: 18, isPunctuation: true },
-  { text: 'are', startOffset: 18, endOffset: 21, isPunctuation: false },
-  { text: ' ', startOffset: 21, endOffset: 22, isPunctuation: true },
-  { text: 'you', startOffset: 22, endOffset: 25, isPunctuation: false },
-  { text: ' ', startOffset: 25, endOffset: 26, isPunctuation: true },
-  { text: 'doing', startOffset: 26, endOffset: 31, isPunctuation: false },
-  { text: ' ', startOffset: 31, endOffset: 32, isPunctuation: true },
-  { text: 'today', startOffset: 32, endOffset: 37, isPunctuation: false },
-  { text: '?', startOffset: 37, endOffset: 38, isPunctuation: true }
+	{ text: 'Hello', startOffset: 0, endOffset: 5, isPunctuation: false },
+	{ text: ',', startOffset: 5, endOffset: 6, isPunctuation: true },
+	{ text: ' ', startOffset: 6, endOffset: 7, isPunctuation: true },
+	{ text: 'world', startOffset: 7, endOffset: 12, isPunctuation: false },
+	{ text: '!', startOffset: 12, endOffset: 13, isPunctuation: true },
+	{ text: ' ', startOffset: 13, endOffset: 14, isPunctuation: true },
+	{ text: 'How', startOffset: 14, endOffset: 17, isPunctuation: false },
+	{ text: ' ', startOffset: 17, endOffset: 18, isPunctuation: true },
+	{ text: 'are', startOffset: 18, endOffset: 21, isPunctuation: false },
+	{ text: ' ', startOffset: 21, endOffset: 22, isPunctuation: true },
+	{ text: 'you', startOffset: 22, endOffset: 25, isPunctuation: false },
+	{ text: ' ', startOffset: 25, endOffset: 26, isPunctuation: true },
+	{ text: 'doing', startOffset: 26, endOffset: 31, isPunctuation: false },
+	{ text: ' ', startOffset: 31, endOffset: 32, isPunctuation: true },
+	{ text: 'today', startOffset: 32, endOffset: 37, isPunctuation: false },
+	{ text: '?', startOffset: 37, endOffset: 38, isPunctuation: true }
 ]
 ```
 
@@ -87,13 +87,13 @@ prints a list of objects, including metadata on each word:
 
 ```ts
 const result: SegmentationResult =
-  await segmentText(`Hello, world! How are you doing today?`)
+	await segmentText(`Hello, world! How are you doing today?`)
 ```
 
 `result` is a nested object containing a breakdown of sentences, phrases and words in the given text. It is described by these TypeScript types:
 ```ts
 interface SegmentationResult {
-	wordSequence: WordSequence
+	words: WordSequence
 	sentences: Sentence[]
 }
 
@@ -101,7 +101,7 @@ interface Sentence {
 	text: string
 	charRange: Range
 	wordRange: Range
-	wordSequence: WordSequence
+	words: WordSequence
 
 	phrases: Phrase[]
 }
@@ -110,7 +110,7 @@ interface Phrase {
 	text: string
 	charRange: Range
 	wordRange: Range
-	wordSequence: WordSequence
+	words: WordSequence
 }
 
 interface Range {
@@ -140,12 +140,17 @@ There are several types of accepted patterns, evaluated in this order:
 * Language-specific suppressions. For example, in English it would include abbreviations like `Mr.`, `Mrs.`, `e.g.`, `i.e`, `etc.`, or contractions like and `'cause`, `'bout` in English, `'n` in Afrikaans
 * Noun suppressions, shared in all languages, like names of brands, misc. abbreviations and programming languages. Examples: `C#`, `F#`, `C++`, `Yahoo!`, `Toys"R"Us`, `Dunkin'`, `Ke$ha`, `Sky+`, `I/O`, `A/C`, `A/V`
 * Top-level domains, like `.com`, `.org`, `.net` (also doubling as the noun `.NET` as in the ".NET framework")
-* Number patterns, consisting of a sequences of digits separated by various separator characters, like decimal separators `3.14`, `3,14`, and thousands separators like `233,421` (`,`), `233.421` (`.`) or `233 421` (` `). optional `-` or `+` signs like `-34,534.123`, `+43 345,344`
-* Date, time, and phone number patterns, consisting of digits separated by characters like `/`, `:`, `-`, like `15:23:23`, `1953/11/06`, `64-534-756`
-* Percentage patterns, which are a subset of number patterns preceded or followed by `%`, like `53.243%` or `%34.12`
+* Number patterns, consisting of a sequences of digits separated by various separator characters, like decimal separators `3.14`, `3,14`, and thousands separators like `233,421` (`,`), `233.421` (`.`) or `233 421` (` `). optional `-` or `+` signs like `-34,534.123`, `+43 345,344`, or scientific notation like `1.554e-34`
+* Time patterns like `15:12` and `06:34:23`
+* Date patterns like `10/15/2021` (currently disabled)
+* Dimension patterns like `53x545x76`
+* Percentage patterns, which are number patterns preceded or followed by `%`, like `53.243%` or `%34.12`
 * Currency patterns like `$101.25`, `€50`, `20£`, `-53.23¥`, which, like percentage patterns, are number patterns preceded or followed by a currency symbol
 * Abbreviation patterns like `Y.M.C.A`: these patterns will be automatically matched (no special suppressions needed) if there is a sequence of single `.` alternating between single letters (like `x.y.z`), optionally, there may be a space between the characters, like `x. y. z.`
-* **And finally** (but most importantly): word character sequences consisting of letter characters (Unicode category `Letter`), mark characters (Unicode category `Mark`) or digit characters (Unicode category `Decimal_Number`), which may include inner apostrophes like `'` and `’` and inner separators like `-`, `_` `·`
+* Dot connected word sequences like `abc.def.g12`
+* Underscore connected word sequence `abc_def_g12`
+* Interpunct connected word sequence `abc·def·g12`
+* **And finally**: word character sequences consisting of letter characters (Unicode category `Letter`), mark characters (Unicode category `Mark`) or digit characters (Unicode category `Decimal_Number`), which may include inner apostrophes like `'` and `’` and inner separators like `-`, `_` `·`
 
 **Current limitations**:
 * No current general identification of preceding or trailing apostrophes, like in English possessive plurals "The brothers' friend" or "The diplomats' contracts". **Reason**: this requires a large lexicon or more sophisticated language understanding, since the apostrophe is generally ambiguous with a single quote.

package/dist/EastAsianCharacterPatterns.d.ts

@@ -1,7 +1 @@
-export declare const chineseCharacterRanges: import("regexp-composer").AnyOf;
-export declare const japaneseHiraganaCharacterRanges: import("regexp-composer").AnyOf;
-export declare const japaneseKatakanaCharacterRanges: import("regexp-composer").AnyOf;
-export declare const thaiLetterRanges: import("regexp-composer").AnyOf;
-export declare const khmerLetterRanges: import("regexp-composer").AnyOf;
-export declare const eastAsianCharRanges: import("regexp-composer").AnyOf;
 export declare const eastAsianCharRangesRegExp: RegExp;

package/dist/EastAsianCharacterPatterns.js

@@ -1,14 +1,14 @@
 import { anyOf, buildRegExp, codepointRange } from 'regexp-composer';
-export const chineseCharacterRanges = anyOf(codepointRange('4E00', '9FFF'), // Main
+const chineseCharacterRanges = anyOf(codepointRange('4E00', '9FFF'), // Main
 codepointRange('3400', '4DBF'), // CJK Unified Ideographs Extension A
 codepointRange('20000', '2A6DF') // CJK Unified Ideographs Extension B
 );
-export const japaneseHiraganaCharacterRanges = anyOf(codepointRange('3040', '309F'), // Hiragana
+const japaneseHiraganaCharacterRanges = anyOf(codepointRange('3040', '309F'), // Hiragana
 codepointRange('1AFF0', '1AFFF'), // Kana Extended-B
 codepointRange('1B000', '1B0FF'), // Kana Supplement
 codepointRange('1B100', '1B12F'), // Kana Extended-A
 codepointRange('1B130', '1B16F'));
-export const japaneseKatakanaCharacterRanges = anyOf(codepointRange('30A0', '30FF'), // Katakana
+const japaneseKatakanaCharacterRanges = anyOf(codepointRange('30A0', '30FF'), // Katakana
 codepointRange('31F0', '31FF'), // Katakana Phonetic Extensions
 codepointRange('3200', '32FF'), // Enclosed CJK Letters and Months
 codepointRange('FF00', 'FFEF'), // Halfwidth and Fullwidth Forms
@@ -16,9 +16,9 @@ codepointRange('1AFF0', '1AFFF'), // Kana Extended-B
 codepointRange('1B000', '1B0FF'), // Kana Supplement
 codepointRange('1B100', '1B12F'), // Kana Extended-A
 codepointRange('1B130', '1B16F'));
-export const thaiLetterRanges = anyOf(codepointRange('0E00', '0E7F'));
-export const khmerLetterRanges = anyOf(codepointRange('1780', '17FF'), // Letters
+const thaiLetterRanges = anyOf(codepointRange('0E00', '0E7F'));
+const khmerLetterRanges = anyOf(codepointRange('1780', '17FF'), // Letters
 codepointRange('19E0', '19FF'));
-export const eastAsianCharRanges = anyOf(chineseCharacterRanges, japaneseHiraganaCharacterRanges, japaneseKatakanaCharacterRanges, thaiLetterRanges, khmerLetterRanges);
+const eastAsianCharRanges = anyOf(chineseCharacterRanges, japaneseHiraganaCharacterRanges, japaneseKatakanaCharacterRanges, thaiLetterRanges, khmerLetterRanges);
 export const eastAsianCharRangesRegExp = buildRegExp(eastAsianCharRanges);
 //# sourceMappingURL=EastAsianCharacterPatterns.js.map

package/dist/EastAsianCharacterPatterns.js.map

@@ -1 +1 @@
-{"version":3,"file":"EastAsianCharacterPatterns.js","sourceRoot":"","sources":["../src/EastAsianCharacterPatterns.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAEpE,MAAM,CAAC,MAAM,sBAAsB,GAAG,KAAK,CAC1C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,OAAO;AACvC,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,qCAAqC;AACrE,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC,qCAAqC;CACtE,CAAA;AAED,MAAM,CAAC,MAAM,+BAA+B,GAAG,KAAK,CACnD,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,WAAW;AAC3C,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAChC,CAAA;AAED,MAAM,CAAC,MAAM,+BAA+B,GAAG,KAAK,CACnD,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,WAAW;AAC3C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,+BAA+B;AAC/D,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,kCAAkC;AAClE,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,gCAAgC;AAChE,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAChC,CAAA;AAED,MAAM,CAAC,MAAM,gBAAgB,GAAG,KAAK,CACpC,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,CAC9B,CAAA;AAED,MAAM,CAAC,MAAM,iBAAiB,GAAG,KAAK,CACrC,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,UAAU;AAC1C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,CAC9B,CAAA;AAED,MAAM,CAAC,MAAM,mBAAmB,GAAG,KAAK,CACvC,sBAAsB,EACtB,+BAA+B,EAC/B,+BAA+B,EAC/B,gBAAgB,EAChB,iBAAiB,CACjB,CAAA;AAED,MAAM,CAAC,MAAM,yBAAyB,GAAG,WAAW,CAAC,mBAAmB,CAAC,CAAA"}
+{"version":3,"file":"EastAsianCharacterPatterns.js","sourceRoot":"","sources":["../src/EastAsianCharacterPatterns.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAEpE,MAAM,sBAAsB,GAAG,KAAK,CACnC,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,OAAO;AACvC,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,qCAAqC;AACrE,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC,qCAAqC;CACtE,CAAA;AAED,MAAM,+BAA+B,GAAG,KAAK,CAC5C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,WAAW;AAC3C,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAChC,CAAA;AAED,MAAM,+BAA+B,GAAG,KAAK,CAC5C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,WAAW;AAC3C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,+BAA+B;AAC/D,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,kCAAkC;AAClE,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,gCAAgC;AAChE,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,kBAAkB;AACpD,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAChC,CAAA;AAED,MAAM,gBAAgB,GAAG,KAAK,CAC7B,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,CAC9B,CAAA;AAED,MAAM,iBAAiB,GAAG,KAAK,CAC9B,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,UAAU;AAC1C,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,CAC9B,CAAA;AAED,MAAM,mBAAmB,GAAG,KAAK,CAChC,sBAAsB,EACtB,+BAA+B,EAC/B,+BAA+B,EAC/B,gBAAgB,EAChB,iBAAiB,CACjB,CAAA;AAED,MAAM,CAAC,MAAM,yBAAyB,GAAG,WAAW,CAAC,mBAAmB,CAAC,CAAA"}

package/dist/Patterns.d.ts

@@ -1,34 +1,9 @@
 export declare function buildWordOrNumberPattern(suppressions: string[]): import("regexp-composer").AnyOf;
 export declare function buildSuppressionPattern(suppressions: string[]): import("regexp-composer").Pattern;
-export declare const punctuationPattern: import("regexp-composer").SpecialToken;
-export declare const digitPattern: import("regexp-composer").SpecialToken;
-export declare const arabicNumeralPattern: import("regexp-composer").SpecialToken;
-export declare const numericSeparatorPattern: import("regexp-composer").Pattern;
-export declare const dateTimeSeparatorPattern: import("regexp-composer").Pattern;
-export declare const dateTimePattern: import("regexp-composer").OneOrMore;
-export declare const spacedThousandsSeparatorPattern: import("regexp-composer").Pattern;
-export declare const numericSignPattern: import("regexp-composer").Pattern;
-export declare const numberPattern: import("regexp-composer").OneOrMore[];
-export declare const prefixPercentageOrCurrencyPattern: import("regexp-composer").Pattern;
-export declare const suffixPercentagePattern: import("regexp-composer").Pattern;
-export declare const percentagePattern: import("regexp-composer").AnyOf;
-export declare const letterPattern: import("regexp-composer").SpecialToken;
-export declare const markPattern: import("regexp-composer").SpecialToken;
-export declare const apostrophPattern: import("regexp-composer").AnyOf;
-export declare const letterOrMarkPattern: import("regexp-composer").AnyOf;
-export declare const dottedAbbreviationSequencePattern: import("regexp-composer").Pattern;
-export declare const wordCharacterPattern: import("regexp-composer").AnyOf;
-export declare const wordSeparatorPattern: import("regexp-composer").Pattern;
-export declare const wordInnerApostrophPattern: import("regexp-composer").Pattern;
-export declare const wordStartApostrophPattern: import("regexp-composer").Pattern;
-export declare const basicWordPattern: import("regexp-composer").OneOrMore;
-export declare const wordSegmentPattern: import("regexp-composer").AnyOf;
-export declare const phraseSeparators: string[];
+export declare const wordCharacterRegExp: RegExp;
+export declare const whitespacePatternRegExp: RegExp;
+export declare const letterPatternGlobalRegExp: RegExp;
 export declare const phraseSeparatorRegExp: RegExp;
-export declare const sentenceSeparators: string[];
+export declare const phraseSeparatorTrailingPunctuationRegExp: RegExp;
 export declare const sentenceSeparatorRegExp: RegExp;
 export declare const sentenceSeparatorTrailingPunctuationRegExp: RegExp;
-export declare const phraseSeparatorTrailingPunctuationRegExp: RegExp;
-export declare const oneOrMoreSpacesRegExp: RegExp;
-export declare const wordCharacterRegExp: RegExp;
-export declare const whitespacePatternRegExp: RegExp;

package/dist/Patterns.js

@@ -1,4 +1,7 @@
-import { anyOf, buildRegExp, charRange, inputEnd, inputStart, matches, oneOrMore, possibly, repeated, tab, unicodeProperty, whitespace } from 'regexp-composer';
+import { anyOf, buildRegExp, charRange, inputEnd, inputStart, matches, oneOrMore, possibly, repeated, tab, unicodeProperty, whitespace, zeroOrMore } from 'regexp-composer';
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Pattern builder methods
+////////////////////////////////////////////////////////////////////////////////////////////////
 export function buildWordOrNumberPattern(suppressions) {
     return anyOf(buildSuppressionPattern(suppressions), wordSegmentPattern);
 }
@@ -7,57 +10,96 @@ export function buildSuppressionPattern(suppressions) {
     return suppressionsPattern;
 }
 ////////////////////////////////////////////////////////////////////////////////////////////////
+// Single character patterns
+////////////////////////////////////////////////////////////////////////////////////////////////
+const letterPattern = unicodeProperty('Letter');
+const markPattern = unicodeProperty('Mark');
+const letterOrMarkPattern = anyOf(letterPattern, markPattern);
+const apostrophPattern = anyOf(`'`, `’`, `‘`);
+const punctuationPattern = unicodeProperty('Punctuation');
+const digitPattern = unicodeProperty('Decimal_Number');
+const arabicNumeralPattern = charRange('0', '9');
+const percentageCharacters = ['%'];
+const currencyCharacters = ['$', '¥', '€', '£', '₩', '₭', '₽', '₫', '฿', '¢', '₮', '؋', '₦', '₱', '₴', '₪'];
+const percentageOrCurrencyCharacterPattern = anyOf(...percentageCharacters, ...currencyCharacters);
+////////////////////////////////////////////////////////////////////////////////////////////////
 // Numeric patterns
 ////////////////////////////////////////////////////////////////////////////////////////////////
-export const punctuationPattern = unicodeProperty('Punctuation');
-export const digitPattern = unicodeProperty('Decimal_Number');
-export const arabicNumeralPattern = charRange('0', '9');
-export const numericSeparatorPattern = matches(anyOf('.', ',', '٬', '_'), {
+const numericSeparatorPattern = matches(anyOf('.', ',', '٬', '_'), {
     ifPrecededBy: arabicNumeralPattern,
     ifFollowedBy: arabicNumeralPattern
 });
-export const dateTimeSeparatorPattern = matches(anyOf('/', '-', ':'), {
-    ifPrecededBy: arabicNumeralPattern,
-    ifFollowedBy: arabicNumeralPattern
+const dimensionsPattern = matches([
+    oneOrMore(arabicNumeralPattern),
+    oneOrMore([
+        'x',
+        oneOrMore(arabicNumeralPattern),
+    ]),
+], {
+    ifPrecededBy: anyOf(whitespace, punctuationPattern, inputStart),
+    ifFollowedBy: anyOf(whitespace, punctuationPattern, inputEnd)
 });
-export const dateTimePattern = oneOrMore(anyOf(arabicNumeralPattern, dateTimeSeparatorPattern));
-export const spacedThousandsSeparatorPattern = matches(' ', {
+const spacedThousandsSeparatorPattern = matches(' ', {
     ifPrecededBy: arabicNumeralPattern,
-    ifFollowedBy: repeated(2, arabicNumeralPattern)
+    ifFollowedBy: [
+        repeated(3, arabicNumeralPattern),
+        anyOf(whitespace, punctuationPattern, inputEnd)
+    ]
 });
-export const numericSignPattern = matches(anyOf('-', '+'), {
-    ifPrecededBy: anyOf(whitespace, punctuationPattern),
+const numericSignPattern = matches(anyOf('-', '+'), {
+    ifPrecededBy: anyOf(whitespace, punctuationPattern, inputStart),
     ifFollowedBy: arabicNumeralPattern,
 });
-export const numberPattern = [
-    oneOrMore(anyOf(digitPattern, numericSeparatorPattern, spacedThousandsSeparatorPattern, numericSignPattern)),
+const numberPattern = [
+    possibly(numericSignPattern),
+    digitPattern,
+    zeroOrMore(anyOf(digitPattern, numericSeparatorPattern, spacedThousandsSeparatorPattern))
+];
+const exponentPattern = [
+    anyOf('e', 'E'),
+    possibly(anyOf('+', '-')),
+    oneOrMore(arabicNumeralPattern),
+];
+const numberPossiblyFollowedByExponentOrLettersPattern = [
+    numberPattern,
+    possibly(anyOf(exponentPattern, matches(zeroOrMore(unicodeProperty('Letter')), {
+        ifNotFollowedBy: digitPattern,
+    })))
 ];
-const percentageChars = ['%'];
-const currencySpecialChars = ['$', '¥', '€', '£', '¥', '₩', '₭', '₽', '₫', '฿', '¢', '₮', '؋', '₦', '₱', '₴', '₪'];
-const percentageOrCurrencyPattern = anyOf(...percentageChars, ...currencySpecialChars);
-export const prefixPercentageOrCurrencyPattern = matches([
-    percentageOrCurrencyPattern,
+const precedingPercentageOrCurrencyPattern = matches([
+    percentageOrCurrencyCharacterPattern,
     numberPattern,
 ], {
     ifNotPrecededBy: digitPattern,
-    ifFollowedBy: anyOf(whitespace, punctuationPattern),
+    ifFollowedBy: anyOf(whitespace, punctuationPattern, inputEnd),
 });
-export const suffixPercentagePattern = matches([
+const followingPercentageOrCurrencyPattern = matches([
     numberPattern,
-    percentageOrCurrencyPattern,
+    percentageOrCurrencyCharacterPattern,
+], {
+    ifPrecededBy: anyOf(whitespace, punctuationPattern, inputStart),
+    ifNotFollowedBy: anyOf(digitPattern),
+});
+const percentageOrCurrencyPattern = anyOf(precedingPercentageOrCurrencyPattern, followingPercentageOrCurrencyPattern);
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Time patterns
+////////////////////////////////////////////////////////////////////////////////////////////////
+const timePattern = matches([
+    anyOf(arabicNumeralPattern, [charRange('0', '1'), arabicNumeralPattern], ['2', charRange('0', '3')]),
+    repeated([1, 2], [
+        ':',
+        anyOf(arabicNumeralPattern, [charRange('0', '5'), arabicNumeralPattern]),
+    ])
 ], {
-    ifPrecededBy: anyOf(whitespace, punctuationPattern),
-    ifNotFollowedBy: digitPattern,
+    ifPrecededBy: anyOf(whitespace, punctuationPattern, inputStart),
+    ifNotPrecededBy: ':',
+    ifFollowedBy: anyOf(whitespace, punctuationPattern, inputEnd),
+    ifNotFollowedBy: ':',
 });
-export const percentagePattern = anyOf(prefixPercentageOrCurrencyPattern, suffixPercentagePattern);
 ////////////////////////////////////////////////////////////////////////////////////////////////
 // Letter patterns
 ////////////////////////////////////////////////////////////////////////////////////////////////
-export const letterPattern = unicodeProperty('Letter');
-export const markPattern = unicodeProperty('Mark');
-export const apostrophPattern = anyOf(`'`, `’`, `‘`);
-export const letterOrMarkPattern = anyOf(letterPattern, markPattern);
-export const dottedAbbreviationSequencePattern = matches(anyOf([
+const dottedAbbreviationSequencePattern = matches(anyOf([
     letterPattern,
     oneOrMore([
         '. ',
@@ -76,51 +118,84 @@ export const dottedAbbreviationSequencePattern = matches(anyOf([
     ifNotPrecededBy: anyOf(letterOrMarkPattern, digitPattern),
     ifNotFollowedBy: anyOf(letterOrMarkPattern, digitPattern)
 });
-export const wordCharacterPattern = anyOf(letterPattern, markPattern, digitPattern);
-export const wordSeparatorPattern = matches(anyOf('-', '_', '·', '‧', '&'), {
-    ifPrecededBy: letterOrMarkPattern,
-    ifFollowedBy: letterOrMarkPattern
-});
-export const wordInnerApostrophPattern = matches(apostrophPattern, {
+const wordCharacterPattern = anyOf(letterPattern, markPattern, digitPattern);
+const wordInnerApostrophPattern = matches(apostrophPattern, {
     ifPrecededBy: letterOrMarkPattern,
     ifFollowedBy: letterOrMarkPattern,
 });
-export const wordStartApostrophPattern = matches(apostrophPattern, {
+const wordStartApostrophPattern = matches(apostrophPattern, {
     ifPrecededBy: whitespace,
-    ifFollowedBy: [letterOrMarkPattern, letterOrMarkPattern, whitespace]
+    ifFollowedBy: [letterOrMarkPattern, letterOrMarkPattern, anyOf(whitespace, inputEnd)]
+});
+const basicWordPattern = oneOrMore(anyOf(wordCharacterPattern, wordInnerApostrophPattern));
+const hyphenatedWordPattern = [
+    basicWordPattern,
+    oneOrMore([
+        '-',
+        basicWordPattern,
+    ]),
+];
+const dotSeparatedWordPattern = matches([
+    basicWordPattern,
+    oneOrMore([
+        '.',
+        basicWordPattern,
+    ]),
+], {
+//ifPrecededBy: anyOf(whitespace, inputStart),
+//ifFollowedBy: anyOf(whitespace, inputEnd)
 });
-export const basicWordPattern = oneOrMore(anyOf(wordCharacterPattern, wordSeparatorPattern, wordInnerApostrophPattern));
-export const wordSegmentPattern = anyOf(dottedAbbreviationSequencePattern, dateTimeSeparatorPattern, percentagePattern, numberPattern, basicWordPattern);
+const interpunctSeparatedWordPattern = [
+    basicWordPattern,
+    oneOrMore([
+        anyOf('·', '‧'),
+        basicWordPattern,
+    ]),
+];
+const underscoreSeparatedWordPattern = [
+    basicWordPattern,
+    oneOrMore([
+        '_',
+        basicWordPattern,
+    ]),
+];
+const wordSegmentPattern = anyOf(timePattern, dimensionsPattern, percentageOrCurrencyPattern, numberPossiblyFollowedByExponentOrLettersPattern, interpunctSeparatedWordPattern, dotSeparatedWordPattern, dottedAbbreviationSequencePattern, underscoreSeparatedWordPattern, basicWordPattern);
 ////////////////////////////////////////////////////////////////////////////////////////////////
-// Prebuilt regular expressions
+// Phrase separation patterns
 ////////////////////////////////////////////////////////////////////////////////////////////////
-export const phraseSeparators = [',', '、', '，', '،', ';', '；', ':', '：', '—'];
-export const phraseSeparatorRegExp = buildRegExp([
+const phraseSeparatorCharacters = [',', '、', '，', '،', ';', '；', ':', '：', '—'];
+const phraseSeparatorPattern = [
     inputStart,
-    anyOf(...phraseSeparators),
+    anyOf(...phraseSeparatorCharacters),
     inputEnd
-]);
-export const sentenceSeparators = ['.', '。', '?', '？', '!', '！', '\n'];
-export const sentenceSeparatorRegExp = buildRegExp([
-    inputStart,
-    anyOf(...sentenceSeparators),
-    inputEnd
-]);
-export const sentenceSeparatorTrailingPunctuationRegExp = buildRegExp([
+];
+const phraseSeparatorTrailingPunctuationPattern = [
     inputStart,
-    anyOf('"', '”', '’', ')', ']', '}', '»', ...sentenceSeparators, ...phraseSeparators, oneOrMore(whitespace)),
+    anyOf(...phraseSeparatorCharacters, ' ', tab),
     inputEnd
-]);
-export const phraseSeparatorTrailingPunctuationRegExp = buildRegExp([
+];
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Sentence separation patterns
+////////////////////////////////////////////////////////////////////////////////////////////////
+const sentenceSeparatorCharacters = ['.', '。', '?', '？', '!', '！', '\n'];
+const sentenceSeparatorPattern = [
     inputStart,
-    anyOf(...phraseSeparators, ' ', tab),
+    anyOf(...sentenceSeparatorCharacters),
     inputEnd
-]);
-export const oneOrMoreSpacesRegExp = buildRegExp([
+];
+const sentenceSeparatorTrailingPunctuationPattern = [
     inputStart,
-    oneOrMore(' '),
+    anyOf('"', '”', '’', ')', ']', '}', '»', ...sentenceSeparatorCharacters, ...phraseSeparatorCharacters, oneOrMore(whitespace)),
     inputEnd
-]);
+];
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Prebuilt regular expressions
+////////////////////////////////////////////////////////////////////////////////////////////////
 export const wordCharacterRegExp = buildRegExp(wordCharacterPattern);
 export const whitespacePatternRegExp = buildRegExp(whitespace);
+export const letterPatternGlobalRegExp = buildRegExp(letterPattern, { global: true });
+export const phraseSeparatorRegExp = buildRegExp(phraseSeparatorPattern);
+export const phraseSeparatorTrailingPunctuationRegExp = buildRegExp(phraseSeparatorTrailingPunctuationPattern);
+export const sentenceSeparatorRegExp = buildRegExp(sentenceSeparatorPattern);
+export const sentenceSeparatorTrailingPunctuationRegExp = buildRegExp(sentenceSeparatorTrailingPunctuationPattern);
 //# sourceMappingURL=Patterns.js.map

package/dist/Patterns.js.map

@@ -1 +1 @@
-{"version":3,"file":"Patterns.js","sourceRoot":"","sources":["../src/Patterns.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,QAAQ,EAAE,GAAG,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAE/J,MAAM,UAAU,wBAAwB,CAAC,YAAsB;IAC9D,OAAO,KAAK,CACX,uBAAuB,CAAC,YAAY,CAAC,EACrC,kBAAkB,CAClB,CAAA;AACF,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,YAAsB;IAC7D,MAAM,mBAAmB,GACxB,OAAO,CACN,KAAK,CAAC,GAAG,YAAY,CAAC,EACtB,EAAE,eAAe,EAAE,oBAAoB,EAAE,CACzC,CAAA;IAEF,OAAO,mBAAmB,CAAA;AAC3B,CAAC;AAED,gGAAgG;AAChG,mBAAmB;AACnB,gGAAgG;AAChG,MAAM,CAAC,MAAM,kBAAkB,GAAG,eAAe,CAAC,aAAa,CAAC,CAAA;AAChE,MAAM,CAAC,MAAM,YAAY,GAAG,eAAe,CAAC,gBAAgB,CAAC,CAAA;AAC7D,MAAM,CAAC,MAAM,oBAAoB,GAAG,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA;AAEvD,MAAM,CAAC,MAAM,uBAAuB,GACnC,OAAO,CACN,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,EAAE;IAC3B,YAAY,EAAE,oBAAoB;IAClC,YAAY,EAAE,oBAAoB;CAClC,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,wBAAwB,GAAG,OAAO,CAC9C,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,EAAE;IACtB,YAAY,EAAE,oBAAoB;IAClC,YAAY,EAAE,oBAAoB;CAClC,CAAC,CAAA;AAEF,MAAM,CAAC,MAAM,eAAe,GAC3B,SAAS,CAAC,KAAK,CACd,oBAAoB,EACpB,wBAAwB,CACxB,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,+BAA+B,GAC3C,OAAO,CACN,GAAG,EAAE;IACL,YAAY,EAAE,oBAAoB;IAClC,YAAY,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC;CAC/C,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,kBAAkB,GAC9B,OAAO,CACN,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE;IACjB,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,CAAC;IACnD,YAAY,EAAE,oBAAoB;CAClC,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,aAAa,GAAG;IAC5B,SAAS,CAAC,KAAK,CACd,YAAY,EACZ,uBAAuB,EACvB,+BAA+B,EAC/B,kBAAkB,CAClB,CAAC;CACF,CAAA;AAED,MAAM,eAAe,GAAG,CAAC,GAAG,CAAC,CAAA;AAC7B,MAAM,oBAAoB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAElH,MAAM,2BAA2B,GAAG,KAAK,CAAC,GAAG,eAAe,EAAE,GAAG,oBAAoB,CAAC,CAAA;AAEtF,MAAM,CAAC,MAAM,iCAAiC,GAC7C,OAAO,CAAC;IACP,2BAA2B;IAC3B,aAAa;CACb,EAAE;IACF,eAAe,EAAE,YAAY;IAC7B,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,CAAC;CACnD,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,uBAAuB,GACnC,OAAO,CAAC;IACP,aAAa;IACb,2BAA2B;CAC3B,EAAE;IACF,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,CAAC;IACnD,eAAe,EAAE,YAAY;CAC7B,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,iBAAiB,GAAG,KAAK,CACrC,iCAAiC,EACjC,uBAAuB,CACvB,CAAA;AAED,gGAAgG;AAChG,kBAAkB;AAClB,gGAAgG;AAChG,MAAM,CAAC,MAAM,aAAa,GAAG,eAAe,CAAC,QAAQ,CAAC,CAAA;AACtD,MAAM,CAAC,MAAM,WAAW,GAAG,eAAe,CAAC,MAAM,CAAC,CAAA;AAClD,MAAM,CAAC,MAAM,gBAAgB,GAAG,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAEpD,MAAM,CAAC,MAAM,mBAAmB,GAAG,KAAK,CACvC,aAAa,EACb,WAAW,CACX,CAAA;AAED,MAAM,CAAC,MAAM,iCAAiC,GAC7C,OAAO,CACN,KAAK,CACJ;IACC,aAAa;IAEb,SAAS,CAAC;QACT,IAAI;QACJ,aAAa;KACb,CAAC;IAEF,QAAQ,CAAC,GAAG,CAAC;CACb,EACD;IACC,aAAa;IAEb,SAAS,CAAC;QACT,GAAG;QACH,QAAQ,CAAC,GAAG,CAAC;QACb,aAAa;KACb,CAAC;IAEF,QAAQ,CAAC,GAAG,CAAC;CACb,CACD,EAAE;IACH,eAAe,EAAE,KAAK,CAAC,mBAAmB,EAAE,YAAY,CAAC;IACzD,eAAe,EAAE,KAAK,CAAC,mBAAmB,EAAE,YAAY,CAAC;CACzD,CAAC,CAAA;AAGH,MAAM,CAAC,MAAM,oBAAoB,GAChC,KAAK,CACJ,aAAa,EACb,WAAW,EACX,YAAY,CACZ,CAAA;AAEF,MAAM,CAAC,MAAM,oBAAoB,GAChC,OAAO,CACN,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,EAAE;IAEhC,YAAY,EAAE,mBAAmB;IACjC,YAAY,EAAE,mBAAmB;CACjC,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,yBAAyB,GACrC,OAAO,CACN,gBAAgB,EAAE;IAElB,YAAY,EAAE,mBAAmB;IACjC,YAAY,EAAE,mBAAmB;CACjC,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,yBAAyB,GACrC,OAAO,CACN,gBAAgB,EAAE;IAElB,YAAY,EAAE,UAAU;IACxB,YAAY,EAAE,CAAC,mBAAmB,EAAE,mBAAmB,EAAE,UAAU,CAAC;CACpE,CAAC,CAAA;AAEH,MAAM,CAAC,MAAM,gBAAgB,GAC5B,SAAS,CACR,KAAK,CACJ,oBAAoB,EACpB,oBAAoB,EACpB,yBAAyB,CACzB,CACD,CAAA;AAEF,MAAM,CAAC,MAAM,kBAAkB,GAAG,KAAK,CACtC,iCAAiC,EACjC,wBAAwB,EACxB,iBAAiB,EACjB,aAAa,EACb,gBAAgB,CAChB,CAAA;AAED,gGAAgG;AAChG,+BAA+B;AAC/B,gGAAgG;AAChG,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAE7E,MAAM,CAAC,MAAM,qBAAqB,GAAG,WAAW,CAAC;IAChD,UAAU;IACV,KAAK,CAAC,GAAG,gBAAgB,CAAC;IAC1B,QAAQ;CACR,CAAC,CAAA;AAEF,MAAM,CAAC,MAAM,kBAAkB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,CAAC,CAAA;AAEtE,MAAM,CAAC,MAAM,uBAAuB,GAAG,WAAW,CAAC;IAClD,UAAU;IACV,KAAK,CAAC,GAAG,kBAAkB,CAAC;IAC5B,QAAQ;CACR,CAAC,CAAA;AAEF,MAAM,CAAC,MAAM,0CAA0C,GAAG,WAAW,CAAC;IACrE,UAAU;IACV,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,kBAAkB,EAAE,GAAG,gBAAgB,EAAE,SAAS,CAAC,UAAU,CAAC,CAAC;IAC3G,QAAQ;CACR,CAAC,CAAA;AAEF,MAAM,CAAC,MAAM,wCAAwC,GAAG,WAAW,CAAC;IACnE,UAAU;IACV,KAAK,CAAC,GAAG,gBAAgB,EAAE,GAAG,EAAE,GAAG,CAAC;IACpC,QAAQ;CACR,CAAC,CAAA;AAEF,MAAM,CAAC,MAAM,qBAAqB,GAAG,WAAW,CAAC;IAChD,UAAU;IACV,SAAS,CAAC,GAAG,CAAC;IACd,QAAQ;CACR,CAAC,CAAA;AAEF,MAAM,CAAC,MAAM,mBAAmB,GAAG,WAAW,CAAC,oBAAoB,CAAC,CAAA;AACpE,MAAM,CAAC,MAAM,uBAAuB,GAAG,WAAW,CAAC,UAAU,CAAC,CAAA"}
+{"version":3,"file":"Patterns.js","sourceRoot":"","sources":["../src/Patterns.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,QAAQ,EAAE,GAAG,EAAE,eAAe,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAE3K,gGAAgG;AAChG,0BAA0B;AAC1B,gGAAgG;AAChG,MAAM,UAAU,wBAAwB,CAAC,YAAsB;IAC9D,OAAO,KAAK,CACX,uBAAuB,CAAC,YAAY,CAAC,EACrC,kBAAkB,CAClB,CAAA;AACF,CAAC;AAED,MAAM,UAAU,uBAAuB,CAAC,YAAsB;IAC7D,MAAM,mBAAmB,GACxB,OAAO,CACN,KAAK,CAAC,GAAG,YAAY,CAAC,EACtB,EAAE,eAAe,EAAE,oBAAoB,EAAE,CACzC,CAAA;IAEF,OAAO,mBAAmB,CAAA;AAC3B,CAAC;AAED,gGAAgG;AAChG,4BAA4B;AAC5B,gGAAgG;AAChG,MAAM,aAAa,GAAG,eAAe,CAAC,QAAQ,CAAC,CAAA;AAC/C,MAAM,WAAW,GAAG,eAAe,CAAC,MAAM,CAAC,CAAA;AAE3C,MAAM,mBAAmB,GAAG,KAAK,CAChC,aAAa,EACb,WAAW,CACX,CAAA;AAED,MAAM,gBAAgB,GAAG,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAE7C,MAAM,kBAAkB,GAAG,eAAe,CAAC,aAAa,CAAC,CAAA;AACzD,MAAM,YAAY,GAAG,eAAe,CAAC,gBAAgB,CAAC,CAAA;AACtD,MAAM,oBAAoB,GAAG,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA;AAEhD,MAAM,oBAAoB,GAAG,CAAC,GAAG,CAAC,CAAA;AAClC,MAAM,kBAAkB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAE3G,MAAM,oCAAoC,GAAG,KAAK,CAAC,GAAG,oBAAoB,EAAE,GAAG,kBAAkB,CAAC,CAAA;AAElG,gGAAgG;AAChG,mBAAmB;AACnB,gGAAgG;AAChG,MAAM,uBAAuB,GAC5B,OAAO,CACN,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,EAAE;IAC3B,YAAY,EAAE,oBAAoB;IAClC,YAAY,EAAE,oBAAoB;CAClC,CAAC,CAAA;AAEH,MAAM,iBAAiB,GAAG,OAAO,CAAC;IACjC,SAAS,CAAC,oBAAoB,CAAC;IAE/B,SAAS,CAAC;QACT,GAAG;QACH,SAAS,CAAC,oBAAoB,CAAC;KAC/B,CAAC;CACF,EAAE;IACF,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,UAAU,CAAC;IAC/D,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,QAAQ,CAAC;CAC7D,CAAC,CAAA;AAEF,MAAM,+BAA+B,GACpC,OAAO,CACN,GAAG,EAAE;IACL,YAAY,EAAE,oBAAoB;IAClC,YAAY,EAAE;QACb,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC;QACjC,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,QAAQ,CAAC;KAC/C;CACD,CAAC,CAAA;AAEH,MAAM,kBAAkB,GACvB,OAAO,CACN,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE;IACjB,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,UAAU,CAAC;IAC/D,YAAY,EAAE,oBAAoB;CAClC,CAAC,CAAA;AAEH,MAAM,aAAa,GAAG;IACrB,QAAQ,CAAC,kBAAkB,CAAC;IAC5B,YAAY;IAEZ,UAAU,CAAC,KAAK,CACf,YAAY,EACZ,uBAAuB,EACvB,+BAA+B,CAC/B,CAAC;CACF,CAAA;AAED,MAAM,eAAe,GAAG;IACvB,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC;IACf,QAAQ,CAAC,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC;IACzB,SAAS,CAAC,oBAAoB,CAAC;CAC/B,CAAA;AAED,MAAM,gDAAgD,GAAG;IACxD,aAAa;IAEb,QAAQ,CAAC,KAAK,CACb,eAAe,EAEf,OAAO,CACN,UAAU,CAAC,eAAe,CAAC,QAAQ,CAAC,CAAC,EAAE;QACvC,eAAe,EAAE,YAAY;KAC7B,CAAC,CACF,CAAC;CACF,CAAA;AAED,MAAM,oCAAoC,GACzC,OAAO,CAAC;IACP,oCAAoC;IACpC,aAAa;CACb,EAAE;IACF,eAAe,EAAE,YAAY;IAC7B,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,QAAQ,CAAC;CAC7D,CAAC,CAAA;AAEH,MAAM,oCAAoC,GACzC,OAAO,CAAC;IACP,aAAa;IACb,oCAAoC;CACpC,EAAE;IACF,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,UAAU,CAAC;IAC/D,eAAe,EAAE,KAAK,CAAC,YAAY,CAAC;CACpC,CAAC,CAAA;AAEH,MAAM,2BAA2B,GAAG,KAAK,CACxC,oCAAoC,EACpC,oCAAoC,CACpC,CAAA;AAED,gGAAgG;AAChG,gBAAgB;AAChB,gGAAgG;AAChG,MAAM,WAAW,GAAG,OAAO,CAAC;IAC3B,KAAK,CAAC,oBAAoB,EAAE,CAAC,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,oBAAoB,CAAC,EAAE,CAAC,GAAG,EAAE,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;IAEpG,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE;QAChB,GAAG;QACH,KAAK,CAAC,oBAAoB,EAAE,CAAC,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,oBAAoB,CAAC,CAAC;KACxE,CAAC;CACF,EAAE;IACF,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,UAAU,CAAC;IAC/D,eAAe,EAAE,GAAG;IACpB,YAAY,EAAE,KAAK,CAAC,UAAU,EAAE,kBAAkB,EAAE,QAAQ,CAAC;IAC7D,eAAe,EAAE,GAAG;CACpB,CAAC,CAAA;AAEF,gGAAgG;AAChG,kBAAkB;AAClB,gGAAgG;AAChG,MAAM,iCAAiC,GACtC,OAAO,CACN,KAAK,CACJ;IACC,aAAa;IAEb,SAAS,CAAC;QACT,IAAI;QACJ,aAAa;KACb,CAAC;IAEF,QAAQ,CAAC,GAAG,CAAC;CACb,EACD;IACC,aAAa;IAEb,SAAS,CAAC;QACT,GAAG;QACH,QAAQ,CAAC,GAAG,CAAC;QACb,aAAa;KACb,CAAC;IAEF,QAAQ,CAAC,GAAG,CAAC;CACb,CACD,EAAE;IACH,eAAe,EAAE,KAAK,CAAC,mBAAmB,EAAE,YAAY,CAAC;IACzD,eAAe,EAAE,KAAK,CAAC,mBAAmB,EAAE,YAAY,CAAC;CACzD,CAAC,CAAA;AAEH,MAAM,oBAAoB,GACzB,KAAK,CACJ,aAAa,EACb,WAAW,EACX,YAAY,CACZ,CAAA;AAEF,MAAM,yBAAyB,GAC9B,OAAO,CACN,gBAAgB,EAAE;IAElB,YAAY,EAAE,mBAAmB;IACjC,YAAY,EAAE,mBAAmB;CACjC,CAAC,CAAA;AAEH,MAAM,yBAAyB,GAC9B,OAAO,CACN,gBAAgB,EAAE;IAElB,YAAY,EAAE,UAAU;IACxB,YAAY,EAAE,CAAC,mBAAmB,EAAE,mBAAmB,EAAE,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC;CACrF,CAAC,CAAA;AAEH,MAAM,gBAAgB,GACrB,SAAS,CACR,KAAK,CACJ,oBAAoB,EACpB,yBAAyB,CACzB,CACD,CAAA;AAEF,MAAM,qBAAqB,GAAG;IAC7B,gBAAgB;IAEhB,SAAS,CAAC;QACT,GAAG;QACH,gBAAgB;KAChB,CAAC;CACF,CAAA;AAED,MAAM,uBAAuB,GAAG,OAAO,CAAC;IACvC,gBAAgB;IAEhB,SAAS,CAAC;QACT,GAAG;QACH,gBAAgB;KAChB,CAAC;CACF,EAAE;AACF,8CAA8C;AAC9C,2CAA2C;CAC3C,CAAC,CAAA;AAEF,MAAM,8BAA8B,GAAG;IACtC,gBAAgB;IAEhB,SAAS,CAAC;QACT,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC;QACf,gBAAgB;KAChB,CAAC;CACF,CAAA;AAED,MAAM,8BAA8B,GAAG;IACtC,gBAAgB;IAEhB,SAAS,CAAC;QACT,GAAG;QACH,gBAAgB;KAChB,CAAC;CACF,CAAA;AAED,MAAM,kBAAkB,GAAG,KAAK,CAC/B,WAAW,EACX,iBAAiB,EACjB,2BAA2B,EAC3B,gDAAgD,EAEhD,8BAA8B,EAC9B,uBAAuB,EACvB,iCAAiC,EACjC,8BAA8B,EAC9B,gBAAgB,CAChB,CAAA;AAED,gGAAgG;AAChG,6BAA6B;AAC7B,gGAAgG;AAChG,MAAM,yBAAyB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAE/E,MAAM,sBAAsB,GAAG;IAC9B,UAAU;IACV,KAAK,CAAC,GAAG,yBAAyB,CAAC;IACnC,QAAQ;CACR,CAAA;AAED,MAAM,yCAAyC,GAAG;IACjD,UAAU;IACV,KAAK,CAAC,GAAG,yBAAyB,EAAE,GAAG,EAAE,GAAG,CAAC;IAC7C,QAAQ;CACR,CAAA;AAED,gGAAgG;AAChG,+BAA+B;AAC/B,gGAAgG;AAChG,MAAM,2BAA2B,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,CAAC,CAAA;AAExE,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,KAAK,CAAC,GAAG,2BAA2B,CAAC;IACrC,QAAQ;CACR,CAAA;AAED,MAAM,2CAA2C,GAAG;IACnD,UAAU;IACV,KAAK,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,2BAA2B,EAAE,GAAG,yBAAyB,EAAE,SAAS,CAAC,UAAU,CAAC,CAAC;IAC7H,QAAQ;CACR,CAAA;AAED,gGAAgG;AAChG,+BAA+B;AAC/B,gGAAgG;AAChG,MAAM,CAAC,MAAM,mBAAmB,GAAG,WAAW,CAAC,oBAAoB,CAAC,CAAA;AACpE,MAAM,CAAC,MAAM,uBAAuB,GAAG,WAAW,CAAC,UAAU,CAAC,CAAA;AAE9D,MAAM,CAAC,MAAM,yBAAyB,GAAG,WAAW,CAAC,aAAa,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAA;AAErF,MAAM,CAAC,MAAM,qBAAqB,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;AACxE,MAAM,CAAC,MAAM,wCAAwC,GAAG,WAAW,CAAC,yCAAyC,CAAC,CAAA;AAE9G,MAAM,CAAC,MAAM,uBAAuB,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAC5E,MAAM,CAAC,MAAM,0CAA0C,GAAG,WAAW,CAAC,2CAA2C,CAAC,CAAA"}

package/dist/Suppressions.d.ts

@@ -1,4 +1,5 @@
 export declare const cldrSuppressions: Record<string, string[]>;
+export declare const additionalSuppressions: Record<string, string[]>;
 export declare const leadingApostropheContractionSuppressions: Record<string, string[]>;
 export declare const nounSuppressions: string[];
 export declare const tldSuppressions: string[];