glost 0.1.0

package/dist/nodes.js

@@ -0,0 +1,181 @@
+// ============================================================================
+// Node Factory Functions
+// ============================================================================
+/**
+ * Create an GLOST word node
+ */
+export function createGLOSTWordNode(value, transcription, metadata, level = "word", lang, script, extras) {
+    return {
+        type: "WordNode",
+        lang,
+        script,
+        transcription,
+        metadata,
+        extras,
+        children: [createGLOSTTextNode(value)],
+    };
+}
+/**
+ * Create an GLOST sentence node
+ */
+export function createGLOSTSentenceNode(originalText, lang, script, children = [], transcription, extras) {
+    return {
+        type: "SentenceNode",
+        originalText,
+        lang,
+        script,
+        transcription,
+        children,
+        extras,
+    };
+}
+/**
+ * Create an GLOST paragraph node
+ */
+export function createGLOSTParagraphNode(children = [], extras) {
+    return {
+        type: "ParagraphNode",
+        children,
+        position: undefined,
+        extras,
+    };
+}
+/**
+ * Create an GLOST root node
+ */
+export function createGLOSTRootNode(lang, script, children = [], metadata, extras) {
+    return {
+        type: "RootNode",
+        lang,
+        script,
+        metadata,
+        children,
+        position: undefined,
+        extras,
+    };
+}
+// ============================================================================
+// Helper Functions for Common Patterns
+// ============================================================================
+/**
+ * Create a simple word node with basic transcription
+ */
+export function createSimpleWord(text, transliteration, system = "ipa", partOfSpeech = "unknown", level = "word") {
+    const transcription = {
+        [system]: {
+            text: transliteration,
+            system: system,
+            syllables: [text],
+        },
+    };
+    const metadata = {
+        partOfSpeech,
+    };
+    return createGLOSTWordNode(text, transcription, metadata, level);
+}
+/**
+ * Create a Thai word node with RTGS transcription
+ */
+export function createThaiWord(text, rtgs, partOfSpeech = "unknown", tone, syllables) {
+    const transcription = {
+        rtgs: {
+            text: rtgs,
+            system: "rtgs",
+            tone,
+            syllables: syllables || [text],
+        },
+    };
+    const metadata = {
+        partOfSpeech,
+    };
+    return createGLOSTWordNode(text, transcription, metadata, "word", "th", "thai");
+}
+/**
+ * Create a Japanese word node with romaji transcription
+ */
+export function createJapaneseWord(text, romaji, partOfSpeech = "unknown", furigana) {
+    const transcription = {
+        romaji: {
+            text: romaji,
+            system: "romaji",
+            syllables: [text],
+        },
+    };
+    if (furigana) {
+        transcription.furigana = {
+            text: furigana,
+            system: "furigana",
+            syllables: [text],
+        };
+    }
+    const metadata = {
+        partOfSpeech,
+    };
+    return createGLOSTWordNode(text, transcription, metadata, "word", "ja", "mixed");
+}
+/**
+ * Create a sentence from an array of words
+ */
+export function createSentenceFromWords(words, lang, script, originalText) {
+    const text = originalText ||
+        words
+            .map((w) => {
+            // Extract text from word's Text node children
+            const textNode = w.children.find((child) => child.type === "TextNode");
+            return textNode ? textNode.value : "";
+        })
+            .join("");
+    return createGLOSTSentenceNode(text, lang, script, words);
+}
+/**
+ * Create a paragraph from an array of sentences
+ */
+export function createParagraphFromSentences(sentences) {
+    return createGLOSTParagraphNode(sentences);
+}
+/**
+ * Create a document from an array of paragraphs
+ */
+export function createDocumentFromParagraphs(paragraphs, lang, script, metadata) {
+    return createGLOSTRootNode(lang, script, paragraphs, metadata);
+}
+// ============================================================================
+// NLCST Node Factory Functions
+// ============================================================================
+/**
+ * Create an GLOST punctuation node
+ */
+export function createGLOSTPunctuationNode(value) {
+    return {
+        type: "PunctuationNode",
+        value,
+    };
+}
+/**
+ * Create an GLOST whitespace node
+ */
+export function createGLOSTWhiteSpaceNode(value) {
+    return {
+        type: "WhiteSpaceNode",
+        value,
+    };
+}
+/**
+ * Create an GLOST symbol node
+ */
+export function createGLOSTSymbolNode(value) {
+    return {
+        type: "SymbolNode",
+        value,
+    };
+}
+/**
+ * Create an GLOST text node
+ */
+export function createGLOSTTextNode(value) {
+    return {
+        type: "TextNode",
+        value,
+    };
+}
+//# sourceMappingURL=nodes.js.map

package/dist/nodes.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"nodes.js","sourceRoot":"","sources":["../src/nodes.ts"],"names":[],"mappings":"AAiBA,+EAA+E;AAC/E,yBAAyB;AACzB,+EAA+E;AAE/E;;GAEG;AACH,MAAM,UAAU,mBAAmB,CACjC,KAAa,EACb,aAAkC,EAClC,QAA4B,EAC5B,QAAyB,MAAM,EAC/B,IAAmB,EACnB,MAAqB,EACrB,MAAoB;IAEpB,OAAO;QACL,IAAI,EAAE,UAAU;QAChB,IAAI;QACJ,MAAM;QACN,aAAa;QACb,QAAQ;QACR,MAAM;QACN,QAAQ,EAAE,CAAC,mBAAmB,CAAC,KAAK,CAAC,CAAC;KACvC,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,uBAAuB,CACrC,YAAoB,EACpB,IAAkB,EAClB,MAAoB,EACpB,WAAwB,EAAE,EAC1B,aAAmC,EACnC,MAAoB;IAEpB,OAAO;QACL,IAAI,EAAE,cAAc;QACpB,YAAY;QACZ,IAAI;QACJ,MAAM;QACN,aAAa;QACb,QAAQ;QACR,MAAM;KACP,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,wBAAwB,CACtC,WAA4B,EAAE,EAC9B,MAAoB;IAEpB,OAAO;QACL,IAAI,EAAE,eAAe;QAErB,QAAQ;QACR,QAAQ,EAAE,SAAS;QACnB,MAAM;KACP,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CACjC,IAAkB,EAClB,MAAoB,EACpB,WAA6B,EAAE,EAC/B,QAKC,EACD,MAAoB;IAEpB,OAAO;QACL,IAAI,EAAE,UAAU;QAChB,IAAI;QACJ,MAAM;QACN,QAAQ;QACR,QAAQ;QACR,QAAQ,EAAE,SAAS;QACnB,MAAM;KACP,CAAC;AACJ,CAAC;AAED,+EAA+E;AAC/E,uCAAuC;AACvC,+EAA+E;AAE/E;;GAEG;AACH,MAAM,UAAU,gBAAgB,CAC9B,IAAY,EACZ,eAAuB,EACvB,SAAiB,KAAK,EACtB,eAAuB,SAAS,EAChC,QAAyB,MAAM;IAE/B,MAAM,aAAa,GAAwB;QACzC,CAAC,MAAM,CAAC,EAAE;YACR,IAAI,EAAE,eAAe;YACrB,MAAM,EAAE,MAAa;YACrB,SAAS,EAAE,CAAC,IAAI,CAAC;SAClB;KACF,CAAC;IAEF,MAAM,QAAQ,GAAuB;QACnC,YAAY;KACb,CAAC;IAEF,OAAO,mBAAmB,CAAC,IAAI,EAAE,aAAa,EAAE,QAAQ,EAAE,KAAK,CAAC,CAAC;AACnE,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAC5B,IAAY,EACZ,IAAY,EACZ,eAAuB,SAAS,EAChC,IAAa,EACb,SAAoB;IAEpB,MAAM,aAAa,GAAwB;QACzC,IAAI,EAAE;YACJ,IAAI,EAAE,IAAI;YACV,MAAM,EAAE,MAAM;YACd,IAAI;YACJ,SAAS,EAAE,SAAS,IAAI,CAAC,IAAI,CAAC;SAC/B;KACF,CAAC;IAEF,MAAM,QAAQ,GAAuB;QACnC,YAAY;KACb,CAAC;IAEF,OAAO,mBAAmB,CACxB,IAAI,EACJ,aAAa,EACb,QAAQ,EACR,MAAM,EACN,IAAI,EACJ,MAAM,CACP,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAChC,IAAY,EACZ,MAAc,EACd,eAAuB,SAAS,EAChC,QAAiB;IAEjB,MAAM,aAAa,GAAwB;QACzC,MAAM,EAAE;YACN,IAAI,EAAE,MAAM;YACZ,MAAM,EAAE,QAAQ;YAChB,SAAS,EAAE,CAAC,IAAI,CAAC;SAClB;KACF,CAAC;IAEF,IAAI,QAAQ,EAAE,CAAC;QACb,aAAa,CAAC,QAAQ,GAAG;YACvB,IAAI,EAAE,QAAQ;YACd,MAAM,EAAE,UAAU;YAClB,SAAS,EAAE,CAAC,IAAI,CAAC;SAClB,CAAC;IACJ,CAAC;IAED,MAAM,QAAQ,GAAuB;QACnC,YAAY;KACb,CAAC;IAEF,OAAO,mBAAmB,CACxB,IAAI,EACJ,aAAa,EACb,QAAQ,EACR,MAAM,EACN,IAAI,EACJ,OAAO,CACR,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,uBAAuB,CACrC,KAAkB,EAClB,IAAkB,EAClB,MAAoB,EACpB,YAAqB;IAErB,MAAM,IAAI,GACR,YAAY;QACZ,KAAK;aACF,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE;YACT,8CAA8C;YAC9C,MAAM,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAC,IAAI,CAC9B,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,KAAK,UAAU,CACxB,CAAC;YACf,OAAO,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;QACxC,CAAC,CAAC;aACD,IAAI,CAAC,EAAE,CAAC,CAAC;IACd,OAAO,uBAAuB,CAAC,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC;AAC5D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,4BAA4B,CAC1C,SAA0B;IAE1B,OAAO,wBAAwB,CAAC,SAAS,CAAC,CAAC;AAC7C,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,4BAA4B,CAC1C,UAA4B,EAC5B,IAAkB,EAClB,MAAoB,EACpB,QAKC;IAED,OAAO,mBAAmB,CAAC,IAAI,EAAE,MAAM,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC;AACjE,CAAC;AAED,+EAA+E;AAC/E,+BAA+B;AAC/B,+EAA+E;AAE/E;;GAEG;AACH,MAAM,UAAU,0BAA0B,CAAC,KAAa;IACtD,OAAO;QACL,IAAI,EAAE,iBAAiB;QACvB,KAAK;KACN,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,yBAAyB,CAAC,KAAa;IACrD,OAAO;QACL,IAAI,EAAE,gBAAgB;QACtB,KAAK;KACN,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,qBAAqB,CAAC,KAAa;IACjD,OAAO;QACL,IAAI,EAAE,YAAY;QAClB,KAAK;KACN,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,KAAa;IAC/C,OAAO;QACL,IAAI,EAAE,UAAU;QAChB,KAAK;KACN,CAAC;AACJ,CAAC"}

package/dist/types.d.ts

@@ -0,0 +1,379 @@
+import type { Literal as NlcstLiteral, Paragraph as NlcstParagraph, Punctuation as NlcstPunctuation, Root as NlcstRoot, Sentence as NlcstSentence, Source as NlcstSource, Symbol as NlcstSymbol, Text as NlcstText, WhiteSpace as NlcstWhiteSpace, Word as NlcstWord } from "nlcst";
+/**
+ * Linguistic level of a text segment
+ */
+export type LinguisticLevel = "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
+/**
+ * Context for pronunciation variants
+ */
+export type PronunciationContext = "formal" | "informal" | "historical" | "regional" | "dialectal";
+/**
+ * Transcription system identifiers
+ */
+export type TranscriptionSystem = "rtgs" | "aua" | "paiboon" | "romaji" | "furigana" | "ipa" | "pinyin" | "hangul" | string;
+/**
+ * Language codes following BCP-47 format (RFC 5646)
+ * Format: language[-script][-region][-variant]
+ * Examples: th-TH, ja-JP, zh-CN, ko-KR, en-US, fr-FR, de-DE
+ */
+export type LanguageCode = "th-TH" | "th" | "ja-JP" | "ja" | "zh-CN" | "zh-TW" | "zh-HK" | "zh" | "ko-KR" | "ko-KP" | "ko" | "en-US" | "en-GB" | "en-CA" | "en-AU" | "en" | "fr-FR" | "fr-CA" | "fr-BE" | "fr" | "de-DE" | "de-AT" | "de-CH" | "de" | "es-ES" | "es-MX" | "es-AR" | "es" | "it-IT" | "it-CH" | "it" | "pt-PT" | "pt-BR" | "pt" | "ru-RU" | "ru" | "ar-SA" | "ar-EG" | "ar" | "hi-IN" | "hi" | string;
+/**
+ * Script system identifiers
+ */
+export type ScriptSystem = "thai" | "hiragana" | "katakana" | "kanji" | "hanzi" | "hangul" | "latin" | "mixed" | string;
+/**
+ * Quick translations in different languages using BCP-47 format
+ */
+export type QuickTranslations = {
+    /** English translations */
+    "en-US"?: string;
+    "en-GB"?: string;
+    "en"?: string;
+    /** Thai translations */
+    "th-TH"?: string;
+    "th"?: string;
+    /** Japanese translations */
+    "ja-JP"?: string;
+    "ja"?: string;
+    /** Chinese translations */
+    "zh-CN"?: string;
+    "zh-TW"?: string;
+    "zh"?: string;
+    /** Korean translations */
+    "ko-KR"?: string;
+    "ko"?: string;
+    /** French translations */
+    "fr-FR"?: string;
+    "fr-CA"?: string;
+    "fr"?: string;
+    /** German translations */
+    "de-DE"?: string;
+    "de"?: string;
+    /** Spanish translations */
+    "es-ES"?: string;
+    "es-MX"?: string;
+    "es"?: string;
+    /** Custom language translations using BCP-47 format */
+    [lang: string]: string | undefined;
+};
+/**
+ * Extended metadata for enhanced functionality
+ */
+export type ExtendedMetadata = {
+    /** Quick translations in multiple languages */
+    translations?: QuickTranslations;
+    /** Difficulty level for learners */
+    difficulty?: "beginner" | "intermediate" | "advanced";
+    /** Frequency in common usage */
+    frequency?: "rare" | "uncommon" | "common" | "very-common";
+    /** Cultural notes */
+    culturalNotes?: string;
+    /** Related words or concepts */
+    related?: string[];
+    /** Example sentences */
+    examples?: string[];
+    /** Custom extensions */
+    [key: string]: any;
+};
+/**
+ * Extras field for extending GLOST nodes
+ */
+export type GLOSTExtras = {
+    /** Quick translations */
+    translations?: QuickTranslations;
+    /** Extended metadata */
+    metadata?: ExtendedMetadata;
+    /** Custom extensions */
+    [key: string]: any;
+};
+/**
+ * Pronunciation variant for a text segment
+ */
+export type PronunciationVariant = {
+    /** The variant text in the transcription system */
+    text: string;
+    /** Context where this variant is used */
+    context: PronunciationContext;
+    /** Additional notes about this variant */
+    notes?: string;
+};
+/**
+ * Transcription information for a text segment
+ */
+export type TranscriptionInfo = {
+    /** The transcription text */
+    text: string;
+    /** The transcription system used */
+    system: TranscriptionSystem;
+    /** Pronunciation variants */
+    variants?: PronunciationVariant[];
+    /** Tone information (for tonal languages) */
+    tone?: number;
+    /** Syllable breakdown */
+    syllables?: string[];
+    /** Additional phonetic information */
+    phonetic?: string;
+};
+/**
+ * Complete transliteration data for a text segment
+ */
+export type TransliterationData = {
+    /** Map of transcription systems to their data */
+    [system: string]: TranscriptionInfo;
+};
+/**
+ * Linguistic metadata for a text segment
+ */
+export type LinguisticMetadata = {
+    /** @deprecated Use extras.translations instead */
+    meaning?: string;
+    /** Part of speech */
+    partOfSpeech: string;
+    /** Usage notes */
+    usage?: string;
+    /** Etymology information */
+    etymology?: string;
+    /** Example usage */
+    examples?: string[];
+    /** Frequency information */
+    frequency?: "high" | "medium" | "low";
+    /** Formality level */
+    formality?: "formal" | "neutral" | "informal";
+    /** Register (academic, colloquial, etc.) */
+    register?: string;
+    /** @deprecated Use extras.translations instead */
+    shortDefinition?: string;
+    /** @deprecated Use extras.translations instead */
+    fullDefinition?: string;
+    /** @deprecated Use metadata enrichment extensions instead */
+    difficulty?: "beginner" | "intermediate" | "advanced";
+};
+/**
+ * Union type for all GLOST nodes
+ */
+export type GLOSTNode = GLOSTWord | GLOSTSentence | GLOSTParagraph | GLOSTRoot | GLOSTText | GLOSTSymbol | GLOSTPunctuation | GLOSTWhiteSpace | GLOSTSource | GLOSTClause | GLOSTPhrase | GLOSTSyllable | GLOSTCharacter;
+/**
+ * GLOST nodes that extend nlcst Literal (have a value property)
+ */
+export type GLOSTLiteral = NlcstLiteral & {
+    /** Language code for this node */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Linguistic level of this segment */
+    level?: LinguisticLevel;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras;
+};
+/**
+ * GLOST Punctuation node (extends nlcst PunctuationNode)
+ */
+export type GLOSTPunctuation = NlcstPunctuation & {};
+/**
+ * GLOST WhiteSpace node (extends nlcst WhiteSpaceNode)
+ */
+export type GLOSTWhiteSpace = NlcstWhiteSpace & {};
+/**
+ * GLOST Symbol node (extends nlcst SymbolNode)
+ */
+export type GLOSTSymbol = NlcstSymbol & {};
+/**
+ * GLOST Text node (extends nlcst TextNode)
+ */
+export type GLOSTText = NlcstText & {};
+/**
+ * GLOST Source node (extends nlcst SourceNode)
+ */
+export type GLOSTSource = NlcstSource & {};
+/**
+ * Extended word node with transcription support
+ * Extends nlcst WordNode and adds GLOST-specific properties
+ */
+export type GLOSTWord = Omit<NlcstWord, "children"> & {
+    /** Transcription data */
+    transcription: TransliterationData;
+    /** Linguistic metadata */
+    metadata: LinguisticMetadata;
+    /** @deprecated Use extras.translations instead */
+    shortDefinition?: string;
+    /** @deprecated Use extras.translations instead */
+    fullDefinition?: string;
+    /** @deprecated Use metadata enrichment extensions instead */
+    difficulty?: "beginner" | "intermediate" | "advanced";
+    /** Language code for this node */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Linguistic level of this segment */
+    level?: LinguisticLevel;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras;
+    /** Children nodes - must contain at least one Text node */
+    children: GLOSTWordContent[];
+};
+/**
+ * Extended sentence node
+ * Extends nlcst SentenceNode and adds GLOST-specific properties
+ */
+export type GLOSTSentence = Omit<NlcstSentence, "children"> & {
+    /** Language of the sentence */
+    lang: LanguageCode;
+    /** Script system used */
+    script: ScriptSystem;
+    /** Original text */
+    originalText: string;
+    /** Transcription data for the entire sentence */
+    transcription?: TransliterationData;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras;
+    /** Children nodes - must be nlcst-compliant */
+    children: GLOSTSentenceContent[];
+};
+/**
+ * Extended paragraph node
+ */
+export type GLOSTParagraph = Omit<NlcstParagraph, "children"> & {
+    /** Language of the paragraph */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras;
+    /** Children nodes - must be nlcst-compliant */
+    children: GLOSTParagraphContent[];
+};
+/**
+ * Extended root node
+ */
+export type GLOSTRoot = Omit<NlcstRoot, "children"> & {
+    /** Primary language of the document */
+    lang: LanguageCode;
+    /** Primary script system */
+    script: ScriptSystem;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras;
+    /** Document metadata */
+    metadata?: {
+        title?: string;
+        author?: string;
+        date?: string;
+        description?: string;
+    };
+    /** Children nodes - must be nlcst-compliant */
+    children: GLOSTRootContent[];
+};
+/**
+ * Clause node - represents grammatical clauses within sentences
+ * Created by ClauseSegmenterExtension transformer
+ */
+export type GLOSTClause = {
+    type: "ClauseNode";
+    /** Type of clause */
+    clauseType: "main" | "subordinate" | "relative" | "adverbial";
+    /** Children nodes - phrases, words, or punctuation */
+    children: (GLOSTPhrase | GLOSTWord | GLOSTPunctuation)[];
+    /** Language code for this clause */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras & {
+        /** Whether this clause has been negated */
+        isNegated?: boolean;
+        /** Grammatical mood */
+        mood?: "declarative" | "interrogative" | "imperative" | "conditional";
+        /** Tense information */
+        tense?: string;
+        /** Original form before transformation */
+        originalForm?: string;
+    };
+};
+/**
+ * Phrase node - groups words into grammatical phrases
+ * Created by PhraseSegmenterExtension transformer
+ */
+export type GLOSTPhrase = {
+    type: "PhraseNode";
+    /** Type of phrase */
+    phraseType: "noun" | "verb" | "prepositional" | "adjectival" | "adverbial";
+    /** Main word of the phrase (head) */
+    headWord?: string;
+    /** Children nodes - words or punctuation */
+    children: (GLOSTWord | GLOSTPunctuation)[];
+    /** Language code for this phrase */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras & {
+        /** Grammatical role in the clause/sentence */
+        role?: "subject" | "object" | "complement" | "modifier";
+    };
+};
+/**
+ * Syllable node - represents phonological syllable structure
+ * Created by SyllableSegmenterExtension transformer (language-specific)
+ */
+export type GLOSTSyllable = {
+    type: "SyllableNode";
+    /** Syllable structure information */
+    structure: {
+        /** Initial consonant(s) - Generic (e.g., "h" in "hello") */
+        onset?: string;
+        /** Vowel - Generic (e.g., "e" in "hello") */
+        nucleus: string;
+        /** Final consonant(s) - Generic (e.g., "l" in "hello") */
+        coda?: string;
+        /** Initial consonant (Thai: พยัญชนะต้น) */
+        Ci?: string;
+        /** Vowel (Thai: สระ) */
+        V?: string;
+        /** Final consonant (Thai: ตัวสะกด) */
+        Cf?: string;
+        /** Tone mark (Thai: วรรณยุกต์) */
+        T?: string;
+    };
+    /** Children nodes - individual characters */
+    children: GLOSTCharacter[];
+    /** Language code for this syllable */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Tone number (for tonal languages like Thai, Mandarin) */
+    tone?: number;
+    /** Stress level (for stress languages like English) */
+    stress?: "primary" | "secondary" | "unstressed";
+    /** Extras field for extensions */
+    extras?: GLOSTExtras;
+};
+/**
+ * Character node - represents individual characters with linguistic roles
+ * Created by SyllableSegmenterExtension or CharacterSegmenterExtension
+ */
+export type GLOSTCharacter = {
+    type: "CharacterNode";
+    /** The character value (single character) */
+    value: string;
+    /** Linguistic role of the character */
+    role?: "consonant" | "vowel" | "tone" | "diacritic" | "modifier";
+    /** Placement in the syllable/word (renamed from 'position' to avoid conflict with unist Position) */
+    placement?: "initial" | "medial" | "final" | "above" | "below" | "before" | "after";
+    /** Language code for this character */
+    lang?: LanguageCode;
+    /** Script system used */
+    script?: ScriptSystem;
+    /** Extras field for extensions */
+    extras?: GLOSTExtras & {
+        /** Unicode code point (e.g., "U+0E04") */
+        unicode?: string;
+        /** Thai consonant class (high/mid/low) */
+        class?: "high" | "mid" | "low";
+        /** Phonological sound class */
+        soundClass?: string;
+    };
+};
+export type GLOSTRootContent = GLOSTParagraph | GLOSTSentence | GLOSTWord | GLOSTText | GLOSTSymbol | GLOSTPunctuation | GLOSTWhiteSpace | GLOSTSource;
+export type GLOSTParagraphContent = GLOSTSentence | GLOSTPunctuation | GLOSTSymbol | GLOSTWhiteSpace | GLOSTSource;
+export type GLOSTSentenceContent = GLOSTClause | GLOSTWord | GLOSTPunctuation | GLOSTSymbol | GLOSTWhiteSpace | GLOSTSource;
+export type GLOSTWordContent = GLOSTSyllable | GLOSTText | GLOSTSymbol | GLOSTPunctuation | GLOSTWhiteSpace | GLOSTSource;
+//# sourceMappingURL=types.d.ts.map

package/dist/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,OAAO,IAAI,YAAY,EAAqB,SAAS,IAAI,cAAc,EAAyB,WAAW,IAAI,gBAAgB,EAAE,IAAI,IAAI,SAAS,EAAmC,QAAQ,IAAI,aAAa,EAAE,MAAM,IAAI,WAAW,EAAE,MAAM,IAAI,WAAW,EAAE,IAAI,IAAI,SAAS,EAAE,UAAU,IAAI,eAAe,EAAE,IAAI,IAAI,SAAS,EAAE,MAAM,OAAO,CAAC;AAU/V;;GAEG;AACH,MAAM,MAAM,eAAe,GACvB,WAAW,GACX,UAAU,GACV,MAAM,GACN,QAAQ,GACR,UAAU,GACV,WAAW,CAAC;AAEhB;;GAEG;AACH,MAAM,MAAM,oBAAoB,GAC5B,QAAQ,GACR,UAAU,GACV,YAAY,GACZ,UAAU,GACV,WAAW,CAAC;AAEhB;;GAEG;AACH,MAAM,MAAM,mBAAmB,GAC3B,MAAM,GACN,KAAK,GACL,SAAS,GACT,QAAQ,GACR,UAAU,GACV,KAAK,GACL,QAAQ,GACR,QAAQ,GACR,MAAM,CAAC;AAEX;;;;GAIG;AACH,MAAM,MAAM,YAAY,GAEpB,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,OAAO,GACP,IAAI,GAGJ,OAAO,GACP,IAAI,GAGJ,MAAM,CAAC;AAEX;;GAEG;AACH,MAAM,MAAM,YAAY,GACpB,MAAM,GACN,UAAU,GACV,UAAU,GACV,OAAO,GACP,OAAO,GACP,QAAQ,GACR,OAAO,GACP,OAAO,GACP,MAAM,CAAC;AAMX;;GAEG;AACH,MAAM,MAAM,iBAAiB,GAAG;IAC9B,2BAA2B;IAC3B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,wBAAwB;IACxB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,4BAA4B;IAC5B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,2BAA2B;IAC3B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,0BAA0B;IAC1B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,0BAA0B;IAC1B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,0BAA0B;IAC1B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,2BAA2B;IAC3B,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,IAAI,CAAC,EAAE,MAAM,CAAC;IAEd,uDAAuD;IACvD,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAAC;CACpC,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,gBAAgB,GAAG;IAC7B,+CAA+C;IAC/C,YAAY,CAAC,EAAE,iBAAiB,CAAC;IACjC,oCAAoC;IACpC,UAAU,CAAC,EAAE,UAAU,GAAG,cAAc,GAAG,UAAU,CAAC;IACtD,gCAAgC;IAChC,SAAS,CAAC,EAAE,MAAM,GAAG,UAAU,GAAG,QAAQ,GAAG,aAAa,CAAC;IAC3D,qBAAqB;IACrB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,gCAAgC;IAChC,OAAO,CAAC,EAAE,MAAM,EAAE,CAAC;IACnB,wBAAwB;IACxB,QAAQ,CAAC,EAAE,MAAM,EAAE,CAAC;IACpB,wBAAwB;IACxB,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;CACpB,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG;IACxB,yBAAyB;IACzB,YAAY,CAAC,EAAE,iBAAiB,CAAC;IACjC,wBAAwB;IACxB,QAAQ,CAAC,EAAE,gBAAgB,CAAC;IAC5B,wBAAwB;IACxB,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG,CAAC;CACpB,CAAC;AAMF;;GAEG;AACH,MAAM,MAAM,oBAAoB,GAAG;IACjC,mDAAmD;IACnD,IAAI,EAAE,MAAM,CAAC;IACb,yCAAyC;IACzC,OAAO,EAAE,oBAAoB,CAAC;IAC9B,0CAA0C;IAC1C,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,iBAAiB,GAAG;IAC9B,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb,oCAAoC;IACpC,MAAM,EAAE,mBAAmB,CAAC;IAC5B,6BAA6B;IAC7B,QAAQ,CAAC,EAAE,oBAAoB,EAAE,CAAC;IAClC,6CAA6C;IAC7C,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,yBAAyB;IACzB,SAAS,CAAC,EAAE,MAAM,EAAE,CAAC;IACrB,sCAAsC;IACtC,QAAQ,CAAC,EAAE,MAAM,CAAC;CACnB,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,mBAAmB,GAAG;IAChC,iDAAiD;IACjD,CAAC,MAAM,EAAE,MAAM,GAAG,iBAAiB,CAAC;CACrC,CAAC;AAMF;;GAEG;AACH,MAAM,MAAM,kBAAkB,GAAG;IAC/B,kDAAkD;IAClD,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,qBAAqB;IACrB,YAAY,EAAE,MAAM,CAAC;IACrB,kBAAkB;IAClB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,4BAA4B;IAC5B,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,oBAAoB;IACpB,QAAQ,CAAC,EAAE,MAAM,EAAE,CAAC;IACpB,4BAA4B;IAC5B,SAAS,CAAC,EAAE,MAAM,GAAG,QAAQ,GAAG,KAAK,CAAC;IACtC,sBAAsB;IACtB,SAAS,CAAC,EAAE,QAAQ,GAAG,SAAS,GAAG,UAAU,CAAC;IAC9C,4CAA4C;IAC5C,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,kDAAkD;IAClD,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,kDAAkD;IAClD,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,6DAA6D;IAC7D,UAAU,CAAC,EAAE,UAAU,GAAG,cAAc,GAAG,UAAU,CAAC;CACvD,CAAC;AAMF;;GAEG;AACH,MAAM,MAAM,SAAS,GACjB,SAAS,GACT,aAAa,GACb,cAAc,GACd,SAAS,GACT,SAAS,GACT,WAAW,GACX,gBAAgB,GAChB,eAAe,GACf,WAAW,GAEX,WAAW,GACX,WAAW,GACX,aAAa,GACb,cAAc,CAAC;AAEnB;;GAEG;AACH,MAAM,MAAM,YAAY,GAAG,YAAY,GAAG;IACxC,kCAAkC;IAClC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,uCAAuC;IACvC,KAAK,CAAC,EAAE,eAAe,CAAC;IACxB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,CAAC;CACtB,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,gBAAgB,GAAG,gBAAgB,GAAG,EAAE,CAAC;AAErD;;GAEG;AACH,MAAM,MAAM,eAAe,GAAG,eAAe,GAAG,EAAE,CAAC;AAEnD;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,WAAW,GAAG,EAAE,CAAC;AAE3C;;GAEG;AACH,MAAM,MAAM,SAAS,GAAG,SAAS,GAAG,EAEnC,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,WAAW,GAAG,EAEvC,CAAC;AAEF;;;GAGG;AACH,MAAM,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,EAAE,UAAU,CAAC,GAAG;IACpD,yBAAyB;IACzB,aAAa,EAAE,mBAAmB,CAAC;IACnC,0BAA0B;IAC1B,QAAQ,EAAE,kBAAkB,CAAC;IAC7B,kDAAkD;IAClD,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,kDAAkD;IAClD,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,6DAA6D;IAC7D,UAAU,CAAC,EAAE,UAAU,GAAG,cAAc,GAAG,UAAU,CAAC;IACtD,kCAAkC;IAClC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,uCAAuC;IACvC,KAAK,CAAC,EAAE,eAAe,CAAC;IACxB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,2DAA2D;IAC3D,QAAQ,EAAE,gBAAgB,EAAE,CAAC;CAC9B,CAAC;AAEF;;;GAGG;AACH,MAAM,MAAM,aAAa,GAAG,IAAI,CAAC,aAAa,EAAE,UAAU,CAAC,GAAG;IAC5D,+BAA+B;IAC/B,IAAI,EAAE,YAAY,CAAC;IACnB,yBAAyB;IACzB,MAAM,EAAE,YAAY,CAAC;IACrB,oBAAoB;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,iDAAiD;IACjD,aAAa,CAAC,EAAE,mBAAmB,CAAC;IACpC,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,+CAA+C;IAC/C,QAAQ,EAAE,oBAAoB,EAAE,CAAC;CAClC,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,cAAc,GAAG,IAAI,CAAC,cAAc,EAAE,UAAU,CAAC,GAAG;IAC9D,gCAAgC;IAChC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,+CAA+C;IAC/C,QAAQ,EAAE,qBAAqB,EAAE,CAAC;CACnC,CAAC;AAEF;;GAEG;AACH,MAAM,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,EAAE,UAAU,CAAC,GAAG;IACpD,uCAAuC;IACvC,IAAI,EAAE,YAAY,CAAC;IACnB,4BAA4B;IAC5B,MAAM,EAAE,YAAY,CAAC;IACrB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB,wBAAwB;IACxB,QAAQ,CAAC,EAAE;QACT,KAAK,CAAC,EAAE,MAAM,CAAC;QACf,MAAM,CAAC,EAAE,MAAM,CAAC;QAChB,IAAI,CAAC,EAAE,MAAM,CAAC;QACd,WAAW,CAAC,EAAE,MAAM,CAAC;KACtB,CAAC;IACF,+CAA+C;IAC/C,QAAQ,EAAE,gBAAgB,EAAE,CAAC;CAC9B,CAAC;AAMF;;;GAGG;AACH,MAAM,MAAM,WAAW,GAAG;IACxB,IAAI,EAAE,YAAY,CAAC;IACnB,qBAAqB;IACrB,UAAU,EAAE,MAAM,GAAG,aAAa,GAAG,UAAU,GAAG,WAAW,CAAC;IAC9D,sDAAsD;IACtD,QAAQ,EAAE,CAAC,WAAW,GAAG,SAAS,GAAG,gBAAgB,CAAC,EAAE,CAAC;IACzD,oCAAoC;IACpC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,GAAG;QACrB,2CAA2C;QAC3C,SAAS,CAAC,EAAE,OAAO,CAAC;QACpB,uBAAuB;QACvB,IAAI,CAAC,EAAE,aAAa,GAAG,eAAe,GAAG,YAAY,GAAG,aAAa,CAAC;QACtE,wBAAwB;QACxB,KAAK,CAAC,EAAE,MAAM,CAAC;QACf,0CAA0C;QAC1C,YAAY,CAAC,EAAE,MAAM,CAAC;KACvB,CAAC;CACH,CAAC;AAEF;;;GAGG;AACH,MAAM,MAAM,WAAW,GAAG;IACxB,IAAI,EAAE,YAAY,CAAC;IACnB,qBAAqB;IACrB,UAAU,EAAE,MAAM,GAAG,MAAM,GAAG,eAAe,GAAG,YAAY,GAAG,WAAW,CAAC;IAC3E,qCAAqC;IACrC,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,4CAA4C;IAC5C,QAAQ,EAAE,CAAC,SAAS,GAAG,gBAAgB,CAAC,EAAE,CAAC;IAC3C,oCAAoC;IACpC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,GAAG;QACrB,8CAA8C;QAC9C,IAAI,CAAC,EAAE,SAAS,GAAG,QAAQ,GAAG,YAAY,GAAG,UAAU,CAAC;KACzD,CAAC;CACH,CAAC;AAEF;;;GAGG;AACH,MAAM,MAAM,aAAa,GAAG;IAC1B,IAAI,EAAE,cAAc,CAAC;IACrB,qCAAqC;IACrC,SAAS,EAAE;QACT,4DAA4D;QAC5D,KAAK,CAAC,EAAE,MAAM,CAAC;QACf,6CAA6C;QAC7C,OAAO,EAAE,MAAM,CAAC;QAChB,0DAA0D;QAC1D,IAAI,CAAC,EAAE,MAAM,CAAC;QAGd,2CAA2C;QAC3C,EAAE,CAAC,EAAE,MAAM,CAAC;QACZ,wBAAwB;QACxB,CAAC,CAAC,EAAE,MAAM,CAAC;QACX,sCAAsC;QACtC,EAAE,CAAC,EAAE,MAAM,CAAC;QACZ,kCAAkC;QAClC,CAAC,CAAC,EAAE,MAAM,CAAC;KACZ,CAAC;IACF,6CAA6C;IAC7C,QAAQ,EAAE,cAAc,EAAE,CAAC;IAC3B,sCAAsC;IACtC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,4DAA4D;IAC5D,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,uDAAuD;IACvD,MAAM,CAAC,EAAE,SAAS,GAAG,WAAW,GAAG,YAAY,CAAC;IAChD,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,CAAC;CACtB,CAAC;AAEF;;;GAGG;AACH,MAAM,MAAM,cAAc,GAAG;IAC3B,IAAI,EAAE,eAAe,CAAC;IACtB,6CAA6C;IAC7C,KAAK,EAAE,MAAM,CAAC;IACd,uCAAuC;IACvC,IAAI,CAAC,EAAE,WAAW,GAAG,OAAO,GAAG,MAAM,GAAG,WAAW,GAAG,UAAU,CAAC;IACjE,qGAAqG;IACrG,SAAS,CAAC,EAAE,SAAS,GAAG,QAAQ,GAAG,OAAO,GAAG,OAAO,GAAG,OAAO,GAAG,QAAQ,GAAG,OAAO,CAAC;IACpF,uCAAuC;IACvC,IAAI,CAAC,EAAE,YAAY,CAAC;IACpB,yBAAyB;IACzB,MAAM,CAAC,EAAE,YAAY,CAAC;IACtB,kCAAkC;IAClC,MAAM,CAAC,EAAE,WAAW,GAAG;QACrB,0CAA0C;QAC1C,OAAO,CAAC,EAAE,MAAM,CAAC;QACjB,0CAA0C;QAC1C,KAAK,CAAC,EAAE,MAAM,GAAG,KAAK,GAAG,KAAK,CAAC;QAC/B,+BAA+B;QAC/B,UAAU,CAAC,EAAE,MAAM,CAAC;KACrB,CAAC;CACH,CAAC;AAEF,MAAM,MAAM,gBAAgB,GAAG,cAAc,GAAG,aAAa,GAAG,SAAS,GAAG,SAAS,GAAG,WAAW,GAAG,gBAAgB,GAAG,eAAe,GAAG,WAAW,CAAC;AACvJ,MAAM,MAAM,qBAAqB,GAAG,aAAa,GAAG,gBAAgB,GAAG,WAAW,GAAG,eAAe,GAAG,WAAW,CAAC;AACnH,MAAM,MAAM,oBAAoB,GAAG,WAAW,GAAG,SAAS,GAAG,gBAAgB,GAAG,WAAW,GAAG,eAAe,GAAG,WAAW,CAAC;AAC5H,MAAM,MAAM,gBAAgB,GAAG,aAAa,GAAG,SAAS,GAAG,WAAW,GAAG,gBAAgB,GAAG,eAAe,GAAG,WAAW,CAAC"}

package/dist/types.js

@@ -0,0 +1,6 @@
+export {};
+// ============================================================================
+// Utility Types
+// ============================================================================
+// Type guards are now implemented in utils.ts using unist-util-is
+//# sourceMappingURL=types.js.map

package/dist/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":";AAgjBA,+EAA+E;AAC/E,gBAAgB;AAChB,+EAA+E;AAE/E,kEAAkE"}