glost 0.1.1 → 0.3.0

package/dist/validators.d.ts

@@ -275,6 +275,7 @@ export declare const GLOSTWordNodeSchema: z.ZodObject<{
     };
     children: any[];
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -287,7 +288,6 @@ export declare const GLOSTWordNodeSchema: z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     position?: any;
     lang?: string | undefined;
@@ -304,6 +304,7 @@ export declare const GLOSTWordNodeSchema: z.ZodObject<{
         register?: string | undefined;
     };
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -316,7 +317,6 @@ export declare const GLOSTWordNodeSchema: z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     children?: any[] | undefined;
     position?: any;
@@ -384,6 +384,7 @@ export declare const GLOSTSentenceNodeSchema: z.ZodObject<{
     script: string;
     originalText: string;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -396,7 +397,6 @@ export declare const GLOSTSentenceNodeSchema: z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 }, {
     type: "GLOSTSentenceNode";
     lang: string;
@@ -404,6 +404,7 @@ export declare const GLOSTSentenceNodeSchema: z.ZodObject<{
     originalText: string;
     children?: any[] | undefined;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -416,7 +417,6 @@ export declare const GLOSTSentenceNodeSchema: z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 }>;
 /**
  * GLOST paragraph node schema
@@ -594,6 +594,7 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
     };
     children: any[];
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -606,7 +607,6 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     position?: any;
     lang?: string | undefined;
@@ -623,6 +623,7 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
         register?: string | undefined;
     };
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -635,7 +636,6 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     children?: any[] | undefined;
     position?: any;
@@ -699,6 +699,7 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
     script: string;
     originalText: string;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -711,7 +712,6 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 }, {
     type: "GLOSTSentenceNode";
     lang: string;
@@ -719,6 +719,7 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
     originalText: string;
     children?: any[] | undefined;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -731,7 +732,6 @@ export declare const GLOSTNodeSchema: z.ZodUnion<[z.ZodObject<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 }>, z.ZodObject<{
     level: z.ZodOptional<z.ZodEnum<["character", "syllable", "word", "phrase", "sentence", "paragraph"]>>;
     position: z.ZodOptional<z.ZodAny>;
@@ -840,6 +840,7 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
         register?: string | undefined;
     };
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -852,7 +853,6 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     children?: any[] | undefined;
     position?: any;
@@ -865,6 +865,7 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
     originalText: string;
     children?: any[] | undefined;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -877,7 +878,6 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 } | {
     type: "GLOSTParagraphNode";
     lang: string;
@@ -911,6 +911,7 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
     };
     children: any[];
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -923,7 +924,6 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     position?: any;
     lang?: string | undefined;
@@ -935,6 +935,7 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
     script: string;
     originalText: string;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -947,7 +948,6 @@ export declare function parseGLOSTNode(data: unknown): z.SafeParseReturnType<{
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 } | {
     children: any[];
     type: "GLOSTParagraphNode";
@@ -984,6 +984,7 @@ export declare function parseGLOSTWordNode(data: unknown): z.SafeParseReturnType
         register?: string | undefined;
     };
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -996,7 +997,6 @@ export declare function parseGLOSTWordNode(data: unknown): z.SafeParseReturnType
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     children?: any[] | undefined;
     position?: any;
@@ -1015,6 +1015,7 @@ export declare function parseGLOSTWordNode(data: unknown): z.SafeParseReturnType
     };
     children: any[];
     type: "GLOSTWordNode";
+    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     transcription: Record<string, {
         text: string;
         system: string;
@@ -1027,7 +1028,6 @@ export declare function parseGLOSTWordNode(data: unknown): z.SafeParseReturnType
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }>;
-    level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
     value: string;
     position?: any;
     lang?: string | undefined;
@@ -1043,6 +1043,7 @@ export declare function parseGLOSTSentenceNode(data: unknown): z.SafeParseReturn
     originalText: string;
     children?: any[] | undefined;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -1055,7 +1056,6 @@ export declare function parseGLOSTSentenceNode(data: unknown): z.SafeParseReturn
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 }, {
     children: any[];
     type: "GLOSTSentenceNode";
@@ -1063,6 +1063,7 @@ export declare function parseGLOSTSentenceNode(data: unknown): z.SafeParseReturn
     script: string;
     originalText: string;
     position?: any;
+    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     transcription?: Record<string, {
         text: string;
         system: string;
@@ -1075,7 +1076,6 @@ export declare function parseGLOSTSentenceNode(data: unknown): z.SafeParseReturn
         syllables?: string[] | undefined;
         phonetic?: string | undefined;
     }> | undefined;
-    level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
 }>;
 /**
  * Parse and validate GLOST paragraph node with error details
@@ -1354,6 +1354,7 @@ export declare const schemas: {
         };
         children: any[];
         type: "GLOSTWordNode";
+        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         transcription: Record<string, {
             text: string;
             system: string;
@@ -1366,7 +1367,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }>;
-        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         value: string;
         position?: any;
         lang?: string | undefined;
@@ -1383,6 +1383,7 @@ export declare const schemas: {
             register?: string | undefined;
         };
         type: "GLOSTWordNode";
+        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         transcription: Record<string, {
             text: string;
             system: string;
@@ -1395,7 +1396,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }>;
-        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         value: string;
         children?: any[] | undefined;
         position?: any;
@@ -1459,6 +1459,7 @@ export declare const schemas: {
         script: string;
         originalText: string;
         position?: any;
+        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
         transcription?: Record<string, {
             text: string;
             system: string;
@@ -1471,7 +1472,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }> | undefined;
-        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     }, {
         type: "GLOSTSentenceNode";
         lang: string;
@@ -1479,6 +1479,7 @@ export declare const schemas: {
         originalText: string;
         children?: any[] | undefined;
         position?: any;
+        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
         transcription?: Record<string, {
             text: string;
             system: string;
@@ -1491,7 +1492,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }> | undefined;
-        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     }>, z.ZodObject<{
         level: z.ZodOptional<z.ZodEnum<["character", "syllable", "word", "phrase", "sentence", "paragraph"]>>;
         position: z.ZodOptional<z.ZodAny>;
@@ -1658,6 +1658,7 @@ export declare const schemas: {
         };
         children: any[];
         type: "GLOSTWordNode";
+        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         transcription: Record<string, {
             text: string;
             system: string;
@@ -1670,7 +1671,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }>;
-        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         value: string;
         position?: any;
         lang?: string | undefined;
@@ -1687,6 +1687,7 @@ export declare const schemas: {
             register?: string | undefined;
         };
         type: "GLOSTWordNode";
+        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         transcription: Record<string, {
             text: string;
             system: string;
@@ -1699,7 +1700,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }>;
-        level: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph";
         value: string;
         children?: any[] | undefined;
         position?: any;
@@ -1764,6 +1764,7 @@ export declare const schemas: {
         script: string;
         originalText: string;
         position?: any;
+        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
         transcription?: Record<string, {
             text: string;
             system: string;
@@ -1776,7 +1777,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }> | undefined;
-        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     }, {
         type: "GLOSTSentenceNode";
         lang: string;
@@ -1784,6 +1784,7 @@ export declare const schemas: {
         originalText: string;
         children?: any[] | undefined;
         position?: any;
+        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
         transcription?: Record<string, {
             text: string;
             system: string;
@@ -1796,7 +1797,6 @@ export declare const schemas: {
             syllables?: string[] | undefined;
             phonetic?: string | undefined;
         }> | undefined;
-        level?: "character" | "syllable" | "word" | "phrase" | "sentence" | "paragraph" | undefined;
     }>;
     GLOSTParagraphNode: z.ZodObject<{
         level: z.ZodOptional<z.ZodEnum<["character", "syllable", "word", "phrase", "sentence", "paragraph"]>>;

package/package.json

@@ -1,7 +1,14 @@
 {
   "name": "glost",
-  "version": "0.1.1",
+  "version": "0.3.0",
   "description": "GLOST - Glossed Syntax Tree for augmenting text with language learning metadata",
+  "author": "fustilio",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/fustilio/glost.git",
+    "directory": "packages/core"
+  },
   "type": "module",
   "main": "./src/index.ts",
   "types": "./src/index.ts",
@@ -38,7 +45,7 @@
     "unist-util-select": "^5.1.0",
     "unist-util-visit": "^5.0.0",
     "zod": "^3.23.8",
-    "glost-common": "0.1.1"
+    "glost-common": "0.1.3"
   },
   "devDependencies": {
     "@types/nlcst": "^2.0.3",
@@ -52,10 +59,6 @@
     "multilingual",
     "language-learning",
     "linguistics",
-    "thai",
-    "japanese",
-    "chinese",
-    "korean",
     "syntax-tree",
     "unist"
   ],

package/src/__tests__/README.md

@@ -0,0 +1,20 @@
+# Test Fixtures
+
+This directory contains mock data and fixtures for testing purposes only.
+
+**Important:** Files in this directory should **NOT** be imported by production code to avoid bundling test data with the main library.
+
+## Usage in Tests
+
+```typescript
+// In your test files
+import { thaiDocumentWithExtras, japaneseDocumentWithExtras } from '../__fixtures__/mock-data';
+```
+
+## Available Fixtures
+
+- `mock-data.ts` - Sample GLOST documents with enhanced Thai and Japanese examples including:
+  - Word-level annotations with multiple transcription systems
+  - Translations in multiple languages
+  - Cultural notes and metadata
+  - Example sentences and usage patterns

package/src/__tests__/example.test.ts

@@ -0,0 +1,43 @@
+// Example test file showing how to use fixtures
+// This demonstrates that fixtures are only imported in test files
+
+import { describe, it, expect } from 'vitest';
+import {
+  thaiDocumentWithExtras,
+  japaneseDocumentWithExtras,
+  getQuickTranslation,
+  getDifficulty
+} from './mock-data.js';
+import { getAllWords } from '../utils.js';
+
+
+describe('Using mock data fixtures', () => {
+  it('should load Thai document with extras', () => {
+    expect(thaiDocumentWithExtras.type).toBe('RootNode');
+    expect(thaiDocumentWithExtras.lang).toBe('th');
+
+    const words = getAllWords(thaiDocumentWithExtras);
+    expect(words.length).toBeGreaterThan(0);
+  });
+
+  it('should load Japanese document with extras', () => {
+    expect(japaneseDocumentWithExtras.type).toBe('RootNode');
+    expect(japaneseDocumentWithExtras.lang).toBe('ja');
+
+    const words = getAllWords(japaneseDocumentWithExtras);
+    expect(words.length).toBeGreaterThan(0);
+  });
+
+  it('should access translation data from fixtures', () => {
+    const words = getAllWords(thaiDocumentWithExtras);
+    const firstWord = words[0];
+
+    if (firstWord) {
+      const englishTranslation = getQuickTranslation(firstWord, 'en');
+      expect(englishTranslation).toBeDefined();
+
+      const difficulty = getDifficulty(firstWord);
+      expect(difficulty).toBeDefined();
+    }
+  });
+});

package/src/{example.ts → __tests__/example.ts}

@@ -1,7 +1,5 @@
 // Example usage of the GLOST package
 import {
-  createThaiWord,
-  createJapaneseWord,
   createSentenceFromWords,
   createParagraphFromSentences,
   createDocumentFromParagraphs,
@@ -9,48 +7,50 @@ import {
   getWordTranscription,
   validateGLOSTTree,
   getWordText
-} from './index';
+} from '../index';
+import { createThaiWord } from 'glost-th';
+import { createJapaneseWord } from 'glost-ja';
 
 // ============================================================================
 // Thai Example: "สวัสดีครับ ผมชื่อสมชาย" (Hello, my name is Somchai)
 // ============================================================================
 
 const thaiWords = [
-  createThaiWord(
-    'สวัสดี',
-    'sà-wàt-dii',
-    'interjection',
-    2, // tone
-    ['sa', 'wat', 'dii']
-  ),
-  createThaiWord(
-    'ครับ',
-    'khráp',
-    'particle',
-    2,
-    ['khrap']
-  ),
-  createThaiWord(
-    'ผม',
-    'phǒm',
-    'pronoun',
-    3,
-    ['phom']
-  ),
-  createThaiWord(
-    'ชื่อ',
-    'chûue',
-    'noun',
-    3,
-    ['chue']
-  ),
-  createThaiWord(
-    'สมชาย',
-    'sǒm-chaai',
-    'proper noun',
-    3,
-    ['som', 'chaai']
-  )
+  createThaiWord({
+    text: 'สวัสดี',
+    rtgs: 'sà-wàt-dii',
+    partOfSpeech: 'interjection',
+    tone: 2,
+    syllables: ['sa', 'wat', 'dii']
+  }),
+  createThaiWord({
+    text: 'ครับ',
+    rtgs: 'khráp',
+    partOfSpeech: 'particle',
+    tone: 2,
+    syllables: ['khrap']
+  }),
+  createThaiWord({
+    text: 'ผม',
+    rtgs: 'phǒm',
+    partOfSpeech: 'pronoun',
+    tone: 3,
+    syllables: ['phom']
+  }),
+  createThaiWord({
+    text: 'ชื่อ',
+    rtgs: 'chûue',
+    partOfSpeech: 'noun',
+    tone: 3,
+    syllables: ['chue']
+  }),
+  createThaiWord({
+    text: 'สมชาย',
+    rtgs: 'sǒm-chaai',
+    partOfSpeech: 'proper noun',
+    tone: 3,
+    syllables: ['som', 'chaai']
+  })
 ];
 
 const thaiSentence = createSentenceFromWords(
@@ -65,39 +65,39 @@ const thaiSentence = createSentenceFromWords(
 // ============================================================================
 
 const japaneseWords = [
-  createJapaneseWord(
-    '私',
-    'watashi',
-    'pronoun',
-    'わたし'
-  ),
-  createJapaneseWord(
-    'の',
-    'no',
-    'particle'
-  ),
-  createJapaneseWord(
-    '名前',
-    'namae',
-    'noun',
-    'なまえ'
-  ),
-  createJapaneseWord(
-    'は',
-    'wa',
-    'particle'
-  ),
-  createJapaneseWord(
-    '田中',
-    'tanaka',
-    'proper noun',
-    'たなか'
-  ),
-  createJapaneseWord(
-    'です',
-    'desu',
-    'copula'
-  )
+  createJapaneseWord({
+    text: '私',
+    romaji: 'watashi',
+    partOfSpeech: 'pronoun',
+    furigana: 'わたし'
+  }),
+  createJapaneseWord({
+    text: 'の',
+    romaji: 'no',
+    partOfSpeech: 'particle'
+  }),
+  createJapaneseWord({
+    text: '名前',
+    romaji: 'namae',
+    partOfSpeech: 'noun',
+    furigana: 'なまえ'
+  }),
+  createJapaneseWord({
+    text: 'は',
+    romaji: 'wa',
+    partOfSpeech: 'particle'
+  }),
+  createJapaneseWord({
+    text: '田中',
+    romaji: 'tanaka',
+    partOfSpeech: 'proper noun',
+    furigana: 'たなか'
+  }),
+  createJapaneseWord({
+    text: 'です',
+    romaji: 'desu',
+    partOfSpeech: 'copula'
+  })
 ];
 
 const japaneseSentence = createSentenceFromWords(