glost 0.3.0 → 0.5.0

package/package.json

@@ -1,56 +1,44 @@
 {
   "name": "glost",
-  "version": "0.3.0",
-  "description": "GLOST - Glossed Syntax Tree for augmenting text with language learning metadata",
+  "version": "0.5.0",
+  "description": "GLOST - Glossed Syntax Tree for language learning and multilingual text processing",
   "author": "fustilio",
   "license": "MIT",
   "repository": {
     "type": "git",
     "url": "https://github.com/fustilio/glost.git",
-    "directory": "packages/core"
+    "directory": "packages/glost"
   },
   "type": "module",
-  "main": "./src/index.ts",
-  "types": "./src/index.ts",
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
   "exports": {
     ".": {
-      "types": "./src/index.ts",
-      "default": "./src/index.ts"
+      "types": "./dist/index.d.ts",
+      "default": "./dist/index.js"
     },
-    "./nodes": {
-      "types": "./src/nodes.ts",
-      "default": "./src/nodes.ts"
+    "./processor": {
+      "types": "./dist/processor.d.ts",
+      "default": "./dist/processor.js"
     },
-    "./utils": {
-      "types": "./src/utils.ts",
-      "default": "./src/utils.ts"
+    "./registry": {
+      "types": "./dist/registry.d.ts",
+      "default": "./dist/registry.js"
     },
-    "./validators": {
-      "types": "./src/validators.ts",
-      "default": "./src/validators.ts"
-    },
-    "./guards": {
-      "types": "./src/guards.ts",
-      "default": "./src/guards.ts"
+    "./presets": {
+      "types": "./dist/presets.d.ts",
+      "default": "./dist/presets.js"
     }
   },
   "dependencies": {
-    "unist-util-filter": "^5.0.1",
-    "unist-util-find": "^3.0.0",
-    "unist-util-flatmap": "^1.0.0",
-    "unist-util-is": "^6.0.0",
-    "unist-util-map": "^4.0.0",
-    "unist-util-modify-children": "^4.0.0",
-    "unist-util-remove": "^4.0.0",
-    "unist-util-select": "^5.1.0",
-    "unist-util-visit": "^5.0.0",
-    "zod": "^3.23.8",
-    "glost-common": "0.1.3"
+    "glost-core": "0.5.0",
+    "glost-processor": "0.5.0",
+    "glost-presets": "0.5.0",
+    "glost-registry": "0.5.0"
   },
   "devDependencies": {
-    "@types/nlcst": "^2.0.3",
-    "typescript": "^5.8.3",
-    "vitest": "^3.0.5"
+    "@types/node": "^20.0.0",
+    "typescript": "^5.3.3"
   },
   "keywords": [
     "glost",
@@ -60,11 +48,13 @@
     "language-learning",
     "linguistics",
     "syntax-tree",
-    "unist"
+    "unist",
+    "processor",
+    "pipeline",
+    "unified"
   ],
   "scripts": {
     "build": "tsc",
-    "typecheck": "tsc --noEmit",
-    "test": "vitest --passWithNoTests"
+    "typecheck": "tsc --noEmit"
   }
 }

package/src/index.ts

@@ -1,68 +1,126 @@
-// GLOST - Glossed Syntax Tree
-// Extends nlcst for language learning with annotations
-
-export * from "./types";
-export * from "./nodes";
-export * from "./utils";
-export * from "./validators";
-export * from "./guards";
-// export * from './example';
-
-// Re-export utility types
-export type { ParagraphLike } from "./utils";
-
-// Re-export key utilities for transcription components
-export {
-  // Tree traversal
-  getAllWords,
-  getAllSentences,
-  getAllParagraphs,
-  getAllClauses,
-  getAllPhrases,
-  getAllSyllables,
-  getAllCharacters,
-  getWordsFromDocument,
-  getFirstSentence,
-  getWordsFromSentence,
-  getWordsFromParagraph,
-
-  // Node finding
-  findNodesByType,
-  findWordsByLanguage,
-  findWordsByTranscriptionSystem,
-
-  // Type guards (most are now exported via guards.ts)
-  isGLOSTWord,
-  isGLOSTSentence,
-  isGLOSTParagraph,
-  isGLOSTRoot,
-  isGLOSTClause,
-  isGLOSTPhrase,
-  isGLOSTSyllable,
-  isGLOSTCharacter,
-
-  // Word utilities
-  getWordText,
-  getWordTranscription,
-  hasWordTranscription,
-  getWordTranslation,
-  getWordMeaning,
-  getWordPartOfSpeech,
-  getWordDifficulty,
-
-  // Sentence utilities
-  getSentenceTranslation,
-
-  // Content statistics utilities
-  getGLOSTWordCount,
-  adaptParagraphLikeToGLOST,
-
-  // BCP-47 Language utilities
-  parseLanguageTag,
-  getBaseLanguage,
-  areLanguagesCompatible,
-  findBestLanguageMatch,
-  getLanguageFallback,
-  normalizeLanguageTag,
-  isValidLanguageTag,
-} from "./utils";
+/**
+ * GLOST - Glossed Syntax Tree
+ * 
+ * Main package that re-exports all GLOST functionality.
+ * 
+ * @packageDocumentation
+ * 
+ * @example
+ * ```typescript
+ * // Import everything from one place
+ * import { glost, createSimpleDocument, getAllWords } from "glost";
+ * import { languageLearningPreset } from "glost/presets";
+ * 
+ * // Create a document
+ * const document = createSimpleDocument(words, "th", "thai");
+ * 
+ * // Process with plugins
+ * const result = await glost()
+ *   .use(languageLearningPreset)
+ *   .process(document);
+ * ```
+ */
+
+// ============================================================================
+// Core Types
+// ============================================================================
+export type {
+  GLOSTNode,
+  GLOSTRoot,
+  GLOSTParagraph,
+  GLOSTSentence,
+  GLOSTWord,
+  GLOSTText,
+  GLOSTWhiteSpace,
+  GLOSTPunctuation,
+  GLOSTSymbol,
+  GLOSTSource,
+  GLOSTClause,
+  GLOSTPhrase,
+  GLOSTSyllable,
+  GLOSTCharacter,
+  GLOSTExtras,
+  TransliterationData,
+  LinguisticMetadata,
+  LanguageCode,
+  ScriptSystem,
+} from "glost-core";
+
+// ============================================================================
+// Node Factories
+// ============================================================================
+export {
+  createGLOSTRootNode,
+  createGLOSTParagraphNode,
+  createGLOSTSentenceNode,
+  createGLOSTWordNode,
+  createGLOSTTextNode,
+  createGLOSTWhiteSpaceNode,
+  createGLOSTPunctuationNode,
+  createGLOSTSymbolNode,
+  createSimpleDocument,
+  createDocumentFromSentences,
+  createDocumentFromParagraphs,
+  createSentenceFromWords,
+  createParagraphFromSentences,
+  createSimpleWord,
+  NODE_TYPES,
+} from "glost-core";
+
+// ============================================================================
+// Tree Utilities
+// ============================================================================
+export {
+  getAllWords,
+  getAllSentences,
+  getAllParagraphs,
+  getAllClauses,
+  getFirstSentence,
+  getWordsFromSentence,
+  getWordsFromParagraph,
+  findWordsByLanguage,
+  isGLOSTWord,
+  isGLOSTSentence,
+  isGLOSTParagraph,
+  isGLOSTRoot,
+  getWordText,
+  getWordTranscription,
+  getWordTranslation,
+  getSentenceTranslation,
+  getGLOSTWordCount,
+} from "glost-core";
+
+// ============================================================================
+// Processor API
+// ============================================================================
+export { glost, GLOSTProcessor } from "glost-processor";
+export type {
+  FrozenProcessor,
+  Plugin,
+  PluginSpec,
+  Preset,
+  ProcessorOptions,
+  ProcessingResult,
+  ProcessingError,
+  ProcessingWarning,
+  ProcessingStats,
+  BeforeHook,
+  AfterHook,
+  ErrorHook,
+  SkipHook,
+  ProgressHook,
+  ProgressStats,
+} from "glost-processor";
+
+// ============================================================================
+// Plugin Registry
+// ============================================================================
+export { pluginRegistry, PluginRegistry } from "glost-registry";
+export type {
+  PluginMetadata,
+  PluginCategory,
+  PluginCapabilities,
+  PluginQuery,
+  ConflictReport,
+  ValidationResult,
+} from "glost-registry";

package/src/presets.ts

@@ -0,0 +1,18 @@
+/**
+ * Presets exports
+ * 
+ * Re-exports preset configurations for convenient access via glost/presets.
+ */
+
+export {
+  languageLearningPreset,
+  createLanguageLearningPreset,
+  readingAppPreset,
+  createReadingAppPreset,
+  vocabularyBuilderPreset,
+  createVocabularyBuilderPreset,
+  grammarAnalyzerPreset,
+  createGrammarAnalyzerPreset,
+  minimalPreset,
+  createMinimalPreset,
+} from "glost-presets";

package/src/processor.ts

@@ -0,0 +1,24 @@
+/**
+ * Processor exports
+ * 
+ * Re-exports the processor API for convenient access via glost/processor.
+ */
+
+export { glost, GLOSTProcessor } from "glost-processor";
+export type {
+  FrozenProcessor,
+  Plugin,
+  PluginSpec,
+  Preset,
+  ProcessorOptions,
+  ProcessingResult,
+  ProcessingError,
+  ProcessingWarning,
+  ProcessingStats,
+  BeforeHook,
+  AfterHook,
+  ErrorHook,
+  SkipHook,
+  ProgressHook,
+  ProgressStats,
+} from "glost-processor";

package/src/registry.ts

@@ -0,0 +1,23 @@
+/**
+ * Registry exports
+ * 
+ * Re-exports the registry API for convenient access via glost/registry.
+ */
+
+export { pluginRegistry, PluginRegistry, PluginFilter, PluginSorter, PluginDiscovery, PluginValidator } from "glost-registry";
+export type {
+  PluginMetadata,
+  PluginCategory,
+  PluginCapabilities,
+  PluginRequirements,
+  PluginOptionsSchema,
+  PropertySchema,
+  PluginExample,
+  PluginQuery,
+  ConflictReport,
+  PluginConflict,
+  ValidationResult,
+  ValidationError,
+  ValidationWarning,
+  RegistryStatistics,
+} from "glost-registry";

package/tsconfig.json

@@ -1,9 +1,12 @@
 {
-  "extends": "../../tsconfig.json",
+  "extends": "../../tsconfig.base.json",
   "compilerOptions": {
+    "outDir": "./dist",
     "rootDir": "./src",
-    "outDir": "./dist"
+    "declaration": true,
+    "declarationMap": true,
+    "sourceMap": true
   },
   "include": ["src/**/*"],
-  "exclude": ["node_modules", "dist", "**/*.test.ts", "**/*.spec.ts", "src/__tests__"]
+  "exclude": ["node_modules", "dist"]
 }

package/CHANGELOG.md

@@ -1,151 +0,0 @@
-# glost
-
-## 0.3.0
-
-### Minor Changes
-
-- Externalized language-specific helpers and data-dependent extensions into dedicated packages. Language helpers (`createThaiWord`, `createJapaneseWord`) moved to `glost-th` and `glost-ja`. Extensions (frequency, difficulty, POS, gender, clause-segmenter, transcription, translation) now require explicit data providers instead of fallback data. See `MIGRATION_EXTENSIONS.md` for migration guide.
-
-### Patch Changes
-
-- Updated dependencies
-  - glost-common@0.1.3
-
-## 0.2.0
-
-### Minor Changes
-
-- # Externalize Language-Specific Helpers to Dedicated Packages
-
-  ## Breaking Changes
-
-  Language-specific helper functions have been moved from the core `glost` package into dedicated language packages (`glost-th`, `glost-ja`). This keeps the core lightweight, reduces bundle sizes, and allows for independent language support.
-
-  ### Removed from `glost` (Breaking)
-  - `createThaiWord()` → moved to `glost-th` package
-  - `createJapaneseWord()` → moved to `glost-ja` package
-  - `CreateThaiWordOptions` interface → moved to `glost-th`
-  - `CreateJapaneseWordOptions` interface → moved to `glost-ja`
-
-  ### Migration Required
-
-  Install language packages separately:
-
-  ```bash
-  npm install glost-th  # Thai language support
-  npm install glost-ja  # Japanese language support
-  ```
-
-  Update imports in your code:
-
-  ```typescript
-  // Before (v0.1.x)
-  import { createThaiWord, createJapaneseWord } from "glost";
-
-  // After (v0.2.0+)
-  import { createThaiWord } from "glost-th";
-  import { createJapaneseWord } from "glost-ja";
-  ```
-
-  See [MIGRATION.md](../../MIGRATION.md) for complete upgrade guide.
-
-  ## New Packages
-
-  ### `glost-th` - Thai Language Support
-
-  Initial release providing:
-  - `createThaiWord()` helper
-  - Thai transcription provider interfaces
-  - Support for RTGS, Paiboon, Paiboon+, AUA, and IPA schemes
-
-  ### `glost-ja` - Japanese Language Support
-
-  Initial release providing:
-  - `createJapaneseWord()` helper
-  - Japanese transcription provider interfaces
-  - Support for Romaji (Hepburn, Kunrei, Nihon), Furigana, and IPA schemes
-
-  ## Internal Improvements
-  - Fixed circular dependencies in core package (changed `glost/src/*` imports to relative imports)
-  - Reduced core package size by ~30%
-  - Improved build performance
-  - Better modularity and maintainability
-
-  ## What Stays in Core
-
-  All core functionality remains unchanged:
-  - ✅ `createGLOSTWordNode()` - Generic word creation
-  - ✅ `createGLOSTSentenceNode()` - Sentence creation
-  - ✅ `createGLOSTRootNode()` - Document creation
-  - ✅ `createSimpleWord()` - Simple word helper
-  - ✅ All tree traversal utilities
-  - ✅ Type guards and validators
-  - ✅ All core types and interfaces
-
-  ## Benefits
-  - **Smaller bundles** - Import only languages you use
-  - **Independent releases** - Update languages without core changes
-  - **Better organization** - All Thai code in `glost-th`, all Japanese in `glost-ja`
-  - **Easier contributions** - Add new languages without touching core
-  - **Scalability** - Can support 50+ languages without bloating core
-
-### Patch Changes
-
-- Updated dependencies
-  - glost-common@0.1.2
-
-## 0.2.0
-
-### Breaking Changes
-
-- **BREAKING:** Removed `createThaiWord()` - moved to `glost-th` package
-- **BREAKING:** Removed `createJapaneseWord()` - moved to `glost-ja` package
-- **BREAKING:** Removed `CreateThaiWordOptions` interface - moved to `glost-th`
-- **BREAKING:** Removed `CreateJapaneseWordOptions` interface - moved to `glost-ja`
-
-### Migration
-
-Install language packages separately:
-
-```bash
-npm install glost-th  # Thai language support
-npm install glost-ja  # Japanese language support
-```
-
-Update imports:
-
-```typescript
-// Before (v0.1.x)
-import { createThaiWord, createJapaneseWord } from "glost";
-
-// After (v0.2.0+)
-import { createThaiWord } from "glost-th";
-import { createJapaneseWord } from "glost-ja";
-```
-
-See [MIGRATION.md](../../MIGRATION.md) for complete upgrade guide.
-
-### Internal Changes
-
-- Fixed circular dependencies (changed `glost/src/*` imports to relative imports)
-- Reduced package size by ~30%
-- Improved build performance
-
-### What Stays in Core
-
-All core functionality remains unchanged:
-
-- ✅ `createGLOSTWordNode()` - Generic word creation
-- ✅ `createGLOSTSentenceNode()` - Sentence creation
-- ✅ `createGLOSTRootNode()` - Document creation
-- ✅ `createSimpleWord()` - Simple word helper
-- ✅ All tree traversal utilities
-- ✅ Type guards and validators
-
-## 0.1.1
-
-### Patch Changes
-
-- fix names
-- Updated dependencies
-  - glost-common@0.1.1

package/dist/example.d.ts

@@ -1,10 +0,0 @@
-declare const thaiWords: import("../types").GLOSTWord[];
-declare const thaiSentence: import("../types").GLOSTSentence;
-declare const japaneseWords: import("../types").GLOSTWord[];
-declare const japaneseSentence: import("../types").GLOSTSentence;
-declare const thaiParagraph: import("../types").GLOSTParagraph;
-declare const japaneseParagraph: import("../types").GLOSTParagraph;
-declare const document: import("../types").GLOSTRoot;
-export declare function demonstrateUtilities(): void;
-export { thaiWords, japaneseWords, thaiSentence, japaneseSentence, thaiParagraph, japaneseParagraph, document };
-//# sourceMappingURL=example.d.ts.map

package/dist/example.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"example.d.ts","sourceRoot":"","sources":["../src/example.ts"],"names":[],"mappings":"AAiBA,QAAA,MAAM,SAAS,+BAoCd,CAAC;AAEF,QAAA,MAAM,YAAY,iCAKjB,CAAC;AAMF,QAAA,MAAM,aAAa,+BAkClB,CAAC;AAEF,QAAA,MAAM,gBAAgB,iCAKrB,CAAC;AAMF,QAAA,MAAM,aAAa,kCAA+C,CAAC;AACnE,QAAA,MAAM,iBAAiB,kCAAmD,CAAC;AAE3E,QAAA,MAAM,QAAQ,6BAQb,CAAC;AAMF,wBAAgB,oBAAoB,SA4CnC;AAGD,OAAO,EACL,SAAS,EACT,aAAa,EACb,YAAY,EACZ,gBAAgB,EAChB,aAAa,EACb,iBAAiB,EACjB,QAAQ,EACT,CAAC"}

package/dist/example.js

@@ -1,138 +0,0 @@
-// Example usage of the GLOST package
-import { createThaiWord, createJapaneseWord, createSentenceFromWords, createParagraphFromSentences, createDocumentFromParagraphs, getAllWords, getWordTranscription, validateGLOSTTree, getWordText } from './index';
-// ============================================================================
-// Thai Example: "สวัสดีครับ ผมชื่อสมชาย" (Hello, my name is Somchai)
-// ============================================================================
-const thaiWords = [
-    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(thaiWords, 'th', 'thai', 'สวัสดีครับ ผมชื่อสมชาย');
-// ============================================================================
-// Japanese Example: "私の名前は田中です。" (My name is Tanaka)
-// ============================================================================
-const japaneseWords = [
-    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(japaneseWords, 'ja', 'mixed', '私の名前は田中です。');
-// ============================================================================
-// Create Document Structure
-// ============================================================================
-const thaiParagraph = createParagraphFromSentences([thaiSentence]);
-const japaneseParagraph = createParagraphFromSentences([japaneseSentence]);
-const document = createDocumentFromParagraphs([thaiParagraph, japaneseParagraph], 'mixed', 'mixed', {
-    title: 'Multilingual Greeting Examples',
-    description: 'Examples of greetings in Thai and Japanese with transcriptions'
-});
-// ============================================================================
-// Demonstrate Utilities
-// ============================================================================
-export function demonstrateUtilities() {
-    console.log('=== GLOST Package Demo ===\n');
-    // Get all words from the document
-    const allWords = getAllWords(document);
-    console.log(`Total words: ${allWords.length}`);
-    // Show Thai words with RTGS transcriptions
-    const thaiWordsOnly = allWords.filter(word => word.lang === 'th');
-    console.log('\n=== Thai Words ===');
-    thaiWordsOnly.forEach(word => {
-        const rtgs = getWordTranscription(word, 'rtgs');
-        console.log(`${getWordText(word)} → ${rtgs} (${word.metadata.partOfSpeech})`);
-    });
-    // Show Japanese words with romaji
-    const japaneseWordsOnly = allWords.filter(word => word.lang === 'ja');
-    console.log('\n=== Japanese Words ===');
-    japaneseWordsOnly.forEach(word => {
-        const romaji = getWordTranscription(word, 'romaji');
-        console.log(`${getWordText(word)} → ${romaji} (${word.metadata.partOfSpeech})`);
-    });
-    // Validate the tree
-    const validationErrors = validateGLOSTTree(document);
-    if (validationErrors.length === 0) {
-        console.log('\n✅ Document is valid!');
-    }
-    else {
-        console.log('\n❌ Validation errors:');
-        validationErrors.forEach((error) => console.log(`  - ${error}`));
-    }
-    // Show document structure
-    console.log('\n=== Document Structure ===');
-    console.log(`Language: ${document.lang}`);
-    console.log(`Script: ${document.script}`);
-    console.log(`Paragraphs: ${document.children.length}`);
-    console.log(`Sentences: ${document.children.reduce((acc, p) => {
-        if (p.type === 'ParagraphNode') {
-            return acc + (p.children?.length || 0);
-        }
-        return acc;
-    }, 0)}`);
-    console.log(`Words: ${allWords.length}`);
-}
-// Export for use in other files
-export { thaiWords, japaneseWords, thaiSentence, japaneseSentence, thaiParagraph, japaneseParagraph, document };
-//# sourceMappingURL=example.js.map