@okf/ootils 1.29.4 → 1.30.0

package/dist/universal.d.ts

@@ -1367,4 +1367,186 @@ declare const UI_CONTENT: {
  */
 declare const genCleanCamelCaseId: (id: string) => string;
 
-export { BASE_BULLMQ_CONFIG, FILTER_IDS, TEMP_removeDuplicateFilters, UI_CONTENT, _self_managed_buildAnnoHierarchyConfig, _self_managed_buildDocHierarchyConfig, _self_managed_getFixedAnnoRollupBlocks, _self_managed_getFixedAnnoTagBlock, autoGenFilterConfigsFromTpl, buildFilterConfigurations, compareAndGroupBlocks, deleteVal, extractAllBlocksFromTpl, extractAndOrganizeBlocks, genCleanCamelCaseId, genTagId, generateFilterKey, getFilterKeyForBlock, getPlatformContextContent, getRollupPossibilities, getRoutePathToContentTypeLanding, getRoutePathToEditContent, getRoutePathToModerateContent, getRoutePathToMyContent, getRoutePathToPublishedContent, getRoutePathToReviewDashboard, getRoutePathToTCI, getRoutePathToTagCategoryLanding, getVal, mergeAnnoDataIntoAnnotationsTags, parseSpecialConfigSyntax, processAuthorAndCommonFilters, _recursExtractBlocks as recursivelyExtractBlocks, segrigateDocs, setVal, toArray };
+/**
+ * BlockDef — single source of truth for a tplBlock's cross-cutting configuration.
+ *
+ * Each block type (e.g. LexicalTextEditor, TextInput) declares its capabilities,
+ * schema types, field paths, validation, and translation config in one place.
+ * Consumers across all repos query the registry instead of maintaining scattered arrays/switches.
+ */
+interface BlockDef {
+    compName: string;
+    category: 'text' | 'selection' | 'tags' | 'date' | 'number' | 'media' | 'structural' | 'special';
+    qualQuant: 'qual' | 'quant' | null;
+    /** Mongoose schema type — the actual value returned to compToTypeMap (e.g. { type: Object }) */
+    mongoSchemaType: Record<string, any>;
+    /** Elasticsearch mapping shape — used directly by generateMappingsFromTpl */
+    esMapping: Record<string, any> | null;
+    capabilities: BlockCapabilities;
+    /** Sub-paths within the block's value. null = value itself is used directly. */
+    fieldPaths: {
+        /** Path to get the plain text string (e.g. 'allText'). null = value itself is the string. */
+        plainTextString?: string | null;
+        /** Path appended to valuePath for ES/listing search (e.g. 'allText'). null = valuePath used directly. */
+        searchField?: string | null;
+        /** Path to get display value for table/card rendering (e.g. 'allText'). null = value itself. */
+        displayValue?: string | null;
+    };
+    validation: {
+        /** Name of the "is populated" validator fn (resolved by consumer) */
+        populatedCheckFn: string;
+        /** Name of form validation fn, if different from populatedCheck */
+        formValidationFn?: string;
+    };
+    translation?: {
+        /** Handler type for auto-translation (e.g. 'lexical', 'text', 'tags', 'rte') */
+        handlerType: string;
+    };
+    tableCell?: {
+        /** The cell component to render in table view (e.g. 'RichTextAsPlainTextLex') */
+        cellComp: string;
+        /** Sub-path appended to valuePath for backend sort (e.g. 'editorState.root.children.0.children.0.text') */
+        sortPathSuffix?: string;
+        /** Extra props to pass to the cell component */
+        cellProps?: Record<string, any>;
+    };
+    csvExport?: {
+        /** Transform function name for CSV export */
+        transformFn: string;
+    };
+    slackFormat?: {
+        /** Handler function name for Slack message formatting */
+        handlerFn: string;
+    };
+    batchImport?: {
+        /** Function name that transforms raw import value into the block's expected shape */
+        valueInjectorFn: string;
+    };
+    /** How this block renders as a selectable option in TCI template builder and direct import UI */
+    contentBlockOption?: {
+        display: string;
+        icon?: string;
+        iconWeight?: string;
+        displayInDirectImportUI?: string;
+    };
+    /** Full chunking config used by okf-sub CreateChunksHandler */
+    chunkingConfig?: {
+        strategy: string;
+        [key: string]: any;
+    };
+}
+interface BlockCapabilities {
+    /** Block's value contains extractable plain text (even if value also has other structure like JSON) */
+    hasPlainText: boolean;
+    /** Block's value is in the Lexical shape — editorState, allText, etc. */
+    hasLexicalShape: boolean;
+    /**
+     * General annotation flag — this block type supports annotation
+     * (human, human-in-the-loop, or AI). Used by general checks: should the
+     * annotation UI render, should we scan this field for anno data, should we
+     * queue an annos rebuild, should this field appear in annotation explorer, etc.
+     */
+    annotation: boolean;
+    /**
+     * AI auto-annotation flag — AI auto-annotation and human-in-the-loop
+     * annotation pipelines should process this block type. Subset of `annotation`
+     * — only meaningful when annotation is also true. Used by AI suggestion services
+     * and auto-annotation pipelines.
+     */
+    aiAnnotation: boolean;
+    /** Supports AI enrichment — categorization, sentiment analysis, NER */
+    aiEnrichment: boolean;
+    /** Treated as qualitative in AI Chat? */
+    aiChatQualField: boolean;
+    /** Treated as quantitative in AI Chat? */
+    aiChatQuantField: boolean;
+    /** Can be searched via ES/listing search? */
+    searchable: boolean;
+    /** Supported in direct data import? */
+    directDataImport: boolean;
+    /** Has CSV export transform? */
+    csvExport: boolean;
+    /** Supports auto-translation? */
+    translatable: boolean;
+    /** Included in document summarizer? */
+    documentSummarizer: boolean;
+    /**
+     * Strip this block's value from `main` when syncing to chunks/annos collections.
+     * Used for large-payload blocks (like Lexical) to avoid duplicating their full text
+     * in chunk/anno metadata — the actual text already lives in the chunks/annos themselves.
+     */
+    stripFromMainOnAnnoChunkSync: boolean;
+}
+
+/**
+ * Shared schema presets for mongoSchemaType and esMapping.
+ *
+ * RULE: All mongo schema types and ES mappings MUST be defined here as presets.
+ * Block defs reference these — never define schema shapes inline in a block def.
+ */
+declare const MONGO_SCHEMA_PRESETS: {
+    readonly object: {
+        readonly type: ObjectConstructor;
+    };
+    readonly string: {
+        readonly type: StringConstructor;
+    };
+};
+declare const ELASTIC_MAPPING_PRESETS: {
+    readonly largeText: {
+        readonly properties: {
+            readonly allText: {
+                readonly type: "text";
+                readonly analyzer: "LargeTextAnalyzer";
+            };
+        };
+    };
+};
+declare const CHUNKING_PRESETS: {
+    readonly lexicalSemantic: {
+        readonly strategy: "semanticChunking";
+        readonly windowSize: 3;
+        readonly minSimilarityScore: 0.7;
+    };
+    readonly simpleText: {
+        readonly strategy: "simpleChunking";
+    };
+};
+
+declare class BlockRegistry {
+    private blocks;
+    constructor();
+    /** Register a block descriptor. */
+    register(descriptor: BlockDef): void;
+    /** Get the full descriptor for a block type. Returns undefined if not registered. */
+    getBlock(compType: string): BlockDef | undefined;
+    /** Check if a block type is registered in the registry. */
+    isRegistered(compType: string): boolean;
+    /**
+     * Get all registered block descriptors that have a given capability set to a truthy value.
+     * Optionally pass a specific value to match (e.g. for enum-style capabilities).
+     */
+    getBlocksByCapability(capability: keyof BlockCapabilities, value?: boolean | string): BlockDef[];
+    /**
+     * Get compType strings for all registered blocks with a given capability.
+     * Replaces scattered hardcoded arrays like:
+     *   const TEXT_FIELD_COMPONENTS = ["TextInput", "LexicalTextEditor", ...]
+     * becomes:
+     *   const TEXT_FIELD_COMPONENTS = blockRegistry.getComps('aiTextExtraction')
+     */
+    getComps(capability: keyof BlockCapabilities, value?: boolean | string): string[];
+    /** Get all registered blocks in a given category. */
+    getBlocksByCategory(category: BlockDef['category']): BlockDef[];
+    /** Get compType strings for all qual blocks. */
+    getQualBlocks(): string[];
+    /** Get compType strings for all quant blocks. */
+    getQuantBlocks(): string[];
+    /** Check if a specific block has a specific capability. */
+    hasCapability(compType: string, capability: keyof BlockCapabilities): boolean;
+    /** Get all registered block descriptors. */
+    getAll(): BlockDef[];
+}
+/** Singleton instance — the one registry shared across the app. */
+declare const blockRegistry: BlockRegistry;
+
+export { BASE_BULLMQ_CONFIG, type BlockCapabilities, type BlockDef, BlockRegistry, CHUNKING_PRESETS, ELASTIC_MAPPING_PRESETS, FILTER_IDS, MONGO_SCHEMA_PRESETS, TEMP_removeDuplicateFilters, UI_CONTENT, _self_managed_buildAnnoHierarchyConfig, _self_managed_buildDocHierarchyConfig, _self_managed_getFixedAnnoRollupBlocks, _self_managed_getFixedAnnoTagBlock, autoGenFilterConfigsFromTpl, blockRegistry, buildFilterConfigurations, compareAndGroupBlocks, deleteVal, extractAllBlocksFromTpl, extractAndOrganizeBlocks, genCleanCamelCaseId, genTagId, generateFilterKey, getFilterKeyForBlock, getPlatformContextContent, getRollupPossibilities, getRoutePathToContentTypeLanding, getRoutePathToEditContent, getRoutePathToModerateContent, getRoutePathToMyContent, getRoutePathToPublishedContent, getRoutePathToReviewDashboard, getRoutePathToTCI, getRoutePathToTagCategoryLanding, getVal, mergeAnnoDataIntoAnnotationsTags, parseSpecialConfigSyntax, processAuthorAndCommonFilters, _recursExtractBlocks as recursivelyExtractBlocks, segrigateDocs, setVal, toArray };

package/dist/universal.js

@@ -21,7 +21,11 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var universal_exports = {};
 __export(universal_exports, {
   BASE_BULLMQ_CONFIG: () => BASE_BULLMQ_CONFIG,
+  BlockRegistry: () => BlockRegistry,
+  CHUNKING_PRESETS: () => CHUNKING_PRESETS,
+  ELASTIC_MAPPING_PRESETS: () => ELASTIC_MAPPING_PRESETS,
   FILTER_IDS: () => FILTER_IDS,
+  MONGO_SCHEMA_PRESETS: () => MONGO_SCHEMA_PRESETS,
   TEMP_removeDuplicateFilters: () => TEMP_removeDuplicateFilters,
   UI_CONTENT: () => UI_CONTENT,
   _self_managed_buildAnnoHierarchyConfig: () => _self_managed_buildAnnoHierarchyConfig,
@@ -29,6 +33,7 @@ __export(universal_exports, {
   _self_managed_getFixedAnnoRollupBlocks: () => _self_managed_getFixedAnnoRollupBlocks,
   _self_managed_getFixedAnnoTagBlock: () => _self_managed_getFixedAnnoTagBlock,
   autoGenFilterConfigsFromTpl: () => autoGenFilterConfigsFromTpl,
+  blockRegistry: () => blockRegistry,
   buildFilterConfigurations: () => buildFilterConfigurations,
   compareAndGroupBlocks: () => compareAndGroupBlocks,
   deleteVal: () => deleteVal,
@@ -2223,10 +2228,172 @@ var genCleanCamelCaseId = (id) => {
   if (/^\d/.test(result)) result = "a" + result;
   return result.slice(0, MAX_LENGTH);
 };
+
+// src/blockRegistry/schemaPresets.ts
+var MONGO_SCHEMA_PRESETS = {
+  object: { type: Object },
+  string: { type: String }
+};
+var ELASTIC_MAPPING_PRESETS = {
+  largeText: {
+    properties: {
+      allText: {
+        type: "text",
+        analyzer: "LargeTextAnalyzer"
+      }
+    }
+  }
+};
+var CHUNKING_PRESETS = {
+  // Lexical-shaped text — uses semantic chunking on allText
+  lexicalSemantic: {
+    strategy: "semanticChunking",
+    windowSize: 3,
+    minSimilarityScore: 0.7
+  },
+  // Plain text input — single chunk per field
+  simpleText: {
+    strategy: "simpleChunking"
+  }
+};
+
+// src/blockRegistry/blocks/LexicalTextEditor.ts
+var LexicalTextEditor = {
+  compName: "LexicalTextEditor",
+  // Identity
+  category: "text",
+  qualQuant: "qual",
+  // Schema
+  mongoSchemaType: MONGO_SCHEMA_PRESETS.object,
+  esMapping: ELASTIC_MAPPING_PRESETS.largeText,
+  // Capabilities
+  capabilities: {
+    hasPlainText: true,
+    hasLexicalShape: true,
+    annotation: true,
+    aiAnnotation: true,
+    aiEnrichment: true,
+    aiChatQualField: true,
+    aiChatQuantField: false,
+    searchable: true,
+    directDataImport: true,
+    csvExport: true,
+    translatable: true,
+    documentSummarizer: true,
+    stripFromMainOnAnnoChunkSync: true
+  },
+  // Field paths
+  fieldPaths: {
+    plainTextString: "allText",
+    searchField: "allText",
+    displayValue: "allText"
+  },
+  // Validation
+  validation: {
+    populatedCheckFn: "lexicalTextEditorHasValue",
+    formValidationFn: "lexicalTextEditorHasValue"
+  },
+  // Translation
+  translation: {
+    handlerType: "LexicalBlockHandler"
+  },
+  // Table rendering
+  tableCell: {
+    cellComp: "RichTextAsPlainTextLex",
+    sortPathSuffix: "editorState.root.children.0.children.0.text"
+  },
+  // CSV export
+  csvExport: {
+    transformFn: "KPRichLexicalEditor"
+  },
+  // Slack
+  slackFormat: {
+    handlerFn: "lexicalRichText"
+  },
+  // Batch import
+  batchImport: {
+    valueInjectorFn: "toLexicalValue"
+  },
+  // Content block option — TCI template builder & direct import UI
+  contentBlockOption: {
+    display: "Rich Text Field",
+    icon: "TextAa"
+  },
+  // Chunking config — used by okf-sub CreateChunksHandler
+  chunkingConfig: CHUNKING_PRESETS.lexicalSemantic
+};
+
+// src/blockRegistry/registry.ts
+var BlockRegistry = class {
+  constructor() {
+    this.blocks = /* @__PURE__ */ new Map();
+    this.register(LexicalTextEditor);
+  }
+  /** Register a block descriptor. */
+  register(descriptor) {
+    this.blocks.set(descriptor.compName, descriptor);
+  }
+  /** Get the full descriptor for a block type. Returns undefined if not registered. */
+  getBlock(compType) {
+    return this.blocks.get(compType);
+  }
+  /** Check if a block type is registered in the registry. */
+  isRegistered(compType) {
+    return this.blocks.has(compType);
+  }
+  /**
+   * Get all registered block descriptors that have a given capability set to a truthy value.
+   * Optionally pass a specific value to match (e.g. for enum-style capabilities).
+   */
+  getBlocksByCapability(capability, value = true) {
+    return Array.from(this.blocks.values()).filter((b) => {
+      const cap = b.capabilities[capability];
+      if (value === true) return !!cap;
+      return cap === value;
+    });
+  }
+  /**
+   * Get compType strings for all registered blocks with a given capability.
+   * Replaces scattered hardcoded arrays like:
+   *   const TEXT_FIELD_COMPONENTS = ["TextInput", "LexicalTextEditor", ...]
+   * becomes:
+   *   const TEXT_FIELD_COMPONENTS = blockRegistry.getComps('aiTextExtraction')
+   */
+  getComps(capability, value = true) {
+    return this.getBlocksByCapability(capability, value).map((b) => b.compName);
+  }
+  /** Get all registered blocks in a given category. */
+  getBlocksByCategory(category) {
+    return Array.from(this.blocks.values()).filter((b) => b.category === category);
+  }
+  /** Get compType strings for all qual blocks. */
+  getQualBlocks() {
+    return Array.from(this.blocks.values()).filter((b) => b.qualQuant === "qual").map((b) => b.compName);
+  }
+  /** Get compType strings for all quant blocks. */
+  getQuantBlocks() {
+    return Array.from(this.blocks.values()).filter((b) => b.qualQuant === "quant").map((b) => b.compName);
+  }
+  /** Check if a specific block has a specific capability. */
+  hasCapability(compType, capability) {
+    const block = this.blocks.get(compType);
+    if (!block) return false;
+    return !!block.capabilities[capability];
+  }
+  /** Get all registered block descriptors. */
+  getAll() {
+    return Array.from(this.blocks.values());
+  }
+};
+var blockRegistry = new BlockRegistry();
 // Annotate the CommonJS export names for ESM import in node:
 0 && (module.exports = {
   BASE_BULLMQ_CONFIG,
+  BlockRegistry,
+  CHUNKING_PRESETS,
+  ELASTIC_MAPPING_PRESETS,
   FILTER_IDS,
+  MONGO_SCHEMA_PRESETS,
   TEMP_removeDuplicateFilters,
   UI_CONTENT,
   _self_managed_buildAnnoHierarchyConfig,
@@ -2234,6 +2401,7 @@ var genCleanCamelCaseId = (id) => {
   _self_managed_getFixedAnnoRollupBlocks,
   _self_managed_getFixedAnnoTagBlock,
   autoGenFilterConfigsFromTpl,
+  blockRegistry,
   buildFilterConfigurations,
   compareAndGroupBlocks,
   deleteVal,

package/dist/universal.mjs

@@ -2162,9 +2162,171 @@ var genCleanCamelCaseId = (id) => {
   if (/^\d/.test(result)) result = "a" + result;
   return result.slice(0, MAX_LENGTH);
 };
+
+// src/blockRegistry/schemaPresets.ts
+var MONGO_SCHEMA_PRESETS = {
+  object: { type: Object },
+  string: { type: String }
+};
+var ELASTIC_MAPPING_PRESETS = {
+  largeText: {
+    properties: {
+      allText: {
+        type: "text",
+        analyzer: "LargeTextAnalyzer"
+      }
+    }
+  }
+};
+var CHUNKING_PRESETS = {
+  // Lexical-shaped text — uses semantic chunking on allText
+  lexicalSemantic: {
+    strategy: "semanticChunking",
+    windowSize: 3,
+    minSimilarityScore: 0.7
+  },
+  // Plain text input — single chunk per field
+  simpleText: {
+    strategy: "simpleChunking"
+  }
+};
+
+// src/blockRegistry/blocks/LexicalTextEditor.ts
+var LexicalTextEditor = {
+  compName: "LexicalTextEditor",
+  // Identity
+  category: "text",
+  qualQuant: "qual",
+  // Schema
+  mongoSchemaType: MONGO_SCHEMA_PRESETS.object,
+  esMapping: ELASTIC_MAPPING_PRESETS.largeText,
+  // Capabilities
+  capabilities: {
+    hasPlainText: true,
+    hasLexicalShape: true,
+    annotation: true,
+    aiAnnotation: true,
+    aiEnrichment: true,
+    aiChatQualField: true,
+    aiChatQuantField: false,
+    searchable: true,
+    directDataImport: true,
+    csvExport: true,
+    translatable: true,
+    documentSummarizer: true,
+    stripFromMainOnAnnoChunkSync: true
+  },
+  // Field paths
+  fieldPaths: {
+    plainTextString: "allText",
+    searchField: "allText",
+    displayValue: "allText"
+  },
+  // Validation
+  validation: {
+    populatedCheckFn: "lexicalTextEditorHasValue",
+    formValidationFn: "lexicalTextEditorHasValue"
+  },
+  // Translation
+  translation: {
+    handlerType: "LexicalBlockHandler"
+  },
+  // Table rendering
+  tableCell: {
+    cellComp: "RichTextAsPlainTextLex",
+    sortPathSuffix: "editorState.root.children.0.children.0.text"
+  },
+  // CSV export
+  csvExport: {
+    transformFn: "KPRichLexicalEditor"
+  },
+  // Slack
+  slackFormat: {
+    handlerFn: "lexicalRichText"
+  },
+  // Batch import
+  batchImport: {
+    valueInjectorFn: "toLexicalValue"
+  },
+  // Content block option — TCI template builder & direct import UI
+  contentBlockOption: {
+    display: "Rich Text Field",
+    icon: "TextAa"
+  },
+  // Chunking config — used by okf-sub CreateChunksHandler
+  chunkingConfig: CHUNKING_PRESETS.lexicalSemantic
+};
+
+// src/blockRegistry/registry.ts
+var BlockRegistry = class {
+  constructor() {
+    this.blocks = /* @__PURE__ */ new Map();
+    this.register(LexicalTextEditor);
+  }
+  /** Register a block descriptor. */
+  register(descriptor) {
+    this.blocks.set(descriptor.compName, descriptor);
+  }
+  /** Get the full descriptor for a block type. Returns undefined if not registered. */
+  getBlock(compType) {
+    return this.blocks.get(compType);
+  }
+  /** Check if a block type is registered in the registry. */
+  isRegistered(compType) {
+    return this.blocks.has(compType);
+  }
+  /**
+   * Get all registered block descriptors that have a given capability set to a truthy value.
+   * Optionally pass a specific value to match (e.g. for enum-style capabilities).
+   */
+  getBlocksByCapability(capability, value = true) {
+    return Array.from(this.blocks.values()).filter((b) => {
+      const cap = b.capabilities[capability];
+      if (value === true) return !!cap;
+      return cap === value;
+    });
+  }
+  /**
+   * Get compType strings for all registered blocks with a given capability.
+   * Replaces scattered hardcoded arrays like:
+   *   const TEXT_FIELD_COMPONENTS = ["TextInput", "LexicalTextEditor", ...]
+   * becomes:
+   *   const TEXT_FIELD_COMPONENTS = blockRegistry.getComps('aiTextExtraction')
+   */
+  getComps(capability, value = true) {
+    return this.getBlocksByCapability(capability, value).map((b) => b.compName);
+  }
+  /** Get all registered blocks in a given category. */
+  getBlocksByCategory(category) {
+    return Array.from(this.blocks.values()).filter((b) => b.category === category);
+  }
+  /** Get compType strings for all qual blocks. */
+  getQualBlocks() {
+    return Array.from(this.blocks.values()).filter((b) => b.qualQuant === "qual").map((b) => b.compName);
+  }
+  /** Get compType strings for all quant blocks. */
+  getQuantBlocks() {
+    return Array.from(this.blocks.values()).filter((b) => b.qualQuant === "quant").map((b) => b.compName);
+  }
+  /** Check if a specific block has a specific capability. */
+  hasCapability(compType, capability) {
+    const block = this.blocks.get(compType);
+    if (!block) return false;
+    return !!block.capabilities[capability];
+  }
+  /** Get all registered block descriptors. */
+  getAll() {
+    return Array.from(this.blocks.values());
+  }
+};
+var blockRegistry = new BlockRegistry();
 export {
   BASE_BULLMQ_CONFIG,
+  BlockRegistry,
+  CHUNKING_PRESETS,
+  ELASTIC_MAPPING_PRESETS,
   FILTER_IDS,
+  MONGO_SCHEMA_PRESETS,
   TEMP_removeDuplicateFilters,
   UI_CONTENT,
   _self_managed_buildAnnoHierarchyConfig,
@@ -2172,6 +2334,7 @@ export {
   _self_managed_getFixedAnnoRollupBlocks,
   _self_managed_getFixedAnnoTagBlock,
   autoGenFilterConfigsFromTpl,
+  blockRegistry,
   buildFilterConfigurations,
   compareAndGroupBlocks,
   deleteVal,

package/package.json

@@ -3,7 +3,7 @@
   "publishConfig": {
     "access": "public"
   },
-  "version": "1.29.4",
+  "version": "1.30.0",
   "description": "Utility functions for both browser and Node.js",
   "main": "dist/index.js",
   "module": "dist/index.mjs",