archy

/** * Entity Extractor Utilities * * This file contains utility functions for extracting entities and relationships * from text descriptions, which are used for generating diagrams. */ /** * Extract entities from a text description. * * This method analyzes the text description to identify potential entities * that can be used in diagrams. * * @param {string} description - The text description to analyze * @returns {string[]} An array of identified entities */ export function extractEntities(description: string): string[] { // Simple entity extraction based on capitalized words // In a real implementation, this would use NLP techniques const words = description.split(/\s+/); const entities = new Set<string>(); // Look for capitalized words that might be entities for (const word of words) { // Check if the word starts with a capital letter and is not at the beginning of a sentence if (/^[A-Z][a-z]+/.test(word) && !word.includes('.')) { // Remove any punctuation const cleanWord = word.replace(/[.,;:!?]$/, ''); if (cleanWord.length > 1) { entities.add(cleanWord); } } } // Look for common entity keywords const entityKeywords = ['class', 'entity', 'component', 'system', 'user', 'database', 'service']; const regex = new RegExp(`\\b(${entityKeywords.join('|')})\\s+([A-Za-z0-9_]+)\\b`, 'gi'); let match; while ((match = regex.exec(description)) !== null) { entities.add(match[2]); } return Array.from(entities); } /** * Extract relationships between entities from a text description. * * This method analyzes the text description to identify potential relationships * between the identified entities. * * @param {string} description - The text description to analyze * @param {string[]} entities - The identified entities * @returns {Array<{from: string, to: string, label: string}>} An array of relationships */ export function extractRelationships(description: string, entities: string[]): Array<{from: string, to: string, label: string}> { const relationships: Array<{from: string, to: string, label: string}> = []; // Look for relationship patterns in the description // This is a simplified approach; a real implementation would use more sophisticated NLP // Create a regex pattern to match sentences containing two entities for (let i = 0; i < entities.length; i++) { for (let j = 0; j < entities.length; j++) { if (i !== j) { const entity1 = entities[i]; const entity2 = entities[j]; // Look for sentences containing both entities const sentenceRegex = new RegExp( `[^.!?]*\\b${entity1}\\b[^.!?]*\\b${entity2}\\b[^.!?]*[.!?]`, 'gi' ); const matches = description.match(sentenceRegex); if (matches) { for (const match of matches) { // Extract relationship label based on verbs or prepositions between entities const betweenEntities = match.substring( match.indexOf(entity1) + entity1.length, match.indexOf(entity2) ); // Look for verbs or relationship words const relationshipWords = betweenEntities.match(/\b(has|have|contains|includes|uses|calls|depends on|relates to|connects to|links to|references|inherits from|extends|implements)\b/i); const label = relationshipWords ? relationshipWords[0] : 'relates to'; relationships.push({ from: entity1, to: entity2, label }); } } } } } // If no relationships were found but we have entities, create some default relationships if (relationships.length === 0 && entities.length > 1) { // Check for specific relationship keywords in the description const hasRelation = description.includes('has') || description.includes('have'); const usesRelation = description.includes('uses') || description.includes('use'); const containsRelation = description.includes('contains') || description.includes('contain'); const relationLabel = hasRelation ? 'has' : (usesRelation ? 'uses' : (containsRelation ? 'contains' : 'relates to')); // Create relationships between consecutive entities for (let i = 0; i < entities.length - 1; i++) { relationships.push({ from: entities[i], to: entities[i + 1], label: relationLabel }); } } return relationships; } /** * Extract process steps from a text description. * * @param {string} description - The text description to analyze * @returns {string[]} An array of process steps */ export function extractProcessSteps(description: string): string[] { const steps: string[] = []; // Look for numbered lists const numberedListRegex = /\b(\d+)[.)][\s]+([^\n.]+)/g; let match; while ((match = numberedListRegex.exec(description)) !== null) { steps.push(match[2].trim()); } // If no numbered list was found, look for steps indicated by keywords if (steps.length === 0) { const stepKeywords = ['first', 'then', 'next', 'finally', 'lastly']; const sentences = description.split(/[.!?]+/); for (const sentence of sentences) { for (const keyword of stepKeywords) { if (sentence.toLowerCase().includes(keyword)) { steps.push(sentence.trim()); break; } } } } // If still no steps, extract sentences that might be actions if (steps.length === 0) { const sentences = description.split(/[.!?]+/); for (const sentence of sentences) { // Look for sentences that start with a verb (action) if (/^\s*[A-Z][a-z]+\s+[a-z]+/.test(sentence)) { const trimmed = sentence.trim(); if (trimmed.length > 10 && trimmed.length < 60) { // Reasonable length for a step steps.push(trimmed); } } } } return steps; }