ProDisco

/** * Extensible schema builder for custom fields */ import type { Schema } from '@orama/orama'; import { baseSchema, createEmptyDocument, type BaseDocument } from './base-schema.js'; import type { ExtractedType, ExtractedMethod, ExtractedFunction, PropertyInfo, } from '../extractor/types.js'; import type { CachedScript } from '../script/types.js'; import { splitCamelCase } from '../extractor/ast-parser.js'; /** * Extension schema type - additional fields to add to base schema */ export type ExtensionSchema = Record<string, 'string' | 'enum' | 'number' | 'boolean'>; /** * Options for building an extended schema */ export interface SchemaBuilderOptions<T extends ExtensionSchema> { /** Additional fields to add to the base schema */ extensions?: T; /** Tokenizer options */ tokenizerOptions?: { stemming?: boolean; stemmerSkipProperties?: string[]; }; } /** * Build a schema with optional extensions */ export function buildSchema<T extends ExtensionSchema = Record<string, never>>( options?: SchemaBuilderOptions<T> ): typeof baseSchema & T { if (!options?.extensions) { return baseSchema as typeof baseSchema & T; } return { ...baseSchema, ...options.extensions, } as typeof baseSchema & T; } /** * Build a document from an extracted type */ export function buildTypeDocument(type: ExtractedType): BaseDocument { const searchTokens = [ type.name, splitCamelCase(type.name), type.kind, type.description, ...type.properties.map((p) => p.name), ...type.nestedTypes, ].join(' '); const typeDefinition = formatTypeDefinition(type); return { ...createEmptyDocument(), id: `type:${type.library}:${type.name}`, documentType: 'type', name: type.name, description: type.description, searchTokens, library: type.library, category: type.kind, properties: JSON.stringify(type.properties), typeDefinition, nestedTypes: type.nestedTypes.join(', '), typeKind: type.kind, }; } /** * Build a document from an extracted method */ export function buildMethodDocument(method: ExtractedMethod): BaseDocument { const searchTokens = [ method.name, splitCamelCase(method.name), method.className, method.description, ...method.parameters.map((p) => p.name), method.returnType, ].join(' '); const signature = buildMethodSignature(method); const category = inferMethodCategory(method.name); return { ...createEmptyDocument(), id: `method:${method.library}:${method.className}:${method.name}`, documentType: 'method', name: method.name, description: method.description, searchTokens, library: method.library, category, parameters: JSON.stringify(method.parameters), returnType: method.returnType, returnTypeDefinition: method.returnTypeDefinition || '', signature, className: method.className, }; } /** * Build a document from an extracted function */ export function buildFunctionDocument(func: ExtractedFunction): BaseDocument { const searchTokens = [ func.name, splitCamelCase(func.name), func.description, ...func.parameters.map((p) => p.name), func.returnType, ].join(' '); const category = inferFunctionCategory(func.name, func.library); return { ...createEmptyDocument(), id: `function:${func.library}:${func.name}`, documentType: 'function', name: func.name, description: func.description, searchTokens, library: func.library, category, parameters: JSON.stringify(func.parameters), returnType: func.returnType, signature: func.signature, }; } /** * Build a document from a cached script */ export function buildScriptDocument(script: CachedScript): BaseDocument { const isAutoGenerated = /^script-\d{4}-\d{2}-\d{2}T\d{2}-\d{2}-\d{2}-[a-f0-9]+\.ts$/.test( script.filename ); const searchTokens = [ 'script', // Only include filename tokens for manually named scripts ...(isAutoGenerated ? [] : [script.filename.replace(/\.ts$/, '').replace(/[-_]/g, ' ')]), ...script.resourceTypes, script.description, ...script.apiClasses, ...script.keywords, ].join(' '); const displayName = isAutoGenerated ? script.apiClasses.length > 0 ? script.apiClasses[0]!.toLowerCase() : 'sandbox-script' : script.filename.replace(/\.ts$/, ''); return { ...createEmptyDocument(), id: `script:${script.filename}`, documentType: 'script', name: displayName, description: script.description, searchTokens, library: script.apiClasses.length > 0 ? script.apiClasses[0]! : 'CachedScript', category: 'script', filePath: script.filePath, keywords: script.keywords.join(' '), }; } /** * Build a document from any extracted item */ export function buildDocument( item: ExtractedType | ExtractedMethod | ExtractedFunction | CachedScript ): BaseDocument { if ('kind' in item) { return buildTypeDocument(item as ExtractedType); } if ('className' in item) { return buildMethodDocument(item as ExtractedMethod); } if ('signature' in item) { return buildFunctionDocument(item as ExtractedFunction); } return buildScriptDocument(item as CachedScript); } /** * Format a type definition for display */ function formatTypeDefinition(type: ExtractedType, maxProperties: number = 20): string { let result = `${type.kind} ${type.name} {\n`; const propsToShow = type.properties.slice(0, maxProperties); const hasMore = type.properties.length > maxProperties; for (const prop of propsToShow) { const optionalMarker = prop.optional ? '?' : ''; result += ` ${prop.name}${optionalMarker}: ${prop.type}\n`; } if (hasMore) { result += ` ... ${type.properties.length - maxProperties} more properties\n`; } result += `}`; return result; } /** * Build a method signature string */ function buildMethodSignature(method: ExtractedMethod): string { const params = method.parameters .map((p) => `${p.name}${p.optional ? '?' : ''}: ${p.type}`) .join(', '); const asyncPrefix = method.isAsync ? 'async ' : ''; const staticPrefix = method.isStatic ? 'static ' : ''; return `${staticPrefix}${asyncPrefix}${method.name}(${params}): ${method.returnType}`; } /** * Infer category from method name */ function inferMethodCategory(methodName: string): string { const lowerName = methodName.toLowerCase(); if (lowerName.startsWith('list')) return 'list'; if (lowerName.startsWith('get') || lowerName.startsWith('read')) return 'read'; if (lowerName.startsWith('create') || lowerName.startsWith('add')) return 'create'; if (lowerName.startsWith('delete') || lowerName.startsWith('remove')) return 'delete'; if (lowerName.startsWith('patch')) return 'patch'; if (lowerName.startsWith('update')) return 'update'; if (lowerName.startsWith('replace') || lowerName.startsWith('set')) return 'replace'; if (lowerName.startsWith('watch')) return 'watch'; if (lowerName.startsWith('query')) return 'query'; return 'other'; } /** * Infer category from function name and library */ function inferFunctionCategory(funcName: string, library: string): string { const lowerName = funcName.toLowerCase(); const lowerLib = library.toLowerCase(); // Statistics functions if (lowerLib.includes('statistics') || lowerLib.includes('stats')) { if (['mean', 'median', 'mode', 'variance', 'std'].some((s) => lowerName.includes(s))) { return 'descriptive'; } if (lowerName.includes('regress')) return 'regression'; if (['normal', 'poisson', 'binomial'].some((s) => lowerName.includes(s))) { return 'distribution'; } } // Math functions if (lowerLib.includes('math')) { if (['matrix', 'transpose', 'inverse', 'det'].some((s) => lowerName.includes(s))) { return 'matrix'; } if (['sin', 'cos', 'tan', 'log', 'exp'].some((s) => lowerName.includes(s))) { return 'math'; } } // Signal processing if (lowerLib.includes('fft') || lowerName.includes('fft')) { return 'signal'; } return 'utility'; }