@stacksjs/ai 0.70.55 → 0.70.56

package/src/personalization.ts

@@ -1,490 +0,0 @@
-/**
- * AI Personalization Module
- *
- * Provides AI-powered personalization features including content recommendations,
- * user profiling, sentiment analysis, content classification, and summarization.
- */
-
-import type { AIResult } from './types'
-
-// ============================================================================
-// Types
-// ============================================================================
-
-export interface UserProfile {
-  id: string
-  preferences: Record<string, number>
-  interactions: Interaction[]
-  segments: string[]
-  metadata?: Record<string, unknown>
-}
-
-export interface Interaction {
-  type: 'view' | 'click' | 'purchase' | 'like' | 'dislike' | 'share' | 'bookmark' | 'custom'
-  itemId: string
-  timestamp: number
-  weight?: number
-  metadata?: Record<string, unknown>
-}
-
-export interface ContentItem {
-  id: string
-  content: string
-  category?: string
-  tags?: string[]
-  metadata?: Record<string, unknown>
-}
-
-export interface RecommendationOptions {
-  provider?: 'anthropic' | 'openai' | 'ollama'
-  model?: string
-  maxTokens?: number
-  temperature?: number
-  limit?: number
-}
-
-export interface RecommendationResult {
-  recommendations: Array<{
-    itemId: string
-    score: number
-    reason: string
-  }>
-  model: string
-  provider: string
-}
-
-export interface SentimentResult {
-  sentiment: 'positive' | 'negative' | 'neutral' | 'mixed'
-  score: number
-  confidence: number
-  aspects?: Array<{
-    aspect: string
-    sentiment: 'positive' | 'negative' | 'neutral'
-    score: number
-  }>
-}
-
-export interface ClassificationResult {
-  label: string
-  confidence: number
-  allLabels: Array<{
-    label: string
-    confidence: number
-  }>
-}
-
-export interface SummaryOptions {
-  provider?: 'anthropic' | 'openai' | 'ollama'
-  model?: string
-  maxLength?: number
-  style?: 'concise' | 'detailed' | 'bullet-points'
-  language?: string
-}
-
-// ============================================================================
-// Sentiment Analysis
-// ============================================================================
-
-/**
- * Analyze the sentiment of text using AI.
- */
-export async function analyzeSentiment(
-  text: string,
-  options: {
-    provider?: 'anthropic' | 'openai' | 'ollama'
-    model?: string
-    aspects?: string[]
-  } = {},
-): Promise<SentimentResult> {
-  const { provider = 'anthropic', aspects } = options
-
-  const aspectInstruction = aspects
-    ? `\nAlso analyze sentiment for these specific aspects: ${aspects.join(', ')}`
-    : ''
-
-  const systemPrompt = `You are a sentiment analysis expert. Analyze the sentiment of the given text and respond with ONLY valid JSON in this exact format:
-{
-  "sentiment": "positive" | "negative" | "neutral" | "mixed",
-  "score": <number from -1.0 to 1.0>,
-  "confidence": <number from 0.0 to 1.0>${aspects ? `,
-  "aspects": [{"aspect": "<name>", "sentiment": "positive" | "negative" | "neutral", "score": <number>}]` : ''}
-}${aspectInstruction}`
-
-  const result = await callProvider(provider, systemPrompt, text, options.model)
-
-  try {
-    return JSON.parse(result.content) as SentimentResult
-  }
-  catch {
-    // Fallback: extract sentiment from text response
-    const lower = result.content.toLowerCase()
-    const isPositive = lower.includes('positive')
-    const isNegative = lower.includes('negative')
-    return {
-      sentiment: isPositive && isNegative ? 'mixed' : isPositive ? 'positive' : isNegative ? 'negative' : 'neutral',
-      score: isPositive ? 0.5 : isNegative ? -0.5 : 0,
-      confidence: 0.5,
-    }
-  }
-}
-
-// ============================================================================
-// Text Classification
-// ============================================================================
-
-/**
- * Classify text into one of the provided labels.
- */
-export async function classifyText(
-  text: string,
-  labels: string[],
-  options: {
-    provider?: 'anthropic' | 'openai' | 'ollama'
-    model?: string
-    multiLabel?: boolean
-  } = {},
-): Promise<ClassificationResult> {
-  const { provider = 'anthropic', multiLabel = false } = options
-
-  const systemPrompt = `You are a text classification expert. Classify the given text into ${multiLabel ? 'one or more of' : 'exactly one of'} these categories: ${labels.join(', ')}.
-
-Respond with ONLY valid JSON in this exact format:
-{
-  "label": "<primary label>",
-  "confidence": <number from 0.0 to 1.0>,
-  "allLabels": [{"label": "<label>", "confidence": <number>}]
-}
-
-Include ALL provided labels in allLabels with their confidence scores, sorted by confidence descending.`
-
-  const result = await callProvider(provider, systemPrompt, text, options.model)
-
-  try {
-    return JSON.parse(result.content) as ClassificationResult
-  }
-  catch {
-    return {
-      label: labels[0],
-      confidence: 0.5,
-      allLabels: labels.map(l => ({ label: l, confidence: 1 / labels.length })),
-    }
-  }
-}
-
-// ============================================================================
-// Smart Summarization
-// ============================================================================
-
-/**
- * Generate an intelligent summary of text.
- *
- * Not exported as a bare top-level identifier because `text.ts` already
- * exposes `summarize`, and `index.ts` re-exports both via `export *`.
- * Reach this multi-provider variant through the `personalization`
- * namespace export below: `personalization.summarize(...)`.
- */
-async function summarize(
-  text: string,
-  options: SummaryOptions = {},
-): Promise<AIResult> {
-  const {
-    provider = 'anthropic',
-    style = 'concise',
-    maxLength,
-    language,
-  } = options
-
-  let styleInstruction: string
-  switch (style) {
-    case 'bullet-points':
-      styleInstruction = 'Use bullet points to organize the key points.'
-      break
-    case 'detailed':
-      styleInstruction = 'Provide a detailed summary covering all major points.'
-      break
-    default:
-      styleInstruction = 'Be concise and focus on the most important points.'
-  }
-
-  const lengthInstruction = maxLength ? ` Keep the summary under ${maxLength} words.` : ''
-  const languageInstruction = language ? ` Write the summary in ${language}.` : ''
-
-  const systemPrompt = `You are an expert summarizer. ${styleInstruction}${lengthInstruction}${languageInstruction}`
-
-  return callProvider(provider, systemPrompt, `Summarize the following text:\n\n${text}`, options.model)
-}
-
-// ============================================================================
-// Content Recommendations
-// ============================================================================
-
-/**
- * Generate personalized content recommendations based on user profile and available items.
- */
-export async function recommend(
-  profile: UserProfile,
-  items: ContentItem[],
-  options: RecommendationOptions = {},
-): Promise<RecommendationResult> {
-  const { provider = 'anthropic', limit = 5 } = options
-
-  const profileSummary = buildProfileSummary(profile)
-  const itemsList = items.map(item =>
-    `ID: ${item.id} | Category: ${item.category || 'none'} | Tags: ${item.tags?.join(', ') || 'none'} | Content: ${item.content.slice(0, 200)}`,
-  ).join('\n')
-
-  const systemPrompt = `You are a recommendation engine. Based on the user profile, recommend the most relevant items. Respond with ONLY valid JSON:
-{
-  "recommendations": [
-    {"itemId": "<id>", "score": <0.0-1.0>, "reason": "<brief reason>"}
-  ]
-}
-
-Return at most ${limit} recommendations, sorted by relevance score descending.`
-
-  const prompt = `User Profile:\n${profileSummary}\n\nAvailable Items:\n${itemsList}`
-  const result = await callProvider(provider, systemPrompt, prompt, options.model)
-
-  try {
-    const parsed = JSON.parse(result.content) as { recommendations: Array<{ itemId: string; score: number; reason: string }> }
-    return {
-      recommendations: parsed.recommendations.slice(0, limit),
-      model: result.model,
-      provider: provider,
-    }
-  }
-  catch {
-    return {
-      recommendations: [],
-      model: result.model,
-      provider: provider,
-    }
-  }
-}
-
-function buildProfileSummary(profile: UserProfile): string {
-  const topPreferences = Object.entries(profile.preferences)
-    .sort(([, a], [, b]) => b - a)
-    .slice(0, 10)
-    .map(([key, value]) => `${key}: ${value.toFixed(2)}`)
-    .join(', ')
-
-  const recentInteractions = profile.interactions
-    .sort((a, b) => b.timestamp - a.timestamp)
-    .slice(0, 10)
-    .map(i => `${i.type} on ${i.itemId}`)
-    .join(', ')
-
-  return `Segments: ${profile.segments.join(', ')}
-Top Preferences: ${topPreferences || 'none'}
-Recent Interactions: ${recentInteractions || 'none'}`
-}
-
-// ============================================================================
-// User Profile Building
-// ============================================================================
-
-/**
- * Create a new empty user profile.
- */
-export function createProfile(id: string, segments: string[] = []): UserProfile {
-  return {
-    id,
-    preferences: {},
-    interactions: [],
-    segments,
-  }
-}
-
-/**
- * Record a user interaction and update preferences.
- */
-export function recordInteraction(profile: UserProfile, interaction: Interaction): UserProfile {
-  profile.interactions.push(interaction)
-
-  // Update preference weights based on interaction type
-  const weights: Record<string, number> = {
-    view: 0.1,
-    click: 0.3,
-    like: 0.5,
-    share: 0.6,
-    bookmark: 0.7,
-    purchase: 1.0,
-    dislike: -0.5,
-    custom: interaction.weight || 0.3,
-  }
-
-  const weight = weights[interaction.type] || 0.1
-  const key = interaction.itemId
-  profile.preferences[key] = (profile.preferences[key] || 0) + weight
-
-  return profile
-}
-
-/**
- * Extract keywords/topics from user interactions using AI.
- */
-export async function extractUserInterests(
-  profile: UserProfile,
-  items: ContentItem[],
-  options: { provider?: 'anthropic' | 'openai' | 'ollama'; model?: string } = {},
-): Promise<string[]> {
-  const { provider = 'anthropic' } = options
-
-  const interactedItemIds = new Set(profile.interactions.map(i => i.itemId))
-  const interactedItems = items.filter(item => interactedItemIds.has(item.id))
-
-  if (interactedItems.length === 0) return profile.segments
-
-  const contentSample = interactedItems
-    .slice(0, 20)
-    .map(item => item.content.slice(0, 200))
-    .join('\n---\n')
-
-  const systemPrompt = `Extract the main interests/topics from the user's interaction history. Return ONLY a JSON array of strings, e.g. ["technology", "cooking", "travel"]. Maximum 10 interests.`
-
-  const result = await callProvider(provider, systemPrompt, contentSample, options.model)
-
-  try {
-    return JSON.parse(result.content) as string[]
-  }
-  catch {
-    return profile.segments
-  }
-}
-
-// ============================================================================
-// Provider Helper
-// ============================================================================
-
-async function callProvider(
-  provider: string,
-  systemPrompt: string,
-  userMessage: string,
-  model?: string,
-): Promise<AIResult> {
-  if (provider === 'anthropic') {
-    const apiKey = process.env.ANTHROPIC_API_KEY
-    if (!apiKey) throw new Error('ANTHROPIC_API_KEY required.')
-
-    const response = await fetch('https://api.anthropic.com/v1/messages', {
-      method: 'POST',
-      headers: {
-        'Content-Type': 'application/json',
-        'x-api-key': apiKey,
-        'anthropic-version': '2023-06-01',
-      },
-      body: JSON.stringify({
-        model: model || 'claude-sonnet-4-20250514',
-        max_tokens: 4096,
-        system: systemPrompt,
-        messages: [{ role: 'user', content: userMessage }],
-      }),
-    })
-
-    if (!response.ok) {
-      const error = await response.text()
-      throw new Error(`Claude API error: ${error}`)
-    }
-
-    const data = (await response.json()) as any
-    return {
-      content: data.content[0].text,
-      model: data.model,
-      usage: {
-        promptTokens: data.usage?.input_tokens || 0,
-        completionTokens: data.usage?.output_tokens || 0,
-        totalTokens: (data.usage?.input_tokens || 0) + (data.usage?.output_tokens || 0),
-      },
-      finishReason: data.stop_reason,
-    }
-  }
-
-  if (provider === 'openai') {
-    const apiKey = process.env.OPENAI_API_KEY
-    if (!apiKey) throw new Error('OPENAI_API_KEY required.')
-
-    const response = await fetch('https://api.openai.com/v1/chat/completions', {
-      method: 'POST',
-      headers: {
-        'Content-Type': 'application/json',
-        'Authorization': `Bearer ${apiKey}`,
-      },
-      body: JSON.stringify({
-        model: model || 'gpt-4o',
-        max_tokens: 4096,
-        messages: [
-          { role: 'system', content: systemPrompt },
-          { role: 'user', content: userMessage },
-        ],
-      }),
-    })
-
-    if (!response.ok) {
-      const error = await response.text()
-      throw new Error(`OpenAI API error: ${error}`)
-    }
-
-    const data = (await response.json()) as any
-    return {
-      content: data.choices[0].message.content,
-      model: data.model,
-      usage: {
-        promptTokens: data.usage?.prompt_tokens || 0,
-        completionTokens: data.usage?.completion_tokens || 0,
-        totalTokens: data.usage?.total_tokens || 0,
-      },
-      finishReason: data.choices[0].finish_reason,
-    }
-  }
-
-  if (provider === 'ollama') {
-    const host = process.env.OLLAMA_HOST || 'http://localhost:11434'
-
-    const response = await fetch(`${host}/api/chat`, {
-      method: 'POST',
-      headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({
-        model: model || 'llama3.2',
-        messages: [
-          { role: 'system', content: systemPrompt },
-          { role: 'user', content: userMessage },
-        ],
-        stream: false,
-      }),
-    })
-
-    if (!response.ok) {
-      const error = await response.text()
-      throw new Error(`Ollama API error: ${error}`)
-    }
-
-    const data = (await response.json()) as any
-    return {
-      content: data.message.content,
-      model: data.model,
-      usage: {
-        promptTokens: data.prompt_eval_count || 0,
-        completionTokens: data.eval_count || 0,
-        totalTokens: (data.prompt_eval_count || 0) + (data.eval_count || 0),
-      },
-      finishReason: data.done_reason || 'stop',
-    }
-  }
-
-  throw new Error(`Provider not supported: ${provider}`)
-}
-
-// ============================================================================
-// Exports
-// ============================================================================
-
-export const personalization = {
-  analyzeSentiment,
-  classifyText,
-  summarize,
-  recommend,
-  createProfile,
-  recordInteraction,
-  extractUserInterests,
-}