Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 | /** * Worksheet Parsing Background Task * * Wraps the LLM-based worksheet parsing in a background task for: * - Real-time progress streaming via Socket.IO * - Survival across page reloads * - Event replay for late-joining clients * - Database persistence of results */ import { eq } from 'drizzle-orm' import { db } from '@/db' import { type ParsingStatus, practiceAttachments } from '@/db/schema/practice-attachments' import { createTaskLLM } from '../llm' import { createTask, type TaskHandle } from '../task-manager' import { buildWorksheetParsingPrompt, computeParsingStats, type StreamParseWorksheetOptions, streamParseWorksheetImage, } from '../worksheet-parsing' import { AiFeature } from '@/lib/ai-usage/features' import { createUsageRecordingMiddleware } from '@/lib/ai-usage/llm-middleware' import type { BoundingBox, WorksheetParsingResult } from '../worksheet-parsing/schemas' import type { WorksheetParseEvent } from './events' /** * Input for worksheet parsing task */ export interface WorksheetParseInput { /** Base64-encoded image data URL */ imageDataUrl: string /** Additional prompt options */ promptOptions?: { focusProblemNumbers?: number[] additionalContext?: string } /** Attachment ID for database association */ attachmentId: string /** Player ID for database association */ playerId: string /** Preserved bounding boxes from user adjustments */ preservedBoundingBoxes?: Record<number, BoundingBox> _userId?: string } /** * Output from worksheet parsing task */ export interface WorksheetParseOutput { /** Parsed worksheet data */ data: WorksheetParsingResult /** Statistics computed from the result */ stats: ReturnType<typeof computeParsingStats> /** Provider used */ provider: string /** Model used */ model: string /** Token usage */ usage: { promptTokens: number completionTokens: number totalTokens: number } } /** * Start a worksheet parsing background task * * @param input - Parsing configuration * @returns Task ID that can be used to track progress * * @example * ```typescript * // Start parsing task * const taskId = await startWorksheetParsing({ * imageDataUrl: 'data:image/jpeg;base64,...', * attachmentId: 'abc123', * playerId: 'player456', * }) * * // Client subscribes via Socket.IO * socket.emit('task:subscribe', taskId) * socket.on('task:event', (event) => { * if (event.eventType === 'reasoning') { * console.log('AI thinking:', event.payload.text) * } * }) * ``` */ export async function startWorksheetParsing(input: WorksheetParseInput): Promise<string> { // Validate required fields if (!input.imageDataUrl || !input.imageDataUrl.startsWith('data:image/')) { throw new Error('Invalid image data URL') } if (!input.attachmentId) { throw new Error('Attachment ID is required') } if (!input.playerId) { throw new Error('Player ID is required') } // Clear any previous error and update timestamp (status will be set on completion) await db .update(practiceAttachments) .set({ parsingError: null, parsedAt: new Date().toISOString(), }) .where(eq(practiceAttachments.id, input.attachmentId)) return createTask<WorksheetParseInput, WorksheetParseOutput, WorksheetParseEvent>( 'worksheet-parse', input, async (handle, config) => { const { imageDataUrl, promptOptions, attachmentId, preservedBoundingBoxes } = config handle.setProgress(5, 'Initializing parser...') handle.emit({ type: 'parse_started', modelConfigId: 'gpt-5.2-thinking', useStreaming: true, attachmentId, }) try { await runStreamingParse(handle, imageDataUrl, { promptOptions, attachmentId, preservedBoundingBoxes, _userId: config._userId, }) } catch (error) { // Update DB with error const errorMessage = error instanceof Error ? error.message : String(error) await db .update(practiceAttachments) .set({ parsingStatus: 'failed', parsingError: errorMessage, }) .where(eq(practiceAttachments.id, attachmentId)) throw error } } ) } interface ParseOptions { promptOptions?: WorksheetParseInput['promptOptions'] attachmentId: string preservedBoundingBoxes?: Record<number, BoundingBox> _userId?: string } /** * Run streaming parse with real-time events * * Uses middleware-enhanced LLM client that automatically: * - Emits transient reasoning/output_delta events via Socket.IO * - Persists reasoning/output snapshots every 3s for page-reload recovery */ async function runStreamingParse( handle: TaskHandle<WorksheetParseOutput, WorksheetParseEvent>, imageDataUrl: string, options: ParseOptions ): Promise<void> { const { attachmentId, preservedBoundingBoxes } = options const streamOptions: StreamParseWorksheetOptions = { promptOptions: options.promptOptions, } // Build prompt for metadata const promptUsed = buildWorksheetParsingPrompt(options.promptOptions ?? {}) // Create task-aware LLM client (middleware handles reasoning/output streaming & snapshots) const usageMw = options._userId ? createUsageRecordingMiddleware({ userId: options._userId, feature: AiFeature.WORKSHEET_PARSE, backgroundTaskId: handle.id, }) : undefined const taskLLM = createTaskLLM(handle, usageMw) const stream = streamParseWorksheetImage(imageDataUrl, streamOptions, taskLLM) for await (const event of stream) { // Check for cancellation if (handle.isCancelled()) { handle.emit({ type: 'cancelled', reason: 'User cancelled' }) await db .update(practiceAttachments) .set({ parsingStatus: null, parsingError: null }) .where(eq(practiceAttachments.id, attachmentId)) return } // Handle domain-specific events (reasoning/output_delta handled by middleware) if (event.type === 'progress') { handle.setProgress(10, event.message) handle.emit({ type: 'parse_progress', stage: event.stage, message: event.message }) } else if (event.type === 'started') { handle.setProgress(15, 'AI analyzing worksheet...') handle.emit({ type: 'parse_llm_started', responseId: event.responseId, model: 'gpt-5.2', provider: 'openai', }) } else if (event.type === 'complete') { handle.setProgress(90, 'Validating results...') // Merge preserved bounding boxes let parsingResult = event.data if (preservedBoundingBoxes && Object.keys(preservedBoundingBoxes).length > 0) { parsingResult = { ...parsingResult, problems: parsingResult.problems.map((problem, index) => { const preservedBox = preservedBoundingBoxes[index] return preservedBox ? { ...problem, problemBoundingBox: preservedBox } : problem }), } } const stats = computeParsingStats(parsingResult) const status: ParsingStatus = parsingResult.needsReview ? 'needs_review' : 'approved' // Save results to database await db .update(practiceAttachments) .set({ parsingStatus: status, parsedAt: new Date().toISOString(), rawParsingResult: parsingResult, confidenceScore: parsingResult.overallConfidence, needsReview: parsingResult.needsReview, parsingError: null, llmProvider: 'openai', llmModel: 'gpt-5.2', llmPromptUsed: promptUsed, llmRawResponse: null, llmJsonSchema: null, llmImageSource: 'cropped', llmAttempts: 1, llmPromptTokens: event.usage.promptTokens, llmCompletionTokens: event.usage.completionTokens, llmTotalTokens: event.usage.promptTokens + event.usage.completionTokens, }) .where(eq(practiceAttachments.id, attachmentId)) handle.emit({ type: 'parse_complete', data: parsingResult, stats, status }) handle.complete({ data: parsingResult, stats, provider: 'openai', model: 'gpt-5.2', usage: { promptTokens: event.usage.promptTokens, completionTokens: event.usage.completionTokens, totalTokens: event.usage.promptTokens + event.usage.completionTokens, }, }) return } else if (event.type === 'error') { // Reasoning text already available to clients via snapshots handle.emit({ type: 'parse_error', error: event.message, reasoningText: '' }) throw new Error(event.message) } // reasoning and output_delta events are handled by middleware } // Stream ended without complete event throw new Error('Parsing stream ended unexpectedly') } |