Files
bodyshop/server/ai/bill-ocr/bill-ocr-helpers.js

160 lines
4.6 KiB
JavaScript

const PDFDocument = require('pdf-lib').PDFDocument;
const logger = require("../../utils/logger");
const TEXTRACT_REDIS_PREFIX = `textract:${process.env?.NODE_ENV}`
const TEXTRACT_JOB_TTL = 10 * 60;
/**
* Generate Redis key for Textract job using textract job ID
* @param {string} textractJobId
* @returns {string}
*/
function getTextractJobKey(textractJobId) {
return `${TEXTRACT_REDIS_PREFIX}:${textractJobId}`;
}
/**
* Store Textract job data in Redis
* @param {string} textractJobId
* @param {Object} redisPubClient
* @param {Object} jobData
*/
async function setTextractJob({ redisPubClient, textractJobId, jobData }) {
if (!redisPubClient) {
throw new Error('Redis client not initialized. Call initializeBillOcr first.');
}
const key = getTextractJobKey(textractJobId);
await redisPubClient.set(key, JSON.stringify(jobData));
await redisPubClient.expire(key, TEXTRACT_JOB_TTL);
}
/**
* Retrieve Textract job data from Redis
* @param {string} textractJobId
* @param {Object} redisPubClient
* @returns {Promise<Object|null>}
*/
async function getTextractJob({ redisPubClient, textractJobId }) {
if (!redisPubClient) {
throw new Error('Redis client not initialized. Call initializeBillOcr first.');
}
const key = getTextractJobKey(textractJobId);
const data = await redisPubClient.get(key);
return data ? JSON.parse(data) : null;
}
/**
* Detect file type based on MIME type and file signature
* @param {Object} file - Multer file object
* @returns {string} 'pdf', 'image', or 'unknown'
*/
function getFileType(file) {
// Check MIME type first
const mimeType = file.mimetype?.toLowerCase();
if (mimeType === 'application/pdf') {
return 'pdf';
}
if (mimeType && mimeType.startsWith('image/')) {
return 'image';
}
// Fallback: Check file signature (magic bytes)
const buffer = file.buffer;
if (buffer && buffer.length > 4) {
// PDF signature: %PDF
if (buffer[0] === 0x25 && buffer[1] === 0x50 && buffer[2] === 0x44 && buffer[3] === 0x46) {
return 'pdf';
}
// JPEG signature: FF D8 FF
if (buffer[0] === 0xFF && buffer[1] === 0xD8 && buffer[2] === 0xFF) {
return 'image';
}
// PNG signature: 89 50 4E 47
if (buffer[0] === 0x89 && buffer[1] === 0x50 && buffer[2] === 0x4E && buffer[3] === 0x47) {
return 'image';
}
// HEIC/HEIF: Check for ftyp followed by heic/heix/hevc/hevx
if (buffer.length > 12) {
const ftypIndex = buffer.indexOf(Buffer.from('ftyp'));
if (ftypIndex > 0 && ftypIndex < 12) {
const brand = buffer.slice(ftypIndex + 4, ftypIndex + 8).toString('ascii');
if (brand.startsWith('heic') || brand.startsWith('heix') ||
brand.startsWith('hevc') || brand.startsWith('hevx') ||
brand.startsWith('mif1')) {
return 'image';
}
}
}
}
return 'unknown';
}
/**
* Get the number of pages in a PDF buffer
* @param {Buffer} pdfBuffer
* @returns {Promise<number>}
*/
async function getPdfPageCount(pdfBuffer) {
try {
const pdfDoc = await PDFDocument.load(pdfBuffer);
return pdfDoc.getPageCount();
} catch (error) {
console.error('Error reading PDF page count:', error);
throw new Error('Failed to read PDF: ' + error.message);
}
}
/**
* Check if there are any jobs in IN_PROGRESS status
* @returns {Promise<boolean>}
*/
async function hasActiveJobs({ redisPubClient }) {
if (!redisPubClient) {
throw new Error('Redis client not initialized.');
}
try {
// Get all textract job keys
const pattern = `${TEXTRACT_REDIS_PREFIX}:*`;
const keys = await redisPubClient.keys(pattern);
if (!keys || keys.length === 0) {
return false;
}
//TODO: Is there a better way to do this that supports clusters?
// Check if any job has IN_PROGRESS status
for (const key of keys) {
const data = await redisPubClient.get(key);
if (data) {
const jobData = JSON.parse(data);
if (jobData.status === 'IN_PROGRESS') {
return true;
}
}
}
return false;
} catch (error) {
logger.log("bill-ocr-job-check-error", "ERROR", "api", null, { error: error.message, stack: error.stack });
return false;
}
}
module.exports = {
getTextractJobKey,
setTextractJob,
getTextractJob,
getFileType,
getPdfPageCount,
hasActiveJobs,
TEXTRACT_REDIS_PREFIX
}