IO-3515 Add translations and logging.

This commit is contained in:
Patrick Fic
2026-02-19 13:54:39 -08:00
parent ae1408012f
commit b2bc19c5c9
7 changed files with 463 additions and 78 deletions

View File

@@ -5,7 +5,7 @@ const { v4: uuidv4 } = require('uuid');
const { getTextractJobKey, setTextractJob, getTextractJob, getFileType, getPdfPageCount, hasActiveJobs } = require("./bill-ocr-helpers");
const { extractInvoiceData, processScanData } = require("./bill-ocr-normalize");
const { generateBillFormData } = require("./bill-ocr-generator");
const logger = require("../../utils/logger");
// Initialize AWS clients
const awsConfig = {
region: process.env.AWS_AI_REGION || "ca-central-1",
@@ -53,24 +53,27 @@ async function jobExists(textractJobId) {
return false;
}
async function handleBillOcr(request, response) {
async function handleBillOcr(req, res) {
// Check if file was uploaded
if (!request.file) {
response.status(400).send({ error: 'No file uploaded.' });
if (!req.file) {
res.status(400).send({ error: 'No file uploaded.' });
return;
}
// The uploaded file is available in request.file
const uploadedFile = request.file;
const { jobid, bodyshopid, partsorderid } = request.body;
// The uploaded file is available in request file
const uploadedFile = req.file;
const { jobid, bodyshopid, partsorderid } = req.body;
logger.log("bill-ocr-start", "DEBUG", req.user.email, jobid, null);
try {
const fileType = getFileType(uploadedFile);
// Images are always processed synchronously (single page)
if (fileType === 'image') {
const processedData = await processSinglePageDocument(uploadedFile.buffer);
const billForm = await generateBillFormData({ processedData: processedData, jobid, bodyshopid, partsorderid, req: request });
response.status(200).send({
const billForm = await generateBillFormData({ processedData: processedData, jobid, bodyshopid, partsorderid, req: req });
logger.log("bill-ocr-single-complete", "DEBUG", req.user.email, jobid, { ...processedData, billForm });
res.status(200).send({
success: true,
status: 'COMPLETED',
data: { ...processedData, billForm },
@@ -83,9 +86,9 @@ async function handleBillOcr(request, response) {
if (pageCount === 1) {
// Process synchronously for single-page documents
const processedData = await processSinglePageDocument(uploadedFile.buffer);
const billForm = await generateBillFormData({ processedData: processedData, jobid, bodyshopid, partsorderid, req: request });
//const billResult = await generateBillFormData({ result, });
response.status(200).send({
const billForm = await generateBillFormData({ processedData: processedData, jobid, bodyshopid, partsorderid, req: req });
logger.log("bill-ocr-single-complete", "DEBUG", req.user.email, jobid, { ...processedData, billForm });
res.status(200).send({
success: true,
status: 'COMPLETED',
data: { ...processedData, billForm },
@@ -94,8 +97,9 @@ async function handleBillOcr(request, response) {
} else {
// Start the Textract job (non-blocking) for multi-page documents
const jobInfo = await startTextractJob(uploadedFile.buffer, { jobid, bodyshopid, partsorderid });
logger.log("bill-ocr-multipage-start", "DEBUG", req.user.email, jobid, jobInfo);
response.status(202).send({
res.status(202).send({
success: true,
textractJobId: jobInfo.jobId,
message: 'Invoice processing started',
@@ -103,32 +107,35 @@ async function handleBillOcr(request, response) {
});
}
} else {
response.status(400).send({
logger.log("bill-ocr-unsupported-filetype", "WARN", req.user.email, jobid, { fileType });
res.status(400).send({
error: 'Unsupported file type',
message: 'Please upload a PDF or supported image file (JPEG, PNG, TIFF)'
});
}
} catch (error) {
console.error('Error starting invoice processing:', error);
response.status(500).send({
logger.log("bill-ocr-error", "ERROR", req.user.email, jobid, { error: error.message, stack: error.stack });
res.status(500).send({
error: 'Failed to start invoice processing',
message: error.message
});
}
}
async function handleBillOcrStatus(request, response) {
const { textractJobId } = request.params;
async function handleBillOcrStatus(req, res) {
const { textractJobId } = req.params;
if (!textractJobId) {
console.log('No textractJobId found in params');
response.status(400).send({ error: 'Job ID is required' });
logger.log("bill-ocr-status-error", "WARN", req.user.email, null, { error: 'No textractJobId found in params' });
res.status(400).send({ error: 'Job ID is required' });
res.status(400).send({ error: 'Job ID is required' });
return;
}
const jobStatus = await getTextractJob({ redisPubClient, textractJobId });
if (!jobStatus) {
response.status(404).send({ error: 'Job not found' });
res.status(404).send({ error: 'Job not found' });
return;
}
@@ -143,8 +150,9 @@ async function handleBillOcrStatus(request, response) {
jobid: jobStatus.context.jobid,
bodyshopid: jobStatus.context.bodyshopid,
partsorderid: jobStatus.context.partsorderid,
req: request // Now we have request context!
req: req // Now we have request context!
});
logger.log("bill-ocr-multipage-complete", "DEBUG", req.user.email, jobStatus.context.jobid, { ...jobStatus.data, billForm });
// Cache the billForm back to Redis for future requests
await setTextractJob({
@@ -159,7 +167,9 @@ async function handleBillOcrStatus(request, response) {
}
});
} catch (error) {
response.status(500).send({
logger.log("bill-ocr-multipage-error", "ERROR", req.user.email, jobStatus.context.jobid, { ...jobStatus.data, error: error.message, stack: error.stack });
res.status(500).send({
status: 'COMPLETED',
error: 'Data processed but failed to generate bill form',
message: error.message,
@@ -169,7 +179,7 @@ async function handleBillOcrStatus(request, response) {
}
}
response.status(200).send({
res.status(200).send({
status: 'COMPLETED',
data: {
...jobStatus.data,
@@ -177,12 +187,14 @@ async function handleBillOcrStatus(request, response) {
}
});
} else if (jobStatus.status === 'FAILED') {
response.status(500).send({
logger.log("bill-ocr-multipage-failed", "ERROR", req.user.email, jobStatus.context.jobid, { ...jobStatus.data, error: jobStatus.error, });
res.status(500).send({
status: 'FAILED',
error: jobStatus.error
});
} else {
response.status(200).send({
res.status(200).send({
status: jobStatus.status
});
}
@@ -307,22 +319,51 @@ async function processSQSMessages() {
console.log('Processing', result.Messages.length, 'messages from SQS');
for (const message of result.Messages) {
try {
//TODO: Add environment level filtering here.
await handleTextractNotification(message);
// Environment-level filtering: check if this message belongs to this environment
const shouldProcess = await shouldProcessMessage(message);
// Delete message after successful processing
const deleteCommand = new DeleteMessageCommand({
QueueUrl: queueUrl,
ReceiptHandle: message.ReceiptHandle
});
await sqsClient.send(deleteCommand);
if (shouldProcess) {
await handleTextractNotification(message);
// Delete message after successful processing
const deleteCommand = new DeleteMessageCommand({
QueueUrl: queueUrl,
ReceiptHandle: message.ReceiptHandle
});
await sqsClient.send(deleteCommand);
} else {
console.log('Ignoring message - job not found in this environment');
}
} catch (error) {
console.error('Error processing message:', error);
logger.log("bill-ocr-sqs-processing-error", "ERROR", "api", null, { message, error: error.message, stack: error.stack });
}
}
}
} catch (error) {
console.error('Error receiving SQS messages:', error);
logger.log("bill-ocr-sqs-receiving-error", "ERROR", "api", null, { error: error.message, stack: error.stack });
}
}
/**
* Check if a message should be processed by this environment
* @param {Object} message - SQS message
* @returns {Promise<boolean>}
*/
async function shouldProcessMessage(message) {
try {
const body = JSON.parse(message.Body);
const snsMessage = JSON.parse(body.Message);
const textractJobId = snsMessage.JobId;
// Check if job exists in Redis for this environment (using environment-specific prefix)
const exists = await jobExists(textractJobId);
return exists;
} catch (error) {
console.error('Error checking if message should be processed:', error);
// If we can't parse the message, don't process it
return false;
}
}
@@ -332,30 +373,22 @@ async function handleTextractNotification(message) {
try {
snsMessage = JSON.parse(body.Message);
} catch (error) {
//Delete the message so it doesn't clog the queue
const deleteCommand = new DeleteMessageCommand({
QueueUrl: process.env.AWS_TEXTRACT_SQS_QUEUE_URL,
ReceiptHandle: message.ReceiptHandle
});
await sqsClient.send(deleteCommand);
console.error('Error parsing SNS message:', error);
console.log('Message Deleted:', body);
console.log('Invalid message format:', body);
return;
}
const textractJobId = snsMessage.JobId;
const status = snsMessage.Status;
// Check if job exists in Redis
const exists = await jobExists(textractJobId);
// Get job info from Redis
const jobInfo = await getTextractJob({ redisPubClient, textractJobId });
if (!exists) {
console.warn(`Job not found for Textract job ID: ${textractJobId}`);
if (!jobInfo) {
console.warn(`Job info not found in Redis for Textract job ID: ${textractJobId}`);
return;
}
const jobInfo = await getTextractJob({ redisPubClient, textractJobId });
if (status === 'SUCCEEDED') {
// Retrieve the results
const { processedData, originalResponse } = await retrieveTextractResults(textractJobId);