admin/bodyshop
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

IO-3515 Minor improvements to Bill AI.

Browse Source
This commit is contained in:
Patrick Fic
2026-03-03 14:56:28 -08:00
parent 5f8a08b0a7
commit 47a9a963fa
4 changed files with 89 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

96
server/ai/bill-ocr/bill-ocr-generator.js
View File

@@ -1,12 +1,12 @@
const Fuse = require('fuse.js');
const { has } = require("lodash");
const { standardizedFieldsnames } = require('./bill-ocr-normalize');
const InstanceManager = require("../../utils/instanceMgr").default;
const PRICE_PERCENT_MARGIN_TOLERANCE = 0.5; //Used to make sure prices and costs are likely.
const PRICE_QUANTITY_MARGIN_TOLERANCE = 0.03; //Used to make sure that if there is a quantity, the price is likely a unit price.
// Helper function to normalize fields
const normalizePartNumber = (str) => {
return str.replace(/[^a-zA-Z0-9]/g, '').toUpperCase();
@@ -17,7 +17,38 @@ const normalizeText = (str) => {
};
const normalizePrice = (str) => {
if (typeof str !== 'string') return str;
return str.replace(/[^0-9.-]+/g, "");
let value = str.trim();
// Handle European-style decimal comma like "292,37".
// Only treat the *last* comma as a decimal separator when:
// - there's no '.' anywhere (so we don't fight normal US formatting like "1,234.56")
// - and the suffix after the last comma is 1-2 digits (so "1,234" stays 1234)
if (!value.includes('.') && value.includes(',')) {
const lastCommaIndex = value.lastIndexOf(',');
const decimalSuffix = value.slice(lastCommaIndex + 1).trim();
if (/^\d{1,2}$/.test(decimalSuffix)) {
const before = value.slice(0, lastCommaIndex).replace(/,/g, '');
value = `${before}.${decimalSuffix}`;
} else {
// Treat commas as thousands separators (or noise) and drop them.
value = value.replace(/,/g, '');
}
}
return value.replace(/[^0-9.-]+/g, "");
};
const roundToIncrement = (value, increment) => {
if (typeof value !== 'number' || !isFinite(value) || typeof increment !== 'number' || !isFinite(increment) || increment <= 0) {
return value;
}
const rounded = Math.round((value + Number.EPSILON) / increment) * increment;
// Prevent float artifacts (e.g. 0.20500000000000002)
const decimals = Math.max(0, Math.ceil(-Math.log10(increment)));
return parseFloat(rounded.toFixed(decimals));
};
//More complex function. Not necessary at the moment, keeping for reference.
@@ -134,6 +165,7 @@ const calculateTextractConfidence = (textractLineItem) => {
const hasActualCost = Object.values(textractLineItem).some(field => field.normalizedLabel === standardizedFieldsnames.actual_cost);
const hasActualPrice = Object.values(textractLineItem).some(field => field.normalizedLabel === standardizedFieldsnames.actual_price);
const hasLineDesc = Object.values(textractLineItem).some(field => field.normalizedLabel === standardizedFieldsnames.line_desc);
const hasQuantity = textractLineItem?.QUANTITY?.value; //We don't normalize quantity, we just use what textract gives us.
// Calculate weighted average, giving more weight to important fields
// If we can identify key fields (ITEM, PRODUCT_CODE, PRICE), weight them higher
@@ -173,10 +205,11 @@ const calculateTextractConfidence = (textractLineItem) => {
if (!hasActualCost) missingCount++;
if (!hasActualPrice) missingCount++;
if (!hasLineDesc) missingCount++;
if (!hasQuantity) missingCount++;
// Each missing field reduces confidence by 15%
// Each missing field reduces confidence by 20%
if (missingCount > 0) {
missingFieldsPenalty = 1.0 - (missingCount * 0.15);
missingFieldsPenalty = 1.0 - (missingCount * 0.20);
}
avgConfidence = avgConfidence * missingFieldsPenalty;
@@ -361,16 +394,16 @@ async function generateBillFormData({ processedData, jobid: jobidFromProps, body
const joblineMatches = joblineFuzzySearch({ fuseToSearch: jobLineDescFuse, processedData });
const vendorFuse = new Fuse(
jobData.vendors,
jobData.vendors.map(v => ({ ...v, name_normalized: normalizeText(v.name) })),
{
keys: ['name'],
threshold: 0.4, //Adjust as needed for matching sensitivity,
keys: [{ name: "name", weight: 3 }, { name: 'name_normalized', weight: 2 }],
threshold: 0.4,
includeScore: true,
},
}
);
const vendorMatches = vendorFuse.search(processedData.summary?.VENDOR_NAME?.value || processedData.summary?.NAME?.value);
const vendorMatches = vendorFuse.search(normalizeText(processedData.summary?.VENDOR_NAME?.value || processedData.summary?.NAME?.value));
let vendorid;
if (vendorMatches.length > 0) {
@@ -381,6 +414,21 @@ async function generateBillFormData({ processedData, jobid: jobidFromProps, body
throw new Error('Job not found for bill form data generation.');
}
//Is there a subtotal level discount? If there is, we need to figure out what the percentage is, and apply that to the actual cost as a reduction
const subtotalDiscountValueRaw = processedData.summary?.DISCOUNT?.value || processedData.summary?.SUBTOTAL_DISCOUNT?.value || 0;
let discountPercentageDecimal = 0;
if (subtotalDiscountValueRaw) {
const subtotal = parseFloat(normalizePrice(processedData.summary?.SUBTOTAL?.value || 0)) || 0;
const subtotalDiscountValue = parseFloat(normalizePrice(subtotalDiscountValueRaw)) || 0;
if (subtotal > 0 && subtotalDiscountValue) {
// Store discount percentage as a decimal (e.g. 20.5% => 0.205),
// but only allow half-percent increments (0.005 steps).
discountPercentageDecimal = Math.abs(subtotalDiscountValue / subtotal);
discountPercentageDecimal = roundToIncrement(discountPercentageDecimal, 0.005);
}
}
//TODO: How do we handle freight lines and core charges?
//Create the form data structure for the bill posting screen.
const billFormData = {
@@ -448,6 +496,31 @@ async function generateBillFormData({ processedData, jobid: jobidFromProps, body
}
}
//If there's nothing, just fall back to seeing if there's a price object from textract.
if (!actualPrice && textractLineItem.PRICE) {
actualPrice = textractLineItem.PRICE.value;
}
if (!actualCost && textractLineItem.PRICE) {
actualCost = textractLineItem.PRICE.value;
}
//If quantity greater than 1, check if the actual cost is a multiple of the actual price, if so, divide it out to get the unit price.
const quantity = parseInt(textractLineItem?.QUANTITY?.value);
if (quantity && quantity > 1) {
if (actualPrice && quantity && Math.abs((actualPrice / quantity) - (parseFloat(matchToUse?.item?.act_price) || 0)) / ((parseFloat(matchToUse?.item?.act_price) || 1)) < PRICE_QUANTITY_MARGIN_TOLERANCE) {
actualPrice = actualPrice / quantity;
}
if (actualCost && quantity && Math.abs((actualCost / quantity) - (parseFloat(matchToUse?.item?.act_price) || 0)) / ((parseFloat(matchToUse?.item?.act_price) || 1)) < PRICE_QUANTITY_MARGIN_TOLERANCE) {
actualCost = actualCost / quantity;
}
}
if (discountPercentageDecimal > 0) {
actualCost = actualCost * (1 - discountPercentageDecimal);
}
const responsibilityCenters = job.bodyshop.md_responsibility_centers
//TODO: Do we need to verify the lines to see if it is a unit price or total price (i.e. quantity * price)
const lineObject = {
@@ -714,5 +787,6 @@ const bodyshopHasDmsKey = (bodyshop) =>
module.exports = {
generateBillFormData
generateBillFormData,
normalizePrice
}

2
server/ai/bill-ocr/bill-ocr-normalize.js
View File

@@ -50,10 +50,12 @@ function normalizeLabelName(labelText) {
'unit_price': standardizedFieldsnames.actual_price,
'list': standardizedFieldsnames.actual_price,
'retail_price': standardizedFieldsnames.actual_price,
'retail': standardizedFieldsnames.actual_price,
'net': standardizedFieldsnames.actual_cost,
'selling_price': standardizedFieldsnames.actual_cost,
'net_price': standardizedFieldsnames.actual_cost,
'net_cost': standardizedFieldsnames.actual_cost,
'total': standardizedFieldsnames.actual_cost,
'po_no': standardizedFieldsnames.ro_number,
'customer_po_no': standardizedFieldsnames.ro_number,
'customer_po_no_': standardizedFieldsnames.ro_number

2
server/ai/bill-ocr/bill-ocr.js
View File

@@ -83,7 +83,7 @@ async function handleBillOcr(req, res) {
// Process synchronously for single-page documents
const processedData = await processSinglePageDocument(uploadedFile.buffer);
const billForm = await generateBillFormData({ processedData: processedData, jobid, bodyshopid, partsorderid, req: req });
logger.log("bill-ocr-single-complete", "DEBUG", req.user.email, jobid, { ...processedData, billForm });
logger.log("bill-ocr-single-complete", "DEBUG", req.user.email, jobid, { ..._.omit(processedData, "originalTextractResponse"), billForm });
return res.status(200).json({
success: true,
status: 'COMPLETED',