@zerocarbon/erp-config-sdk 1.0.28 → 1.0.29

package/dist/index.js

@@ -37900,6 +37900,355 @@ const buildBulkBillAgentPromptConfig = (context = {}, documentText = '') => {
         },
     };
 };
+const numericPattern = /-?(?:(?:\d{1,3}(?:,\d{3})+)|\d+)(?:\.\d+)?/g;
+const physicalUnitPattern = /\b(kwh|kw\s*h|kvah|kva\s*h|mwh|kg|kgs|kilograms?|tonnes?|tons?|mt|kl|litres?|liters?|ltr?s?|m3|m³|scm|ncm|nm3|km|passenger[-\s]?km|room[-\s]?nights?|nights?|cyl(?:inders?)?|nos?|pcs|pieces?)\b/i;
+const metadataOnlyPattern = /(?:₹|rs\.?|inr|amount|payable|charge|tax|gst|duty|cess|rate|tariff|arrear|rebate|subsidy|surcharge|rent|total\s+amount|invoice\s*(?:no|number)|account\s*(?:no|number)|consumer\s*(?:no|number))/i;
+const compactKey = (value) => normalize(value).replace(/\s+/g, '');
+const canonicalActivityUnit = (unit) => {
+    const value = compactKey(unit);
+    if (value === 'kwh' || value === 'unit' || value === 'units')
+        return 'kWh';
+    if (value === 'kvah')
+        return 'kVAh';
+    if (value === 'mwh')
+        return 'MWh';
+    if (value === 'kg' || value === 'kgs' || value === 'kilogram' || value === 'kilograms')
+        return 'Kg';
+    if (value === 'tonne' || value === 'tonnes' || value === 'ton' || value === 'tons' || value === 'mt')
+        return 'Tonnes';
+    if (value === 'litre' || value === 'litres' || value === 'liter' || value === 'liters' || value === 'ltr' || value === 'ltrs')
+        return 'Litre';
+    if (value === 'kl')
+        return 'KL';
+    if (value === 'm3' || value === 'm³' || value === 'scm' || value === 'ncm' || value === 'nm3')
+        return 'M3';
+    if (value === 'km')
+        return 'km';
+    if (value === 'passengerkm')
+        return 'passenger-km';
+    if (value === 'night' || value === 'nights' || value === 'roomnight' || value === 'roomnights')
+        return 'night';
+    if (value === 'cyl' || value === 'cylinder' || value === 'cylinders')
+        return 'Cyl';
+    if (value === 'nos' || value === 'no' || value === 'pcs' || value === 'pieces')
+        return 'Nos';
+    return unit;
+};
+const parsePositiveNumbers = (line) => [...line.matchAll(numericPattern)]
+    .map((match) => ({
+    value: match[0],
+    index: match.index || 0,
+    number: Number(match[0].replace(/,/g, '')),
+}))
+    .filter((entry) => Number.isFinite(entry.number) && entry.number > 0 && !/[A-Za-z]/.test(line[entry.index - 1] || ''));
+const adjacentActivityUnit = (line, numberIndex, valueLength) => {
+    var _a, _b, _c;
+    const after = line.slice(numberIndex + valueLength, numberIndex + valueLength + 18);
+    const right = ((_a = after.match(/^\s*([A-Za-z³][-A-Za-z0-9³ ]{0,16})/i)) === null || _a === void 0 ? void 0 : _a[1]) || '';
+    const rightUnit = (_b = right.match(physicalUnitPattern)) === null || _b === void 0 ? void 0 : _b[1];
+    if (rightUnit)
+        return canonicalActivityUnit(rightUnit);
+    const before = line.slice(Math.max(0, numberIndex - 18), numberIndex);
+    const beforeUnit = (_c = before.match(new RegExp(`(${physicalUnitPattern.source})\\s*$`, 'i'))) === null || _c === void 0 ? void 0 : _c[1];
+    return beforeUnit ? canonicalActivityUnit(beforeUnit) : '';
+};
+const candidateAllowedScore = (query, item) => {
+    const queryKey = normalize(query);
+    const text = itemText(item);
+    if (!queryKey || !text)
+        return 0;
+    const itemNameKey = normalize(item.itemName);
+    if (itemNameKey && (queryKey.includes(itemNameKey) || itemNameKey.includes(queryKey)))
+        return 1;
+    const queryTokens = queryKey.split(' ').filter((token) => token.length > 1);
+    const itemTokens = text.split(' ').filter((token) => token.length > 1);
+    if (!queryTokens.length || !itemTokens.length)
+        return 0;
+    const matched = itemTokens.filter((token) => queryTokens.some((candidate) => token === candidate || token.includes(candidate) || candidate.includes(token))).length;
+    return matched / itemTokens.length;
+};
+const bestAllowedItemForQuery = (query, allowedItems) => {
+    const best = allowedItems
+        .map((item) => ({ item, score: candidateAllowedScore(query, item) }))
+        .sort((left, right) => right.score - left.score)[0];
+    return best && best.score >= 0.34 ? best : null;
+};
+const profileTextLines = (pages = [], group, documentText = '') => {
+    const wanted = new Set(group.pageNumbers.map(Number));
+    const selectedPages = pages.filter((page) => wanted.has(Number(page.pageNumber)));
+    if (!selectedPages.length && documentText) {
+        return documentText.split(/\n+/).map((text, lineIndex) => ({
+            pageNumber: 1,
+            lineIndex,
+            text: text.replace(/\s+/g, ' ').trim(),
+        }));
+    }
+    return selectedPages.flatMap((page) => String(page.text || '').split(/\n+/).map((text, lineIndex) => ({
+        pageNumber: Number(page.pageNumber),
+        lineIndex,
+        text: text.replace(/\s+/g, ' ').trim(),
+    })));
+};
+const buildProfileCandidate = ({ group, profileId, pageNumber, lineIndex, numberIndex, itemName, quantity, unit, evidenceText, allowedMatch, role = 'item_table_quantity', confidence = 0.76, }) => ({
+    candidateId: `${group.groupId}-${profileId}-${pageNumber}-${lineIndex}-${numberIndex}-${compactKey(itemName).slice(0, 24)}`,
+    groupId: group.groupId,
+    profileId,
+    itemName: (allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.item.itemName) || itemName,
+    quantity: quantity.replace(/,/g, ''),
+    unit,
+    role,
+    confidence: Math.min(0.98, confidence + ((allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.score) || 0) * 0.14),
+    pageNumber,
+    evidenceText,
+    matchedAllowedItemName: allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.item.itemName,
+    matchedAllowedItemKey: allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.item.itemKey,
+    matchedAllowedItemSourceId: allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.item.sourceId,
+    mappingConfidence: (allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.score) || undefined,
+});
+const refrigerantLabel = (line) => {
+    const match = line.match(/\b(?:hfc[-\s]*)?r[-\s]?(?:410\s*a?|32|134\s*a?|22|404\s*a?|125|407\s*c?)\b/i);
+    if (!match)
+        return '';
+    return match[0].toUpperCase().replace(/\s+/g, ' ').replace(/^R\s*/, 'R-').replace(/R-410A$/, 'R-410 A');
+};
+const industrialGasLabel = (line) => {
+    if (/\b(?:o2|0?2\s*gas|oxygen)\b/i.test(line))
+        return 'Oxygen Gas';
+    if (/\b(?:co2|c0?2|carbon\s+dioxide)\b/i.test(line))
+        return 'CO2 Gas';
+    if (/\bnitrogen\b/i.test(line))
+        return 'Nitrogen Gas';
+    if (/\bargon\b/i.test(line))
+        return 'Argon Gas';
+    if (/\bacetylene\b/i.test(line))
+        return 'Acetylene';
+    return '';
+};
+const fuelLabel = (line) => {
+    if (/\b(?:hsd|diesel)\b/i.test(line))
+        return 'Diesel';
+    if (/\bpetrol\b/i.test(line))
+        return 'Petrol';
+    if (/\bcng\b/i.test(line))
+        return 'CNG';
+    if (/\bfurnace\s+oil\b/i.test(line))
+        return 'Furnace Oil';
+    if (/\b(?:petroleum\s+coke|petcoke|coke)\b/i.test(line))
+        return 'Petroleum Coke';
+    if (/\bcoal\b/i.test(line))
+        return 'Coal';
+    if (/\b(?:lpg|liquefied\s+petroleum)\b/i.test(line))
+        return 'LPG';
+    if (/\b(?:lubrication\s+oil|lube\s+oil|grease)\b/i.test(line))
+        return 'Lubrication Oil';
+    if (/\bkerosene\b/i.test(line))
+        return 'Kerosene';
+    return '';
+};
+const fireExtinguisherLabel = (line) => /\b(?:fire\s+extinguisher|extinguisher|refill|refilling)\b/i.test(line) ? 'Fire Extinguisher Refilling' : '';
+const sf6Label = (line) => /\b(?:sf6|sulph?ur\s+hexafluoride)\b/i.test(line) ? 'SF6 Usage' : '';
+const waterLabel = (line) => {
+    if (/\b(?:tanker)\b/i.test(line))
+        return 'Tanker Water';
+    if (/\b(?:municipal|corporation|utility)\b/i.test(line))
+        return 'Municipal Water';
+    if (/\b(?:ground\s*water|borewell|bore\s*well)\b/i.test(line))
+        return 'Groundwater';
+    if (/\b(?:wastewater|effluent|sewage)\b/i.test(line))
+        return 'Wastewater';
+    return /\bwater\b/i.test(line) ? 'Water' : '';
+};
+const wasteLabel = (line) => {
+    if (/\bhazardous\b/i.test(line))
+        return 'Hazardous Waste';
+    if (/\b(?:recycl|scrap)\b/i.test(line))
+        return 'Recycled Waste';
+    if (/\bcompost\b/i.test(line))
+        return 'Composted Waste';
+    if (/\blandfill\b/i.test(line))
+        return 'Landfill Waste';
+    return /\bwaste\b/i.test(line) ? 'Waste Generated in Operations' : '';
+};
+const businessTravelLabel = (line) => {
+    if (/\b(?:flight|airline|air\s+travel)\b/i.test(line))
+        return 'Air Travel';
+    if (/\b(?:train|rail|metro)\b/i.test(line))
+        return 'Rail Travel';
+    if (/\b(?:hotel|room\s*night|nights?)\b/i.test(line))
+        return 'Hotel Stay';
+    if (/\b(?:taxi|cab|rental\s+car|car\s+rental)\b/i.test(line))
+        return 'Rental Car';
+    return '';
+};
+const processEmissionLabel = (line) => {
+    if (/\bclinker\b/i.test(line))
+        return 'Clinker';
+    if (/\bcalcination\b/i.test(line))
+        return 'Calcination Process';
+    if (/\b(?:limestone|dolomite|flux)\b/i.test(line))
+        return 'Flux calcination';
+    if (/\b(?:hot\s+metal|ironmaking)\b/i.test(line))
+        return 'Process Emissions- Ironmaking';
+    if (/\bsteelmaking\b/i.test(line))
+        return 'Process Emissions- Steelmaking';
+    return '';
+};
+const genericMaterialLabel = (line, allowedItems) => {
+    if (metadataOnlyPattern.test(line) && !physicalUnitPattern.test(line))
+        return '';
+    const best = bestAllowedItemForQuery(line, allowedItems);
+    return (best === null || best === void 0 ? void 0 : best.item.itemName) || line.split('|').map((part) => part.trim()).find((part) => /[A-Za-z]/.test(part) && !metadataOnlyPattern.test(part)) || '';
+};
+const commuteModeLabel = (line) => {
+    const text = normalize(line);
+    if (/\b(bike|bikt|motorcycle)\b/.test(text))
+        return 'Motorcycle';
+    if (/\b(scooty|scoow|scooter)\b/.test(text))
+        return 'Scooter';
+    if (/\b(auto|byauto|auio)\b/.test(text))
+        return 'Small Car Gasoline';
+    if (/\b(car|eycar|hatchback)\b/.test(text))
+        return 'Hatchback Gasoline';
+    if (/\b(metro|14etro|train|rail)\b/.test(text))
+        return 'Train Domestic Electric';
+    if (/\b(cycle|cycl|cyct)\b/.test(text))
+        return 'Employee commute cycle';
+    if (/\b(bus)\b/.test(text))
+        return 'Bus';
+    return '';
+};
+const extractNamedQuantityCandidates = ({ profileId, group, lines, allowedItems, labelForLine, }) => lines.flatMap(({ pageNumber, lineIndex, text }) => {
+    if (!text || (metadataOnlyPattern.test(text) && !physicalUnitPattern.test(text)))
+        return [];
+    const itemName = labelForLine(text, allowedItems);
+    if (!itemName)
+        return [];
+    return parsePositiveNumbers(text).flatMap((entry, numberIndex) => {
+        const unit = adjacentActivityUnit(text, entry.index, entry.value.length);
+        if (!unit)
+            return [];
+        const allowedMatch = bestAllowedItemForQuery(`${itemName} ${text}`, allowedItems);
+        return [buildProfileCandidate({
+                group,
+                profileId,
+                pageNumber,
+                lineIndex,
+                numberIndex,
+                itemName,
+                quantity: entry.value,
+                unit,
+                evidenceText: text,
+                allowedMatch,
+            })];
+    });
+});
+const extractElectricityCandidates = (group, lines, allowedItems) => {
+    const electricityLine = /\b(?:net\s+cons\.?\s+for\s+billing|billed\s+units?|billing\s+units?|units\s+consumed|energy\s+consumption|import|export|generation|solar|renewable)\b/i;
+    return lines.flatMap(({ pageNumber, lineIndex, text }) => {
+        if (!electricityLine.test(text) || metadataOnlyPattern.test(text))
+            return [];
+        return parsePositiveNumbers(text).flatMap((entry, numberIndex) => {
+            const unit = adjacentActivityUnit(text, entry.index, entry.value.length) || (/\bunits?\b/i.test(text) ? 'kWh' : '');
+            if (!unit || !/\b(?:kwh|kvah|mwh)\b/i.test(unit))
+                return [];
+            const role = /\b(?:solar|renewable|generation|export)\b/i.test(text) ? 'component_consumption' : 'bill_level_consumption';
+            const allowedMatch = bestAllowedItemForQuery(text, allowedItems);
+            return [buildProfileCandidate({
+                    group,
+                    profileId: 'electricity',
+                    pageNumber,
+                    lineIndex,
+                    numberIndex,
+                    itemName: (allowedMatch === null || allowedMatch === void 0 ? void 0 : allowedMatch.item.itemName) || 'Electricity Purchased from Grid',
+                    quantity: entry.value,
+                    unit,
+                    evidenceText: text,
+                    allowedMatch,
+                    role,
+                    confidence: role === 'bill_level_consumption' ? 0.86 : 0.78,
+                })];
+        });
+    });
+};
+const extractEmployeeCommutingCandidates = (group, lines, allowedItems) => {
+    const totals = new Map();
+    for (const { pageNumber, text } of lines) {
+        const mode = commuteModeLabel(text);
+        if (!mode)
+            continue;
+        const distanceValues = parsePositiveNumbers(text)
+            .filter((entry) => /\bkm\b/i.test(text.slice(entry.index, entry.index + 24)) || /\bkm\b/i.test(text));
+        const value = distanceValues.length ? distanceValues[distanceValues.length - 1].number : null;
+        if (!value)
+            continue;
+        const current = totals.get(mode) || { total: 0, evidence: [], pageNumber };
+        current.total += value;
+        if (current.evidence.length < 4)
+            current.evidence.push(text);
+        totals.set(mode, current);
+    }
+    return [...totals.entries()].map(([mode, aggregate], index) => {
+        const allowedMatch = bestAllowedItemForQuery(mode, allowedItems);
+        return buildProfileCandidate({
+            group,
+            profileId: 'employee_commuting',
+            pageNumber: aggregate.pageNumber,
+            lineIndex: index,
+            numberIndex: 0,
+            itemName: mode,
+            quantity: String(Number(aggregate.total.toFixed(5))),
+            unit: 'km',
+            evidenceText: aggregate.evidence.join(' | '),
+            allowedMatch,
+            role: 'aggregated_distance',
+            confidence: 0.88,
+        });
+    });
+};
+const extractForProfile = (profileId, group, lines, allowedItems) => {
+    if (profileId === 'electricity')
+        return extractElectricityCandidates(group, lines, allowedItems);
+    if (profileId === 'employee_commuting')
+        return extractEmployeeCommutingCandidates(group, lines, allowedItems);
+    const labelForProfile = {
+        fuel_combustion: (line) => fuelLabel(line),
+        lpg_usage: (line) => fuelLabel(line),
+        refrigerants: (line) => refrigerantLabel(line),
+        industrial_gases: (line) => industrialGasLabel(line),
+        business_travel: (line) => businessTravelLabel(line),
+        fire_extinguisher: (line) => fireExtinguisherLabel(line),
+        sf6_usage: (line) => sf6Label(line),
+        water: (line) => waterLabel(line),
+        waste: (line) => wasteLabel(line),
+        process_emissions: (line) => processEmissionLabel(line),
+        purchased_goods: genericMaterialLabel,
+        generic_material: genericMaterialLabel,
+    };
+    const labelForLine = labelForProfile[profileId] || genericMaterialLabel;
+    return extractNamedQuantityCandidates({ profileId, group, lines, allowedItems, labelForLine });
+};
+const extractBulkBillAgentProfileCandidates = ({ mappingContext = {}, pages = [], group, documentText = '', }) => {
+    const allowedItems = Array.isArray(mappingContext.allowedItems) ? mappingContext.allowedItems : [];
+    const lines = profileTextLines(pages, group, documentText);
+    const groupText = documentText || lines.map((line) => line.text).join('\n');
+    const primaryProfile = getPrimaryBulkBillAgentProfile(mappingContext, groupText);
+    const profiles = getBulkBillAgentProfilesForMappingContext(mappingContext, groupText)
+        .map((entry) => entry.profile.id)
+        .filter((profileId, index, profileIds) => profileIds.indexOf(profileId) === index);
+    const extractionProfiles = profiles.length ? profiles : [primaryProfile.id];
+    const candidates = extractionProfiles.flatMap((profileId) => extractForProfile(profileId, group, lines, allowedItems));
+    const seen = new Set();
+    return candidates
+        .sort((left, right) => right.confidence - left.confidence)
+        .filter((candidate) => {
+        const key = [candidate.profileId, candidate.matchedAllowedItemKey || candidate.itemName, candidate.quantity, candidate.unit, candidate.pageNumber].join('|');
+        if (seen.has(key))
+            return false;
+        seen.add(key);
+        return true;
+    })
+        .slice(0, 30);
+};
 
 // User-specific configurations mapping
 const USER_CONFIGS = {
@@ -44747,6 +45096,7 @@ exports.createCapitalizedMonthMapping = createCapitalizedMonthMapping;
 exports.createEmissionSourceGenerator = createEmissionSourceGenerator;
 exports.createMonthMapping = createMonthMapping;
 exports.debugCarbonIntensityMapping = debugCarbonIntensityMapping;
+exports.extractBulkBillAgentProfileCandidates = extractBulkBillAgentProfileCandidates;
 exports.extractCarbonIntensityData = extractCarbonIntensityData;
 exports.formatEmissionValue = formatEmissionValue;
 exports.formatIntensityValue = formatIntensityValue;