@parseo/appraisals 1.0.0

package/dist/form-1073/parse-sales.js

@@ -0,0 +1,477 @@
+import { toBBox } from "@parseo/shared";
+// ── Utilities ────────────────────────────────────────────────────────────
+function parseNum(raw) {
+    if (!raw)
+        return null;
+    const cleaned = raw.replace(/[$,%]/g, "").replace(/,/g, "").trim();
+    if (!cleaned || /^n\/?a$/i.test(cleaned))
+        return null;
+    const n = Number(cleaned);
+    return Number.isNaN(n) ? null : n;
+}
+function parseCurrency(raw) {
+    const match = raw.match(/\$?\s*([\d,]+(?:\.\d+)?)/);
+    return match ? parseNum(match[1]) : null;
+}
+// Column x-boundaries for Form 1073 sales comparison grid
+// Derived from segment positions in the sample PDF
+const SUBJECT_X = { min: 114, max: 180 };
+const COMP_COLS = [
+    { desc: { min: 180, max: 260 }, adj: { min: 260, max: 313 } },
+    { desc: { min: 313, max: 392 }, adj: { min: 392, max: 444 } },
+    { desc: { min: 444, max: 524 }, adj: { min: 524, max: 580 } },
+];
+function segInRange(line, min, max) {
+    const segs = line.segments.filter((s) => s.x >= min && s.x < max);
+    return segs.map((s) => s.text.trim()).join(" ").trim();
+}
+function segWithBBox(line, min, max) {
+    const seg = line.segments.find((s) => s.x >= min && s.x < max);
+    return { text: seg ? seg.text.trim() : "", seg: seg ?? null };
+}
+function findLine(lines, pattern) {
+    return lines.find((l) => pattern.test(l.fullText));
+}
+// ── Parse single comparable from a column ────────────────────────────────
+function parseComparable(lines, compNum, col) {
+    const bb = {};
+    const prefix = `comp${compNum}`;
+    function getDesc(line) {
+        return line ? segInRange(line, col.desc.min, col.adj.min) : "";
+    }
+    function getAdj(line) {
+        if (!line)
+            return null;
+        const text = segInRange(line, col.adj.min, col.adj.min + 60);
+        return parseNum(text);
+    }
+    function getDescWithBB(line, key) {
+        if (!line)
+            return "";
+        const r = segWithBBox(line, col.desc.min, col.adj.min);
+        if (r.seg)
+            bb[`${prefix}_${key}`] = toBBox(r.seg, line);
+        return r.text;
+    }
+    const addrLine = findLine(lines, /^Address/i);
+    const address = getDescWithBB(addrLine, "address");
+    const addrIdx = addrLine ? lines.indexOf(addrLine) : -1;
+    const cityLine = addrIdx >= 0 ? lines[addrIdx + 1] : undefined;
+    const cityVal = cityLine ? segInRange(cityLine, col.desc.min, col.adj.min) : "";
+    const projLine = findLine(lines, /^Project Name/i);
+    const projectNamePhase = getDesc(projLine);
+    const proxLine = findLine(lines, /^Proximity to Subject/i);
+    const proximityToSubject = getDescWithBB(proxLine, "proximity");
+    const priceLine = findLine(lines, /^Sale Price\b/i);
+    let salePrice = null;
+    if (priceLine) {
+        const priceText = segInRange(priceLine, col.desc.min - 30, col.adj.min);
+        salePrice = parseCurrency(priceText);
+        const pSeg = segWithBBox(priceLine, col.desc.min, col.adj.min);
+        if (pSeg.seg)
+            bb[`${prefix}_salePrice`] = toBBox(pSeg.seg, priceLine);
+    }
+    const ppsLine = findLine(lines, /^Sale Price\/Gross Liv/i);
+    let salePricePerSqft = null;
+    if (ppsLine) {
+        const txt = segInRange(ppsLine, col.desc.min - 30, col.adj.min);
+        const m = txt.match(/([\d,.]+)\s*sq\.?\s*ft/i);
+        if (m)
+            salePricePerSqft = parseNum(m[1]);
+    }
+    const dsLine = findLine(lines, /^Data Source\(s\)/i);
+    const dataSources = getDescWithBB(dsLine, "dataSources");
+    const vsLine = findLine(lines, /^Verification Source/i);
+    const verificationSources = getDesc(vsLine);
+    const sfLine = findLine(lines, /^Sales or Financing/i);
+    const salesOrFinancing = getDesc(sfLine);
+    const concLine = findLine(lines, /^Concessions/i);
+    const concessions = getDesc(concLine);
+    const dateLine = findLine(lines, /^Date of Sale\/Time/i);
+    const dateOfSaleTime = getDescWithBB(dateLine, "dateOfSale");
+    const locLine = findLine(lines, /^Location\b/i);
+    const location = getDesc(locLine);
+    const feeLine = findLine(lines, /^Leasehold\/Fee Simple/i);
+    const leaseholdFeeSimple = getDesc(feeLine);
+    const hoaLine = findLine(lines, /^HOA Mo\. Assessment/i);
+    const hoaMoAssessment = hoaLine ? parseNum(getDesc(hoaLine)) : null;
+    const commLine = findLine(lines, /^Common Elements/i);
+    const commonElements = getDesc(commLine);
+    const floorLine = findLine(lines, /^Floor Location/i);
+    const floorLocation = getDesc(floorLine);
+    const viewLine = findLine(lines, /^View\b/i);
+    const view = getDesc(viewLine);
+    const styleLine = findLine(lines, /^Design \(Style\)/i);
+    const designStyle = getDesc(styleLine);
+    const qualLine = findLine(lines, /^Quality of Construction/i);
+    const qualityOfConstruction = getDesc(qualLine);
+    const ageLine = findLine(lines, /^Actual Age\b/i);
+    const actualAge = parseNum(getDesc(ageLine));
+    const condLine = findLine(lines, /^Condition\b/i);
+    const condition = getDescWithBB(condLine, "condition");
+    const conditionAdjustment = getAdj(condLine);
+    const rcLine = findLine(lines, /^Room Count/i);
+    let roomCountTotal = null, roomCountBedrooms = null, roomCountBaths = null, roomCountAdjustment = null;
+    if (rcLine) {
+        const text = segInRange(rcLine, col.desc.min, col.adj.min);
+        const parts = text.split(/\s+/).map((s) => parseNum(s)).filter((n) => n !== null);
+        if (parts.length >= 3) {
+            roomCountTotal = parts[0];
+            roomCountBedrooms = parts[1];
+            roomCountBaths = parts[2];
+        }
+        roomCountAdjustment = getAdj(rcLine);
+    }
+    const glaLine = findLine(lines, /^Gross Living Area/i);
+    let grossLivingArea = null, grossLivingAreaAdjustment = null;
+    if (glaLine) {
+        const text = segInRange(glaLine, col.desc.min - 30, col.adj.min);
+        const m = text.match(/([\d,]+)\s*sq/i);
+        if (m)
+            grossLivingArea = parseNum(m[1]);
+        grossLivingAreaAdjustment = getAdj(glaLine);
+        const glaSeg = segWithBBox(glaLine, col.desc.min, col.adj.min);
+        if (glaSeg.seg)
+            bb[`${prefix}_gla`] = toBBox(glaSeg.seg, glaLine);
+    }
+    const bfLine = findLine(lines, /^Basement & Finished/i);
+    const basementFinished = getDesc(bfLine);
+    const fuLine = findLine(lines, /^Functional Utility/i);
+    const functionalUtility = getDesc(fuLine);
+    const hcLine = findLine(lines, /^Heating\/Cooling/i);
+    const heatingCooling = getDesc(hcLine);
+    const eeLine = findLine(lines, /^Energy Efficient/i);
+    const energyEfficientItems = getDesc(eeLine);
+    const gcLine = findLine(lines, /^Garage\/Carport/i);
+    const garageCarport = getDesc(gcLine);
+    const ppLine = findLine(lines, /^Porch\/Patio\/Deck/i);
+    const porchPatioDeck = getDesc(ppLine);
+    // Net/Gross adjustment
+    const netLine = findLine(lines, /^Net Adjustment \(Total\)/i);
+    let netAdjustmentTotal = null;
+    if (netLine) {
+        const text = segInRange(netLine, col.adj.min - 10, col.adj.min + 60);
+        netAdjustmentTotal = parseNum(text);
+    }
+    const adjPriceLine = findLine(lines, /^Adjusted Sale Price/i);
+    let netAdjustmentPercent = null;
+    if (adjPriceLine) {
+        const text = segInRange(adjPriceLine, col.desc.min, col.adj.min + 20);
+        const pctMatch = text.match(/([\d.]+)\s*%/);
+        if (pctMatch)
+            netAdjustmentPercent = parseNum(pctMatch[1]);
+    }
+    const grossLine = findLine(lines, /^of Comparables/i);
+    let grossAdjustmentPercent = null, adjustedSalePrice = null;
+    if (grossLine) {
+        const text = segInRange(grossLine, col.desc.min, col.adj.min + 60);
+        const pctMatch = text.match(/([\d.]+)\s*%/);
+        if (pctMatch)
+            grossAdjustmentPercent = parseNum(pctMatch[1]);
+        const priceMatch = text.match(/\$\s*([\d,]+)/);
+        if (priceMatch)
+            adjustedSalePrice = parseNum(priceMatch[1]);
+    }
+    return {
+        number: compNum,
+        address: address + (cityVal ? `, ${cityVal}` : ""),
+        projectNamePhase, proximityToSubject, salePrice, salePricePerSqft,
+        dataSources, verificationSources, salesOrFinancing, concessions, dateOfSaleTime,
+        location, leaseholdFeeSimple, hoaMoAssessment, commonElements, floorLocation,
+        view, designStyle, qualityOfConstruction, actualAge,
+        condition, conditionAdjustment,
+        roomCountTotal, roomCountBedrooms, roomCountBaths, roomCountAdjustment,
+        grossLivingArea, grossLivingAreaAdjustment, basementFinished,
+        functionalUtility, heatingCooling, energyEfficientItems, garageCarport, porchPatioDeck,
+        netAdjustmentTotal, netAdjustmentPercent, grossAdjustmentPercent, adjustedSalePrice,
+        boundingBoxes: bb,
+    };
+}
+// ── Sales Comparison Section ─────────────────────────────────────────────
+export function parseSalesComparisonSection(lines, compStartNum = 1) {
+    const bb = {};
+    // Active listings / comparable sales ranges
+    const alLine = findLine(lines, /comparable properties currently offered/i);
+    let activeListingsCount = null, activeListingsLow = null, activeListingsHigh = null;
+    if (alLine) {
+        const countMatch = alLine.fullText.match(/are\s+(\d+)\s+comparable properties/i);
+        if (countMatch)
+            activeListingsCount = parseNum(countMatch[1]);
+        const rangeMatch = alLine.fullText.match(/\$\s*([\d,]+).*?to\s*\$\s*([\d,]+)/i);
+        if (rangeMatch) {
+            activeListingsLow = parseNum(rangeMatch[1]);
+            activeListingsHigh = parseNum(rangeMatch[2]);
+        }
+    }
+    const csLine = findLine(lines, /comparable sales in the subject/i);
+    let comparableSalesCount = null, comparableSalesLow = null, comparableSalesHigh = null;
+    if (csLine) {
+        const countMatch = csLine.fullText.match(/are\s+(\d+)\s+comparable sales/i);
+        if (countMatch)
+            comparableSalesCount = parseNum(countMatch[1]);
+        const rangeMatch = csLine.fullText.match(/\$\s*([\d,]+).*?to\s*\$\s*([\d,]+)/i);
+        if (rangeMatch) {
+            comparableSalesLow = parseNum(rangeMatch[1]);
+            comparableSalesHigh = parseNum(rangeMatch[2]);
+        }
+    }
+    // Subject column
+    const subBB = {};
+    function getSubject(line) {
+        return line ? segInRange(line, SUBJECT_X.min, SUBJECT_X.max) : "";
+    }
+    function getSubjectWithBB(line, key) {
+        if (!line)
+            return "";
+        const r = segWithBBox(line, SUBJECT_X.min, SUBJECT_X.max);
+        if (r.seg)
+            subBB[key] = toBBox(r.seg, line);
+        return r.text;
+    }
+    const addrLine = findLine(lines, /^Address/i);
+    const subjAddress = getSubjectWithBB(addrLine, "address");
+    const projLine = findLine(lines, /^Project Name/i);
+    const subjProject = getSubject(projLine);
+    const priceLine = findLine(lines, /^Sale Price\b/i);
+    const subjPrice = priceLine ? parseCurrency(segInRange(priceLine, SUBJECT_X.min - 20, SUBJECT_X.max)) : null;
+    const ppsLine = findLine(lines, /^Sale Price\/Gross Liv/i);
+    let subjPPS = null;
+    if (ppsLine) {
+        const m = segInRange(ppsLine, SUBJECT_X.min, SUBJECT_X.max).match(/([\d,.]+)/);
+        if (m)
+            subjPPS = parseNum(m[1]);
+    }
+    const hoaLine = findLine(lines, /^HOA Mo\. Assessment/i);
+    const subjHOA = hoaLine ? parseNum(getSubject(hoaLine)) : null;
+    const rcLine = findLine(lines, /^Room Count/i);
+    let subjRCTotal = null, subjRCBed = null, subjRCBath = null;
+    if (rcLine) {
+        const text = segInRange(rcLine, SUBJECT_X.min, SUBJECT_X.max);
+        const parts = text.split(/\s+/).map((s) => parseNum(s)).filter((n) => n !== null);
+        if (parts.length >= 3) {
+            subjRCTotal = parts[0];
+            subjRCBed = parts[1];
+            subjRCBath = parts[2];
+        }
+    }
+    const glaLine = findLine(lines, /^Gross Living Area/i);
+    let subjGLA = null;
+    if (glaLine) {
+        const m = segInRange(glaLine, SUBJECT_X.min, SUBJECT_X.max).match(/([\d,]+)/);
+        if (m)
+            subjGLA = parseNum(m[1]);
+    }
+    const subject = {
+        address: subjAddress,
+        projectNamePhase: subjProject,
+        salePrice: subjPrice,
+        salePricePerSqft: subjPPS,
+        location: getSubject(findLine(lines, /^Location\b/i)),
+        leaseholdFeeSimple: getSubject(findLine(lines, /^Leasehold\/Fee/i)),
+        hoaMoAssessment: subjHOA,
+        commonElements: getSubject(findLine(lines, /^Common Elements/i)),
+        floorLocation: getSubject(findLine(lines, /^Floor Location/i)),
+        view: getSubject(findLine(lines, /^View\b/i)),
+        designStyle: getSubject(findLine(lines, /^Design \(Style\)/i)),
+        qualityOfConstruction: getSubject(findLine(lines, /^Quality of Construction/i)),
+        actualAge: parseNum(getSubject(findLine(lines, /^Actual Age/i))),
+        condition: getSubjectWithBB(findLine(lines, /^Condition\b/i), "condition"),
+        roomCountTotal: subjRCTotal,
+        roomCountBedrooms: subjRCBed,
+        roomCountBaths: subjRCBath,
+        grossLivingArea: subjGLA,
+        basementFinished: getSubject(findLine(lines, /^Basement & Finished/i)),
+        functionalUtility: getSubject(findLine(lines, /^Functional Utility/i)),
+        heatingCooling: getSubject(findLine(lines, /^Heating\/Cooling/i)),
+        energyEfficientItems: getSubject(findLine(lines, /^Energy Efficient/i)),
+        garageCarport: getSubject(findLine(lines, /^Garage\/Carport/i)),
+        porchPatioDeck: getSubject(findLine(lines, /^Porch\/Patio/i)),
+        boundingBoxes: subBB,
+    };
+    // Parse 3 comparables
+    const comparables = [];
+    for (let i = 0; i < 3; i++) {
+        comparables.push(parseComparable(lines, compStartNum + i, COMP_COLS[i]));
+    }
+    // Summary of Sales Comparison
+    const sumLine = findLine(lines, /^Summary of Sales Comparison/i);
+    let summaryOfSalesComparison = "";
+    if (sumLine) {
+        const idx = lines.indexOf(sumLine);
+        const parts = [];
+        const valueSeg = sumLine.segments.find((s) => s.x > 150);
+        if (valueSeg)
+            parts.push(valueSeg.text.trim());
+        for (let i = idx + 1; i < lines.length; i++) {
+            if (/^Indicated Value|^INCOME/i.test(lines[i].fullText))
+                break;
+            const t = lines[i].fullText.trim();
+            if (t)
+                parts.push(t);
+        }
+        summaryOfSalesComparison = parts.join(" ").trim();
+        if (sumLine.segments[0])
+            bb.summaryOfSalesComparison = toBBox(sumLine.segments[0], sumLine);
+    }
+    // Indicated Value by Sales Comparison
+    const indLine = findLine(lines, /^Indicated Value by Sales Comparison Approach\s*\$/i);
+    let indicatedValueBySalesComparison = null;
+    if (indLine) {
+        const seg = indLine.segments.find((s) => s.x > 150 && /^\d/.test(s.text.trim()));
+        if (seg) {
+            indicatedValueBySalesComparison = parseNum(seg.text);
+            bb.indicatedBySales = toBBox(seg, indLine);
+        }
+    }
+    return {
+        activeListingsCount, activeListingsLow, activeListingsHigh,
+        comparableSalesCount, comparableSalesLow, comparableSalesHigh,
+        subject, comparables, summaryOfSalesComparison, indicatedValueBySalesComparison,
+        boundingBoxes: bb,
+    };
+}
+// ── Reconciliation ───────────────────────────────────────────────────────
+export function parseReconciliationSection(lines) {
+    const bb = {};
+    const indAllLine = findLine(lines, /^Indicated Value by: Sales Comparison/i);
+    let indicatedValueBySalesComparison = null, indicatedValueByIncomeApproach = null;
+    if (indAllLine) {
+        const salesSeg = indAllLine.segments.find((s) => s.x > 180 && s.x < 260 && /^\d/.test(s.text.trim()));
+        if (salesSeg) {
+            indicatedValueBySalesComparison = parseNum(salesSeg.text);
+            bb.indicatedBySales = toBBox(salesSeg, indAllLine);
+        }
+        for (const seg of indAllLine.segments) {
+            if (/Income Approach.*\$\s*([\d,]+)/i.test(seg.text)) {
+                const m = seg.text.match(/\$\s*([\d,]+)/);
+                if (m)
+                    indicatedValueByIncomeApproach = parseNum(m[1]);
+            }
+        }
+    }
+    // Reconciliation comments
+    const recLine = findLine(lines, /The Direct Sales Comparison|ALL WEIGHT GIVEN|reconciliation/i);
+    let reconciliationComments = "";
+    if (recLine) {
+        const idx = lines.indexOf(recLine);
+        const parts = [recLine.fullText.trim()];
+        for (let i = idx + 1; i < lines.length; i++) {
+            if (/^This appraisal is made|^RECONCILIATION/i.test(lines[i].fullText))
+                break;
+            const t = lines[i].fullText.trim();
+            if (t)
+                parts.push(t);
+        }
+        reconciliationComments = parts.join(" ").trim();
+    }
+    // Appraisal basis
+    const basisLine = findLine(lines, /This appraisal is made/i);
+    let appraisalBasis = "";
+    if (basisLine) {
+        const idx = lines.indexOf(basisLine);
+        const parts = [basisLine.fullText.replace(/^RECONCILIATION/i, "").trim()];
+        for (let i = idx + 1; i < lines.length; i++) {
+            if (/^Based on a complete|^\$/i.test(lines[i].fullText))
+                break;
+            parts.push(lines[i].fullText.trim());
+        }
+        appraisalBasis = parts.join(" ").trim();
+    }
+    // Final value and effective date
+    const valueLine = findLine(lines, /^\$\s*[\d,]+\s*,\s*as of/i);
+    let finalValue = null, effectiveDate = "";
+    if (valueLine) {
+        for (const seg of valueLine.segments) {
+            if (/^\d[\d,]*$/.test(seg.text.trim())) {
+                finalValue = parseNum(seg.text);
+                bb.finalValue = toBBox(seg, valueLine);
+            }
+            if (/^\d{2}\/\d{2}\/\d{4}$/.test(seg.text.trim())) {
+                effectiveDate = seg.text.trim();
+                bb.effectiveDate = toBBox(seg, valueLine);
+            }
+        }
+    }
+    return { indicatedValueBySalesComparison, indicatedValueByIncomeApproach, reconciliationComments, appraisalBasis, finalValue, effectiveDate, boundingBoxes: bb };
+}
+// ── Appraiser Info (Page 6) ──────────────────────────────────────────────
+export function parseAppraiserInfo(lines, supervisory) {
+    const bb = {};
+    const xMin = supervisory ? 290 : 0;
+    const xMax = supervisory ? 600 : 290;
+    function findFieldInRange(label) {
+        for (const l of lines) {
+            const seg = l.segments.find((s) => s.x >= xMin && s.x < xMax && label.test(s.text));
+            if (seg) {
+                const val = seg.text.replace(label, "").trim();
+                if (val)
+                    return val;
+                const next = l.segments.find((s) => s.x > seg.x && s.x < xMax && !label.test(s.text));
+                if (next)
+                    return next.text.trim();
+            }
+        }
+        return "";
+    }
+    const name = findFieldInRange(/^(Supervisory )?Appraiser\s*$|^Name\s+/i);
+    // For supervisory: if name is empty, there is no supervisory appraiser
+    if (supervisory && !name)
+        return null;
+    const nameLine = lines.find((l) => l.segments.some((s) => s.x >= xMin && s.x < xMax && /^Name\s/i.test(s.text)));
+    let parsedName = "";
+    if (nameLine) {
+        const seg = nameLine.segments.find((s) => s.x >= xMin && s.x < xMax && /^Name\s/i.test(s.text));
+        if (seg) {
+            parsedName = seg.text.replace(/^Name\s+/i, "").trim();
+            bb.name = toBBox(seg, nameLine);
+        }
+    }
+    return {
+        name: parsedName || name,
+        companyName: findFieldInRange(/^Company Name\s*/i),
+        companyAddress: findFieldInRange(/^Company Address\s*/i),
+        telephoneNumber: findFieldInRange(/^Telephone Number\s*/i),
+        emailAddress: findFieldInRange(/^Email Address\s*/i),
+        dateOfSignature: findFieldInRange(/^Date of Signature( and Report)?\s*/i),
+        effectiveDateOfAppraisal: findFieldInRange(/^Effective Date of Appraisal\s*/i),
+        stateCertification: findFieldInRange(/^State Certification #\s*/i),
+        stateOrLicense: findFieldInRange(/^or State License #\s*/i),
+        state: findFieldInRange(/^State\s+(?!Certification|License)/i),
+        expirationDate: findFieldInRange(/^Expiration Date of Certification or License\s*/i),
+        boundingBoxes: bb,
+    };
+}
+// ── Lender/Client Info (Page 6) ──────────────────────────────────────────
+export function parseLenderClientInfo(lines) {
+    const bb = {};
+    // Lender info is in the lower-left area of page 6
+    const lenderHeader = findLine(lines, /^LENDER\/CLIENT$/i);
+    if (!lenderHeader)
+        return { name: "", companyName: "", companyAddress: "", emailAddress: "", boundingBoxes: bb };
+    const startY = lenderHeader.y;
+    const lenderLines = lines.filter((l) => l.y > startY && l.segments.some((s) => s.x < 290));
+    function getField(label) {
+        for (const l of lenderLines) {
+            for (const seg of l.segments) {
+                if (seg.x >= 290)
+                    continue;
+                if (label.test(seg.text)) {
+                    const val = seg.text.replace(label, "").trim();
+                    if (val)
+                        return val;
+                    const next = l.segments.find((s) => s.x > seg.x && s.x < 290 && !label.test(s.text));
+                    if (next)
+                        return next.text.trim();
+                }
+            }
+        }
+        return "";
+    }
+    return {
+        name: getField(/^Name\s*/i),
+        companyName: getField(/^Company Name\s*/i),
+        companyAddress: getField(/^Company Address\s*/i),
+        emailAddress: getField(/^Email Address\s*/i),
+        boundingBoxes: bb,
+    };
+}

package/dist/form-1073/parser.d.ts

@@ -0,0 +1,5 @@
+import type { TextLine } from "@parseo/shared";
+import type { Form1073Report } from "./types.js";
+export declare function parseForm1073(buffer: Buffer): Promise<Form1073Report>;
+export declare function parseForm1073FromLines(lines: TextLine[], buffer?: Buffer, pageOffset?: number): Promise<Form1073Report>;
+//# sourceMappingURL=parser.d.ts.map

package/dist/form-1073/parser.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"parser.d.ts","sourceRoot":"","sources":["../../src/form-1073/parser.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,QAAQ,EAAc,MAAM,gBAAgB,CAAC;AAC3D,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,YAAY,CAAC;AAKjD,wBAAsB,aAAa,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,cAAc,CAAC,CAG3E;AAED,wBAAsB,sBAAsB,CAAC,KAAK,EAAE,QAAQ,EAAE,EAAE,MAAM,CAAC,EAAE,MAAM,EAAE,UAAU,SAAI,GAAG,OAAO,CAAC,cAAc,CAAC,CAyGxH"}

package/dist/form-1073/parser.js

@@ -0,0 +1,102 @@
+import { extractLines, extractFilledRects, UnrecognizedFormatError } from "@parseo/shared";
+import { parseSubjectSection, parseContractSection, parseNeighborhoodSection, parseProjectSiteSection, parseProjectInfoSection } from "./parse-page1.js";
+import { parseProjectAnalysisSection, parseUnitDescriptionSection, parsePriorSaleHistorySection } from "./parse-page2.js";
+import { parseSalesComparisonSection, parseReconciliationSection, parseAppraiserInfo, parseLenderClientInfo } from "./parse-sales.js";
+export async function parseForm1073(buffer) {
+    const lines = await extractLines(buffer);
+    return parseForm1073FromLines(lines, buffer);
+}
+export async function parseForm1073FromLines(lines, buffer, pageOffset = 0) {
+    // Format fingerprint: must have "Individual Condominium Unit Appraisal Report"
+    // or "Form 1073" in the first 30 lines
+    const head = lines.slice(0, 30).map((l) => l.fullText).join("\n");
+    if (!/Individual Condominium Unit Appraisal Report|Form 1073/i.test(head)) {
+        throw new UnrecognizedFormatError("Form1073", "first 30 lines do not contain a Form 1073 / Individual Condominium Unit Appraisal Report signature");
+    }
+    // Find the page that starts the main form (has the title)
+    let formStartPage = 1;
+    for (const l of lines) {
+        if (/Individual Condominium Unit Appraisal Report/i.test(l.fullText)) {
+            formStartPage = l.page;
+            break;
+        }
+    }
+    // If the form doesn't start on page 1, remap page numbers
+    let internalOffset = 0;
+    let workLines = lines;
+    if (formStartPage > 1) {
+        internalOffset = formStartPage - 1;
+        workLines = lines
+            .filter((l) => l.page >= formStartPage)
+            .map((l) => ({ ...l, page: l.page - internalOffset }));
+    }
+    // ── Extract checkbox rects from the PDF graphics layer ──
+    let checkboxRects = [];
+    if (buffer) {
+        // The real PDF page accounts for both server-level skipped pages and internal form offset
+        const pdfPage1 = pageOffset + internalOffset + 1;
+        checkboxRects = await extractFilledRects(buffer, [pdfPage1], { minSize: 3, maxSize: 10 });
+    }
+    // ── Page 1: Subject, Contract, Neighborhood, Project Site, Project Info ──
+    const page1 = workLines.filter((l) => l.page === 1);
+    const subject = parseSubjectSection(page1);
+    const contract = parseContractSection(page1);
+    const neighborhood = parseNeighborhoodSection(page1, checkboxRects);
+    const projectSite = parseProjectSiteSection(page1);
+    const projectInfo = parseProjectInfoSection(page1);
+    // ── Page 2: Project Analysis, Unit Description, Prior Sale History ──
+    const page2 = workLines.filter((l) => l.page === 2);
+    const projectAnalysis = parseProjectAnalysisSection(page2);
+    const unitDescription = parseUnitDescriptionSection(page2);
+    const priorSaleHistory = parsePriorSaleHistorySection(page2);
+    // ── Page 3: Sales Comparison (comps 1-3), Reconciliation ──
+    const page3 = workLines.filter((l) => l.page === 3);
+    const salesComparison = parseSalesComparisonSection(page3);
+    const reconciliation = parseReconciliationSection(page3);
+    // ── Pages 4-5: Definitions/Certifications (boilerplate, skip) ──
+    // ── Page 6: Appraiser info ──
+    const page6 = workLines.filter((l) => l.page === 6);
+    const appraiser = parseAppraiserInfo(page6, false) ?? {
+        name: "", companyName: "", companyAddress: "", telephoneNumber: "",
+        emailAddress: "", dateOfSignature: "", effectiveDateOfAppraisal: "",
+        stateCertification: "", stateOrLicense: "", state: "", expirationDate: "",
+        boundingBoxes: {},
+    };
+    const supervisoryAppraiser = parseAppraiserInfo(page6, true);
+    const lenderClient = parseLenderClientInfo(page6);
+    // ── Additional comparables pages (7, 8, ...) ──
+    for (let pageNum = 7; pageNum <= Math.max(...workLines.map((l) => l.page)); pageNum++) {
+        const pageLines = workLines.filter((l) => l.page === pageNum);
+        if (pageLines.length === 0)
+            continue;
+        // Check if this page has a comparable sales grid
+        const hasCompGrid = pageLines.some((l) => /^FEATURE\s+SUBJECT\s+COMPARABLE SALE/i.test(l.fullText));
+        if (!hasCompGrid)
+            continue;
+        // Determine the comp start number from the header
+        const headerLine = pageLines.find((l) => /COMPARABLE SALE #\s*(\d+)/i.test(l.fullText));
+        let compStart = salesComparison.comparables.length + 1;
+        if (headerLine) {
+            const m = headerLine.fullText.match(/COMPARABLE SALE #\s*(\d+)/i);
+            if (m)
+                compStart = parseInt(m[1], 10);
+        }
+        const additionalComps = parseSalesComparisonSection(pageLines, compStart);
+        salesComparison.comparables.push(...additionalComps.comparables);
+    }
+    return {
+        subject,
+        contract,
+        neighborhood,
+        projectSite,
+        projectInfo,
+        projectAnalysis,
+        unitDescription,
+        priorSaleHistory,
+        salesComparison,
+        reconciliation,
+        appraiser,
+        supervisoryAppraiser,
+        lenderClient,
+    };
+}