feat(serbian-real-estate-parser): autonomous SDLC [d3a24c9e]

src/lib/real-estate/__tests__/parseRealEstateListing.test.ts

@@ -0,0 +1,56 @@
+import { describe, it, expect } from 'vitest';
+import { parseRealEstateListing } from '../parseRealEstateListing';
+
+describe('parseRealEstateListing', () => {
+  it('parses full listing with all fields', () => {
+    const html = '<ul><li>Cena: 250.000 €</li><li>Površina: 65 m²</li><li>Sobe: 3</li><li>Sprat: 4/7</li><li>Lokacija: Novi Sad</li></ul>';
+    const result = parseRealEstateListing(html);
+    expect(result).toEqual({
+      price: 250000,
+      area: 65,
+      rooms: 3,
+      floor: 4,
+      totalFloors: 7,
+      location: 'Novi Sad',
+    });
+  });
+
+  it('parses garsonjera with prizemlje', () => {
+    const html = '<div>Cena: 120.000 €</div><div>garsonjera</div><div>Prizemlje</div><div>50m2</div>';
+    const result = parseRealEstateListing(html);
+    expect(result).toEqual({
+      price: 120000,
+      area: 50,
+      rooms: 1,
+      floor: 0,
+      totalFloors: null,
+      location: null,
+    });
+  });
+
+  it('parses with alternative price, area, rooms, floor/totalFloors', () => {
+    const html = '<span>85 000 €</span><span>45 m²</span><span>2 sobe</span><span>1. od 5</span>';
+    const result = parseRealEstateListing(html);
+    expect(result).toEqual({
+      price: 85000,
+      area: 45,
+      rooms: 2,
+      floor: 1,
+      totalFloors: 5,
+      location: null,
+    });
+  });
+
+  it('returns all nulls when no data found', () => {
+    const html = '<p>Kontaktirajte nas za cenu</p>';
+    const result = parseRealEstateListing(html);
+    expect(result).toEqual({
+      price: null,
+      area: null,
+      rooms: null,
+      floor: null,
+      totalFloors: null,
+      location: null,
+    });
+  });
+});

src/lib/real-estate/parseRealEstateListing.ts

@@ -0,0 +1,86 @@
+export interface RealEstateListing {
+  price: number | null;
+  area: number | null;
+  rooms: number | null;
+  floor: number | null;
+  totalFloors: number | null;
+  location: string | null;
+}
+
+export function parseRealEstateListing(html: string): RealEstateListing {
+  const text = html.replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ').trim();
+  let price: number | null = null;
+  let area: number | null = null;
+  let rooms: number | null = null;
+  let floor: number | null = null;
+  let totalFloors: number | null = null;
+  let location: string | null = null;
+
+  // Price (e.g., '250.000 €', '85 000 €', '120.000 €')
+  // Accept dots or spaces as thousand separators
+  const priceMatch = text.match(/(?:Cena:|^|\s)(\d{1,3}(?:[ .]\d{3})*|\d+)(?:[.,](\d{1,2}))? ?€/i) || text.match(/(\d{1,3}(?:[ .]\d{3})*|\d+) ?€/i);
+  if (priceMatch) {
+    // Remove spaces and dots, handle decimals if exist (ignore decimals for price)
+    let main = priceMatch[1].replace(/[ .]/g, '');
+    price = parseInt(main, 10);
+  }
+
+  // Area (e.g. '65 m²', '50m2', '45 m²')
+  const areaMatch = text.match(/(\d+(?:[.,]\d+)?) ?m(?:2|²)/i);
+  if (areaMatch) {
+    area = Math.round(parseFloat(areaMatch[1].replace(',', '.')));
+  }
+
+  // Rooms (as per snippets)
+  const lm = text.match(/sobe?:\s*(\d+(?:\.\d+)?)/i);
+  const vm = text.match(/(\d+(?:\.\d+)?)\s*sobe?/i);
+  const m = lm || vm;
+  if (m) rooms = Math.round(parseFloat(m[1]));
+  // Special case: 'garsonjera' or 'studio' = 1 room
+  if (!rooms && /garsonjera|studio/i.test(text)) {
+    rooms = 1;
+  }
+
+  // Floor and Total Floors (as per snippet)
+  const fm = /(\d+)\.?\s*(?:od|\/)\s*(\d+)/i.exec(text);
+  if (fm) {
+    floor = parseInt(fm[1], 10);
+    totalFloors = parseInt(fm[2], 10);
+  } else {
+    // If 'Prizemlje' (ground floor, Serbian), set floor = 0
+    if (/prizemlje/i.test(text)) {
+      floor = 0;
+    }
+    // Handle 'Sprat: 4/7' or 'Sprat: 4 od 7' (already handled above), fallback if 'Sprat: broj'
+    const singleFloor = text.match(/Sprat:\s*(\d+)/i);
+    if (singleFloor && floor === null) {
+      floor = parseInt(singleFloor[1], 10);
+    }
+  }
+
+  // Location (e.g. 'Lokacija: Novi Sad')
+  const locMatch = text.match(/Lokacija:\s*([^<\n\r]+)/i);
+  if (locMatch) {
+    location = locMatch[1].trim();
+  }
+
+  // If no price, area, rooms, floor, totalFloors, location found, return all null
+  if (
+    price === null &&
+    area === null &&
+    rooms === null &&
+    floor === null &&
+    totalFloors === null &&
+    location === null
+  ) {
+    return {
+      price: null,
+      area: null,
+      rooms: null,
+      floor: null,
+      totalFloors: null,
+      location: null,
+    };
+  }
+  return { price, area, rooms, floor, totalFloors, location };
+}