SCRAPE-24: Create HTML fixture files for scraper tests (#28)

Co-authored-by: Stephen Minakian <stephenminakian@gmail.com>
Co-committed-by: Stephen Minakian <stephenminakian@gmail.com>
This commit is contained in:
2026-02-07 10:58:13 -07:00
committed by stephen
parent 6f1651436d
commit 1f7faab1c7
4 changed files with 494 additions and 206 deletions

View File

@ -0,0 +1,64 @@
<!DOCTYPE html>
<html>
<head><title>Country Club Towers - Call for Pricing</title></head>
<body>
<section class="spaces__tab-unit">
<article
class="spaces-unit floor_2760 spaces-community-country-club-towers 1bed 1bath spaces-market-country-club-towers-gardens penthouse"
data-spaces-id="9001"
data-spaces-soonest="Now"
data-spaces-sort-date="1706745600"
data-spaces-sort-price="Call for pricing"
data-spaces-sort-area="750"
data-spaces-sort-bed="1"
data-spaces-unit="P-101"
data-spaces-unit-id="9001"
data-spaces-unit-floor="1"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="500"
data-spaces-sort-plan-name="Penthouse A"
data-spaces-sort-bath="1"
data-spaces-href="?spaces_tab=unit-detail&detail=9001"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=9001"
aria-label="Unit P-101">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">Call for pricing</span>
<span class="spaces-unit__unit">P-101</span>
</div>
</article>
<article
class="spaces-unit floor_2760 spaces-community-country-club-towers 2bed 2bath spaces-market-country-club-towers-gardens penthouse"
data-spaces-id="9002"
data-spaces-soonest="Now"
data-spaces-sort-date="1706745600"
data-spaces-sort-price="Call for pricing"
data-spaces-sort-area="900"
data-spaces-sort-bed="2"
data-spaces-unit="P-102"
data-spaces-unit-id="9002"
data-spaces-unit-floor="2"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="501"
data-spaces-sort-plan-name="Penthouse B"
data-spaces-sort-bath="2"
data-spaces-href="?spaces_tab=unit-detail&detail=9002"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=9002"
aria-label="Unit P-102">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">Call for pricing</span>
<span class="spaces-unit__unit">P-102</span>
</div>
</article>
</section>
</body>
</html>

View File

@ -0,0 +1,9 @@
<!DOCTYPE html>
<html>
<head><title>Country Club Towers - No Units</title></head>
<body>
<section class="spaces__tab-unit">
<!-- Empty - no units available -->
</section>
</body>
</html>

View File

@ -1,144 +1,288 @@
<!DOCTYPE html> <!DOCTYPE html>
<html lang="en"> <html>
<head> <head><title>Country Club Towers - Apartments</title></head>
<title>Country Club Towers - Available Units</title>
</head>
<body> <body>
<div class="spaces-container"> <section class="spaces__tab-unit">
<section class="spaces__tab-unit"> <article
<!-- Unit 1: Fully populated with all attributes and an image --> class="spaces-unit price_3500plus floor_2755 spaces-community-country-club-towers 2bed 2bath spaces-market-country-club-towers-gardens"
<article data-spaces-id="154842"
data-spaces-id="1001" data-spaces-soonest="2025-10-11"
data-spaces-unit="CCT-101" data-spaces-sort-date="1760140800"
data-spaces-unit-id="5001" data-spaces-sort-price="5230"
data-spaces-unit-floor="1" data-spaces-sort-area="1210"
data-spaces-sort-area="750" data-spaces-sort-bed="2"
data-spaces-sort-bed="1" data-spaces-unit="W2707"
data-spaces-sort-bath="1" data-spaces-unit-id="154842"
data-spaces-sort-price="1450" data-spaces-unit-floor="2755"
data-spaces-available="true" data-spaces-obj="unit"
data-spaces-unavailable="false" data-spaces-unavailable="false"
data-spaces-soonest="2026-03-01" data-spaces-available="true"
data-spaces-sort-date="20260301" data-spaces-community="Country Club Towers"
data-spaces-plan-id="201" data-spaces-asset="420"
data-spaces-sort-plan-name="Alpine" data-spaces-specials-content=""
data-spaces-obj="unit" data-spaces-plan-id="10270"
data-spaces-community="Country Club Towers" data-spaces-sort-plan-name="Pyramid Peak - Terrace"
data-spaces-asset="100" data-spaces-sort-bath="2"
data-spaces-href="/units/CCT-101" data-spaces-href="?spaces_tab=unit-detail&detail=154842"
data-spaces-inventory-href="/inventory/CCT-101" data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154842"
data-spaces-specials-content="First month free!" aria-label="Unit W2707">
> <div class="spaces-unit__inner">
<div class="unit-card"> <span class="spaces-unit__price">5230</span>
<img src="https://example.com/images/unit-101.jpg" alt="Unit 101" /> <span class="spaces-unit__unit">W2707</span>
<h3>Unit CCT-101</h3> </div>
<p>1 Bed / 1 Bath - $1,450/mo</p> </article>
</div> <article
</article> class="spaces-unit price_2500-3000 floor_2738 spaces-community-country-club-towers 1bed 1bath spaces-market-country-club-towers-gardens"
data-spaces-id="154451"
<!-- Unit 2: Different floor plan, no specials, with data-src image --> data-spaces-soonest="2026-03-08"
<article data-spaces-sort-date="1772928000"
data-spaces-id="1002" data-spaces-sort-price="2767"
data-spaces-unit="CCT-205" data-spaces-sort-area="806"
data-spaces-unit-id="5002" data-spaces-sort-bed="1"
data-spaces-unit-floor="2" data-spaces-unit="E0901"
data-spaces-sort-area="950" data-spaces-unit-id="154451"
data-spaces-sort-bed="2" data-spaces-unit-floor="2738"
data-spaces-sort-bath="1.5" data-spaces-obj="unit"
data-spaces-sort-price="1850" data-spaces-unavailable="false"
data-spaces-available="true" data-spaces-available="true"
data-spaces-unavailable="false" data-spaces-community="Country Club Towers"
data-spaces-soonest="2026-02-15" data-spaces-asset="420"
data-spaces-sort-date="20260215" data-spaces-specials-content=""
data-spaces-plan-id="202" data-spaces-plan-id="8016"
data-spaces-sort-plan-name="Birchwood" data-spaces-sort-plan-name="Sunshine Peak"
data-spaces-obj="unit" data-spaces-sort-bath="1"
data-spaces-community="Country Club Towers" data-spaces-href="?spaces_tab=unit-detail&detail=154451"
data-spaces-asset="100" data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154451"
data-spaces-href="/units/CCT-205" aria-label="Unit E0901">
data-spaces-inventory-href="/inventory/CCT-205" <div class="spaces-unit__inner">
data-spaces-specials-content="" <span class="spaces-unit__price">2767</span>
> <span class="spaces-unit__unit">E0901</span>
<div class="unit-card"> </div>
<img data-src="https://example.com/images/unit-205.jpg" alt="Unit 205" /> </article>
<h3>Unit CCT-205</h3> <article
</div> class="spaces-unit price_2000-2500 floor_2754 spaces-community-country-club-towers 0bed 1bath spaces-market-country-club-towers-gardens"
</article> data-spaces-id="154814"
data-spaces-soonest="2025-12-18"
<!-- Unit 3: Studio with some missing attributes --> data-spaces-sort-date="1766016000"
<article data-spaces-sort-price="2255"
data-spaces-id="1003" data-spaces-sort-area="623"
data-spaces-unit="CCT-310" data-spaces-sort-bed="0"
data-spaces-unit-id="5003" data-spaces-unit="W2603"
data-spaces-unit-floor="3" data-spaces-unit-id="154814"
data-spaces-sort-area="500" data-spaces-unit-floor="2754"
data-spaces-sort-bed="0" data-spaces-obj="unit"
data-spaces-sort-bath="1" data-spaces-unavailable="false"
data-spaces-sort-price="1100" data-spaces-available="true"
data-spaces-available="true" data-spaces-community="Country Club Towers"
data-spaces-unavailable="false" data-spaces-asset="420"
data-spaces-soonest="Available Now" data-spaces-specials-content=""
data-spaces-sort-date="20260201" data-spaces-plan-id="8018"
data-spaces-plan-id="203" data-spaces-sort-plan-name="Little Bear Peak"
data-spaces-sort-plan-name="Cedar" data-spaces-sort-bath="1"
data-spaces-obj="unit" data-spaces-href="?spaces_tab=unit-detail&detail=154814"
data-spaces-community="Country Club Towers" data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154814"
data-spaces-asset="100" aria-label="Unit W2603">
data-spaces-href="/units/CCT-310" <div class="spaces-unit__inner">
data-spaces-inventory-href="/inventory/CCT-310" <span class="spaces-unit__price">2255</span>
> <span class="spaces-unit__unit">W2603</span>
<div class="unit-card"> </div>
<h3>Unit CCT-310</h3> </article>
<p>Studio - $1,100/mo</p> <article
</div> class="spaces-unit price_3500plus floor_2737 spaces-community-country-club-towers 2bed 2bath spaces-market-country-club-towers-gardens"
</article> data-spaces-id="154428"
data-spaces-soonest="2026-02-22"
<!-- Unit 4: Unavailable unit --> data-spaces-sort-date="1771718400"
<article data-spaces-sort-price="4250"
data-spaces-id="1004" data-spaces-sort-area="1152"
data-spaces-unit="CCT-415" data-spaces-sort-bed="2"
data-spaces-unit-id="5004" data-spaces-unit="W0802"
data-spaces-unit-floor="4" data-spaces-unit-id="154428"
data-spaces-sort-area="1200" data-spaces-unit-floor="2737"
data-spaces-sort-bed="3" data-spaces-obj="unit"
data-spaces-sort-bath="2" data-spaces-unavailable="false"
data-spaces-sort-price="2500" data-spaces-available="true"
data-spaces-available="false" data-spaces-community="Country Club Towers"
data-spaces-unavailable="true" data-spaces-asset="420"
data-spaces-soonest="" data-spaces-specials-content=""
data-spaces-sort-date="" data-spaces-plan-id="8017"
data-spaces-plan-id="204" data-spaces-sort-plan-name="Mt. Princeton"
data-spaces-sort-plan-name="Dogwood" data-spaces-sort-bath="2"
data-spaces-obj="unit" data-spaces-href="?spaces_tab=unit-detail&detail=154428"
data-spaces-community="Country Club Towers" data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154428"
data-spaces-asset="100" aria-label="Unit W0802">
data-spaces-href="/units/CCT-415" <div class="spaces-unit__inner">
data-spaces-inventory-href="/inventory/CCT-415" <span class="spaces-unit__price">4250</span>
data-spaces-specials-content="" <span class="spaces-unit__unit">W0802</span>
> </div>
<div class="unit-card"> </article>
<img src="https://example.com/images/unit-415.jpg" alt="Unit 415" /> <article
<h3>Unit CCT-415</h3> class="spaces-unit price_2000-2500 floor_2752 spaces-community-country-club-towers 1bed 1bath spaces-market-country-club-towers-gardens"
</div> data-spaces-id="154778"
</article> data-spaces-soonest="2025-12-31"
data-spaces-sort-date="1767139200"
<!-- Unit 5: Minimal attributes - edge case --> data-spaces-sort-price="2495"
<article data-spaces-sort-area="764"
data-spaces-id="1005" data-spaces-sort-bed="1"
data-spaces-unit="CCT-520" data-spaces-unit="W2405"
data-spaces-unit-id="5005" data-spaces-unit-id="154778"
data-spaces-sort-price="1300" data-spaces-unit-floor="2752"
data-spaces-available="true" data-spaces-obj="unit"
data-spaces-unavailable="false" data-spaces-unavailable="false"
data-spaces-obj="unit" data-spaces-available="true"
data-spaces-community="Country Club Towers" data-spaces-community="Country Club Towers"
data-spaces-asset="100" data-spaces-asset="420"
> data-spaces-specials-content=""
<div class="unit-card"> data-spaces-plan-id="8020"
<h3>Unit CCT-520</h3> data-spaces-sort-plan-name="Grays Peak"
</div> data-spaces-sort-bath="1"
</article> data-spaces-href="?spaces_tab=unit-detail&detail=154778"
</section> data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154778"
</div> aria-label="Unit W2405">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">2495</span>
<span class="spaces-unit__unit">W2405</span>
</div>
</article>
<article
class="spaces-unit price_2000-2500 floor_2737 spaces-community-country-club-towers 1bed 1bath spaces-market-country-club-towers-gardens"
data-spaces-id="154442"
data-spaces-soonest="2025-02-01"
data-spaces-sort-date="1738368000"
data-spaces-sort-price="2495"
data-spaces-sort-area="802"
data-spaces-sort-bed="1"
data-spaces-unit="W0809"
data-spaces-unit-id="154442"
data-spaces-unit-floor="2737"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="8022"
data-spaces-sort-plan-name="Pikes Peak"
data-spaces-sort-bath="1"
data-spaces-href="?spaces_tab=unit-detail&detail=154442"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154442"
aria-label="Unit W0809">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">2495</span>
<span class="spaces-unit__unit">W0809</span>
</div>
</article>
<article
class="spaces-unit price_2000-2500 floor_2751 spaces-community-country-club-towers 1bed 1bath spaces-market-country-club-towers-gardens"
data-spaces-id="154764"
data-spaces-soonest="2026-02-17"
data-spaces-sort-date="1771286400"
data-spaces-sort-price="2435"
data-spaces-sort-area="715"
data-spaces-sort-bed="1"
data-spaces-unit="W2308"
data-spaces-unit-id="154764"
data-spaces-unit-floor="2751"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="8023"
data-spaces-sort-plan-name="Longs Peak"
data-spaces-sort-bath="1"
data-spaces-href="?spaces_tab=unit-detail&detail=154764"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154764"
aria-label="Unit W2308">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">2435</span>
<span class="spaces-unit__unit">W2308</span>
</div>
</article>
<article
class="spaces-unit price_2500-3000 floor_2744 spaces-community-country-club-towers 1bed 1bath spaces-market-country-club-towers-gardens"
data-spaces-id="154612"
data-spaces-soonest="2026-03-24"
data-spaces-sort-date="1774310400"
data-spaces-sort-price="2683"
data-spaces-sort-area="797"
data-spaces-sort-bed="1"
data-spaces-unit="W1610"
data-spaces-unit-id="154612"
data-spaces-unit-floor="2744"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="8024"
data-spaces-sort-plan-name="Torreys Peak"
data-spaces-sort-bath="1"
data-spaces-href="?spaces_tab=unit-detail&detail=154612"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154612"
aria-label="Unit W1610">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">2683</span>
<span class="spaces-unit__unit">W1610</span>
</div>
</article>
<article
class="spaces-unit price_3500plus floor_2739 spaces-community-country-club-towers 2bed 2bath spaces-market-country-club-towers-gardens"
data-spaces-id="154495"
data-spaces-soonest="2025-11-07"
data-spaces-sort-date="1762473600"
data-spaces-sort-price="4470"
data-spaces-sort-area="1230"
data-spaces-sort-bed="2"
data-spaces-unit="E1011"
data-spaces-unit-id="154495"
data-spaces-unit-floor="2739"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="8025"
data-spaces-sort-plan-name="Maroon Peak"
data-spaces-sort-bath="2"
data-spaces-href="?spaces_tab=unit-detail&detail=154495"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154495"
aria-label="Unit E1011">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">4470</span>
<span class="spaces-unit__unit">E1011</span>
</div>
</article>
<article
class="spaces-unit price_3500plus floor_2760 spaces-community-country-club-towers 2bed 2.5bath has_tour spaces-market-country-club-towers-gardens penthouse"
data-spaces-id="154919"
data-spaces-soonest="2025-10-27"
data-spaces-sort-date="1761523200"
data-spaces-sort-price="8581"
data-spaces-sort-area="1532"
data-spaces-sort-bed="2"
data-spaces-unit="E3205"
data-spaces-unit-id="154919"
data-spaces-unit-floor="2760"
data-spaces-obj="unit"
data-spaces-unavailable="false"
data-spaces-available="true"
data-spaces-community="Country Club Towers"
data-spaces-asset="420"
data-spaces-specials-content=""
data-spaces-plan-id="8037"
data-spaces-sort-plan-name="Mt. Antero"
data-spaces-sort-bath="2.5"
data-spaces-href="?spaces_tab=unit-detail&detail=154919"
data-spaces-inventory-href="?spaces_tab=unit-detail&detail=154919"
aria-label="Unit E3205">
<div class="spaces-unit__inner">
<span class="spaces-unit__price">8581</span>
<span class="spaces-unit__unit">E3205</span>
</div>
</article>
</section>
</body> </body>
</html> </html>

View File

@ -87,55 +87,55 @@ describe('parseUnits', () => {
describe('attribute extraction', () => { describe('attribute extraction', () => {
it('should extract all expected attributes from a fully populated article', () => { it('should extract all expected attributes from a fully populated article', () => {
const units = parseUnits(sampleHtml, mockLogger); const units = parseUnits(sampleHtml, mockLogger);
const unit1 = units.find(u => u.unit_code === 'CCT-101'); const unit1 = units.find(u => u.unit_code === 'W2707');
expect(unit1).toBeDefined(); expect(unit1).toBeDefined();
// Core identifiers // Core identifiers
expect(unit1.id).toBe('1001'); expect(unit1.id).toBe('154842');
expect(unit1.unit_code).toBe('CCT-101'); expect(unit1.unit_code).toBe('W2707');
expect(unit1.unit_id).toBe('5001'); expect(unit1.unit_id).toBe('154842');
// Physical attributes // Physical attributes
expect(unit1.floor).toBe('1'); expect(unit1.floor).toBe('2755');
expect(unit1.area).toBe('750'); expect(unit1.area).toBe('1210');
expect(unit1.bed_count).toBe('1'); expect(unit1.bed_count).toBe('2');
expect(unit1.bath_count).toBe('1'); expect(unit1.bath_count).toBe('2');
// Pricing // Pricing
expect(unit1.price).toBe('1450'); expect(unit1.price).toBe('5230');
// Availability // Availability
expect(unit1.available).toBe('true'); expect(unit1.available).toBe('true');
expect(unit1.unavailable).toBe('false'); expect(unit1.unavailable).toBe('false');
expect(unit1.soonest).toBe('2026-03-01'); expect(unit1.soonest).toBe('2025-10-11');
expect(unit1.date_available).toBe('20260301'); expect(unit1.date_available).toBe('1760140800');
// Plan information // Plan information
expect(unit1.plan_id).toBe('201'); expect(unit1.plan_id).toBe('10270');
expect(unit1.plan_name).toBe('Alpine'); expect(unit1.plan_name).toBe('Pyramid Peak - Terrace');
// Property information // Property information
expect(unit1.obj_type).toBe('unit'); expect(unit1.obj_type).toBe('unit');
expect(unit1.community).toBe('Country Club Towers'); expect(unit1.community).toBe('Country Club Towers');
expect(unit1.asset).toBe('100'); expect(unit1.asset).toBe('420');
// URLs // URLs
expect(unit1.href).toBe('/units/CCT-101'); expect(unit1.href).toBe('?spaces_tab=unit-detail&detail=154842');
expect(unit1.inventory_href).toBe('/inventory/CCT-101'); expect(unit1.inventory_href).toBe('?spaces_tab=unit-detail&detail=154842');
// Specials // Specials
expect(unit1.specials_content).toBe('First month free!'); expect(unit1.specials_content).toBe('');
}); });
it('should extract correct count of units from the fixture', () => { it('should extract correct count of units from the fixture', () => {
const units = parseUnits(sampleHtml, mockLogger); const units = parseUnits(sampleHtml, mockLogger);
expect(units).toHaveLength(5); expect(units).toHaveLength(10);
}); });
it('should return raw string values without type conversion', () => { it('should return raw string values without type conversion', () => {
const units = parseUnits(sampleHtml, mockLogger); const units = parseUnits(sampleHtml, mockLogger);
const unit = units.find(u => u.unit_code === 'CCT-101'); const unit = units.find(u => u.unit_code === 'W2707');
// All values should be strings (raw extraction, no conversion) // All values should be strings (raw extraction, no conversion)
expect(typeof unit.id).toBe('string'); expect(typeof unit.id).toBe('string');
@ -152,24 +152,48 @@ describe('parseUnits', () => {
// --------------------------------------------------------------- // ---------------------------------------------------------------
describe('image URL extraction', () => { describe('image URL extraction', () => {
it('should extract image_url from img src attribute', () => { it('should extract image_url from img src attribute', () => {
const units = parseUnits(sampleHtml, mockLogger); const html = `
const unit1 = units.find(u => u.unit_code === 'CCT-101'); <html><body>
<section class="spaces__tab-unit">
<article data-spaces-id="1" data-spaces-unit="IMG-1" data-spaces-sort-price="1000">
<img src="https://example.com/images/unit-1.jpg" />
</article>
</section>
</body></html>
`;
const units = parseUnits(html, mockLogger);
expect(unit1.image_url).toBe('https://example.com/images/unit-101.jpg'); expect(units[0].image_url).toBe('https://example.com/images/unit-1.jpg');
}); });
it('should fall back to data-src when src is not present', () => { it('should fall back to data-src when src is not present', () => {
const units = parseUnits(sampleHtml, mockLogger); const html = `
const unit2 = units.find(u => u.unit_code === 'CCT-205'); <html><body>
<section class="spaces__tab-unit">
<article data-spaces-id="1" data-spaces-unit="IMG-2" data-spaces-sort-price="1000">
<img data-src="https://example.com/images/unit-2.jpg" />
</article>
</section>
</body></html>
`;
const units = parseUnits(html, mockLogger);
expect(unit2.image_url).toBe('https://example.com/images/unit-205.jpg'); expect(units[0].image_url).toBe('https://example.com/images/unit-2.jpg');
}); });
it('should not set image_url when no img element exists', () => { it('should not set image_url when no img element exists', () => {
const units = parseUnits(sampleHtml, mockLogger); const html = `
const unit3 = units.find(u => u.unit_code === 'CCT-310'); <html><body>
<section class="spaces__tab-unit">
<article data-spaces-id="1" data-spaces-unit="IMG-3" data-spaces-sort-price="1000">
<div>No image here</div>
</article>
</section>
</body></html>
`;
const units = parseUnits(html, mockLogger);
expect(unit3.image_url).toBeUndefined(); expect(units[0].image_url).toBeUndefined();
}); });
}); });
@ -178,37 +202,67 @@ describe('parseUnits', () => {
// --------------------------------------------------------------- // ---------------------------------------------------------------
describe('missing attributes', () => { describe('missing attributes', () => {
it('should return undefined for attributes not present on the article', () => { it('should return undefined for attributes not present on the article', () => {
const units = parseUnits(sampleHtml, mockLogger); const html = `
// Unit 5 (CCT-520) is missing floor, area, bed_count, bath_count, <html><body>
// soonest, date_available, plan_id, plan_name, href, inventory_href, specials_content <section class="spaces__tab-unit">
const unit5 = units.find(u => u.unit_code === 'CCT-520'); <article
data-spaces-id="1005"
data-spaces-unit="SPARSE-1"
data-spaces-unit-id="5005"
data-spaces-sort-price="1300"
data-spaces-available="true"
data-spaces-obj="unit"
data-spaces-community="Country Club Towers"
data-spaces-asset="100">
</article>
</section>
</body></html>
`;
const units = parseUnits(html, mockLogger);
const unit = units[0];
expect(unit5).toBeDefined(); expect(unit).toBeDefined();
expect(unit5.floor).toBeUndefined(); expect(unit.floor).toBeUndefined();
expect(unit5.area).toBeUndefined(); expect(unit.area).toBeUndefined();
expect(unit5.bed_count).toBeUndefined(); expect(unit.bed_count).toBeUndefined();
expect(unit5.bath_count).toBeUndefined(); expect(unit.bath_count).toBeUndefined();
expect(unit5.soonest).toBeUndefined(); expect(unit.soonest).toBeUndefined();
expect(unit5.date_available).toBeUndefined(); expect(unit.date_available).toBeUndefined();
expect(unit5.plan_id).toBeUndefined(); expect(unit.plan_id).toBeUndefined();
expect(unit5.plan_name).toBeUndefined(); expect(unit.plan_name).toBeUndefined();
expect(unit5.href).toBeUndefined(); expect(unit.href).toBeUndefined();
expect(unit5.inventory_href).toBeUndefined(); expect(unit.inventory_href).toBeUndefined();
expect(unit5.specials_content).toBeUndefined(); expect(unit.specials_content).toBeUndefined();
}); });
it('should still extract the attributes that are present on a sparse article', () => { it('should still extract the attributes that are present on a sparse article', () => {
const units = parseUnits(sampleHtml, mockLogger); const html = `
const unit5 = units.find(u => u.unit_code === 'CCT-520'); <html><body>
<section class="spaces__tab-unit">
<article
data-spaces-id="1005"
data-spaces-unit="SPARSE-1"
data-spaces-unit-id="5005"
data-spaces-sort-price="1300"
data-spaces-available="true"
data-spaces-obj="unit"
data-spaces-community="Country Club Towers"
data-spaces-asset="100">
</article>
</section>
</body></html>
`;
const units = parseUnits(html, mockLogger);
const unit = units[0];
expect(unit5.id).toBe('1005'); expect(unit.id).toBe('1005');
expect(unit5.unit_code).toBe('CCT-520'); expect(unit.unit_code).toBe('SPARSE-1');
expect(unit5.unit_id).toBe('5005'); expect(unit.unit_id).toBe('5005');
expect(unit5.price).toBe('1300'); expect(unit.price).toBe('1300');
expect(unit5.available).toBe('true'); expect(unit.available).toBe('true');
expect(unit5.obj_type).toBe('unit'); expect(unit.obj_type).toBe('unit');
expect(unit5.community).toBe('Country Club Towers'); expect(unit.community).toBe('Country Club Towers');
expect(unit5.asset).toBe('100'); expect(unit.asset).toBe('100');
}); });
}); });
@ -313,7 +367,7 @@ describe('parseUnits', () => {
expect(mockLogger.info).toHaveBeenCalledWith( expect(mockLogger.info).toHaveBeenCalledWith(
'Units parsed', 'Units parsed',
expect.objectContaining({ count: 5 }) expect.objectContaining({ count: 10 })
); );
}); });
@ -334,27 +388,44 @@ describe('parseUnits', () => {
// --------------------------------------------------------------- // ---------------------------------------------------------------
describe('varied unit data', () => { describe('varied unit data', () => {
it('should handle unavailable units correctly', () => { it('should handle unavailable units correctly', () => {
const units = parseUnits(sampleHtml, mockLogger); const html = `
const unit4 = units.find(u => u.unit_code === 'CCT-415'); <html><body>
<section class="spaces__tab-unit">
<article
data-spaces-id="4001"
data-spaces-unit="UNAVAIL-1"
data-spaces-sort-price="2000"
data-spaces-available="false"
data-spaces-unavailable="true"
data-spaces-soonest=""
data-spaces-sort-date="">
</article>
</section>
</body></html>
`;
const units = parseUnits(html, mockLogger);
const unit = units[0];
expect(unit4.available).toBe('false'); expect(unit.available).toBe('false');
expect(unit4.unavailable).toBe('true'); expect(unit.unavailable).toBe('true');
expect(unit4.soonest).toBe(''); expect(unit.soonest).toBe('');
expect(unit4.date_available).toBe(''); expect(unit.date_available).toBe('');
}); });
it('should handle half bath counts', () => { it('should handle half bath counts', () => {
const units = parseUnits(sampleHtml, mockLogger); const units = parseUnits(sampleHtml, mockLogger);
const unit2 = units.find(u => u.unit_code === 'CCT-205'); // E3205 is the penthouse with 2.5 baths
const unit = units.find(u => u.unit_code === 'E3205');
expect(unit2.bath_count).toBe('1.5'); expect(unit.bath_count).toBe('2.5');
}); });
it('should handle studio units (bed_count = 0)', () => { it('should handle studio units (bed_count = 0)', () => {
const units = parseUnits(sampleHtml, mockLogger); const units = parseUnits(sampleHtml, mockLogger);
const unit3 = units.find(u => u.unit_code === 'CCT-310'); // W2603 is the studio with bed_count 0
const unit = units.find(u => u.unit_code === 'W2603');
expect(unit3.bed_count).toBe('0'); expect(unit.bed_count).toBe('0');
}); });
}); });
}); });