Compare commits
27 Commits
main
...
1080121386
| Author | SHA1 | Date | |
|---|---|---|---|
| 1080121386 | |||
| e70f7429ea | |||
| e654d59fea | |||
| a1ee25ef17 | |||
| f389970d16 | |||
| d0538334a5 | |||
| 9d0b9debbb | |||
| daa234428e | |||
| c6d480a870 | |||
| 0c07baa977 | |||
| 9d789d38fd | |||
| dd0269eb30 | |||
| d818296eeb | |||
| 3509da8185 | |||
| 0ad4c98abe | |||
| d1f717891a | |||
| b4978caf31 | |||
| d8adfbf90c | |||
| 2b53288fbe | |||
| 4863dc1824 | |||
| 0cbeaaf9d5 | |||
| 072e80d069 | |||
| c6beaf8333 | |||
| 2993d019c5 | |||
| 3af5a90c93 | |||
| 9dba679221 | |||
| 6daacf4d12 |
54
.github/workflows/deploy.yml
vendored
54
.github/workflows/deploy.yml
vendored
@ -44,11 +44,8 @@ jobs:
|
|||||||
set +e
|
set +e
|
||||||
OUTPUT=$(npm run lint 2>&1)
|
OUTPUT=$(npm run lint 2>&1)
|
||||||
EXIT_CODE=$?
|
EXIT_CODE=$?
|
||||||
echo "$OUTPUT"
|
|
||||||
# Truncate before writing to GITHUB_OUTPUT to prevent
|
|
||||||
# "argument list too long" in downstream jobs
|
|
||||||
echo "lint_output<<EOF" >> $GITHUB_OUTPUT
|
echo "lint_output<<EOF" >> $GITHUB_OUTPUT
|
||||||
echo "${OUTPUT:0:10000}" >> $GITHUB_OUTPUT
|
echo "$OUTPUT" >> $GITHUB_OUTPUT
|
||||||
echo "EOF" >> $GITHUB_OUTPUT
|
echo "EOF" >> $GITHUB_OUTPUT
|
||||||
exit $EXIT_CODE
|
exit $EXIT_CODE
|
||||||
|
|
||||||
@ -59,11 +56,8 @@ jobs:
|
|||||||
set +e
|
set +e
|
||||||
OUTPUT=$(npm test -- --runInBand 2>&1)
|
OUTPUT=$(npm test -- --runInBand 2>&1)
|
||||||
EXIT_CODE=$?
|
EXIT_CODE=$?
|
||||||
echo "$OUTPUT"
|
|
||||||
# Truncate before writing to GITHUB_OUTPUT to prevent
|
|
||||||
# "argument list too long" in downstream jobs
|
|
||||||
echo "test_output<<EOF" >> $GITHUB_OUTPUT
|
echo "test_output<<EOF" >> $GITHUB_OUTPUT
|
||||||
echo "${OUTPUT:0:10000}" >> $GITHUB_OUTPUT
|
echo "$OUTPUT" >> $GITHUB_OUTPUT
|
||||||
echo "EOF" >> $GITHUB_OUTPUT
|
echo "EOF" >> $GITHUB_OUTPUT
|
||||||
exit $EXIT_CODE
|
exit $EXIT_CODE
|
||||||
env:
|
env:
|
||||||
@ -77,7 +71,7 @@ jobs:
|
|||||||
name: Send Webhook Notification
|
name: Send Webhook Notification
|
||||||
runs-on: ubuntu-latest
|
runs-on: ubuntu-latest
|
||||||
needs: [ci]
|
needs: [ci]
|
||||||
if: always() && github.ref != 'refs/heads/main'
|
if: always()
|
||||||
|
|
||||||
steps:
|
steps:
|
||||||
- name: Send results to n8n webhook
|
- name: Send results to n8n webhook
|
||||||
@ -104,9 +98,9 @@ jobs:
|
|||||||
OVERALL_STATUS="failure"
|
OVERALL_STATUS="failure"
|
||||||
fi
|
fi
|
||||||
|
|
||||||
# Outputs are already truncated at the source (ci job)
|
# Truncate outputs if too long (max 10000 chars each)
|
||||||
LINT_OUTPUT="$RAW_LINT_OUTPUT"
|
LINT_OUTPUT="${RAW_LINT_OUTPUT:0:10000}"
|
||||||
TEST_OUTPUT="$RAW_TEST_OUTPUT"
|
TEST_OUTPUT="${RAW_TEST_OUTPUT:0:10000}"
|
||||||
|
|
||||||
# Build JSON payload
|
# Build JSON payload
|
||||||
PAYLOAD=$(jq -n \
|
PAYLOAD=$(jq -n \
|
||||||
@ -200,12 +194,8 @@ jobs:
|
|||||||
build:
|
build:
|
||||||
name: Build & Push Image
|
name: Build & Push Image
|
||||||
runs-on: ubuntu-latest
|
runs-on: ubuntu-latest
|
||||||
needs: [ci, scan-deps]
|
needs: [ci, scan-deps, notify]
|
||||||
if: >-
|
if: github.ref == 'refs/heads/main' && github.event_name != 'pull_request'
|
||||||
github.ref == 'refs/heads/main' &&
|
|
||||||
github.event_name != 'pull_request' &&
|
|
||||||
needs.ci.outputs.lint_status == 'success' &&
|
|
||||||
needs.ci.outputs.test_status == 'success'
|
|
||||||
|
|
||||||
outputs:
|
outputs:
|
||||||
image_tag: ${{ steps.set-tag.outputs.tag }}
|
image_tag: ${{ steps.set-tag.outputs.tag }}
|
||||||
@ -270,13 +260,12 @@ jobs:
|
|||||||
# ============================================================
|
# ============================================================
|
||||||
# SBOM Generation with Syft
|
# SBOM Generation with Syft
|
||||||
# ============================================================
|
# ============================================================
|
||||||
- name: Install Syft
|
|
||||||
run: |
|
|
||||||
curl -sSfL https://raw.githubusercontent.com/anchore/syft/main/install.sh | sh -s -- -b /usr/local/bin
|
|
||||||
|
|
||||||
- name: Generate SBOM with Syft
|
- name: Generate SBOM with Syft
|
||||||
run: |
|
uses: anchore/sbom-action@v0
|
||||||
syft ${{ steps.set-tag.outputs.full_image }} -o spdx-json=sbom.spdx.json
|
with:
|
||||||
|
image: ${{ steps.set-tag.outputs.full_image }}
|
||||||
|
format: spdx-json
|
||||||
|
output-file: sbom.spdx.json
|
||||||
|
|
||||||
# ============================================================
|
# ============================================================
|
||||||
# Image Signing with Cosign
|
# Image Signing with Cosign
|
||||||
@ -317,21 +306,8 @@ jobs:
|
|||||||
if: github.ref == 'refs/heads/main' && github.event_name != 'pull_request'
|
if: github.ref == 'refs/heads/main' && github.event_name != 'pull_request'
|
||||||
|
|
||||||
steps:
|
steps:
|
||||||
- name: Checkout code
|
|
||||||
uses: actions/checkout@v4
|
|
||||||
|
|
||||||
- name: Sync docker-compose.yml to server
|
|
||||||
uses: appleboy/scp-action@v0.1.7
|
|
||||||
with:
|
|
||||||
host: ${{ secrets.SSH_HOST }}
|
|
||||||
username: ${{ secrets.SSH_USER }}
|
|
||||||
key: ${{ secrets.SSH_PRIVATE_KEY }}
|
|
||||||
port: ${{ secrets.SSH_PORT || 22 }}
|
|
||||||
source: "docker-compose.yml"
|
|
||||||
target: ${{ secrets.DEPLOY_PATH }}
|
|
||||||
|
|
||||||
- name: Deploy via SSH
|
- name: Deploy via SSH
|
||||||
uses: appleboy/ssh-action@v1.2.4
|
uses: appleboy/ssh-action@v1.0.3
|
||||||
with:
|
with:
|
||||||
host: ${{ secrets.SSH_HOST }}
|
host: ${{ secrets.SSH_HOST }}
|
||||||
username: ${{ secrets.SSH_USER }}
|
username: ${{ secrets.SSH_USER }}
|
||||||
@ -373,7 +349,7 @@ jobs:
|
|||||||
fi
|
fi
|
||||||
|
|
||||||
- name: Verify deployment
|
- name: Verify deployment
|
||||||
uses: appleboy/ssh-action@v1.2.4
|
uses: appleboy/ssh-action@v1.0.3
|
||||||
with:
|
with:
|
||||||
host: ${{ secrets.SSH_HOST }}
|
host: ${{ secrets.SSH_HOST }}
|
||||||
username: ${{ secrets.SSH_USER }}
|
username: ${{ secrets.SSH_USER }}
|
||||||
|
|||||||
@ -15,15 +15,14 @@
|
|||||||
const { MongoClient } = require('mongodb');
|
const { MongoClient } = require('mongodb');
|
||||||
const { MongoMemoryServer } = require('mongodb-memory-server');
|
const { MongoMemoryServer } = require('mongodb-memory-server');
|
||||||
|
|
||||||
// Will require insertPrices and getNow after implementation
|
// Will require insertPrices after implementation
|
||||||
let insertPrices;
|
let insertPrices;
|
||||||
let getNow;
|
|
||||||
|
|
||||||
let mongoServer;
|
let mongoServer;
|
||||||
let client;
|
let client;
|
||||||
let db;
|
let db;
|
||||||
|
|
||||||
const PRICES_COLLECTION = 'unit_prices_scraper';
|
const PRICES_COLLECTION = 'unit_prices_migration_test';
|
||||||
|
|
||||||
// Mock logger for capturing log calls
|
// Mock logger for capturing log calls
|
||||||
function createMockLogger() {
|
function createMockLogger() {
|
||||||
@ -42,7 +41,7 @@ beforeAll(async () => {
|
|||||||
db = client.db('test_apartments');
|
db = client.db('test_apartments');
|
||||||
|
|
||||||
// Dynamically require to pick up implementation
|
// Dynamically require to pick up implementation
|
||||||
({ insertPrices, getNow } = require('../../services/scraperService'));
|
({ insertPrices } = require('../../services/scraperService'));
|
||||||
});
|
});
|
||||||
|
|
||||||
afterAll(async () => {
|
afterAll(async () => {
|
||||||
@ -242,12 +241,12 @@ describe('insertPrices', () => {
|
|||||||
|
|
||||||
it('should include last_updated as an ISO timestamp string', async () => {
|
it('should include last_updated as an ISO timestamp string', async () => {
|
||||||
const logger = createMockLogger();
|
const logger = createMockLogger();
|
||||||
const beforeTime = getNow();
|
const beforeTime = new Date().toISOString();
|
||||||
|
|
||||||
const units = [{ unit_code: 'F401', price: 1600 }];
|
const units = [{ unit_code: 'F401', price: 1600 }];
|
||||||
await insertPrices(db, units, TODAY, logger);
|
await insertPrices(db, units, TODAY, logger);
|
||||||
|
|
||||||
const afterTime = getNow();
|
const afterTime = new Date().toISOString();
|
||||||
|
|
||||||
const doc = await db.collection(PRICES_COLLECTION).findOne({});
|
const doc = await db.collection(PRICES_COLLECTION).findOne({});
|
||||||
expect(doc.last_updated).toBeDefined();
|
expect(doc.last_updated).toBeDefined();
|
||||||
|
|||||||
@ -21,7 +21,7 @@ let mongoServer;
|
|||||||
let client;
|
let client;
|
||||||
let db;
|
let db;
|
||||||
|
|
||||||
const UNITS_COLLECTION = 'units_scraper';
|
const UNITS_COLLECTION = 'units_migration_test';
|
||||||
|
|
||||||
// Mock logger for capturing log calls
|
// Mock logger for capturing log calls
|
||||||
function createMockLogger() {
|
function createMockLogger() {
|
||||||
|
|||||||
@ -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 === 'W2707');
|
const unit1 = units.find(u => u.unit_code === 'CCT-101');
|
||||||
|
|
||||||
expect(unit1).toBeDefined();
|
expect(unit1).toBeDefined();
|
||||||
|
|
||||||
// Core identifiers
|
// Core identifiers
|
||||||
expect(unit1.id).toBe('154842');
|
expect(unit1.id).toBe('1001');
|
||||||
expect(unit1.unit_code).toBe('W2707');
|
expect(unit1.unit_code).toBe('CCT-101');
|
||||||
expect(unit1.unit_id).toBe('154842');
|
expect(unit1.unit_id).toBe('5001');
|
||||||
|
|
||||||
// Physical attributes
|
// Physical attributes
|
||||||
expect(unit1.floor).toBe('2755');
|
expect(unit1.floor).toBe('1');
|
||||||
expect(unit1.area).toBe('1210');
|
expect(unit1.area).toBe('750');
|
||||||
expect(unit1.bed_count).toBe('2');
|
expect(unit1.bed_count).toBe('1');
|
||||||
expect(unit1.bath_count).toBe('2');
|
expect(unit1.bath_count).toBe('1');
|
||||||
|
|
||||||
// Pricing
|
// Pricing
|
||||||
expect(unit1.price).toBe('5230');
|
expect(unit1.price).toBe('1450');
|
||||||
|
|
||||||
// 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('2025-10-11');
|
expect(unit1.soonest).toBe('2026-03-01');
|
||||||
expect(unit1.date_available).toBe('1760140800');
|
expect(unit1.date_available).toBe('20260301');
|
||||||
|
|
||||||
// Plan information
|
// Plan information
|
||||||
expect(unit1.plan_id).toBe('10270');
|
expect(unit1.plan_id).toBe('201');
|
||||||
expect(unit1.plan_name).toBe('Pyramid Peak - Terrace');
|
expect(unit1.plan_name).toBe('Alpine');
|
||||||
|
|
||||||
// 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('420');
|
expect(unit1.asset).toBe('100');
|
||||||
|
|
||||||
// URLs
|
// URLs
|
||||||
expect(unit1.href).toBe('?spaces_tab=unit-detail&detail=154842');
|
expect(unit1.href).toBe('/units/CCT-101');
|
||||||
expect(unit1.inventory_href).toBe('?spaces_tab=unit-detail&detail=154842');
|
expect(unit1.inventory_href).toBe('/inventory/CCT-101');
|
||||||
|
|
||||||
// Specials
|
// Specials
|
||||||
expect(unit1.specials_content).toBe('');
|
expect(unit1.specials_content).toBe('First month free!');
|
||||||
});
|
});
|
||||||
|
|
||||||
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(10);
|
expect(units).toHaveLength(5);
|
||||||
});
|
});
|
||||||
|
|
||||||
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 === 'W2707');
|
const unit = units.find(u => u.unit_code === 'CCT-101');
|
||||||
|
|
||||||
// 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');
|
||||||
@ -152,48 +152,24 @@ 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 html = `
|
const units = parseUnits(sampleHtml, mockLogger);
|
||||||
<html><body>
|
const unit1 = units.find(u => u.unit_code === 'CCT-101');
|
||||||
<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(units[0].image_url).toBe('https://example.com/images/unit-1.jpg');
|
expect(unit1.image_url).toBe('https://example.com/images/unit-101.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 html = `
|
const units = parseUnits(sampleHtml, mockLogger);
|
||||||
<html><body>
|
const unit2 = units.find(u => u.unit_code === 'CCT-205');
|
||||||
<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(units[0].image_url).toBe('https://example.com/images/unit-2.jpg');
|
expect(unit2.image_url).toBe('https://example.com/images/unit-205.jpg');
|
||||||
});
|
});
|
||||||
|
|
||||||
it('should not set image_url when no img element exists', () => {
|
it('should not set image_url when no img element exists', () => {
|
||||||
const html = `
|
const units = parseUnits(sampleHtml, mockLogger);
|
||||||
<html><body>
|
const unit3 = units.find(u => u.unit_code === 'CCT-310');
|
||||||
<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(units[0].image_url).toBeUndefined();
|
expect(unit3.image_url).toBeUndefined();
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|
||||||
@ -202,67 +178,37 @@ 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 html = `
|
const units = parseUnits(sampleHtml, mockLogger);
|
||||||
<html><body>
|
// Unit 5 (CCT-520) is missing floor, area, bed_count, bath_count,
|
||||||
<section class="spaces__tab-unit">
|
// soonest, date_available, plan_id, plan_name, href, inventory_href, specials_content
|
||||||
<article
|
const unit5 = units.find(u => u.unit_code === 'CCT-520');
|
||||||
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(unit).toBeDefined();
|
expect(unit5).toBeDefined();
|
||||||
expect(unit.floor).toBeUndefined();
|
expect(unit5.floor).toBeUndefined();
|
||||||
expect(unit.area).toBeUndefined();
|
expect(unit5.area).toBeUndefined();
|
||||||
expect(unit.bed_count).toBeUndefined();
|
expect(unit5.bed_count).toBeUndefined();
|
||||||
expect(unit.bath_count).toBeUndefined();
|
expect(unit5.bath_count).toBeUndefined();
|
||||||
expect(unit.soonest).toBeUndefined();
|
expect(unit5.soonest).toBeUndefined();
|
||||||
expect(unit.date_available).toBeUndefined();
|
expect(unit5.date_available).toBeUndefined();
|
||||||
expect(unit.plan_id).toBeUndefined();
|
expect(unit5.plan_id).toBeUndefined();
|
||||||
expect(unit.plan_name).toBeUndefined();
|
expect(unit5.plan_name).toBeUndefined();
|
||||||
expect(unit.href).toBeUndefined();
|
expect(unit5.href).toBeUndefined();
|
||||||
expect(unit.inventory_href).toBeUndefined();
|
expect(unit5.inventory_href).toBeUndefined();
|
||||||
expect(unit.specials_content).toBeUndefined();
|
expect(unit5.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 html = `
|
const units = parseUnits(sampleHtml, mockLogger);
|
||||||
<html><body>
|
const unit5 = units.find(u => u.unit_code === 'CCT-520');
|
||||||
<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(unit.id).toBe('1005');
|
expect(unit5.id).toBe('1005');
|
||||||
expect(unit.unit_code).toBe('SPARSE-1');
|
expect(unit5.unit_code).toBe('CCT-520');
|
||||||
expect(unit.unit_id).toBe('5005');
|
expect(unit5.unit_id).toBe('5005');
|
||||||
expect(unit.price).toBe('1300');
|
expect(unit5.price).toBe('1300');
|
||||||
expect(unit.available).toBe('true');
|
expect(unit5.available).toBe('true');
|
||||||
expect(unit.obj_type).toBe('unit');
|
expect(unit5.obj_type).toBe('unit');
|
||||||
expect(unit.community).toBe('Country Club Towers');
|
expect(unit5.community).toBe('Country Club Towers');
|
||||||
expect(unit.asset).toBe('100');
|
expect(unit5.asset).toBe('100');
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|
||||||
@ -367,7 +313,7 @@ describe('parseUnits', () => {
|
|||||||
|
|
||||||
expect(mockLogger.info).toHaveBeenCalledWith(
|
expect(mockLogger.info).toHaveBeenCalledWith(
|
||||||
'Units parsed',
|
'Units parsed',
|
||||||
expect.objectContaining({ count: 10 })
|
expect.objectContaining({ count: 5 })
|
||||||
);
|
);
|
||||||
});
|
});
|
||||||
|
|
||||||
@ -384,134 +330,31 @@ describe('parseUnits', () => {
|
|||||||
});
|
});
|
||||||
|
|
||||||
// ---------------------------------------------------------------
|
// ---------------------------------------------------------------
|
||||||
// 7. Fixture-based tests (sample-listing-empty.html, sample-listing-call.html)
|
// 7. Multiple units with varied data
|
||||||
// ---------------------------------------------------------------
|
|
||||||
describe('fixture: sample-listing-empty.html', () => {
|
|
||||||
let emptyHtml;
|
|
||||||
|
|
||||||
beforeAll(() => {
|
|
||||||
emptyHtml = fs.readFileSync(
|
|
||||||
path.join(fixturesDir, 'sample-listing-empty.html'),
|
|
||||||
'utf-8'
|
|
||||||
);
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should return empty array when fixture has container but no articles', () => {
|
|
||||||
const units = parseUnits(emptyHtml, mockLogger);
|
|
||||||
|
|
||||||
expect(units).toEqual([]);
|
|
||||||
// Container exists so no error about missing container
|
|
||||||
expect(mockLogger.error).not.toHaveBeenCalled();
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should log zero parsed units for empty fixture', () => {
|
|
||||||
parseUnits(emptyHtml, mockLogger);
|
|
||||||
|
|
||||||
expect(mockLogger.info).toHaveBeenCalledWith(
|
|
||||||
'Units parsed',
|
|
||||||
expect.objectContaining({ count: 0 })
|
|
||||||
);
|
|
||||||
});
|
|
||||||
});
|
|
||||||
|
|
||||||
describe('fixture: sample-listing-call.html', () => {
|
|
||||||
let callHtml;
|
|
||||||
|
|
||||||
beforeAll(() => {
|
|
||||||
callHtml = fs.readFileSync(
|
|
||||||
path.join(fixturesDir, 'sample-listing-call.html'),
|
|
||||||
'utf-8'
|
|
||||||
);
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should extract both units from the call-for-pricing fixture', () => {
|
|
||||||
const units = parseUnits(callHtml, mockLogger);
|
|
||||||
|
|
||||||
expect(units).toHaveLength(2);
|
|
||||||
expect(units.map(u => u.unit_code)).toEqual(['P-101', 'P-102']);
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should preserve "Call for pricing" as raw string in the price field', () => {
|
|
||||||
const units = parseUnits(callHtml, mockLogger);
|
|
||||||
|
|
||||||
expect(units[0].price).toBe('Call for pricing');
|
|
||||||
expect(units[1].price).toBe('Call for pricing');
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should extract all attributes correctly from call-for-pricing units', () => {
|
|
||||||
const units = parseUnits(callHtml, mockLogger);
|
|
||||||
const unit = units.find(u => u.unit_code === 'P-101');
|
|
||||||
|
|
||||||
expect(unit.id).toBe('9001');
|
|
||||||
expect(unit.unit_id).toBe('9001');
|
|
||||||
expect(unit.floor).toBe('1');
|
|
||||||
expect(unit.area).toBe('750');
|
|
||||||
expect(unit.bed_count).toBe('1');
|
|
||||||
expect(unit.bath_count).toBe('1');
|
|
||||||
expect(unit.available).toBe('true');
|
|
||||||
expect(unit.unavailable).toBe('false');
|
|
||||||
expect(unit.soonest).toBe('Now');
|
|
||||||
expect(unit.date_available).toBe('1706745600');
|
|
||||||
expect(unit.plan_id).toBe('500');
|
|
||||||
expect(unit.plan_name).toBe('Penthouse A');
|
|
||||||
expect(unit.community).toBe('Country Club Towers');
|
|
||||||
expect(unit.asset).toBe('420');
|
|
||||||
expect(unit.href).toBe('?spaces_tab=unit-detail&detail=9001');
|
|
||||||
expect(unit.inventory_href).toBe('?spaces_tab=unit-detail&detail=9001');
|
|
||||||
expect(unit.specials_content).toBe('');
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should not have image_url when fixture units lack img elements', () => {
|
|
||||||
const units = parseUnits(callHtml, mockLogger);
|
|
||||||
|
|
||||||
expect(units[0].image_url).toBeUndefined();
|
|
||||||
expect(units[1].image_url).toBeUndefined();
|
|
||||||
});
|
|
||||||
});
|
|
||||||
|
|
||||||
// ---------------------------------------------------------------
|
|
||||||
// 8. Multiple units with varied data
|
|
||||||
// ---------------------------------------------------------------
|
// ---------------------------------------------------------------
|
||||||
describe('varied unit data', () => {
|
describe('varied unit data', () => {
|
||||||
it('should handle unavailable units correctly', () => {
|
it('should handle unavailable units correctly', () => {
|
||||||
const html = `
|
const units = parseUnits(sampleHtml, mockLogger);
|
||||||
<html><body>
|
const unit4 = units.find(u => u.unit_code === 'CCT-415');
|
||||||
<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(unit.available).toBe('false');
|
expect(unit4.available).toBe('false');
|
||||||
expect(unit.unavailable).toBe('true');
|
expect(unit4.unavailable).toBe('true');
|
||||||
expect(unit.soonest).toBe('');
|
expect(unit4.soonest).toBe('');
|
||||||
expect(unit.date_available).toBe('');
|
expect(unit4.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);
|
||||||
// E3205 is the penthouse with 2.5 baths
|
const unit2 = units.find(u => u.unit_code === 'CCT-205');
|
||||||
const unit = units.find(u => u.unit_code === 'E3205');
|
|
||||||
|
|
||||||
expect(unit.bath_count).toBe('2.5');
|
expect(unit2.bath_count).toBe('1.5');
|
||||||
});
|
});
|
||||||
|
|
||||||
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);
|
||||||
// W2603 is the studio with bed_count 0
|
const unit3 = units.find(u => u.unit_code === 'CCT-310');
|
||||||
const unit = units.find(u => u.unit_code === 'W2603');
|
|
||||||
|
|
||||||
expect(unit.bed_count).toBe('0');
|
expect(unit3.bed_count).toBe('0');
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|||||||
@ -15,7 +15,6 @@ const { MongoMemoryServer } = require('mongodb-memory-server');
|
|||||||
|
|
||||||
// Will require after implementation
|
// Will require after implementation
|
||||||
let recordScraperRun;
|
let recordScraperRun;
|
||||||
let getNow;
|
|
||||||
|
|
||||||
let mongoServer;
|
let mongoServer;
|
||||||
let client;
|
let client;
|
||||||
@ -30,7 +29,7 @@ beforeAll(async () => {
|
|||||||
db = client.db('test_apartments');
|
db = client.db('test_apartments');
|
||||||
|
|
||||||
// Dynamically require to pick up implementation
|
// Dynamically require to pick up implementation
|
||||||
({ recordScraperRun, getNow } = require('../../services/scraperService'));
|
({ recordScraperRun } = require('../../services/scraperService'));
|
||||||
});
|
});
|
||||||
|
|
||||||
afterAll(async () => {
|
afterAll(async () => {
|
||||||
@ -270,26 +269,26 @@ describe('recordScraperRun', () => {
|
|||||||
// 6. recordedAt is a valid Date object
|
// 6. recordedAt is a valid Date object
|
||||||
// ---------------------------------------------------------------
|
// ---------------------------------------------------------------
|
||||||
describe('recordedAt field', () => {
|
describe('recordedAt field', () => {
|
||||||
it('should set recordedAt as a string timestamp', async () => {
|
it('should set recordedAt as a Date instance', async () => {
|
||||||
const runData = createSampleRunData();
|
const runData = createSampleRunData();
|
||||||
|
|
||||||
await recordScraperRun(db, runData, logger);
|
await recordScraperRun(db, runData, logger);
|
||||||
|
|
||||||
const doc = await db.collection(SCRAPER_RUNS_COLLECTION).findOne({});
|
const doc = await db.collection(SCRAPER_RUNS_COLLECTION).findOne({});
|
||||||
expect(typeof doc.recordedAt).toBe('string');
|
expect(doc.recordedAt).toBeInstanceOf(Date);
|
||||||
});
|
});
|
||||||
|
|
||||||
it('should set recordedAt close to current time', async () => {
|
it('should set recordedAt close to current time', async () => {
|
||||||
const beforeTime = getNow();
|
const beforeTime = new Date();
|
||||||
|
|
||||||
const runData = createSampleRunData();
|
const runData = createSampleRunData();
|
||||||
await recordScraperRun(db, runData, logger);
|
await recordScraperRun(db, runData, logger);
|
||||||
|
|
||||||
const afterTime = getNow();
|
const afterTime = new Date();
|
||||||
|
|
||||||
const doc = await db.collection(SCRAPER_RUNS_COLLECTION).findOne({});
|
const doc = await db.collection(SCRAPER_RUNS_COLLECTION).findOne({});
|
||||||
expect(doc.recordedAt >= beforeTime).toBe(true);
|
expect(doc.recordedAt.getTime()).toBeGreaterThanOrEqual(beforeTime.getTime());
|
||||||
expect(doc.recordedAt <= afterTime).toBe(true);
|
expect(doc.recordedAt.getTime()).toBeLessThanOrEqual(afterTime.getTime());
|
||||||
});
|
});
|
||||||
|
|
||||||
it('should not overwrite any existing runData fields with recordedAt', async () => {
|
it('should not overwrite any existing runData fields with recordedAt', async () => {
|
||||||
@ -302,7 +301,7 @@ describe('recordScraperRun', () => {
|
|||||||
// Verify the original fields still exist alongside recordedAt
|
// Verify the original fields still exist alongside recordedAt
|
||||||
expect(doc.jobId).toBe(runData.jobId);
|
expect(doc.jobId).toBe(runData.jobId);
|
||||||
expect(doc.status).toBe(runData.status);
|
expect(doc.status).toBe(runData.status);
|
||||||
expect(typeof doc.recordedAt).toBe('string');
|
expect(doc.recordedAt).toBeInstanceOf(Date);
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|||||||
@ -50,15 +50,15 @@ describe('config/scraper', () => {
|
|||||||
});
|
});
|
||||||
|
|
||||||
describe('SCRAPER_TIMEZONE', () => {
|
describe('SCRAPER_TIMEZONE', () => {
|
||||||
test('should default to "America/Denver" when env var not set', () => {
|
test('should default to "UTC" when env var not set', () => {
|
||||||
const config = require('../../config/scraper');
|
const config = require('../../config/scraper');
|
||||||
expect(config.SCRAPER_TIMEZONE).toBe('America/Denver');
|
expect(config.SCRAPER_TIMEZONE).toBe('UTC');
|
||||||
});
|
});
|
||||||
|
|
||||||
test('should use env var override when set', () => {
|
test('should use env var override when set', () => {
|
||||||
process.env.SCRAPER_TIMEZONE = 'Europe/London';
|
process.env.SCRAPER_TIMEZONE = 'America/Denver';
|
||||||
const config = require('../../config/scraper');
|
const config = require('../../config/scraper');
|
||||||
expect(config.SCRAPER_TIMEZONE).toBe('Europe/London');
|
expect(config.SCRAPER_TIMEZONE).toBe('America/Denver');
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|
||||||
@ -142,14 +142,14 @@ describe('config/scraper', () => {
|
|||||||
|
|
||||||
describe('COLLECTIONS', () => {
|
describe('COLLECTIONS', () => {
|
||||||
describe('default values', () => {
|
describe('default values', () => {
|
||||||
test('UNITS should default to "units_scraper"', () => {
|
test('UNITS should default to "units_migration_test"', () => {
|
||||||
const config = require('../../config/scraper');
|
const config = require('../../config/scraper');
|
||||||
expect(config.COLLECTIONS.UNITS).toBe('units_scraper');
|
expect(config.COLLECTIONS.UNITS).toBe('units_migration_test');
|
||||||
});
|
});
|
||||||
|
|
||||||
test('PRICES should default to "unit_prices_scraper"', () => {
|
test('PRICES should default to "unit_prices_migration_test"', () => {
|
||||||
const config = require('../../config/scraper');
|
const config = require('../../config/scraper');
|
||||||
expect(config.COLLECTIONS.PRICES).toBe('unit_prices_scraper');
|
expect(config.COLLECTIONS.PRICES).toBe('unit_prices_migration_test');
|
||||||
});
|
});
|
||||||
|
|
||||||
test('DAILY_SUMMARIES should default to "daily_summaries"', () => {
|
test('DAILY_SUMMARIES should default to "daily_summaries"', () => {
|
||||||
|
|||||||
@ -118,8 +118,8 @@ jest.mock('../../config/scraper', () => ({
|
|||||||
RETRY_CONFIG: { maxRetries: 3, baseDelay: 1000, timeout: 30000 },
|
RETRY_CONFIG: { maxRetries: 3, baseDelay: 1000, timeout: 30000 },
|
||||||
SHUTDOWN_TIMEOUT: 30000,
|
SHUTDOWN_TIMEOUT: 30000,
|
||||||
COLLECTIONS: {
|
COLLECTIONS: {
|
||||||
UNITS: 'units_scraper',
|
UNITS: 'units_migration_test',
|
||||||
PRICES: 'unit_prices_scraper',
|
PRICES: 'unit_prices_migration_test',
|
||||||
DAILY_SUMMARIES: 'daily_summaries',
|
DAILY_SUMMARIES: 'daily_summaries',
|
||||||
SCRAPER_RUNS: 'scraper_runs'
|
SCRAPER_RUNS: 'scraper_runs'
|
||||||
}
|
}
|
||||||
|
|||||||
@ -20,10 +20,9 @@ let mongoServer;
|
|||||||
let client;
|
let client;
|
||||||
let db;
|
let db;
|
||||||
|
|
||||||
// We will require runScrape, recordScraperRun, and createScraperIndexes after implementation
|
// We will require runScrape and recordScraperRun after implementation
|
||||||
let runScrape;
|
let runScrape;
|
||||||
let recordScraperRun;
|
let recordScraperRun;
|
||||||
let createScraperIndexes;
|
|
||||||
|
|
||||||
// Store original module references so we can mock individual functions
|
// Store original module references so we can mock individual functions
|
||||||
let scraperService;
|
let scraperService;
|
||||||
@ -37,7 +36,7 @@ beforeAll(async () => {
|
|||||||
|
|
||||||
// Dynamically require to pick up implementation
|
// Dynamically require to pick up implementation
|
||||||
scraperService = require('../../services/scraperService');
|
scraperService = require('../../services/scraperService');
|
||||||
({ runScrape, recordScraperRun, createScraperIndexes } = scraperService);
|
({ runScrape, recordScraperRun } = scraperService);
|
||||||
});
|
});
|
||||||
|
|
||||||
afterAll(async () => {
|
afterAll(async () => {
|
||||||
@ -121,7 +120,7 @@ describe('recordScraperRun', () => {
|
|||||||
const saved = await db.collection('scraper_runs').findOne({ jobId: 'test-job-001' });
|
const saved = await db.collection('scraper_runs').findOne({ jobId: 'test-job-001' });
|
||||||
expect(saved).toBeTruthy();
|
expect(saved).toBeTruthy();
|
||||||
expect(saved.status).toBe('success');
|
expect(saved.status).toBe('success');
|
||||||
expect(typeof saved.recordedAt).toBe('string');
|
expect(saved.recordedAt).toBeInstanceOf(Date);
|
||||||
});
|
});
|
||||||
|
|
||||||
it('should not throw when insert fails', async () => {
|
it('should not throw when insert fails', async () => {
|
||||||
@ -220,11 +219,11 @@ describe('runScrape', () => {
|
|||||||
expect(result.staleUnitsCount).toBe(0);
|
expect(result.staleUnitsCount).toBe(0);
|
||||||
|
|
||||||
// Verify no units were written to the units collection
|
// Verify no units were written to the units collection
|
||||||
const unitsCount = await db.collection('units_scraper').countDocuments();
|
const unitsCount = await db.collection('units_migration_test').countDocuments();
|
||||||
expect(unitsCount).toBe(0);
|
expect(unitsCount).toBe(0);
|
||||||
|
|
||||||
// Verify no prices were written
|
// Verify no prices were written
|
||||||
const pricesCount = await db.collection('unit_prices_scraper').countDocuments();
|
const pricesCount = await db.collection('unit_prices_migration_test').countDocuments();
|
||||||
expect(pricesCount).toBe(0);
|
expect(pricesCount).toBe(0);
|
||||||
});
|
});
|
||||||
|
|
||||||
@ -263,7 +262,7 @@ describe('runScrape', () => {
|
|||||||
const runRecord = await db.collection('scraper_runs').findOne({ jobId: 'test-record-success' });
|
const runRecord = await db.collection('scraper_runs').findOne({ jobId: 'test-record-success' });
|
||||||
expect(runRecord).toBeTruthy();
|
expect(runRecord).toBeTruthy();
|
||||||
expect(runRecord.status).toBe('success');
|
expect(runRecord.status).toBe('success');
|
||||||
expect(typeof runRecord.recordedAt).toBe('string');
|
expect(runRecord.recordedAt).toBeInstanceOf(Date);
|
||||||
expect(runRecord.unitsProcessed).toBe(1);
|
expect(runRecord.unitsProcessed).toBe(1);
|
||||||
});
|
});
|
||||||
|
|
||||||
@ -278,7 +277,7 @@ describe('runScrape', () => {
|
|||||||
const faultyDb = {
|
const faultyDb = {
|
||||||
collection: (name) => {
|
collection: (name) => {
|
||||||
const realCollection = db.collection(name);
|
const realCollection = db.collection(name);
|
||||||
if (name === 'units_scraper') {
|
if (name === 'units_migration_test') {
|
||||||
return new Proxy(realCollection, {
|
return new Proxy(realCollection, {
|
||||||
get(target, prop) {
|
get(target, prop) {
|
||||||
if (prop === 'bulkWrite') {
|
if (prop === 'bulkWrite') {
|
||||||
@ -344,7 +343,7 @@ describe('runScrape', () => {
|
|||||||
const faultyDb = {
|
const faultyDb = {
|
||||||
collection: (name) => {
|
collection: (name) => {
|
||||||
const realCollection = db.collection(name);
|
const realCollection = db.collection(name);
|
||||||
if (name === 'unit_prices_scraper') {
|
if (name === 'unit_prices_migration_test') {
|
||||||
return new Proxy(realCollection, {
|
return new Proxy(realCollection, {
|
||||||
get(target, prop) {
|
get(target, prop) {
|
||||||
if (prop === 'bulkWrite') {
|
if (prop === 'bulkWrite') {
|
||||||
@ -389,7 +388,7 @@ describe('runScrape', () => {
|
|||||||
const yesterdayStr = yesterday.toISOString().split('T')[0];
|
const yesterdayStr = yesterday.toISOString().split('T')[0];
|
||||||
|
|
||||||
// Yesterday had units: OLD-A, OLD-B, OLD-C
|
// Yesterday had units: OLD-A, OLD-B, OLD-C
|
||||||
await db.collection('unit_prices_scraper').insertMany([
|
await db.collection('unit_prices_migration_test').insertMany([
|
||||||
{ unit_code: 'OLD-A', date_checked: yesterdayStr, price: 1000 },
|
{ unit_code: 'OLD-A', date_checked: yesterdayStr, price: 1000 },
|
||||||
{ unit_code: 'OLD-B', date_checked: yesterdayStr, price: 1100 },
|
{ unit_code: 'OLD-B', date_checked: yesterdayStr, price: 1100 },
|
||||||
{ unit_code: 'OLD-C', date_checked: yesterdayStr, price: 1200 }
|
{ unit_code: 'OLD-C', date_checked: yesterdayStr, price: 1200 }
|
||||||
@ -632,7 +631,7 @@ describe('sanitizeError', () => {
|
|||||||
const faultyDb = {
|
const faultyDb = {
|
||||||
collection: (name) => {
|
collection: (name) => {
|
||||||
const realCollection = db.collection(name);
|
const realCollection = db.collection(name);
|
||||||
if (name === 'units_scraper') {
|
if (name === 'units_migration_test') {
|
||||||
return new Proxy(realCollection, {
|
return new Proxy(realCollection, {
|
||||||
get(target, prop) {
|
get(target, prop) {
|
||||||
if (prop === 'bulkWrite') {
|
if (prop === 'bulkWrite') {
|
||||||
@ -675,65 +674,3 @@ describe('sanitizeError', () => {
|
|||||||
});
|
});
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|
||||||
// ============================================================
|
|
||||||
// Test: createScraperIndexes
|
|
||||||
// ============================================================
|
|
||||||
describe('createScraperIndexes', () => {
|
|
||||||
const mockLogger = { info: jest.fn(), warn: jest.fn(), error: jest.fn() };
|
|
||||||
|
|
||||||
beforeEach(() => {
|
|
||||||
mockLogger.info.mockClear();
|
|
||||||
mockLogger.warn.mockClear();
|
|
||||||
mockLogger.error.mockClear();
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should create the status_startedAt compound index', async () => {
|
|
||||||
await createScraperIndexes(db, mockLogger);
|
|
||||||
|
|
||||||
const indexes = await db.collection('scraper_runs').indexes();
|
|
||||||
const statusIndex = indexes.find(idx => idx.name === 'status_startedAt');
|
|
||||||
expect(statusIndex).toBeTruthy();
|
|
||||||
expect(statusIndex.key).toEqual({ status: 1, startedAt: -1 });
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should create the startedAt_desc index', async () => {
|
|
||||||
await createScraperIndexes(db, mockLogger);
|
|
||||||
|
|
||||||
const indexes = await db.collection('scraper_runs').indexes();
|
|
||||||
const startedAtIndex = indexes.find(idx => idx.name === 'startedAt_desc');
|
|
||||||
expect(startedAtIndex).toBeTruthy();
|
|
||||||
expect(startedAtIndex.key).toEqual({ startedAt: -1 });
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should log success after creating indexes', async () => {
|
|
||||||
await createScraperIndexes(db, mockLogger);
|
|
||||||
|
|
||||||
expect(mockLogger.info).toHaveBeenCalledWith('Scraper indexes created successfully');
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should be idempotent (no error on second call)', async () => {
|
|
||||||
await createScraperIndexes(db, mockLogger);
|
|
||||||
|
|
||||||
// Calling again should not throw
|
|
||||||
await expect(createScraperIndexes(db, mockLogger)).resolves.not.toThrow();
|
|
||||||
|
|
||||||
// Indexes should still exist
|
|
||||||
const indexes = await db.collection('scraper_runs').indexes();
|
|
||||||
const statusIndex = indexes.find(idx => idx.name === 'status_startedAt');
|
|
||||||
const startedAtIndex = indexes.find(idx => idx.name === 'startedAt_desc');
|
|
||||||
expect(statusIndex).toBeTruthy();
|
|
||||||
expect(startedAtIndex).toBeTruthy();
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should handle database errors gracefully', async () => {
|
|
||||||
// Pass an object that will throw when collection() is called
|
|
||||||
const faultyDb = {
|
|
||||||
collection: () => {
|
|
||||||
throw new Error('Connection lost');
|
|
||||||
}
|
|
||||||
};
|
|
||||||
|
|
||||||
await expect(createScraperIndexes(faultyDb, mockLogger)).rejects.toThrow('Connection lost');
|
|
||||||
});
|
|
||||||
});
|
|
||||||
|
|||||||
@ -1,177 +0,0 @@
|
|||||||
/**
|
|
||||||
* Tests for scraper integration in server.js startup
|
|
||||||
*
|
|
||||||
* Covers:
|
|
||||||
* - Server initializes scraper scheduler on startup
|
|
||||||
* - Server does not crash if scheduler init fails
|
|
||||||
* - Indexes are created before scheduler starts
|
|
||||||
* - Scheduler not initialized when SCRAPER_ENABLED=false
|
|
||||||
*
|
|
||||||
* Strategy: Instead of importing server.js directly (which starts Express and
|
|
||||||
* connects to MongoDB), we replicate the scraper initialization logic from
|
|
||||||
* connectToMongoDB() and test it with mocked dependencies. This verifies the
|
|
||||||
* integration contract without needing a full server boot.
|
|
||||||
*/
|
|
||||||
|
|
||||||
const { MongoClient } = require('mongodb');
|
|
||||||
|
|
||||||
describe('Server Scraper Integration', () => {
|
|
||||||
let connection;
|
|
||||||
let db;
|
|
||||||
|
|
||||||
beforeAll(async () => {
|
|
||||||
const uri = process.env.MONGO_URI;
|
|
||||||
connection = await MongoClient.connect(uri);
|
|
||||||
db = connection.db('apartments_server_integration_test');
|
|
||||||
});
|
|
||||||
|
|
||||||
afterAll(async () => {
|
|
||||||
if (connection) {
|
|
||||||
await connection.close();
|
|
||||||
}
|
|
||||||
});
|
|
||||||
|
|
||||||
describe('scraper initialization during startup', () => {
|
|
||||||
let mockCreateScraperIndexes;
|
|
||||||
let mockInitializeScheduler;
|
|
||||||
let mockCreateLogger;
|
|
||||||
let mockLogger;
|
|
||||||
|
|
||||||
beforeEach(() => {
|
|
||||||
mockLogger = {
|
|
||||||
info: jest.fn(),
|
|
||||||
warn: jest.fn(),
|
|
||||||
error: jest.fn(),
|
|
||||||
debug: jest.fn()
|
|
||||||
};
|
|
||||||
mockCreateLogger = jest.fn(() => mockLogger);
|
|
||||||
mockCreateScraperIndexes = jest.fn().mockResolvedValue(undefined);
|
|
||||||
mockInitializeScheduler = jest.fn();
|
|
||||||
});
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Replicates the scraper initialization logic from server.js connectToMongoDB().
|
|
||||||
* This is the exact pattern used in production:
|
|
||||||
*
|
|
||||||
* try {
|
|
||||||
* const scraperLogger = createLogger('server-init');
|
|
||||||
* await createScraperIndexes(db, scraperLogger);
|
|
||||||
* if (scraperConfig.SCRAPER_ENABLED) {
|
|
||||||
* initializeScheduler(db);
|
|
||||||
* }
|
|
||||||
* } catch (error) {
|
|
||||||
* console.error('Scraper initialization failed (server continues):', error.message);
|
|
||||||
* }
|
|
||||||
*/
|
|
||||||
async function runScraperInit(testDb, config) {
|
|
||||||
try {
|
|
||||||
const scraperLogger = mockCreateLogger('server-init');
|
|
||||||
await mockCreateScraperIndexes(testDb, scraperLogger);
|
|
||||||
if (config.SCRAPER_ENABLED) {
|
|
||||||
mockInitializeScheduler(testDb);
|
|
||||||
}
|
|
||||||
return { success: true };
|
|
||||||
} catch (error) {
|
|
||||||
return { success: false, error: error.message };
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
it('should initialize scraper scheduler on startup when enabled', async () => {
|
|
||||||
const config = { SCRAPER_ENABLED: true };
|
|
||||||
|
|
||||||
const result = await runScraperInit(db, config);
|
|
||||||
|
|
||||||
expect(result.success).toBe(true);
|
|
||||||
expect(mockInitializeScheduler).toHaveBeenCalledTimes(1);
|
|
||||||
expect(mockInitializeScheduler).toHaveBeenCalledWith(db);
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should not crash if createScraperIndexes throws an error', async () => {
|
|
||||||
mockCreateScraperIndexes.mockRejectedValue(new Error('Index creation failed'));
|
|
||||||
const config = { SCRAPER_ENABLED: true };
|
|
||||||
|
|
||||||
const result = await runScraperInit(db, config);
|
|
||||||
|
|
||||||
// The server should continue (error is caught, not thrown)
|
|
||||||
expect(result.success).toBe(false);
|
|
||||||
expect(result.error).toBe('Index creation failed');
|
|
||||||
// initializeScheduler should NOT have been called since the error occurred before it
|
|
||||||
expect(mockInitializeScheduler).not.toHaveBeenCalled();
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should create indexes before initializing the scheduler', async () => {
|
|
||||||
const callOrder = [];
|
|
||||||
mockCreateScraperIndexes.mockImplementation(async () => {
|
|
||||||
callOrder.push('createScraperIndexes');
|
|
||||||
});
|
|
||||||
mockInitializeScheduler.mockImplementation(() => {
|
|
||||||
callOrder.push('initializeScheduler');
|
|
||||||
});
|
|
||||||
|
|
||||||
const config = { SCRAPER_ENABLED: true };
|
|
||||||
await runScraperInit(db, config);
|
|
||||||
|
|
||||||
expect(callOrder).toEqual(['createScraperIndexes', 'initializeScheduler']);
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should not initialize scheduler when SCRAPER_ENABLED is false', async () => {
|
|
||||||
const config = { SCRAPER_ENABLED: false };
|
|
||||||
|
|
||||||
const result = await runScraperInit(db, config);
|
|
||||||
|
|
||||||
expect(result.success).toBe(true);
|
|
||||||
expect(mockCreateScraperIndexes).toHaveBeenCalledTimes(1);
|
|
||||||
expect(mockInitializeScheduler).not.toHaveBeenCalled();
|
|
||||||
});
|
|
||||||
});
|
|
||||||
|
|
||||||
describe('server.js imports and exports verification', () => {
|
|
||||||
it('should be able to import createScraperIndexes from scraperService', () => {
|
|
||||||
const { createScraperIndexes } = require('../../services/scraperService');
|
|
||||||
expect(typeof createScraperIndexes).toBe('function');
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should be able to import initializeScheduler from scraperJob', () => {
|
|
||||||
const { initializeScheduler } = require('../../jobs/scraperJob');
|
|
||||||
expect(typeof initializeScheduler).toBe('function');
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should be able to import registerSignalHandlers from scraperJob', () => {
|
|
||||||
const { registerSignalHandlers } = require('../../jobs/scraperJob');
|
|
||||||
expect(typeof registerSignalHandlers).toBe('function');
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should be able to import SCRAPER_ENABLED from scraper config', () => {
|
|
||||||
const scraperConfig = require('../../config/scraper');
|
|
||||||
expect(typeof scraperConfig.SCRAPER_ENABLED).toBe('boolean');
|
|
||||||
});
|
|
||||||
});
|
|
||||||
|
|
||||||
describe('createScraperIndexes error handling with real db', () => {
|
|
||||||
it('should handle index creation gracefully with a valid db', async () => {
|
|
||||||
const { createScraperIndexes } = require('../../services/scraperService');
|
|
||||||
const { createLogger } = require('../../services/scraperLogger');
|
|
||||||
const logger = createLogger('test-server-init');
|
|
||||||
|
|
||||||
// Should not throw - creates indexes on the test db
|
|
||||||
await expect(createScraperIndexes(db, logger)).resolves.not.toThrow();
|
|
||||||
});
|
|
||||||
|
|
||||||
it('should throw when given an invalid db object', async () => {
|
|
||||||
const { createScraperIndexes } = require('../../services/scraperService');
|
|
||||||
const mockLogger = { info: jest.fn(), error: jest.fn(), warn: jest.fn(), debug: jest.fn() };
|
|
||||||
|
|
||||||
// Passing null should throw (simulates what the try/catch in server.js protects against)
|
|
||||||
await expect(createScraperIndexes(null, mockLogger)).rejects.toThrow();
|
|
||||||
});
|
|
||||||
});
|
|
||||||
|
|
||||||
describe('registerSignalHandlers', () => {
|
|
||||||
it('should register SIGTERM and SIGINT handlers without throwing', () => {
|
|
||||||
const { registerSignalHandlers } = require('../../jobs/scraperJob');
|
|
||||||
|
|
||||||
// Should not throw when called
|
|
||||||
expect(() => registerSignalHandlers()).not.toThrow();
|
|
||||||
});
|
|
||||||
});
|
|
||||||
});
|
|
||||||
@ -554,8 +554,8 @@ describe('updateDailySummary', () => {
|
|||||||
const doc = await db.collection(DAILY_SUMMARIES_COLLECTION).findOne({ date: '2026-02-05' });
|
const doc = await db.collection(DAILY_SUMMARIES_COLLECTION).findOne({ date: '2026-02-05' });
|
||||||
expect(doc.timestamp).toBeDefined();
|
expect(doc.timestamp).toBeDefined();
|
||||||
expect(typeof doc.timestamp).toBe('string');
|
expect(typeof doc.timestamp).toBe('string');
|
||||||
// Verify it matches the Denver-local timestamp format (YYYY-MM-DDTHH:mm:ss.mmm)
|
// Verify it is a valid ISO timestamp
|
||||||
expect(doc.timestamp).toMatch(/^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{3}$/);
|
expect(new Date(doc.timestamp).toISOString()).toBe(doc.timestamp);
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|||||||
@ -15,9 +15,8 @@
|
|||||||
const { MongoClient } = require('mongodb');
|
const { MongoClient } = require('mongodb');
|
||||||
const { MongoMemoryServer } = require('mongodb-memory-server');
|
const { MongoMemoryServer } = require('mongodb-memory-server');
|
||||||
|
|
||||||
// We will require upsertUnits and getNow after implementation
|
// We will require upsertUnits after implementation
|
||||||
let upsertUnits;
|
let upsertUnits;
|
||||||
let getNow;
|
|
||||||
|
|
||||||
let mongoServer;
|
let mongoServer;
|
||||||
let client;
|
let client;
|
||||||
@ -40,7 +39,7 @@ beforeAll(async () => {
|
|||||||
db = client.db('test_apartments');
|
db = client.db('test_apartments');
|
||||||
|
|
||||||
// Dynamically require to pick up implementation
|
// Dynamically require to pick up implementation
|
||||||
({ upsertUnits, getNow } = require('../../services/scraperService'));
|
({ upsertUnits } = require('../../services/scraperService'));
|
||||||
});
|
});
|
||||||
|
|
||||||
afterAll(async () => {
|
afterAll(async () => {
|
||||||
@ -57,7 +56,7 @@ beforeEach(async () => {
|
|||||||
});
|
});
|
||||||
|
|
||||||
describe('upsertUnits', () => {
|
describe('upsertUnits', () => {
|
||||||
const UNITS_COLLECTION = 'units_scraper';
|
const UNITS_COLLECTION = 'units_migration_test';
|
||||||
|
|
||||||
// ---------------------------------------------------------------
|
// ---------------------------------------------------------------
|
||||||
// 1. bulkWrite is called with correct updateOne operations
|
// 1. bulkWrite is called with correct updateOne operations
|
||||||
@ -146,12 +145,12 @@ describe('upsertUnits', () => {
|
|||||||
describe('$set fields', () => {
|
describe('$set fields', () => {
|
||||||
it('should set last_scraped timestamp on every upsert', async () => {
|
it('should set last_scraped timestamp on every upsert', async () => {
|
||||||
const logger = createMockLogger();
|
const logger = createMockLogger();
|
||||||
const beforeTime = getNow();
|
const beforeTime = new Date().toISOString();
|
||||||
|
|
||||||
const units = [{ unit_code: 'T100', price: 1000 }];
|
const units = [{ unit_code: 'T100', price: 1000 }];
|
||||||
await upsertUnits(db, units, logger);
|
await upsertUnits(db, units, logger);
|
||||||
|
|
||||||
const afterTime = getNow();
|
const afterTime = new Date().toISOString();
|
||||||
|
|
||||||
const doc = await db.collection(UNITS_COLLECTION).findOne({ unit_code: 'T100' });
|
const doc = await db.collection(UNITS_COLLECTION).findOne({ unit_code: 'T100' });
|
||||||
expect(doc.last_scraped).toBeDefined();
|
expect(doc.last_scraped).toBeDefined();
|
||||||
@ -222,7 +221,7 @@ describe('upsertUnits', () => {
|
|||||||
describe('$setOnInsert for first_seen', () => {
|
describe('$setOnInsert for first_seen', () => {
|
||||||
it('should set first_seen on initial insert', async () => {
|
it('should set first_seen on initial insert', async () => {
|
||||||
const logger = createMockLogger();
|
const logger = createMockLogger();
|
||||||
const beforeTime = getNow();
|
const beforeTime = new Date().toISOString();
|
||||||
|
|
||||||
await upsertUnits(db, [{ unit_code: 'F100', price: 1000 }], logger);
|
await upsertUnits(db, [{ unit_code: 'F100', price: 1000 }], logger);
|
||||||
|
|
||||||
|
|||||||
@ -13,7 +13,7 @@ module.exports = {
|
|||||||
|
|
||||||
// Scheduling configuration
|
// Scheduling configuration
|
||||||
SCRAPER_SCHEDULE: process.env.SCRAPER_SCHEDULE || '0 6 * * *',
|
SCRAPER_SCHEDULE: process.env.SCRAPER_SCHEDULE || '0 6 * * *',
|
||||||
SCRAPER_TIMEZONE: process.env.SCRAPER_TIMEZONE || 'America/Denver',
|
SCRAPER_TIMEZONE: process.env.SCRAPER_TIMEZONE || 'UTC',
|
||||||
SCRAPER_ENABLED: process.env.SCRAPER_ENABLED !== 'false',
|
SCRAPER_ENABLED: process.env.SCRAPER_ENABLED !== 'false',
|
||||||
|
|
||||||
// HTTP settings
|
// HTTP settings
|
||||||
@ -30,11 +30,10 @@ module.exports = {
|
|||||||
// Graceful shutdown timeout (how long to wait for running job before force-stopping)
|
// Graceful shutdown timeout (how long to wait for running job before force-stopping)
|
||||||
SHUTDOWN_TIMEOUT: parseInt(process.env.SCRAPER_SHUTDOWN_TIMEOUT, 10) || 30000,
|
SHUTDOWN_TIMEOUT: parseInt(process.env.SCRAPER_SHUTDOWN_TIMEOUT, 10) || 30000,
|
||||||
|
|
||||||
// MongoDB collection names - defaults to validation collections for scraper validation;
|
// MongoDB collection names (environment variable overrides for development isolation)
|
||||||
// switch to production collections via env vars (SCRAPER_UNITS_COLLECTION, SCRAPER_PRICES_COLLECTION) when ready
|
|
||||||
COLLECTIONS: {
|
COLLECTIONS: {
|
||||||
UNITS: process.env.SCRAPER_UNITS_COLLECTION || 'units_scraper',
|
UNITS: process.env.SCRAPER_UNITS_COLLECTION || 'units_migration_test',
|
||||||
PRICES: process.env.SCRAPER_PRICES_COLLECTION || 'unit_prices_scraper',
|
PRICES: process.env.SCRAPER_PRICES_COLLECTION || 'unit_prices_migration_test',
|
||||||
DAILY_SUMMARIES: process.env.SCRAPER_SUMMARIES_COLLECTION || 'daily_summaries',
|
DAILY_SUMMARIES: process.env.SCRAPER_SUMMARIES_COLLECTION || 'daily_summaries',
|
||||||
SCRAPER_RUNS: process.env.SCRAPER_RUNS_COLLECTION || 'scraper_runs'
|
SCRAPER_RUNS: process.env.SCRAPER_RUNS_COLLECTION || 'scraper_runs'
|
||||||
}
|
}
|
||||||
|
|||||||
@ -1,5 +1,3 @@
|
|||||||
name: apartment-api
|
|
||||||
|
|
||||||
services:
|
services:
|
||||||
apartment-api:
|
apartment-api:
|
||||||
image: ${IMAGE:-apartment-api:latest}
|
image: ${IMAGE:-apartment-api:latest}
|
||||||
@ -22,7 +20,9 @@ services:
|
|||||||
- "traefik.enable=true"
|
- "traefik.enable=true"
|
||||||
- "traefik.docker.network=traefik"
|
- "traefik.docker.network=traefik"
|
||||||
- "traefik.http.routers.apartment-api.rule=Host(`apartments.maverickapplications.com`) && PathPrefix(`/api`)"
|
- "traefik.http.routers.apartment-api.rule=Host(`apartments.maverickapplications.com`) && PathPrefix(`/api`)"
|
||||||
- "traefik.http.routers.apartment-api.entrypoints=web"
|
- "traefik.http.routers.apartment-api.entrypoints=websecure"
|
||||||
|
- "traefik.http.routers.apartment-api.tls=true"
|
||||||
|
- "traefik.http.routers.apartment-api.tls.certresolver=letsencrypt"
|
||||||
- "traefik.http.routers.apartment-api.priority=100"
|
- "traefik.http.routers.apartment-api.priority=100"
|
||||||
- "traefik.http.services.apartment-api.loadbalancer.server.port=8080" # Updated to match
|
- "traefik.http.services.apartment-api.loadbalancer.server.port=8080" # Updated to match
|
||||||
- "traefik.http.middlewares.apartment-api-stripprefix.stripprefix.prefixes=/api"
|
- "traefik.http.middlewares.apartment-api-stripprefix.stripprefix.prefixes=/api"
|
||||||
|
|||||||
16
package-lock.json
generated
16
package-lock.json
generated
@ -1,15 +1,15 @@
|
|||||||
{
|
{
|
||||||
"name": "apartment-api",
|
"name": "apartment-api",
|
||||||
"version": "1.0.1",
|
"version": "1.0.0",
|
||||||
"lockfileVersion": 3,
|
"lockfileVersion": 3,
|
||||||
"requires": true,
|
"requires": true,
|
||||||
"packages": {
|
"packages": {
|
||||||
"": {
|
"": {
|
||||||
"name": "apartment-api",
|
"name": "apartment-api",
|
||||||
"version": "1.0.1",
|
"version": "1.0.0",
|
||||||
"license": "MIT",
|
"license": "MIT",
|
||||||
"dependencies": {
|
"dependencies": {
|
||||||
"axios": "^1.13.5",
|
"axios": "^1.13.4",
|
||||||
"cheerio": "^1.2.0",
|
"cheerio": "^1.2.0",
|
||||||
"cookie-parser": "^1.4.7",
|
"cookie-parser": "^1.4.7",
|
||||||
"cors": "^2.8.5",
|
"cors": "^2.8.5",
|
||||||
@ -2032,13 +2032,13 @@
|
|||||||
"license": "MIT"
|
"license": "MIT"
|
||||||
},
|
},
|
||||||
"node_modules/axios": {
|
"node_modules/axios": {
|
||||||
"version": "1.13.5",
|
"version": "1.13.4",
|
||||||
"resolved": "https://registry.npmjs.org/axios/-/axios-1.13.5.tgz",
|
"resolved": "https://registry.npmjs.org/axios/-/axios-1.13.4.tgz",
|
||||||
"integrity": "sha512-cz4ur7Vb0xS4/KUN0tPWe44eqxrIu31me+fbang3ijiNscE129POzipJJA6zniq2C/Z6sJCjMimjS8Lc/GAs8Q==",
|
"integrity": "sha512-1wVkUaAO6WyaYtCkcYCOx12ZgpGf9Zif+qXa4n+oYzK558YryKqiL6UWwd5DqiH3VRW0GYhTZQ/vlgJrCoNQlg==",
|
||||||
"license": "MIT",
|
"license": "MIT",
|
||||||
"dependencies": {
|
"dependencies": {
|
||||||
"follow-redirects": "^1.15.11",
|
"follow-redirects": "^1.15.6",
|
||||||
"form-data": "^4.0.5",
|
"form-data": "^4.0.4",
|
||||||
"proxy-from-env": "^1.1.0"
|
"proxy-from-env": "^1.1.0"
|
||||||
}
|
}
|
||||||
},
|
},
|
||||||
|
|||||||
@ -1,6 +1,6 @@
|
|||||||
{
|
{
|
||||||
"name": "apartment-api",
|
"name": "apartment-api",
|
||||||
"version": "1.0.1",
|
"version": "1.0.0",
|
||||||
"description": "API backend for Country Club Towers & Gardens apartment dashboard",
|
"description": "API backend for Country Club Towers & Gardens apartment dashboard",
|
||||||
"main": "server.js",
|
"main": "server.js",
|
||||||
"scripts": {
|
"scripts": {
|
||||||
@ -21,7 +21,7 @@
|
|||||||
"author": "Stephen",
|
"author": "Stephen",
|
||||||
"license": "MIT",
|
"license": "MIT",
|
||||||
"dependencies": {
|
"dependencies": {
|
||||||
"axios": "^1.13.5",
|
"axios": "^1.13.4",
|
||||||
"cheerio": "^1.2.0",
|
"cheerio": "^1.2.0",
|
||||||
"cookie-parser": "^1.4.7",
|
"cookie-parser": "^1.4.7",
|
||||||
"cors": "^2.8.5",
|
"cors": "^2.8.5",
|
||||||
|
|||||||
38
server.js
38
server.js
@ -11,11 +11,6 @@ const activityRoutes = require('./routes/activity');
|
|||||||
const adminRoutes = require('./routes/admin');
|
const adminRoutes = require('./routes/admin');
|
||||||
const { createIndexes } = require('./models/user');
|
const { createIndexes } = require('./models/user');
|
||||||
const { createActivityIndexes } = require('./services/activityLogger');
|
const { createActivityIndexes } = require('./services/activityLogger');
|
||||||
const { createScraperIndexes } = require('./services/scraperService');
|
|
||||||
const { initializeScheduler, registerSignalHandlers } = require('./jobs/scraperJob');
|
|
||||||
const scraperConfig = require('./config/scraper');
|
|
||||||
|
|
||||||
const { version: API_VERSION } = require('./package.json');
|
|
||||||
|
|
||||||
const app = express();
|
const app = express();
|
||||||
const PORT = process.env.PORT || 3000;
|
const PORT = process.env.PORT || 3000;
|
||||||
@ -85,32 +80,14 @@ async function connectToMongoDB() {
|
|||||||
await createIndexes(db);
|
await createIndexes(db);
|
||||||
await createActivityIndexes(db);
|
await createActivityIndexes(db);
|
||||||
console.log('✅ Passport configured and indexes created');
|
console.log('✅ Passport configured and indexes created');
|
||||||
|
|
||||||
// Initialize scraper (indexes + scheduler) - errors logged but don't crash server
|
|
||||||
try {
|
|
||||||
const { createLogger } = require('./services/scraperLogger');
|
|
||||||
const scraperLogger = createLogger('server-init');
|
|
||||||
await createScraperIndexes(db, scraperLogger);
|
|
||||||
if (scraperConfig.SCRAPER_ENABLED) {
|
|
||||||
initializeScheduler(db);
|
|
||||||
console.log('✅ Scraper scheduler initialized');
|
|
||||||
} else {
|
|
||||||
console.log('ℹ️ Scraper scheduling disabled');
|
|
||||||
}
|
|
||||||
} catch (error) {
|
|
||||||
console.error('⚠️ Scraper initialization failed (server continues):', error.message);
|
|
||||||
}
|
|
||||||
} catch (error) {
|
} catch (error) {
|
||||||
console.error('❌ Failed to connect to MongoDB:', error);
|
console.error('❌ Failed to connect to MongoDB:', error);
|
||||||
process.exit(1);
|
process.exit(1);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// Register scraper signal handlers for graceful shutdown
|
// Helper function to get today's date in UTC
|
||||||
registerSignalHandlers();
|
const getTodayDateUTC = () => new Date().toISOString().split('T')[0];
|
||||||
|
|
||||||
// Helper function to get today's date in configured timezone
|
|
||||||
const getTodayDate = () => new Date().toLocaleDateString('en-CA', { timeZone: scraperConfig.SCRAPER_TIMEZONE });
|
|
||||||
|
|
||||||
// Cache for the most recent date with data
|
// Cache for the most recent date with data
|
||||||
let cachedLatestDate = null;
|
let cachedLatestDate = null;
|
||||||
@ -145,7 +122,7 @@ const getLatestDateWithData = async () => {
|
|||||||
}
|
}
|
||||||
|
|
||||||
// Fallback to UTC today if no data found
|
// Fallback to UTC today if no data found
|
||||||
return getTodayDate();
|
return getTodayDateUTC();
|
||||||
};
|
};
|
||||||
|
|
||||||
// Helper function to get yesterday relative to the latest date with data
|
// Helper function to get yesterday relative to the latest date with data
|
||||||
@ -177,7 +154,6 @@ app.use('/admin', adminRoutes);
|
|||||||
app.get('/health', (req, res) => {
|
app.get('/health', (req, res) => {
|
||||||
res.json({
|
res.json({
|
||||||
status: 'ok',
|
status: 'ok',
|
||||||
version: API_VERSION,
|
|
||||||
timestamp: new Date().toISOString(),
|
timestamp: new Date().toISOString(),
|
||||||
service: 'apartment-api'
|
service: 'apartment-api'
|
||||||
});
|
});
|
||||||
@ -644,10 +620,10 @@ app.get('/analytics', requireAuth, async (req, res) => {
|
|||||||
]).toArray();
|
]).toArray();
|
||||||
|
|
||||||
// 4. Historical Comparison - current vs 30/90/365 day averages
|
// 4. Historical Comparison - current vs 30/90/365 day averages
|
||||||
const tz = scraperConfig.SCRAPER_TIMEZONE;
|
const now = new Date();
|
||||||
const date30 = new Date(Date.now() - 30 * 24 * 60 * 60 * 1000).toLocaleDateString('en-CA', { timeZone: tz });
|
const date30 = new Date(now.getTime() - 30 * 24 * 60 * 60 * 1000).toISOString().split('T')[0];
|
||||||
const date90 = new Date(Date.now() - 90 * 24 * 60 * 60 * 1000).toLocaleDateString('en-CA', { timeZone: tz });
|
const date90 = new Date(now.getTime() - 90 * 24 * 60 * 60 * 1000).toISOString().split('T')[0];
|
||||||
const date365 = new Date(Date.now() - 365 * 24 * 60 * 60 * 1000).toLocaleDateString('en-CA', { timeZone: tz });
|
const date365 = new Date(now.getTime() - 365 * 24 * 60 * 60 * 1000).toISOString().split('T')[0];
|
||||||
|
|
||||||
const [currentAvg, avg30, avg90, avg365] = await Promise.all([
|
const [currentAvg, avg30, avg90, avg365] = await Promise.all([
|
||||||
db.collection(PRICES_COLLECTION).aggregate([
|
db.collection(PRICES_COLLECTION).aggregate([
|
||||||
|
|||||||
@ -403,7 +403,7 @@ function getYesterday(dateStr) {
|
|||||||
*/
|
*/
|
||||||
async function upsertUnits(db, units, logger) {
|
async function upsertUnits(db, units, logger) {
|
||||||
const collection = db.collection(config.COLLECTIONS.UNITS);
|
const collection = db.collection(config.COLLECTIONS.UNITS);
|
||||||
const now = getNow();
|
const now = new Date().toISOString();
|
||||||
|
|
||||||
const operations = units.map(unit => ({
|
const operations = units.map(unit => ({
|
||||||
updateOne: {
|
updateOne: {
|
||||||
@ -461,7 +461,7 @@ async function upsertUnits(db, units, logger) {
|
|||||||
*/
|
*/
|
||||||
async function insertPrices(db, units, date, logger) {
|
async function insertPrices(db, units, date, logger) {
|
||||||
const collection = db.collection(config.COLLECTIONS.PRICES);
|
const collection = db.collection(config.COLLECTIONS.PRICES);
|
||||||
const now = getNow();
|
const now = new Date().toISOString();
|
||||||
|
|
||||||
// Filter out units without prices
|
// Filter out units without prices
|
||||||
const unitsWithPrices = units.filter(u => u.price !== null);
|
const unitsWithPrices = units.filter(u => u.price !== null);
|
||||||
@ -581,7 +581,7 @@ async function updateDailySummary(db, summaryData, logger) {
|
|||||||
|
|
||||||
const summary = {
|
const summary = {
|
||||||
date,
|
date,
|
||||||
timestamp: getNow(),
|
timestamp: new Date().toISOString(),
|
||||||
new_units: newUnits,
|
new_units: newUnits,
|
||||||
rented_units: rentedUnits,
|
rented_units: rentedUnits,
|
||||||
stale_units: [], // Stale units list is not tracked per PRD
|
stale_units: [], // Stale units list is not tracked per PRD
|
||||||
@ -626,31 +626,11 @@ async function updateDailySummary(db, summaryData, logger) {
|
|||||||
// ============================================================
|
// ============================================================
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* Get today's date in YYYY-MM-DD format using the configured timezone.
|
* Get today's date in YYYY-MM-DD format (UTC).
|
||||||
* @returns {string} Today's date string
|
* @returns {string} Today's date string
|
||||||
*/
|
*/
|
||||||
function getToday() {
|
function getTodayUTC() {
|
||||||
return new Date().toLocaleDateString('en-CA', { timeZone: config.SCRAPER_TIMEZONE });
|
return new Date().toISOString().split('T')[0];
|
||||||
}
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Get current timestamp in the configured timezone as an ISO-like string.
|
|
||||||
* Returns format: YYYY-MM-DDTHH:mm:ss.mmm (no Z suffix, local time).
|
|
||||||
* @returns {string} Current timestamp in configured timezone
|
|
||||||
*/
|
|
||||||
function getNow() {
|
|
||||||
const now = new Date();
|
|
||||||
const formatter = new Intl.DateTimeFormat('en-CA', {
|
|
||||||
timeZone: config.SCRAPER_TIMEZONE,
|
|
||||||
year: 'numeric', month: '2-digit', day: '2-digit',
|
|
||||||
hour: '2-digit', minute: '2-digit', second: '2-digit',
|
|
||||||
hour12: false
|
|
||||||
});
|
|
||||||
const parts = Object.fromEntries(
|
|
||||||
formatter.formatToParts(now).map(({ type, value }) => [type, value])
|
|
||||||
);
|
|
||||||
const ms = String(now.getMilliseconds()).padStart(3, '0');
|
|
||||||
return `${parts.year}-${parts.month}-${parts.day}T${parts.hour}:${parts.minute}:${parts.second}.${ms}`;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
@ -738,7 +718,7 @@ async function recordScraperRun(db, runData, logger) {
|
|||||||
const collection = db.collection(config.COLLECTIONS.SCRAPER_RUNS);
|
const collection = db.collection(config.COLLECTIONS.SCRAPER_RUNS);
|
||||||
const result = await collection.insertOne({
|
const result = await collection.insertOne({
|
||||||
...runData,
|
...runData,
|
||||||
recordedAt: getNow()
|
recordedAt: new Date()
|
||||||
});
|
});
|
||||||
|
|
||||||
return result;
|
return result;
|
||||||
@ -750,37 +730,6 @@ async function recordScraperRun(db, runData, logger) {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// ============================================================
|
|
||||||
// Index Management
|
|
||||||
// ============================================================
|
|
||||||
|
|
||||||
/**
|
|
||||||
* Create indexes on the scraper_runs collection.
|
|
||||||
* Should be called once during application startup.
|
|
||||||
* Idempotent - safe to call multiple times.
|
|
||||||
*
|
|
||||||
* @param {Db} db - MongoDB database instance
|
|
||||||
* @param {Object} logger - Logger instance
|
|
||||||
* @returns {Promise<void>}
|
|
||||||
*/
|
|
||||||
async function createScraperIndexes(db, logger) {
|
|
||||||
const collection = db.collection(config.COLLECTIONS.SCRAPER_RUNS);
|
|
||||||
|
|
||||||
// Compound index for querying runs by status sorted by most recent
|
|
||||||
await collection.createIndex(
|
|
||||||
{ status: 1, startedAt: -1 },
|
|
||||||
{ name: 'status_startedAt' }
|
|
||||||
);
|
|
||||||
|
|
||||||
// Index for sorting all runs by start time (most recent first)
|
|
||||||
await collection.createIndex(
|
|
||||||
{ startedAt: -1 },
|
|
||||||
{ name: 'startedAt_desc' }
|
|
||||||
);
|
|
||||||
|
|
||||||
logger.info('Scraper indexes created successfully');
|
|
||||||
}
|
|
||||||
|
|
||||||
// ============================================================
|
// ============================================================
|
||||||
// Main Orchestration Function
|
// Main Orchestration Function
|
||||||
// ============================================================
|
// ============================================================
|
||||||
@ -810,7 +759,7 @@ async function runScrape(db, options = {}) {
|
|||||||
trigger,
|
trigger,
|
||||||
dryRun,
|
dryRun,
|
||||||
status: 'running',
|
status: 'running',
|
||||||
startedAt: getNow(),
|
startedAt: new Date().toISOString(),
|
||||||
completedAt: null,
|
completedAt: null,
|
||||||
duration: null,
|
duration: null,
|
||||||
unitsProcessed: 0,
|
unitsProcessed: 0,
|
||||||
@ -840,7 +789,7 @@ async function runScrape(db, options = {}) {
|
|||||||
result.unitsProcessed = units.length;
|
result.unitsProcessed = units.length;
|
||||||
|
|
||||||
// Step 4: Database operations
|
// Step 4: Database operations
|
||||||
const today = getToday();
|
const today = getTodayUTC();
|
||||||
|
|
||||||
// Get yesterday's unit codes for comparison
|
// Get yesterday's unit codes for comparison
|
||||||
const yesterdayUnits = await getYesterdayUnitCodes(db, today);
|
const yesterdayUnits = await getYesterdayUnitCodes(db, today);
|
||||||
@ -895,7 +844,7 @@ async function runScrape(db, options = {}) {
|
|||||||
result.errors.push(cleanError.message);
|
result.errors.push(cleanError.message);
|
||||||
|
|
||||||
} finally {
|
} finally {
|
||||||
result.completedAt = getNow();
|
result.completedAt = new Date().toISOString();
|
||||||
result.duration = Date.now() - startTime;
|
result.duration = Date.now() - startTime;
|
||||||
|
|
||||||
// Record run to history (always runs, even on failure)
|
// Record run to history (always runs, even on failure)
|
||||||
@ -922,10 +871,8 @@ module.exports = {
|
|||||||
markStaleUnits,
|
markStaleUnits,
|
||||||
updateDailySummary,
|
updateDailySummary,
|
||||||
recordScraperRun,
|
recordScraperRun,
|
||||||
createScraperIndexes,
|
|
||||||
// Export helpers for testing
|
// Export helpers for testing
|
||||||
getToday,
|
getTodayUTC,
|
||||||
getNow,
|
|
||||||
getYesterdayUnitCodes,
|
getYesterdayUnitCodes,
|
||||||
getYesterday,
|
getYesterday,
|
||||||
parseInteger,
|
parseInteger,
|
||||||
|
|||||||
Reference in New Issue
Block a user