Added send notification service, and queried unsent marketalerts, fixed some issues with crawler, and added proper logging
This commit is contained in:
@@ -87,14 +87,14 @@ module.exports = class OlxCrawler {
|
||||
|
||||
const data = {
|
||||
realEstateType: this.getCategoryId(realEstateType),
|
||||
email : email,
|
||||
email: email,
|
||||
olxId: olxId,
|
||||
// category: category,
|
||||
url,
|
||||
title,
|
||||
price: isNaN(parsedPrice) ? 0 : parsedPrice,
|
||||
size: parseFloat(size),
|
||||
gardenSize: isNaN(parseFloat(gardenSize)) ? 0 : parseFloat(gardenSize),
|
||||
gardenSize: isNaN(parseFloat(gardenSize)) ? 0 : parseFloat(gardenSize),
|
||||
address,
|
||||
region,
|
||||
municipality,
|
||||
@@ -151,22 +151,22 @@ module.exports = class OlxCrawler {
|
||||
}
|
||||
}
|
||||
|
||||
getCategoryId (category) {
|
||||
getCategoryId(category) {
|
||||
|
||||
switch(category) {
|
||||
switch (category) {
|
||||
case 'Stanovi':
|
||||
return 'stan';
|
||||
return 'stan';
|
||||
|
||||
case 'Vikendice':
|
||||
return 'vikendica'
|
||||
return 'vikendica'
|
||||
|
||||
case 'Kuće':
|
||||
return 'kuca';
|
||||
|
||||
default:
|
||||
return '';
|
||||
}
|
||||
}
|
||||
return '';
|
||||
}
|
||||
}
|
||||
|
||||
async indexPages(urls, start, end, maxResults = 1000) {
|
||||
//TODO fix paging
|
||||
@@ -186,16 +186,18 @@ module.exports = class OlxCrawler {
|
||||
}
|
||||
|
||||
async crawl() {
|
||||
console.log("OLX CRAWLER: start crawl");
|
||||
|
||||
const filteredResults = [];
|
||||
const realestateRequests = await allRERequest()
|
||||
const realestateRequests = await allRERequest();
|
||||
console.log("OLX CRAWLER: found " + realestateRequests.length + "subscribed RealEstateRequests");
|
||||
const urls = this.createRequestUrls(realestateRequests);
|
||||
let results = await this.indexPages(urls, this.fromPage, this.toPage, this.maxResults);
|
||||
|
||||
for (const result of results) {
|
||||
for (const finalResult of result) {
|
||||
if (finalResult.lat !== undefined && finalResult.lat !== null && finalResult.lat !== "") {
|
||||
const pointInsideBoundingBox = await findPointInsideBoundingBox([finalResult.lng, finalResult.lat]);
|
||||
const pointInsideBoundingBox = await findPointInsideBoundingBox([finalResult.lng, finalResult.lat], finalResult.email);
|
||||
|
||||
if (pointInsideBoundingBox[0].length !== 0) {
|
||||
filteredResults.push(finalResult);
|
||||
@@ -203,8 +205,7 @@ module.exports = class OlxCrawler {
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
console.log(filteredResults);
|
||||
console.log("OLX CRAWLER: number of olx crawler results, after geo location filtering: " + filteredResults.length);
|
||||
return filteredResults;
|
||||
}
|
||||
|
||||
|
||||
@@ -1,15 +1,62 @@
|
||||
const db = require('../../models/index');
|
||||
|
||||
// TODO Fetch only subscribed realestate requests
|
||||
/**
|
||||
* Find all subscribed RealEstateRequests
|
||||
*/
|
||||
const allRERequest = async () => {
|
||||
return await db.RealEstateRequest.findAll();
|
||||
return await db.RealEstateRequest.findAll({
|
||||
where: {
|
||||
subscribed: true
|
||||
}
|
||||
});
|
||||
}
|
||||
|
||||
const findPointInsideBoundingBox = async (latLng) => {
|
||||
return await db.sequelize.query("SELECT * FROM \"RealEstateRequests\" WHERE ST_Contains(\"RealEstateRequests\".bounding_box, ST_GEOMFROMTEXT(\'POINT (" + latLng[0] + " " + latLng[1]+ ")\'))");
|
||||
|
||||
/**
|
||||
* Find all unnotified marketalerts, and order them by email
|
||||
*
|
||||
* @param notified bolean
|
||||
*
|
||||
* @returns array of MarketAlerts
|
||||
*/
|
||||
const allMarketAlerts = async (notified) => {
|
||||
|
||||
let queryObject = {
|
||||
order: [
|
||||
['email', 'DESC'],
|
||||
]
|
||||
}
|
||||
|
||||
if (notified){
|
||||
queryObject.where = {
|
||||
notified: notified
|
||||
}
|
||||
}
|
||||
return await db.MarketAlert.findAll(queryObject);
|
||||
|
||||
// return await db.MarketAlerts.findAll({
|
||||
// where: {
|
||||
// notified: notified
|
||||
// },
|
||||
// order: [
|
||||
// ['email', 'DESC'],
|
||||
// ]
|
||||
// });
|
||||
}
|
||||
|
||||
/**
|
||||
* Find all unnotified marketalerts
|
||||
* @param latLng array
|
||||
* @param email strig
|
||||
*
|
||||
* @returns array of MarketAlerts
|
||||
*/
|
||||
const findPointInsideBoundingBox = async (latLng, email) => {
|
||||
return await db.sequelize.query(`SELECT * FROM "RealEstateRequests" WHERE email = '${email}' AND subscribed = true AND ST_Contains("RealEstateRequests".bounding_box, ST_GEOMFROMTEXT('POINT (${latLng[0]} ${latLng[1]})'))`);
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
allRERequest,
|
||||
allMarketAlerts,
|
||||
findPointInsideBoundingBox
|
||||
};
|
||||
|
||||
@@ -7,17 +7,6 @@ const currentRERequest = async (req) => {
|
||||
const request = await db.RealEstateRequest.findOne({ where: {uniqueId} });
|
||||
return request;
|
||||
};
|
||||
// TODO Fetch only subscribed realestate requests
|
||||
const allRERequest = async () => {
|
||||
return await db.RealEstateRequest.findAll();
|
||||
}
|
||||
|
||||
const findPointInsideBoundingBox = async (latLng) => {
|
||||
return await db.sequelize.query("SELECT * FROM \"RealEstateRequests\" WHERE ST_Contains(\"RealEstateRequests\".bounding_box, ST_GEOMFROMTEXT(\'POINT (" + latLng[0] + " " + latLng[1]+ ")\'))");
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
currentRERequest,
|
||||
allRERequest,
|
||||
findPointInsideBoundingBox
|
||||
};
|
||||
|
||||
Reference in New Issue
Block a user