Compare commits
128 Commits
more-detai
...
master
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
70bd952ee1 | ||
|
|
14039975c2 | ||
|
|
92e4f4ed5a | ||
|
|
88f9d10586 | ||
|
|
2b1cbcaa47 | ||
|
|
bf8d131025 | ||
|
|
698acb010a | ||
|
|
ade3eb307d | ||
|
|
8031f6f8a3 | ||
|
|
d7a680a3ac | ||
|
|
8018caab47 | ||
|
|
d871d9ad1f | ||
|
|
dfbefc20cd | ||
|
|
a481ecfe37 | ||
|
|
8df94da48c | ||
|
|
d4fcd1950d | ||
|
|
e8115a9215 | ||
|
|
160efdf6ab | ||
|
|
c9b8c2e1a5 | ||
|
|
855b93ca41 | ||
|
|
70779b24c0 | ||
|
|
ba873f9f4e | ||
|
|
e4775158fc | ||
|
|
26377c485c | ||
|
|
b30b0f45a6 | ||
|
|
9c1a029ff1 | ||
|
|
9b49759485 | ||
|
|
0c2f8d11ee | ||
|
|
b27d9d3499 | ||
|
|
dd3f30ef0e | ||
|
|
78c6056db4 | ||
|
|
ecf27f2ba1 | ||
|
|
1229b3fa6c | ||
|
|
542ff56123 | ||
|
|
0aa851015b | ||
|
|
c033b2e47c | ||
|
|
0895654db2 | ||
|
|
8925eb9f4e | ||
|
|
52201af3ba | ||
|
|
1505c07363 | ||
|
|
159fedbc2d | ||
|
|
65068932ad | ||
|
|
820227827e | ||
|
|
d35a113baa | ||
|
|
ba60f8749d | ||
|
|
f1d45fed26 | ||
|
|
ff923605ad | ||
|
|
692577fb8c | ||
|
|
df5e38092d | ||
|
|
feb2d04ed6 | ||
|
|
90e171d07d | ||
|
|
747f56941a | ||
|
|
2a13ab55ed | ||
|
|
441f905b29 | ||
|
|
39f9383ae2 | ||
|
|
edca7f91af | ||
|
|
44402a9cc4 | ||
|
|
b913daa1f7 | ||
|
|
a508f72d7c | ||
|
|
08ad9edfe1 | ||
|
|
ce857ddce9 | ||
|
|
148b2ea863 | ||
|
|
d436d4a37b | ||
|
|
6791a509d0 | ||
|
|
edc6e2bbf7 | ||
|
|
4f230020d7 | ||
|
|
f62a7200c7 | ||
|
|
cff7cc2e9c | ||
|
|
f56cd5b549 | ||
|
|
bc7ce9d708 | ||
|
|
df2a962d0f | ||
|
|
be4508ebea | ||
|
|
22bffc126d | ||
|
|
06f80296f3 | ||
|
|
81fa3f046d | ||
|
|
addd8c1344 | ||
|
|
5bdc8e149a | ||
|
|
fc7fe3c0b3 | ||
|
|
b3007123a5 | ||
|
|
f7d4a9cd07 | ||
|
|
ab6812889a | ||
|
|
b82134e280 | ||
|
|
be378883c8 | ||
|
|
8a87b9e253 | ||
|
|
43bc23b164 | ||
|
|
fc6351af46 | ||
|
|
6267b2cab4 | ||
|
|
97724a47a1 | ||
|
|
91a1c6a91e | ||
|
|
eb4ab2e341 | ||
|
|
2d0a00b967 | ||
|
|
74def9c059 | ||
|
|
d29b3eb1b3 | ||
|
|
41b59e8c7c | ||
|
|
b933fa96d4 | ||
|
|
824db4fbc3 | ||
|
|
0f91841c43 | ||
|
|
12b4a8f6ec | ||
|
|
17621ad310 | ||
|
|
c461525959 | ||
|
|
aec9c1e1d5 | ||
|
|
2d672f4660 | ||
|
|
bc15cf65a5 | ||
|
|
ce11d57ab2 | ||
|
|
4a6bcf262e | ||
|
|
712cde1632 | ||
|
|
1ba7cf8531 | ||
|
|
7a7aecb3ee | ||
|
|
78c4054cde | ||
|
|
94ffc2d6d2 | ||
|
|
b11f18696f | ||
|
|
fa46f75dd3 | ||
|
|
470f53d29b | ||
|
|
40509d2836 | ||
|
|
b2c102bc1a | ||
|
|
98263364c7 | ||
|
|
5b3491fdba | ||
|
|
c6f0e039a5 | ||
|
|
8d3f001678 | ||
|
|
42eddb3aa5 | ||
|
|
0a181f742f | ||
|
|
d117383802 | ||
|
|
870b71a3c7 | ||
|
|
e6725355a0 | ||
|
|
4fd4018bf6 | ||
|
|
b9122f8f00 | ||
|
|
d5d3a1f306 | ||
|
|
42ff1f762f |
1
.prettierignore
Normal file
1
.prettierignore
Normal file
@@ -0,0 +1 @@
|
||||
*.ejs
|
||||
@@ -7,7 +7,42 @@ const PRICE_SLIDER_OPTIONS_SALE = {
|
||||
step: 1000,
|
||||
connect: true
|
||||
};
|
||||
|
||||
const FLAT_PRICE_SLIDER_OPTIONS_SALE = {
|
||||
start: [50000, 150000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [800000]
|
||||
},
|
||||
step: 5000,
|
||||
connect: true
|
||||
};
|
||||
const HOUSE_PRICE_SLIDER_OPTIONS_SALE = {
|
||||
start: [50000, 150000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [1500000]
|
||||
},
|
||||
step: 10000,
|
||||
connect: true
|
||||
};
|
||||
const OFFICE_PRICE_SLIDER_OPTIONS_SALE = {
|
||||
start: [15000, 50000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [2000000]
|
||||
},
|
||||
step: 2000,
|
||||
connect: true
|
||||
};
|
||||
const LAND_PRICE_SLIDER_OPTIONS_SALE = {
|
||||
start: [40000, 80000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [2000000]
|
||||
},
|
||||
step: 10000,
|
||||
connect: true
|
||||
};
|
||||
const PRICE_SLIDER_OPTIONS_RENT = {
|
||||
start: [300, 500],
|
||||
range: {
|
||||
@@ -17,18 +52,62 @@ const PRICE_SLIDER_OPTIONS_RENT = {
|
||||
step: 50,
|
||||
connect: true
|
||||
};
|
||||
|
||||
const FLAT_PRICE_SLIDER_OPTIONS_RENT = {
|
||||
start: [300, 600],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [4000]
|
||||
},
|
||||
step: 100,
|
||||
connect: true
|
||||
};
|
||||
const HOUSE_PRICE_SLIDER_OPTIONS_RENT = {
|
||||
start: [500, 1000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [10000]
|
||||
},
|
||||
step: 100,
|
||||
connect: true
|
||||
};
|
||||
const OFFICE_PRICE_SLIDER_OPTIONS_RENT = {
|
||||
start: [200, 1000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [20000]
|
||||
},
|
||||
step: 100,
|
||||
connect: true
|
||||
};
|
||||
const LAND_PRICE_SLIDER_OPTIONS_RENT = {
|
||||
start: [500, 1000],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [20000]
|
||||
},
|
||||
step: 100,
|
||||
connect: true
|
||||
};
|
||||
//This will be used for Flats, Apartments, Houses
|
||||
const HOME_SIZE_SLIDER_OPTIONS = {
|
||||
start: [30, 75],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [400]
|
||||
max: [500]
|
||||
},
|
||||
step: 5,
|
||||
connect: true
|
||||
};
|
||||
|
||||
const OFFICE_SIZE_SLIDER_OPTIONS = {
|
||||
start: [30, 150],
|
||||
range: {
|
||||
min: [0],
|
||||
max: [1200]
|
||||
},
|
||||
step: 10,
|
||||
connect: true
|
||||
};
|
||||
const GARDEN_SIZE_SLIDER_OPTIONS = {
|
||||
start: [100, 1000],
|
||||
range: {
|
||||
@@ -111,8 +190,8 @@ const AD_CATEGORY = {
|
||||
hasNumberOfRoom: true,
|
||||
hasNumberOfFloors: false,
|
||||
hasFloorProp: true,
|
||||
priceSliderOptionsSale: PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: PRICE_SLIDER_OPTIONS_RENT,
|
||||
priceSliderOptionsSale: FLAT_PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: FLAT_PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: HOME_SIZE_SLIDER_OPTIONS
|
||||
},
|
||||
HOUSE: {
|
||||
@@ -126,8 +205,8 @@ const AD_CATEGORY = {
|
||||
hasNumberOfRoom: true,
|
||||
hasNumberOfFloors: true,
|
||||
hasFloorProp: false,
|
||||
priceSliderOptionsSale: PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: PRICE_SLIDER_OPTIONS_RENT,
|
||||
priceSliderOptionsSale: HOUSE_PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: HOUSE_PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: HOME_SIZE_SLIDER_OPTIONS,
|
||||
gardenSizeSliderOptions: GARDEN_SIZE_SLIDER_OPTIONS
|
||||
},
|
||||
@@ -142,9 +221,9 @@ const AD_CATEGORY = {
|
||||
hasNumberOfRoom: true,
|
||||
hasNumberOfFloors: false,
|
||||
hasFloorProp: true,
|
||||
priceSliderOptionsSale: PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: HOME_SIZE_SLIDER_OPTIONS
|
||||
priceSliderOptionsSale: OFFICE_PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: OFFICE_PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: OFFICE_SIZE_SLIDER_OPTIONS
|
||||
},
|
||||
LAND: {
|
||||
id: "LAND",
|
||||
@@ -157,8 +236,8 @@ const AD_CATEGORY = {
|
||||
hasNumberOfRoom: false,
|
||||
hasNumberOfFloors: false,
|
||||
hasFloorProp: false,
|
||||
priceSliderOptionsSale: PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: PRICE_SLIDER_OPTIONS_RENT,
|
||||
priceSliderOptionsSale: LAND_PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: LAND_PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: LAND_SIZE_SLIDER_OPTIONS
|
||||
},
|
||||
APARTMENT: {
|
||||
@@ -172,8 +251,8 @@ const AD_CATEGORY = {
|
||||
hasNumberOfRoom: true,
|
||||
hasNumberOfFloors: false,
|
||||
hasFloorProp: true,
|
||||
priceSliderOptionsSale: PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: PRICE_SLIDER_OPTIONS_RENT,
|
||||
priceSliderOptionsSale: FLAT_PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: FLAT_PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: HOME_SIZE_SLIDER_OPTIONS
|
||||
},
|
||||
GARAGE: {
|
||||
@@ -202,8 +281,8 @@ const AD_CATEGORY = {
|
||||
hasNumberOfRoom: true,
|
||||
hasNumberOfFloors: true,
|
||||
hasFloorProp: false,
|
||||
priceSliderOptionsSale: PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: PRICE_SLIDER_OPTIONS_RENT,
|
||||
priceSliderOptionsSale: HOUSE_PRICE_SLIDER_OPTIONS_SALE,
|
||||
priceSliderOptionsRent: HOUSE_PRICE_SLIDER_OPTIONS_RENT,
|
||||
sizeSliderOptions: HOME_SIZE_SLIDER_OPTIONS,
|
||||
gardenSizeSliderOptions: GARDEN_SIZE_SLIDER_OPTIONS
|
||||
}
|
||||
@@ -224,7 +303,8 @@ const AD_AGENCY = {
|
||||
OLX: "OLX",
|
||||
RENTAL: "RENTAL",
|
||||
PROSTOR: "PROSTOR",
|
||||
AKTIDO: "AKTIDO"
|
||||
AKTIDO: "AKTIDO",
|
||||
SALJIC: "SALJIC"
|
||||
};
|
||||
|
||||
const CRAWLER_AD_TYPE = {
|
||||
|
||||
@@ -9,11 +9,15 @@ const APP_URL =
|
||||
? process.env.APP_URL || "http://market-alarm"
|
||||
: process.env.APP_URL || `${APP_BASE_URL}:${APP_PORT}`;
|
||||
|
||||
const STAGING = process.env.ENVIRONMENT !== "production";
|
||||
|
||||
const DEFAULT_TIMEZONE = "Europe/Sarajevo";
|
||||
|
||||
const CRAWLER_INTERVAL = parseInt(process.env.CRAWLER_INTERVAL) || 60;
|
||||
const STOP_CRAWLER = !!parseInt(process.env.STOP_CRAWLER);
|
||||
|
||||
const CHECK_UP_DAYS = parseInt(process.env.CHECK_UP_DAYS) || 10;
|
||||
|
||||
const AWS_EMAIL_CONFIG = {
|
||||
REGION: process.env.AWS_REGION || "",
|
||||
CREDENTIALS: {
|
||||
@@ -37,6 +41,15 @@ const PROSTOR_LOGIN = {
|
||||
PASSWORD: process.env.PROSTOR_LOGIN_PASS
|
||||
};
|
||||
|
||||
const USER_AGENT =
|
||||
process.env.USER_AGENT ||
|
||||
"Mozilla/5.0 (Windows NT 6.1; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/65.0.3325.181 Safari/537.36";
|
||||
|
||||
const USE_SCRAPER_API = process.env.USE_SCRAPER_API === undefined ? 1 : parseInt(process.env.USE_SCRAPER_API);
|
||||
const SCRAPER_API_KEY = process.env.SCRAPER_API_KEY || "";
|
||||
const SCRAPER_API_BASE_URL = process.env.SCRAPER_API_BASE_URL || "";
|
||||
const NODE_FETCH_TIMEOUT_MS = parseInt(process.env.NODE_FETCH_TIMEOUT_MS) || 60000
|
||||
|
||||
module.exports = {
|
||||
APP_PORT,
|
||||
APP_URL,
|
||||
@@ -48,5 +61,12 @@ module.exports = {
|
||||
MAX_REAL_ESTATES_IN_FIRST_EMAIL,
|
||||
PRINT_CRAWLER_DEBUG,
|
||||
API_MAP_KEY,
|
||||
PROSTOR_LOGIN
|
||||
STAGING,
|
||||
CHECK_UP_DAYS,
|
||||
PROSTOR_LOGIN,
|
||||
USER_AGENT,
|
||||
USE_SCRAPER_API,
|
||||
SCRAPER_API_KEY,
|
||||
SCRAPER_API_BASE_URL,
|
||||
NODE_FETCH_TIMEOUT_MS
|
||||
};
|
||||
|
||||
@@ -35,7 +35,8 @@ const getFilters = async (req, res) => {
|
||||
balcony,
|
||||
elevator,
|
||||
newBuilding,
|
||||
accessRoadType
|
||||
accessRoadType,
|
||||
includeWithoutPrice
|
||||
} = searchRequest;
|
||||
const category = AD_CATEGORY[realEstateType] || AD_CATEGORY.FLAT;
|
||||
|
||||
@@ -115,7 +116,8 @@ const getFilters = async (req, res) => {
|
||||
advancedSegmentSelectFilterValues,
|
||||
advancedRangeFilterObjects,
|
||||
advancedRangeFilterValues,
|
||||
includeIncompleteAds
|
||||
includeIncompleteAds,
|
||||
includeWithoutPrice
|
||||
});
|
||||
};
|
||||
|
||||
@@ -191,6 +193,7 @@ const postFilters = async (req, res) => {
|
||||
});
|
||||
|
||||
const includeIncompleteAds = req.body.includeIncompleteAds === "on";
|
||||
const includeWithoutPrice = req.body.includeWithoutPrice === "on";
|
||||
|
||||
const balcony = req.body.balcony === "on";
|
||||
const elevator = req.body.elevator === "on";
|
||||
@@ -217,6 +220,7 @@ const postFilters = async (req, res) => {
|
||||
searchRequest.newBuilding = newBuilding;
|
||||
|
||||
searchRequest.includeIncompleteAds = includeIncompleteAds;
|
||||
searchRequest.includeWithoutPrice = includeWithoutPrice;
|
||||
|
||||
searchRequest.accessRoadType = accessRoadType;
|
||||
|
||||
|
||||
@@ -9,12 +9,15 @@ const OlxCrawler = require("./specificCrawlers/olx");
|
||||
const RentalCrawler = require("./specificCrawlers/rental");
|
||||
const ProstorCrawler = require("./specificCrawlers/prostor");
|
||||
const AktidoCrawler = require("./specificCrawlers/aktido");
|
||||
const SaljicCrawler = require("./specificCrawlers/saljic");
|
||||
const { logDebug } = require("../helpers/log");
|
||||
|
||||
const {
|
||||
OLX_CONFIG,
|
||||
RENTAL_CONFIG,
|
||||
PROSTOR_CONFIG,
|
||||
AKTIDO_CONFIG
|
||||
AKTIDO_CONFIG,
|
||||
SALJIC_CONFIG
|
||||
} = require("./crawlerConfig");
|
||||
const PostgresSaver = require("./savers/postgres");
|
||||
|
||||
@@ -57,6 +60,15 @@ async function crawlAll() {
|
||||
AKTIDO_CONFIG.AKTIDO_MAX_RESULTS_PER_PAGE,
|
||||
AKTIDO_CONFIG.AKTIDO_IGNORED_USERNAMES,
|
||||
AKTIDO_CONFIG.AKTIDO_DELAY_BETWEEN_PAGES
|
||||
),
|
||||
new SaljicCrawler(
|
||||
[postgresSaver],
|
||||
SALJIC_CONFIG.SALJIC_CRAWLER_AD_TYPE,
|
||||
SALJIC_CONFIG.SALJIC_CRAWLER_AD_CATEGORIES,
|
||||
SALJIC_CONFIG.SALJIC_MAX_PAGES,
|
||||
SALJIC_CONFIG.SALJIC_MAX_RESULTS_PER_PAGE,
|
||||
SALJIC_CONFIG.SALJIC_IGNORED_USERNAMES,
|
||||
SALJIC_CONFIG.SALJIC_DELAY_BETWEEN_PAGES
|
||||
)
|
||||
];
|
||||
|
||||
@@ -64,7 +76,9 @@ async function crawlAll() {
|
||||
|
||||
for (const crawler of crawlers) {
|
||||
try {
|
||||
logDebug('Starting crawler: ', crawler);
|
||||
const newRealEstatesFromSingleCrawler = await crawler.crawl();
|
||||
logDebug('Crawler done: ', crawler);
|
||||
if (Array.isArray(newRealEstatesFromSingleCrawler)) {
|
||||
newRealEstates.push(...newRealEstatesFromSingleCrawler);
|
||||
}
|
||||
|
||||
@@ -5,10 +5,12 @@ const OLX_CONFIG = require("./specificConfigs/olx");
|
||||
const RENTAL_CONFIG = require("./specificConfigs/rental");
|
||||
const PROSTOR_CONFIG = require("./specificConfigs/prostor");
|
||||
const AKTIDO_CONFIG = require("./specificConfigs/aktido");
|
||||
const SALJIC_CONFIG = require("./specificConfigs/saljic");
|
||||
|
||||
module.exports = {
|
||||
OLX_CONFIG,
|
||||
RENTAL_CONFIG,
|
||||
PROSTOR_CONFIG,
|
||||
AKTIDO_CONFIG
|
||||
AKTIDO_CONFIG,
|
||||
SALJIC_CONFIG
|
||||
};
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
const moment = require("moment");
|
||||
|
||||
const { bulkUpsertRealEstates } = require("../../helpers/db/realEstate");
|
||||
const { bulkUpsertPriceHistory } = require("../../helpers/db/priceHistory");
|
||||
|
||||
class PostgresSaver {
|
||||
connect() {
|
||||
@@ -11,6 +12,21 @@ class PostgresSaver {
|
||||
|
||||
async save(results) {
|
||||
const savedRecords = await bulkUpsertRealEstates(results);
|
||||
//Extruding data for price history table
|
||||
const resultPrices = savedRecords.map(realEstate => {
|
||||
//Null values canot be recognized by ignore duplicates in sequalize
|
||||
//Value price = 0 indicates 'cijena na upit'
|
||||
const priceTmp =
|
||||
realEstate.dataValues.price === null ? 0 : realEstate.dataValues.price;
|
||||
|
||||
return {
|
||||
realEstateId: realEstate.dataValues.id,
|
||||
price: priceTmp,
|
||||
createdAt: realEstate.dataValues.createdAt,
|
||||
updatedAt: realEstate.dataValues.updatedAt
|
||||
};
|
||||
});
|
||||
const savedPrices = await bulkUpsertPriceHistory(resultPrices);
|
||||
|
||||
if (Array.isArray(savedRecords)) {
|
||||
const newRealEstates = [];
|
||||
|
||||
34
app/crawler/specificConfigs/saljic.js
Normal file
34
app/crawler/specificConfigs/saljic.js
Normal file
@@ -0,0 +1,34 @@
|
||||
"use strict";
|
||||
const { CRAWLER_AD_TYPE, AD_CATEGORY } = require("../../common/enums");
|
||||
|
||||
const saljicCrawlerAdType =
|
||||
process.env.SALJIC_CRAWLER_AD_TYPE !== undefined
|
||||
? CRAWLER_AD_TYPE[process.env.SALJIC_CRAWLER_AD_TYPE]
|
||||
: null;
|
||||
|
||||
const saljicParsedCrawlerAdCategories =
|
||||
process.env.SALJIC_CRAWLER_AD_CATEGORIES !== undefined
|
||||
? process.env.SALJIC_CRAWLER_AD_CATEGORIES.split(",").map(category =>
|
||||
category.trim()
|
||||
)
|
||||
: ["FLAT", "HOUSE"];
|
||||
|
||||
const saljicIgnoredUsernames = [];
|
||||
|
||||
const transformedSaljicCrawlerAdCategories = saljicParsedCrawlerAdCategories
|
||||
.map(categoryName =>
|
||||
AD_CATEGORY[categoryName] ? AD_CATEGORY[categoryName].id : undefined
|
||||
)
|
||||
.filter(category => !!category);
|
||||
|
||||
module.exports = {
|
||||
SALJIC_MAX_PAGES: parseInt(process.env.SALJIC_MAX_PAGES) || 100,
|
||||
SALJIC_MAX_RESULTS_PER_PAGE:
|
||||
parseInt(process.env.SALJIC_MAX_RESULTS_PER_PAGE) || 5000,
|
||||
SALJIC_CRAWLER_AD_TYPE: saljicCrawlerAdType || CRAWLER_AD_TYPE.NONE,
|
||||
SALJIC_CRAWLER_AD_CATEGORIES: transformedSaljicCrawlerAdCategories,
|
||||
SALJIC_IGNORED_USERNAMES: saljicIgnoredUsernames || [],
|
||||
SALJIC_DELAY_BETWEEN_PAGES:
|
||||
parseInt(process.env.SALJIC_DELAY_BETWEEN_PAGES) || 1000,
|
||||
SALJIC_FORCE_CRAWL: !!parseInt(process.env.SALJIC_FORCE_CRAWL)
|
||||
};
|
||||
@@ -1,6 +1,6 @@
|
||||
"use strict";
|
||||
|
||||
const fetch = require("node-fetch");
|
||||
const fetch = require("../../helpers/fetchWrapper");
|
||||
const cheerio = require("cheerio");
|
||||
const Promise = require("bluebird");
|
||||
const moment = require("moment-timezone");
|
||||
@@ -159,7 +159,7 @@ class AktidoCrawler {
|
||||
}
|
||||
|
||||
try {
|
||||
const res = await fetch(url);
|
||||
const res = await fetch(url, {}, false);
|
||||
const body = await res.text();
|
||||
const $ = cheerio.load(body);
|
||||
let hrefs = [];
|
||||
@@ -202,6 +202,10 @@ class AktidoCrawler {
|
||||
const body = await adPageSource.text();
|
||||
const $ = cheerio.load(body);
|
||||
|
||||
if (body.indexOf('<html') === -1) {
|
||||
throw { message: 'Failed to fetch page !' }
|
||||
}
|
||||
|
||||
const mapElementParent = $(".box-map").parent();
|
||||
const scriptElement = $("script", mapElementParent);
|
||||
if (
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
"use strict";
|
||||
|
||||
const fetch = require("node-fetch");
|
||||
const fetch = require("../../helpers/fetchWrapper");
|
||||
const { logDebug } = require("../../helpers/log");
|
||||
const cheerio = require("cheerio");
|
||||
const Promise = require("bluebird");
|
||||
const moment = require("moment-timezone");
|
||||
@@ -44,6 +45,16 @@ const OLX_ENUMS = {
|
||||
|
||||
const { OLX_FORCE_CRAWL } = require("../specificConfigs/olx");
|
||||
|
||||
const chunk = (array, size = 10) => {
|
||||
let i, j ,temparray;
|
||||
const result = []
|
||||
for (i=0,j=array.length; i<j; i+=size) {
|
||||
temparray = array.slice(i,i+size);
|
||||
result.push(temparray);
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
class OlxCrawler {
|
||||
constructor(
|
||||
savers = [],
|
||||
@@ -52,7 +63,7 @@ class OlxCrawler {
|
||||
maxPages = 1000,
|
||||
maxResultsPerPage = 100,
|
||||
ignoredUsernames = [],
|
||||
delayBetweenPages = 1000
|
||||
delayBetweenPages = 500
|
||||
) {
|
||||
this.savers = savers;
|
||||
this.baseUrl = "https://www.olx.ba/pretraga?sort_order=desc&sort_po=datum";
|
||||
@@ -65,6 +76,7 @@ class OlxCrawler {
|
||||
}
|
||||
|
||||
async crawl() {
|
||||
logDebug("Starting OLX crawl");
|
||||
const crawlAdCategories = this.crawlerAdCategories;
|
||||
|
||||
const newRealEstates = [];
|
||||
@@ -88,14 +100,32 @@ class OlxCrawler {
|
||||
const entries = singlePageResults.entries();
|
||||
|
||||
for (const [index, { value: singlePageResult }] of entries) {
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log("================================");
|
||||
console.log("Category Indexer index : ", index);
|
||||
}
|
||||
|
||||
if (singlePageResult) {
|
||||
console.log("\tTotal entries : ", singlePageResult.length)
|
||||
const saveResults = await this.saveCrawledResults(singlePageResult);
|
||||
const { newRecords, existingRecords } = saveResults;
|
||||
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log("--------------------------");
|
||||
console.log("\tNew record URLs [", newRecords.length, "] :");
|
||||
|
||||
for(const newRecord of newRecords) {
|
||||
console.log("\t\t",newRecord.url);
|
||||
}
|
||||
|
||||
console.log("\t-------------------------");
|
||||
console.log("\tExisting record URLs [", existingRecords.length, "] :");
|
||||
}
|
||||
|
||||
newRealEstates.push(...newRecords);
|
||||
|
||||
for (const existingRecord of existingRecords) {
|
||||
const { publishedDate, renewedDate } = existingRecord;
|
||||
const { publishedDate, renewedDate, url } = existingRecord;
|
||||
|
||||
const publishedDateMoment = moment.utc(publishedDate);
|
||||
const renewedDateMoment = moment.utc(renewedDate);
|
||||
@@ -105,13 +135,25 @@ class OlxCrawler {
|
||||
"minute"
|
||||
);
|
||||
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log("\t\t", url);
|
||||
console.log("\t\t\tPublished date : ", publishedDate);
|
||||
console.log("\t\t\tRenewed date : ", renewedDate);
|
||||
console.log("\t\t\tIs same (up to minute) : ", stopCrawlingThisCategory);
|
||||
}
|
||||
|
||||
if (stopCrawlingThisCategory && !OLX_FORCE_CRAWL) {
|
||||
generatorsToRemove[index] = true;
|
||||
// console.log("\tGenerator ", index + 1, "has no more new ads");
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log("\t\t\tStopping this category indexer");
|
||||
}
|
||||
break;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log("\tNo more entries in this category, stopping!");
|
||||
}
|
||||
//Generator returned undefined, remove this generator from array
|
||||
generatorsToRemove[index] = true;
|
||||
// console.log("Generator ", index + 1, "has no more pages");
|
||||
@@ -136,31 +178,36 @@ class OlxCrawler {
|
||||
}
|
||||
|
||||
async *categoryIndexer(adCategory) {
|
||||
let pageToIndex = 1;
|
||||
try {
|
||||
let pageToIndex = 1;
|
||||
|
||||
const urlAdTypePart = OLX_ENUMS.OLX_AD_TYPE[this.crawlerAdTypes];
|
||||
const urlCategoryPart = OLX_ENUMS.OLX_AD_CATEGORY[adCategory];
|
||||
if (urlAdTypePart !== undefined && urlCategoryPart !== undefined) {
|
||||
while (true) {
|
||||
const urlPageToCrawl = `${this.baseUrl}${urlAdTypePart}${urlCategoryPart}&stranica=${pageToIndex}`;
|
||||
const singlePageResults = await this.indexSinglePage(
|
||||
urlPageToCrawl,
|
||||
this.maxResultsPerPage
|
||||
);
|
||||
const urlAdTypePart = OLX_ENUMS.OLX_AD_TYPE[this.crawlerAdTypes];
|
||||
const urlCategoryPart = OLX_ENUMS.OLX_AD_CATEGORY[adCategory];
|
||||
if (urlAdTypePart !== undefined && urlCategoryPart !== undefined) {
|
||||
while (true) {
|
||||
const urlPageToCrawl = `${this.baseUrl}${urlAdTypePart}${urlCategoryPart}&stranica=${pageToIndex}`;
|
||||
const singlePageResults = await this.indexSinglePage(
|
||||
urlPageToCrawl,
|
||||
this.maxResultsPerPage
|
||||
);
|
||||
|
||||
if (Array.isArray(singlePageResults) && singlePageResults.length > 0) {
|
||||
yield singlePageResults;
|
||||
} else {
|
||||
return undefined;
|
||||
}
|
||||
|
||||
++pageToIndex;
|
||||
if (pageToIndex === this.maxPages) {
|
||||
return undefined;
|
||||
await this.sleep(this.delayBetweenPages);
|
||||
if (Array.isArray(singlePageResults) && singlePageResults.length > 0) {
|
||||
yield singlePageResults;
|
||||
} else {
|
||||
return undefined;
|
||||
}
|
||||
|
||||
++pageToIndex;
|
||||
if (pageToIndex === this.maxPages) {
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
return undefined;
|
||||
}
|
||||
} else {
|
||||
return undefined;
|
||||
} catch (e) {
|
||||
console.log('Error inside generator: ', e);
|
||||
}
|
||||
}
|
||||
|
||||
@@ -170,8 +217,10 @@ class OlxCrawler {
|
||||
}
|
||||
|
||||
try {
|
||||
const res = await fetch(url);
|
||||
const res = await fetch(url, {}, false);
|
||||
logDebug("Got category results for: ", url);
|
||||
const body = await res.text();
|
||||
logDebug("Got category results text for: ", url);
|
||||
const $ = cheerio.load(body);
|
||||
let hrefs = [];
|
||||
|
||||
@@ -192,26 +241,44 @@ class OlxCrawler {
|
||||
|
||||
const asyncScraping = [];
|
||||
for (let i = 0; i < actualNoOfResults; i++) {
|
||||
asyncScraping.push(this.scrapeAd(hrefs[i]));
|
||||
asyncScraping.push(hrefs[i]);
|
||||
}
|
||||
|
||||
const scrapedData = await Promise.all(asyncScraping);
|
||||
const filteredScrapedData = scrapedData.filter(adData => !!adData);
|
||||
const allChunks = chunk(asyncScraping, 2);
|
||||
const dataResults = []
|
||||
for (let i = 0; i < allChunks.length; i++) {
|
||||
const singleChunk = allChunks[i];
|
||||
const promises = singleChunk.map(c => this.scrapeAd(c))
|
||||
const chunkResults = await Promise.all(promises);
|
||||
await this.sleep(this.delayBetweenPages);
|
||||
dataResults.push(...chunkResults);
|
||||
logDebug("Chunk results len:", chunkResults.length);
|
||||
}
|
||||
|
||||
|
||||
const filteredScrapedData = dataResults.filter(adData => !!adData);
|
||||
logDebug("Filtered scraped data length: ", filteredScrapedData.length);
|
||||
|
||||
return filteredScrapedData;
|
||||
} catch (e) {
|
||||
console.error("Exception caught:" + e);
|
||||
console.error("Exception caught, index single page: " + e);
|
||||
return [];
|
||||
}
|
||||
}
|
||||
|
||||
async scrapeAd(url) {
|
||||
// console.log("Scraping : ", url);
|
||||
logDebug("Scraping : ", url);
|
||||
try {
|
||||
const adPageSource = await fetch(url);
|
||||
const body = await adPageSource.text();
|
||||
const $ = cheerio.load(body);
|
||||
let status = AD_STATUS.STATUS_NORMAL;
|
||||
|
||||
if (body.indexOf('<html') === -1) {
|
||||
console.error("This is the body: ", body);
|
||||
throw { message: 'Failed to fetch page !' }
|
||||
}
|
||||
|
||||
const propertySelectors = {
|
||||
username:
|
||||
"#lg > div.desno2.profil > div:nth-child(2) > div.vrsta1.vrsta_desno > a > div.username > span",
|
||||
@@ -238,37 +305,25 @@ class OlxCrawler {
|
||||
|
||||
//====== PRICE DETECTION AND EXTRACTION =====
|
||||
let price = null;
|
||||
const normalPriceValue = $("#pc > p:nth-child(2)").text();
|
||||
const urgentPriceValue = $(
|
||||
"#artikal_glavni_div > div.artikal_lijevo > div:nth-child(5) > p"
|
||||
)
|
||||
.text()
|
||||
.trim();
|
||||
|
||||
if (normalPriceValue && normalPriceValue.length > 0) {
|
||||
price = normalPriceValue;
|
||||
if (
|
||||
$("#pc > p.n")
|
||||
.text()
|
||||
.indexOf("Hitna") !== -1
|
||||
) {
|
||||
status = AD_STATUS.STATUS_URGENT;
|
||||
} else {
|
||||
status = AD_STATUS.STATUS_NORMAL;
|
||||
}
|
||||
} else if (urgentPriceValue && urgentPriceValue.length > 0) {
|
||||
const priceValues = urgentPriceValue.split("KM");
|
||||
//priceValues will contain values like ["100000", "90000", ...], second element is urgent price
|
||||
if (priceValues.length > 1) {
|
||||
price = priceValues[1].trim();
|
||||
status = AD_STATUS.STATUS_DISCOUNTED;
|
||||
} else {
|
||||
throw { message: "Can't find urgent price" };
|
||||
}
|
||||
} else {
|
||||
throw {
|
||||
message: "Can't find price (it is not normal nor urgent price ?)"
|
||||
};
|
||||
const priceHeader = $("#pc > p.n").text().trim();
|
||||
const priceValue = $("#pc > p:nth-child(2)").text().trim();
|
||||
price = priceValue;
|
||||
|
||||
if (priceHeader.indexOf('Hitn') !== -1) {
|
||||
// Urgent price
|
||||
status = AD_STATUS.STATUS_URGENT;
|
||||
}
|
||||
|
||||
const discountPriceTag = $("#artikal_glavni_div > div.artikal_lijevo > p:nth-child(4)").text().trim();
|
||||
if (discountPriceTag.indexOf('Akcij') !== -1) {
|
||||
status = AD_STATUS.STATUS_DISCOUNTED;
|
||||
const discountPriceValues = $("#artikal_glavni_div > div.artikal_lijevo > div:nth-child(5) > p").text().trim();
|
||||
// discountPriceValues contain string like "10.000 KM 7.500 KM"
|
||||
// First price is regular, second is currently active (discounted) price
|
||||
const bothPrices = discountPriceValues.split('KM');
|
||||
// Now, currently active price is second element of bothPrices array
|
||||
price = bothPrices[1] ? bothPrices[1].trim() : null;
|
||||
}
|
||||
|
||||
//====== OTHER AD INFORMATION ===============
|
||||
@@ -278,7 +333,7 @@ class OlxCrawler {
|
||||
|
||||
let otherInformationDivId;
|
||||
//We need to locate DIV ID where other information are stored
|
||||
for (let possibleId = 10; possibleId <= 20; possibleId++) {
|
||||
for (let possibleId = 1; possibleId <= 30; possibleId++) {
|
||||
const adTypeFieldTitle = $(
|
||||
`#artikal_glavni_div > div.artikal_lijevo > div:nth-child(${possibleId}) > div:nth-child(2) > div.df1`
|
||||
)
|
||||
@@ -650,10 +705,12 @@ class OlxCrawler {
|
||||
distanceToRiver,
|
||||
numberOfViewsAgency
|
||||
};
|
||||
//
|
||||
//console.log("Scraped data:", data);
|
||||
|
||||
return data;
|
||||
} catch (e) {
|
||||
console.error("Exception caught: " + e.message, "\r\nURL:", url);
|
||||
console.error("Exception caught scrapeAd : " + e.message, "\r\nURL:", url);
|
||||
}
|
||||
return null;
|
||||
}
|
||||
@@ -768,6 +825,9 @@ class OlxCrawler {
|
||||
if (!priceText) {
|
||||
return NaN;
|
||||
}
|
||||
if (priceText === "Po dogovoru") {
|
||||
return null;
|
||||
}
|
||||
const formattedPriceText = priceText.replace(".", "").replace(",", ".");
|
||||
return parseFloat(formattedPriceText);
|
||||
}
|
||||
|
||||
@@ -1,9 +1,10 @@
|
||||
"use strict";
|
||||
|
||||
const fetch = require("node-fetch");
|
||||
const fetch = require("../../helpers/fetchWrapper");
|
||||
const cheerio = require("cheerio");
|
||||
const moment = require("moment-timezone");
|
||||
const FormData = require("form-data");
|
||||
const nodeFetch = require("node-fetch");
|
||||
|
||||
const {
|
||||
AD_TYPE,
|
||||
@@ -62,13 +63,19 @@ class ProstorCrawler {
|
||||
|
||||
async crawl() {
|
||||
const crawlAdCategories = this.crawlerAdCategories;
|
||||
const crawlAdTypes = this.crawlerAdTypes;
|
||||
if (!crawlAdCategories || !crawlAdTypes) {
|
||||
return []
|
||||
}
|
||||
|
||||
const newRealEstates = [];
|
||||
//We need session cookie to use login privileges
|
||||
const prostorCookie = await this.getCookies();
|
||||
//New tag to check if crawler loged in
|
||||
//New tag to check if crawler logged in
|
||||
const login = await this.loginForScraping(PROSTOR_LOGIN, prostorCookie);
|
||||
const newRealEstates = [];
|
||||
|
||||
//Crawl only if login was successful
|
||||
if (crawlAdCategories && login) {
|
||||
if (login) {
|
||||
const indexGenerators = [];
|
||||
for (const adCategory of crawlAdCategories) {
|
||||
indexGenerators.push(this.categoryIndexer(adCategory, prostorCookie));
|
||||
@@ -134,6 +141,11 @@ class ProstorCrawler {
|
||||
prostorCookie
|
||||
);
|
||||
|
||||
if (!Array.isArray(listOfAllRealEstates)){
|
||||
console.log('[PROSTOR] Could not find real estate JSON data, check selector !');
|
||||
return undefined;
|
||||
}
|
||||
|
||||
let elementToStartIndexFrom = 0;
|
||||
while (true) {
|
||||
const realEstatesForSinglePage = listOfAllRealEstates.slice(
|
||||
@@ -191,24 +203,22 @@ class ProstorCrawler {
|
||||
const { lat, lng, property_name, price, size, link, status } = realEstate;
|
||||
|
||||
//Status information is given already in realestate list
|
||||
//For VIP Ads status ='' canot be used, but no VIP ads are crawled
|
||||
//We will make "fake" vip ad for RE that have size=55
|
||||
//It is weird because yesterday it said 'VIP ponuda' ???
|
||||
const adStatus =
|
||||
size === "55"
|
||||
? ProstorCrawler.getStatusId("VIP ponuda")
|
||||
: ProstorCrawler.getStatusId(status);
|
||||
const adStatus = ProstorCrawler.getStatusId(status);
|
||||
|
||||
const url = `https://prostor.ba${link}`;
|
||||
|
||||
// console.log("[PROSTOR] Scraping : ", url);
|
||||
try {
|
||||
const adPageSource = await fetch(url, {
|
||||
const adPageSource = await nodeFetch(url, {
|
||||
headers: { Cookie: prostorCookie }
|
||||
});
|
||||
const body = await adPageSource.text();
|
||||
const $ = cheerio.load(body);
|
||||
|
||||
if (body.indexOf('<html') === -1) {
|
||||
throw { message: 'Failed to fetch page !' }
|
||||
}
|
||||
|
||||
// link contains part of the URL in the format of : /prodaja/stan/stup/9556
|
||||
// general form is : /actionType/realEstateType/location/realEstateID
|
||||
// linkParts contains : ['', 'actionType', 'realEstateType', 'location', 'realEstateID']
|
||||
@@ -433,14 +443,14 @@ class ProstorCrawler {
|
||||
}
|
||||
|
||||
try {
|
||||
const res = await fetch(url, {
|
||||
const res = await nodeFetch(url, {
|
||||
headers: { Cookie: prostorCookie }
|
||||
});
|
||||
const body = await res.text();
|
||||
const $ = cheerio.load(body);
|
||||
|
||||
const scriptElement = $(
|
||||
"body > div > div.container-fluid > script:nth-child(7)"
|
||||
"body > div.content > div.container-fluid > script:nth-child(6)"
|
||||
);
|
||||
|
||||
if (
|
||||
@@ -597,7 +607,7 @@ class ProstorCrawler {
|
||||
formData.append("email", PROSTOR_LOGIN.EMAIL);
|
||||
formData.append("password", PROSTOR_LOGIN.PASSWORD);
|
||||
|
||||
return fetch("https://prostor.ba/moj-prostor/prijava", {
|
||||
return nodeFetch("https://prostor.ba/moj-prostor/prijava", {
|
||||
method: "POST",
|
||||
body: formData,
|
||||
headers: { Cookie: prostorCookie }
|
||||
@@ -624,9 +634,12 @@ class ProstorCrawler {
|
||||
});
|
||||
}
|
||||
async getCookies() {
|
||||
const getResponse = await fetch("https://prostor.ba/moj-prostor/prijava", {
|
||||
headers: { Cookie: "" }
|
||||
});
|
||||
const getResponse = await nodeFetch(
|
||||
"https://prostor.ba/moj-prostor/prijava",
|
||||
{
|
||||
headers: { Cookie: "" }
|
||||
}
|
||||
);
|
||||
const raw = getResponse.headers.raw()["set-cookie"];
|
||||
const cookie = raw
|
||||
.map(datastring => {
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
"use strict";
|
||||
|
||||
const fetch = require("node-fetch");
|
||||
const fetch = require("../../helpers/fetchWrapper");
|
||||
const cheerio = require("cheerio");
|
||||
const Promise = require("bluebird");
|
||||
const moment = require("moment-timezone");
|
||||
@@ -159,7 +159,7 @@ class RentalCrawler {
|
||||
}
|
||||
|
||||
try {
|
||||
const res = await fetch(url);
|
||||
const res = await fetch(url, {} , false);
|
||||
const body = await res.text();
|
||||
const $ = cheerio.load(body);
|
||||
let hrefs = [];
|
||||
@@ -202,6 +202,10 @@ class RentalCrawler {
|
||||
const body = await adPageSource.text();
|
||||
const $ = cheerio.load(body);
|
||||
|
||||
if (body.indexOf('<html') === -1) {
|
||||
throw { message: 'Failed to fetch page !' }
|
||||
}
|
||||
|
||||
const mapElementParent = $(".box-map").parent();
|
||||
const scriptElement = $("script", mapElementParent);
|
||||
if (
|
||||
@@ -399,7 +403,9 @@ class RentalCrawler {
|
||||
);
|
||||
if (!publishedDateMoment.isValid()) {
|
||||
throw {
|
||||
message: `Invalid published date : ${extractedData["re_realEstates_inserted"]}`
|
||||
message: `Invalid published date : ${
|
||||
extractedData["re_realEstates_inserted"]
|
||||
}`
|
||||
};
|
||||
}
|
||||
|
||||
@@ -410,7 +416,9 @@ class RentalCrawler {
|
||||
);
|
||||
if (!renewedDateMoment.isValid()) {
|
||||
throw {
|
||||
message: `Invalid renewed date : ${extractedData["re_realEstates_edited"]}`
|
||||
message: `Invalid renewed date : ${
|
||||
extractedData["re_realEstates_edited"]
|
||||
}`
|
||||
};
|
||||
}
|
||||
|
||||
|
||||
687
app/crawler/specificCrawlers/saljic.js
Normal file
687
app/crawler/specificCrawlers/saljic.js
Normal file
@@ -0,0 +1,687 @@
|
||||
"use strict";
|
||||
|
||||
const fetch = require("../../helpers/fetchWrapper");
|
||||
const { getUrlParams } = require("../../helpers/url");
|
||||
const cheerio = require("cheerio");
|
||||
const moment = require("moment-timezone");
|
||||
const PromisePool = require('@supercharge/promise-pool');
|
||||
|
||||
const {
|
||||
AD_TYPE,
|
||||
AD_CATEGORY,
|
||||
AD_AGENCY,
|
||||
AD_STATUS,
|
||||
CRAWLER_AD_TYPE,
|
||||
FURNISHING_TYPE,
|
||||
HEATING_TYPE
|
||||
} = require("../../common/enums");
|
||||
|
||||
const {
|
||||
PRINT_CRAWLER_DEBUG,
|
||||
DEFAULT_TIMEZONE
|
||||
} = require("../../config/appConfig");
|
||||
const { SALJIC_FORCE_CRAWL } = require("../specificConfigs/saljic");
|
||||
|
||||
const SALJIC_ENUMS = {
|
||||
SALJIC_AD_TYPE: {
|
||||
[CRAWLER_AD_TYPE.ALL]: "&input_vrsta=",
|
||||
[CRAWLER_AD_TYPE.ONLY_SELL]: "&input_vrsta=1",
|
||||
[CRAWLER_AD_TYPE.ONLY_RENT]: "&input_vrsta=2"
|
||||
},
|
||||
SALJIC_AD_CATEGORY: {
|
||||
[AD_CATEGORY.ALL.id]: "&input_kategorija=",
|
||||
[AD_CATEGORY.FLAT.id]: "&input_kategorija=15",
|
||||
[AD_CATEGORY.HOUSE.id]: "&input_kategorija=9",
|
||||
[AD_CATEGORY.LAND.id]: "&input_kategorija=5", //3 and 4 also gradjevinsko
|
||||
[AD_CATEGORY.OFFICE.id]: "&input_kategorija=8",
|
||||
[AD_CATEGORY.APARTMENT.id]: "&input_kategorija=1",
|
||||
[AD_CATEGORY.GARAGE.id]: "&input_kategorija=2"
|
||||
//[AD_CATEGORY.COTTAGE.id]: ""
|
||||
}
|
||||
};
|
||||
|
||||
class SaljicCrawler {
|
||||
constructor(
|
||||
savers = [],
|
||||
crawlerAdTypes = CRAWLER_AD_TYPE.ALL,
|
||||
crawlerAdCategories = [AD_CATEGORY.FLAT, AD_CATEGORY.HOUSE],
|
||||
maxPages = 5000,
|
||||
maxResultsPerPage = 5000,
|
||||
ignoredUsernames = [],
|
||||
delayBetweenPages = 500
|
||||
) {
|
||||
this.savers = savers;
|
||||
this.baseUrl = "https://www.saljicnekretnine.ba/v2/nekretnine_search";
|
||||
this.crawlerAdTypes = crawlerAdTypes;
|
||||
this.crawlerAdCategories = crawlerAdCategories;
|
||||
this.maxPages = maxPages
|
||||
this.maxResultsPerPage = maxResultsPerPage;
|
||||
this.delayBetweenPages = delayBetweenPages;
|
||||
}
|
||||
|
||||
async crawl() {
|
||||
const crawlAdCategories = this.crawlerAdCategories;
|
||||
|
||||
const newRealEstates = [];
|
||||
|
||||
if (crawlAdCategories) {
|
||||
const indexGenerators = [];
|
||||
for (const adCategory of crawlAdCategories) {
|
||||
indexGenerators.push(this.categoryIndexer(adCategory));
|
||||
}
|
||||
//
|
||||
//console.log(indexGenerators);
|
||||
//
|
||||
let done = false;
|
||||
while (!done) {
|
||||
const categoryIndexerPromises = [];
|
||||
const generatorsToRemove = [];
|
||||
for (const indexGenerator of indexGenerators) {
|
||||
categoryIndexerPromises.push(indexGenerator.next());
|
||||
generatorsToRemove.push(false);
|
||||
}
|
||||
|
||||
const singlePageResults = await Promise.all(categoryIndexerPromises);
|
||||
const entries = singlePageResults.entries();
|
||||
|
||||
for (const [index, { value: singlePageResult }] of entries) {
|
||||
if (singlePageResult) {
|
||||
const saveResults = await this.saveCrawledResults(singlePageResult);
|
||||
const { newRecords } = saveResults;
|
||||
|
||||
newRealEstates.push(...newRecords);
|
||||
|
||||
if (
|
||||
Array.isArray(newRecords) &&
|
||||
newRecords.length === 0 &&
|
||||
!SALJIC_FORCE_CRAWL
|
||||
) {
|
||||
generatorsToRemove[index] = true;
|
||||
}
|
||||
} else {
|
||||
//Generator returned undefined, remove this generator from array
|
||||
generatorsToRemove[index] = true;
|
||||
// console.log("Generator ", index + 1, "has no more pages");
|
||||
}
|
||||
}
|
||||
|
||||
// console.log("Generators state : ", generatorsToRemove);
|
||||
for (let i = generatorsToRemove.length - 1; i >= 0; i--) {
|
||||
if (generatorsToRemove[i]) {
|
||||
// console.log("\tRemove generator ", i + 1);
|
||||
indexGenerators.splice(i, 1);
|
||||
}
|
||||
}
|
||||
if (indexGenerators.length === 0) {
|
||||
done = true;
|
||||
}
|
||||
|
||||
await this.sleep(this.delayBetweenPages);
|
||||
}
|
||||
}
|
||||
return newRealEstates;
|
||||
}
|
||||
|
||||
async *categoryIndexer(adCategory) {
|
||||
let pageToIndex = 1;
|
||||
|
||||
const urlAdTypePart = SALJIC_ENUMS.SALJIC_AD_TYPE[this.crawlerAdTypes];
|
||||
const urlCategoryPart = SALJIC_ENUMS.SALJIC_AD_CATEGORY[adCategory];
|
||||
|
||||
if (urlAdTypePart !== undefined && urlCategoryPart !== undefined) {
|
||||
while (true) {
|
||||
const urlPagePart = pageToIndex === 1 ? "" : (pageToIndex - 1) * 2 * 11;
|
||||
const urlPageToCrawl = `${this.baseUrl}?order_by=${urlAdTypePart}${urlCategoryPart}&per_page=${urlPagePart}`;
|
||||
|
||||
const singlePageResults = await this.indexSinglePage(
|
||||
urlPageToCrawl,
|
||||
this.maxResultsPerPage
|
||||
);
|
||||
|
||||
if (Array.isArray(singlePageResults) && singlePageResults.length > 0) {
|
||||
yield singlePageResults;
|
||||
} else {
|
||||
return undefined;
|
||||
}
|
||||
|
||||
++pageToIndex;
|
||||
if (pageToIndex === this.maxPages) {
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
|
||||
async indexSinglePage(url, maxResultsPerPage) {
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log("[SALJIC] Index page : ", url);
|
||||
}
|
||||
|
||||
try {
|
||||
const res = await fetch(url, {}, false);
|
||||
const body = await res.text();
|
||||
const $ = cheerio.load(body);
|
||||
let hrefs = [];
|
||||
|
||||
$("#shop")
|
||||
.find(".product")
|
||||
.each((i, elem) => {
|
||||
const href = $(elem)
|
||||
.find("a")
|
||||
.first()
|
||||
.attr("href");
|
||||
if (href) {
|
||||
hrefs.push(href);
|
||||
}
|
||||
});
|
||||
|
||||
let adTypesTmp = [];
|
||||
|
||||
$("#shop")
|
||||
.find(".product")
|
||||
.each((i, elem) => {
|
||||
const adType = $(elem)
|
||||
.find(".trakica-search-page")
|
||||
.text()
|
||||
.trim();
|
||||
if (adType) {
|
||||
adTypesTmp.push(adType);
|
||||
}
|
||||
});
|
||||
|
||||
//Converting to AD_TYPE
|
||||
const adTypes = adTypesTmp.map(adTypeText => {
|
||||
return this.getAdTypeId(adTypeText);
|
||||
});
|
||||
|
||||
//Converting to absolute URLs
|
||||
const hrefsAbs = hrefs.map(link => {
|
||||
return "https://www.saljicnekretnine.ba" + link;
|
||||
});
|
||||
|
||||
let actualNoOfResults =
|
||||
hrefsAbs.length <= maxResultsPerPage
|
||||
? hrefsAbs.length
|
||||
: maxResultsPerPage;
|
||||
|
||||
const asyncScraping = [];
|
||||
for (let i = 0; i < actualNoOfResults; i++) {
|
||||
asyncScraping.push([hrefsAbs[i], adTypes[i]]);
|
||||
}
|
||||
|
||||
|
||||
|
||||
const dataResults = []
|
||||
const { scrapedData, errors } = await PromisePool
|
||||
.withConcurrency(2)
|
||||
.for(asyncScraping)
|
||||
.process(async data => {
|
||||
const result = await this.scrapeAd(...data)
|
||||
await this.sleep(this.delayBetweenPages);
|
||||
dataResults.push(result)
|
||||
return result; //TODO: this does not work, scrapedData is null, dataResults works
|
||||
})
|
||||
|
||||
const filteredScrapedData = dataResults.filter(adData => !!adData);
|
||||
return filteredScrapedData;
|
||||
} catch (e) {
|
||||
console.error("[SALJIC] Exception caught:" + e);
|
||||
return [];
|
||||
}
|
||||
}
|
||||
|
||||
async scrapeAd(url, adType) {
|
||||
// console.log("[SALJIC] Scraping : ", url);
|
||||
try {
|
||||
const adPageSource = await fetch(url);
|
||||
const body = await adPageSource.text();
|
||||
const $ = cheerio.load(body);
|
||||
|
||||
if (body.indexOf('<html') === -1) {
|
||||
throw { message: 'Failed to fetch page !' }
|
||||
}
|
||||
|
||||
// No information for status ex. PRODAN
|
||||
const status = AD_STATUS.STATUS_NORMAL;
|
||||
//Extracting agency ID from url
|
||||
const agencyObjectId = url
|
||||
? parseInt(url.substring(46, url.length))
|
||||
: null;
|
||||
|
||||
if (!agencyObjectId) {
|
||||
throw { message : 'No agency object ID - URL changed?'}
|
||||
}
|
||||
|
||||
//Extracting main properties
|
||||
const propertySelectors = {
|
||||
title:
|
||||
"div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-title > h2",
|
||||
price:
|
||||
"div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.topmargin-sm.single-product > div.product > div.product-price > ins",
|
||||
streetName:
|
||||
"div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-content.topmargin > p",
|
||||
descriptions:
|
||||
"div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-content.topmargin > div.toggle.toggle-bg > div.togglec >p:nth-child(1)",
|
||||
latAndLong:
|
||||
"iframe"
|
||||
};
|
||||
const title = $(propertySelectors.title)
|
||||
.text()
|
||||
.replace(/(\r\n|\n|\r)/gm, "")
|
||||
.replace(/ {1,}/g, " ")
|
||||
.trim();
|
||||
|
||||
const priceText = $(propertySelectors.price)
|
||||
.text()
|
||||
.replace(/(\r\n|\n|\r)/gm, "")
|
||||
.replace(/ {1,}/g, " ")
|
||||
.trim();
|
||||
const price =
|
||||
priceText === "CIJENA NA UPIT"
|
||||
? null
|
||||
: parseFloat(
|
||||
priceText.substring(8, priceText.length - 3).replace(",", "")
|
||||
);
|
||||
|
||||
const streetName = $(propertySelectors.streetName)
|
||||
.text()
|
||||
.replace(/(\r\n|\n|\r)/gm, "")
|
||||
.trim();
|
||||
|
||||
const descriptions = $(propertySelectors.descriptions)
|
||||
.text()
|
||||
.replace(/\"/g, "")
|
||||
.trim();
|
||||
|
||||
const latAndLongSrc = $(propertySelectors.latAndLong).attr("src");
|
||||
let latText;
|
||||
let longText;
|
||||
if (latAndLongSrc){
|
||||
const mapParams = getUrlParams(latAndLongSrc);
|
||||
if (mapParams) {
|
||||
if (mapParams['marker']){
|
||||
const marker = mapParams['marker'].split(',');
|
||||
latText = marker[0] ? marker[0] : undefined;
|
||||
longText = marker[1] ? marker[1] : undefined;
|
||||
}else{
|
||||
if (mapParams['mlat']) {
|
||||
latText = mapParams['mlat'];
|
||||
}
|
||||
if (mapParams['mlon']) {
|
||||
longText = mapParams['mlon'];
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
const locationLat = parseFloat(latText) || null;
|
||||
const locationLong = parseFloat(longText) || null;
|
||||
|
||||
//====== DETAIL INFORMATION FIELDS ==========
|
||||
let area = null,
|
||||
gardenSize = null,
|
||||
numberOfRooms = null,
|
||||
numberOfFloors = null,
|
||||
floor = null,
|
||||
accessRoadType = null,
|
||||
heatingType = null,
|
||||
furnishingType = null,
|
||||
balcony = null,
|
||||
newBuilding = null,
|
||||
elevator = null,
|
||||
water = null,
|
||||
electricity = null,
|
||||
drainageSystem = null,
|
||||
registeredInZkBooks = null,
|
||||
recentlyAdapted = null,
|
||||
parking = null,
|
||||
garage = null,
|
||||
gas = null,
|
||||
antiTheftDoor = null,
|
||||
airCondition = null,
|
||||
phoneConnection = null,
|
||||
cableTV = null,
|
||||
internet = null,
|
||||
basementAttic = null,
|
||||
storeRoom = null,
|
||||
videoSurveillance = null,
|
||||
alarm = null,
|
||||
suitableForStudents = null,
|
||||
includingBills = null,
|
||||
animalsAllowed = null,
|
||||
pool = null,
|
||||
exchange = null,
|
||||
urbanPlanPermit = null,
|
||||
buildingPermit = null,
|
||||
utilityConnection = null,
|
||||
distanceToRiver = null;
|
||||
let publishedDate = null;
|
||||
let renewedDate = null;
|
||||
let realEstateType;
|
||||
let numberOfViewsAgency = null;
|
||||
let numberOfViewsKivi = null;
|
||||
let streetNumber = 0;
|
||||
let adStatus = status;
|
||||
let shortDescription = descriptions
|
||||
? descriptions.substring(0, descriptions.indexOf("."))
|
||||
: "";
|
||||
let longDescription = descriptions || "";
|
||||
//Extracting data - Glavne karakteristike
|
||||
let mainFieldIndex = 1;
|
||||
do {
|
||||
const mainFieldSelector = `div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-content.topmargin > div.col-md-12.bottommargin > ul > li.list-group-item:nth-child(${mainFieldIndex})`;
|
||||
|
||||
const mainField = $(mainFieldSelector)
|
||||
.text()
|
||||
.replace(/[\n\r\t]/gm, "")
|
||||
.trim();
|
||||
|
||||
const mainFieldTitle = mainField
|
||||
? mainField.substring(0, mainField.indexOf(" "))
|
||||
: "";
|
||||
const mainFieldValue = mainField
|
||||
? mainField
|
||||
.substring(mainField.indexOf(" "), mainField.length)
|
||||
.trim()
|
||||
: "";
|
||||
|
||||
switch (mainFieldTitle) {
|
||||
case "Površina":
|
||||
area = parseFloat(
|
||||
mainFieldValue.substring(0, mainFieldValue.indexOf(" "))
|
||||
);
|
||||
break;
|
||||
case "Okućnica":
|
||||
gardenSize = parseFloat(
|
||||
mainFieldValue.substring(0, mainFieldValue.indexOf(" "))
|
||||
);
|
||||
break;
|
||||
case "Broj soba":
|
||||
numberOfRooms = parseInt(mainFieldValue);
|
||||
break;
|
||||
case "Broj spratova":
|
||||
numberOfFloors = this.parseNumberOfFloors(mainFieldValue);
|
||||
break;
|
||||
case "Sprat":
|
||||
floor = parseInt(mainFieldValue);
|
||||
break;
|
||||
case "Godina renoviranja":
|
||||
recentlyAdapted = true;
|
||||
break;
|
||||
case "Broj parking mjesta":
|
||||
parking = true;
|
||||
break;
|
||||
case "Dostupno od":
|
||||
const day = mainFieldValue.substring(0, 2);
|
||||
const month = mainFieldValue.substring(3, 5);
|
||||
const year = mainFieldValue.substring(6, mainFieldValue.length);
|
||||
publishedDate = new Date(`${month}/${day}/${year}`);
|
||||
break;
|
||||
default:
|
||||
break;
|
||||
}
|
||||
|
||||
if (mainFieldTitle === "") {
|
||||
break;
|
||||
}
|
||||
mainFieldIndex++;
|
||||
} while (true);
|
||||
|
||||
//Extracting data - Sadrzaji
|
||||
let additionalFieldIndex = 1;
|
||||
do {
|
||||
const additionalFieldSelector = `div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-content.topmargin > div.col-md-12.bottommargin > ul > li.border-color.col-md-5.col-md-offset-1.col-md-pull-1.list-group-item-bottom:nth-child(${additionalFieldIndex})`;
|
||||
|
||||
const additionalField = $(additionalFieldSelector)
|
||||
.text()
|
||||
.trim();
|
||||
|
||||
if (additionalFieldIndex === 1) {
|
||||
//Extracting data of real estate type
|
||||
const categoryTmp = additionalField
|
||||
.replace(/[\n\r\t]/gm, "")
|
||||
.substring(
|
||||
additionalField.indexOf("Kategorija") + 10,
|
||||
additionalField.length
|
||||
)
|
||||
.trim();
|
||||
realEstateType = this.getAdCategoryId(categoryTmp);
|
||||
if (!realEstateType) {
|
||||
throw { message: 'No real estate type - page body not loaded correctly or page changed?' }
|
||||
}
|
||||
} else {
|
||||
switch (additionalField) {
|
||||
case "Internet":
|
||||
internet = true;
|
||||
break;
|
||||
case "Garaža":
|
||||
garage = true;
|
||||
break;
|
||||
case "Klima":
|
||||
airCondition = true;
|
||||
break;
|
||||
case "Balkon":
|
||||
balcony = true;
|
||||
break;
|
||||
case "Ostava":
|
||||
storeRoom = true;
|
||||
break;
|
||||
case "Podrum":
|
||||
basementAttic = true;
|
||||
break;
|
||||
case "Blindirana vrata":
|
||||
antiTheftDoor = true;
|
||||
break;
|
||||
case "Voda":
|
||||
water = true;
|
||||
break;
|
||||
case "Kablovska":
|
||||
cableTV = true;
|
||||
break;
|
||||
case "Uknjiženo":
|
||||
registeredInZkBooks = true;
|
||||
break;
|
||||
case "Grijanje - centralno":
|
||||
heatingType = HEATING_TYPE.CENTRAL_CITY.id;
|
||||
break;
|
||||
case "Grijanje - plin":
|
||||
heatingType = HEATING_TYPE.GAS.id;
|
||||
break;
|
||||
case "Grijanje - struja":
|
||||
heatingType = HEATING_TYPE.ELECTRICITY.id;
|
||||
break;
|
||||
case "Grijanje":
|
||||
heatingType = HEATING_TYPE.OTHER.id;
|
||||
break;
|
||||
case "Plin":
|
||||
gas = true;
|
||||
break;
|
||||
case "Namješten":
|
||||
furnishingType = FURNISHING_TYPE.FURNISHED.id;
|
||||
break;
|
||||
case "Alarm":
|
||||
alarm = true;
|
||||
break;
|
||||
case "Video nadzor":
|
||||
videoSurveillance = true;
|
||||
break;
|
||||
case "Lift":
|
||||
elevator = true;
|
||||
break;
|
||||
case "Novogradnja":
|
||||
newBuilding = true;
|
||||
break;
|
||||
|
||||
default:
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (additionalField === "") {
|
||||
break;
|
||||
}
|
||||
additionalFieldIndex++;
|
||||
} while (true);
|
||||
|
||||
//If no published date it takes current date of crawling
|
||||
if (publishedDate) {
|
||||
renewedDate = new Date();
|
||||
} else {
|
||||
publishedDate = new Date();
|
||||
renewedDate = new Date();
|
||||
}
|
||||
|
||||
const originAgencyName = AD_AGENCY.SALJIC;
|
||||
const locality = "";
|
||||
const municipality = "";
|
||||
const city = "";
|
||||
const region = "";
|
||||
const entity = "";
|
||||
const country = "";
|
||||
|
||||
const data = {
|
||||
url,
|
||||
agencyObjectId,
|
||||
originAgencyName,
|
||||
realEstateType,
|
||||
adType,
|
||||
title,
|
||||
price,
|
||||
area,
|
||||
gardenSize,
|
||||
shortDescription,
|
||||
longDescription,
|
||||
streetNumber,
|
||||
streetName,
|
||||
locality,
|
||||
municipality,
|
||||
city,
|
||||
region,
|
||||
entity,
|
||||
country,
|
||||
locationLat,
|
||||
locationLong,
|
||||
adStatus,
|
||||
publishedDate,
|
||||
renewedDate,
|
||||
numberOfRooms,
|
||||
numberOfFloors,
|
||||
floor,
|
||||
accessRoadType,
|
||||
heatingType,
|
||||
furnishingType,
|
||||
balcony,
|
||||
newBuilding,
|
||||
elevator,
|
||||
water,
|
||||
electricity,
|
||||
drainageSystem,
|
||||
registeredInZkBooks,
|
||||
recentlyAdapted,
|
||||
parking,
|
||||
garage,
|
||||
gas,
|
||||
antiTheftDoor,
|
||||
airCondition,
|
||||
phoneConnection,
|
||||
cableTV,
|
||||
internet,
|
||||
basementAttic,
|
||||
storeRoom,
|
||||
videoSurveillance,
|
||||
alarm,
|
||||
suitableForStudents,
|
||||
includingBills,
|
||||
animalsAllowed,
|
||||
pool,
|
||||
exchange,
|
||||
urbanPlanPermit,
|
||||
buildingPermit,
|
||||
utilityConnection,
|
||||
distanceToRiver,
|
||||
numberOfViewsAgency,
|
||||
numberOfViewsKivi
|
||||
};
|
||||
|
||||
return data;
|
||||
} catch (e) {
|
||||
console.error("[SALJIC] Exception caught: " + e.message, "\r\nURL:", url);
|
||||
}
|
||||
return null;
|
||||
}
|
||||
|
||||
//======= HELPER FUNCTIONS =============
|
||||
|
||||
getAdCategoryId(categoryText) {
|
||||
switch (categoryText) {
|
||||
case "Stan":
|
||||
return AD_CATEGORY.FLAT.id;
|
||||
case "Građevinsko zemljiste":
|
||||
return AD_CATEGORY.LAND.id;
|
||||
case "Industrijsko zemljiste":
|
||||
return AD_CATEGORY.LAND.id;
|
||||
case "Poljoprivredno zemljiste":
|
||||
return AD_CATEGORY.LAND.id;
|
||||
case "Kuća":
|
||||
return AD_CATEGORY.HOUSE.id;
|
||||
case "Poslovni prostor":
|
||||
return AD_CATEGORY.OFFICE.id;
|
||||
case "Kancelarije":
|
||||
return AD_CATEGORY.OFFICE.id;
|
||||
case "Apartmani":
|
||||
return AD_CATEGORY.APARTMENT.id;
|
||||
case "Garaža":
|
||||
return AD_CATEGORY.GARAGE.id;
|
||||
case "Vikendica":
|
||||
return AD_CATEGORY.COTTAGE.id;
|
||||
default:
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
|
||||
getAdTypeId(adTypeText) {
|
||||
switch (adTypeText) {
|
||||
case "PRODAJA":
|
||||
return AD_TYPE.AD_TYPE_SALE.stringId;
|
||||
case "NAJAM":
|
||||
return AD_TYPE.AD_TYPE_RENT.stringId;
|
||||
default:
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
|
||||
parseNumberOfFloors(numberOfFloorsText) {
|
||||
const tryNumericalValue = parseInt(numberOfFloorsText);
|
||||
if (!isNaN(tryNumericalValue)){
|
||||
return tryNumericalValue;
|
||||
}
|
||||
|
||||
// Guess number of floors based on number of + sign concatenations
|
||||
// e.g. P+S+Pt -> 3 floors
|
||||
if (typeof numberOfFloorsText === 'string' && numberOfFloorsText.indexOf('+') > 0) {
|
||||
return numberOfFloorsText.split('+').length + 1
|
||||
}
|
||||
|
||||
return null
|
||||
}
|
||||
|
||||
async sleep(ms) {
|
||||
return new Promise(resolve => setTimeout(resolve, ms));
|
||||
}
|
||||
|
||||
async saveCrawledResults(results) {
|
||||
const savers = this.savers;
|
||||
|
||||
// for (const saver of savers) {
|
||||
// await saver.save(results);
|
||||
// }
|
||||
|
||||
//For now, we use only Postgres saver, so ...
|
||||
return savers[0].save(results);
|
||||
//so that we can use some sequelize options and information when data is inserted
|
||||
}
|
||||
}
|
||||
|
||||
module.exports = SaljicCrawler;
|
||||
20
app/helpers/db/priceHistory.js
Normal file
20
app/helpers/db/priceHistory.js
Normal file
@@ -0,0 +1,20 @@
|
||||
"use strict";
|
||||
const db = require("../../models/index");
|
||||
const sequelize = require("sequelize");
|
||||
|
||||
const bulkUpsertPriceHistory = async priceHistoryData => {
|
||||
try {
|
||||
const order = [["realEstateId", "desc"]];
|
||||
|
||||
return await db.PriceHistory.bulkCreate(priceHistoryData, {
|
||||
order,
|
||||
ignoreDuplicates: true
|
||||
});
|
||||
} catch (e) {
|
||||
console.log("Error bulk upserting priceHistory : ", e);
|
||||
}
|
||||
};
|
||||
|
||||
module.exports = {
|
||||
bulkUpsertPriceHistory
|
||||
};
|
||||
@@ -2,6 +2,8 @@
|
||||
const db = require("../../models/index");
|
||||
const sequelize = require("sequelize");
|
||||
const Op = sequelize.Op;
|
||||
const { AD_CATEGORY } = require("../../common/enums");
|
||||
|
||||
const bulkUpsertRealEstates = async realEstateData => {
|
||||
try {
|
||||
const fieldsToUpdateIfDuplicate = [
|
||||
@@ -96,12 +98,16 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
floorMin,
|
||||
floorMax,
|
||||
includeIncompleteAds,
|
||||
includeWithoutPrice,
|
||||
balcony,
|
||||
elevator,
|
||||
newBuilding,
|
||||
accessRoadType
|
||||
} = searchRequest;
|
||||
|
||||
//Needed for defining which attribute should exist or not
|
||||
const realEstateTypeObject = AD_CATEGORY[realEstateType];
|
||||
|
||||
const longitudeColumn = sequelize.col("locationLong");
|
||||
const latitudeColumn = sequelize.col("locationLat");
|
||||
|
||||
@@ -134,15 +140,6 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
const query = {
|
||||
adType,
|
||||
realEstateType,
|
||||
price: {
|
||||
[Op.or]: {
|
||||
[Op.and]: {
|
||||
[Op.lte]: priceMax,
|
||||
[Op.gte]: priceMin
|
||||
},
|
||||
[Op.is]: null
|
||||
}
|
||||
},
|
||||
area: {
|
||||
[Op.lte]: sizeMax,
|
||||
[Op.gte]: sizeMin
|
||||
@@ -154,15 +151,6 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
const queryIncludeIncomplete = {
|
||||
adType,
|
||||
realEstateType,
|
||||
price: {
|
||||
[Op.or]: {
|
||||
[Op.and]: {
|
||||
[Op.lte]: priceMax,
|
||||
[Op.gte]: priceMin
|
||||
},
|
||||
[Op.is]: null
|
||||
}
|
||||
},
|
||||
area: {
|
||||
[Op.or]: {
|
||||
[Op.and]: {
|
||||
@@ -175,8 +163,49 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
[Op.and]: geoSearchQueryPart
|
||||
};
|
||||
|
||||
//Every other attribute is checked separately and included in query only if it is defined
|
||||
if (gardenSizeMax && gardenSizeMin) {
|
||||
//Is user unchecked includeWithoutPrice FALSE then it shouldn't return null values of price
|
||||
//If not then null values are accepted (this is DEFAULT)
|
||||
//includeIncpompleteAds does not have effect on price query
|
||||
if (includeWithoutPrice) {
|
||||
query.price = {
|
||||
[Op.or]: {
|
||||
[Op.and]: {
|
||||
[Op.lte]: priceMax,
|
||||
[Op.gte]: priceMin
|
||||
},
|
||||
[Op.is]: null
|
||||
}
|
||||
};
|
||||
queryIncludeIncomplete.price = {
|
||||
[Op.or]: {
|
||||
[Op.and]: {
|
||||
[Op.lte]: priceMax,
|
||||
[Op.gte]: priceMin
|
||||
},
|
||||
[Op.is]: null
|
||||
}
|
||||
};
|
||||
} else {
|
||||
query.price = {
|
||||
[Op.and]: {
|
||||
[Op.lte]: priceMax,
|
||||
[Op.gte]: priceMin
|
||||
}
|
||||
};
|
||||
queryIncludeIncomplete.price = {
|
||||
[Op.and]: {
|
||||
[Op.lte]: priceMax,
|
||||
[Op.gte]: priceMin
|
||||
}
|
||||
};
|
||||
}
|
||||
//Every other attribute is checked separately and included in query only if it is defined for real estate type
|
||||
|
||||
if (
|
||||
realEstateTypeObject.hasGardenSize &&
|
||||
gardenSizeMax != null &&
|
||||
gardenSizeMin != null
|
||||
) {
|
||||
query.gardenSize = {
|
||||
[Op.lte]: gardenSizeMax,
|
||||
[Op.gte]: gardenSizeMin
|
||||
@@ -192,7 +221,11 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
};
|
||||
}
|
||||
|
||||
if (numberOfRoomsMin && numberOfRoomsMax) {
|
||||
if (
|
||||
realEstateTypeObject.hasNumberOfRoom &&
|
||||
numberOfRoomsMin != null &&
|
||||
numberOfRoomsMax != null
|
||||
) {
|
||||
query.numberOfRooms = {
|
||||
[Op.lte]: numberOfRoomsMax,
|
||||
[Op.gte]: numberOfRoomsMin
|
||||
@@ -208,7 +241,11 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
};
|
||||
}
|
||||
|
||||
if (numberOfFloorsMin && numberOfFloorsMax) {
|
||||
if (
|
||||
realEstateTypeObject.hasNumberOfFloors &&
|
||||
numberOfFloorsMin != null &&
|
||||
numberOfFloorsMax != null
|
||||
) {
|
||||
query.numberOfFloors = {
|
||||
[Op.lte]: numberOfFloorsMax,
|
||||
[Op.gte]: numberOfFloorsMin
|
||||
@@ -224,7 +261,11 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
};
|
||||
}
|
||||
|
||||
if (floorMin && floorMax) {
|
||||
if (
|
||||
realEstateTypeObject.hasFloorProp &&
|
||||
floorMin != null &&
|
||||
floorMax != null
|
||||
) {
|
||||
query.floor = {
|
||||
[Op.lte]: floorMax,
|
||||
[Op.gte]: floorMin
|
||||
@@ -239,8 +280,10 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
if (balcony) {
|
||||
//Logic for balcony, newBuilding and elevator from users side
|
||||
//If true is checked, then I want characteristic to be true but,
|
||||
//if it is not checked, then I dont care - it can be null or false or true
|
||||
if (realEstateTypeObject.hasBalconyProp && balcony === true) {
|
||||
query.balcony = {
|
||||
[Op.eq]: balcony
|
||||
};
|
||||
@@ -252,7 +295,7 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
};
|
||||
}
|
||||
|
||||
if (newBuilding) {
|
||||
if (realEstateTypeObject.hasNewBuildingProp && newBuilding === true) {
|
||||
query.newBuilding = {
|
||||
[Op.eq]: newBuilding
|
||||
};
|
||||
@@ -264,7 +307,7 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
};
|
||||
}
|
||||
|
||||
if (elevator) {
|
||||
if (realEstateTypeObject.hasElevatorProp && elevator === true) {
|
||||
query.elevator = {
|
||||
[Op.eq]: elevator
|
||||
};
|
||||
@@ -275,7 +318,8 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
//If user wants 'ANY' road type acces then it is not included in query -
|
||||
//returns every road type and null values
|
||||
if (accessRoadType !== "ANY") {
|
||||
query.accessRoadType = {
|
||||
[Op.eq]: accessRoadType
|
||||
@@ -288,10 +332,14 @@ const findRealEstatesForSearchRequest = async (searchRequest, maxResults) => {
|
||||
};
|
||||
}
|
||||
|
||||
//When includeIncompleteAds are not defined - null it will consider it true
|
||||
const order = [["updatedAt", "desc"]];
|
||||
|
||||
return db.RealEstate.findAll({
|
||||
where: includeIncompleteAds ? queryIncludeIncomplete : query,
|
||||
where:
|
||||
includeIncompleteAds || includeIncompleteAds == null
|
||||
? queryIncludeIncomplete
|
||||
: query,
|
||||
limit: maxResults,
|
||||
order
|
||||
});
|
||||
|
||||
@@ -3,6 +3,7 @@ const db = require("../../models/index");
|
||||
const sequelize = require("sequelize");
|
||||
const Op = sequelize.Op;
|
||||
const { AD_CATEGORY } = require("../../common/enums");
|
||||
const { CHECK_UP_DAYS } = require("../../config/appConfig");
|
||||
|
||||
const getSearchRequest = async searchRequestId => {
|
||||
try {
|
||||
@@ -16,6 +17,22 @@ const getSearchRequest = async searchRequestId => {
|
||||
const createSearchRequest = async (searchRequestFields = {}) => {
|
||||
return await db.SearchRequest.create(searchRequestFields);
|
||||
};
|
||||
const findAllRequestsForCheckUp = async () => {
|
||||
const checkUpOffset = 24 * 60 * 60 * 1000 * CHECK_UP_DAYS; //in miliseconds
|
||||
const checkupDate = new Date();
|
||||
checkupDate.setTime(checkupDate.getTime() - checkUpOffset);
|
||||
|
||||
const dateQuery = {
|
||||
notifiedAt: {
|
||||
[Op.lte]: checkupDate
|
||||
}
|
||||
};
|
||||
const allRequestsForCheckUp = await db.SearchRequest.findAll({
|
||||
where: dateQuery
|
||||
});
|
||||
|
||||
return allRequestsForCheckUp;
|
||||
};
|
||||
|
||||
const findSearchRequestsForRealEstate = async realEstate => {
|
||||
const {
|
||||
@@ -49,132 +66,416 @@ const findSearchRequestsForRealEstate = async realEstate => {
|
||||
|
||||
const geoSearchQueryPart = sequelize.where(contains, true);
|
||||
|
||||
//General query contains only attributes that are defined for every RealEstate - not null
|
||||
const query = {
|
||||
adType,
|
||||
realEstateType,
|
||||
subscribed: true,
|
||||
[Op.and]: geoSearchQueryPart
|
||||
};
|
||||
//Needed for defining which attribute should exist or not
|
||||
const realEstateTypeObject = AD_CATEGORY[realEstateType];
|
||||
//Needed to decide on including incomplete RealEstates data
|
||||
|
||||
// ?? Needed to decide on including incomplete RealEstates data
|
||||
let checkForIncompleteWanted = false;
|
||||
|
||||
//Attributes are checked separately and included in query only if defined
|
||||
//Price and area should be defined for every property
|
||||
//Attributes are checked separately to make different query parts
|
||||
|
||||
if (price) {
|
||||
query.priceMin = {
|
||||
[Op.lte]: price
|
||||
//If real estate price is number then it searches for req that have priceMin and priceMax
|
||||
//If real estate price is null it searches for req that accept ads without price
|
||||
//User always defines price and area (sliders) - not null in search req
|
||||
let priceQuery = {};
|
||||
if (price != null) {
|
||||
priceQuery = {
|
||||
[Op.and]: [
|
||||
{
|
||||
priceMin: {
|
||||
[Op.lte]: price
|
||||
}
|
||||
},
|
||||
{
|
||||
priceMax: {
|
||||
[Op.gte]: price
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
query.priceMax = {
|
||||
[Op.gte]: price
|
||||
} else {
|
||||
priceQuery = {
|
||||
includeWithoutPrice: {
|
||||
[Op.eq]: true
|
||||
}
|
||||
};
|
||||
}
|
||||
|
||||
if (area) {
|
||||
query.sizeMin = {
|
||||
[Op.lte]: area
|
||||
};
|
||||
query.sizeMax = {
|
||||
[Op.gte]: area
|
||||
let areaQuery = {};
|
||||
if (area != null) {
|
||||
areaQuery = {
|
||||
[Op.and]: [
|
||||
{
|
||||
sizeMin: {
|
||||
[Op.lte]: area
|
||||
}
|
||||
},
|
||||
{
|
||||
sizeMax: {
|
||||
[Op.gte]: area
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
} else {
|
||||
checkForIncompleteWanted = true;
|
||||
}
|
||||
|
||||
//Other attributes can be defined or not depending on RealEstate type
|
||||
if (gardenSize) {
|
||||
query.gardenSizeMin = {
|
||||
[Op.lte]: gardenSize
|
||||
};
|
||||
query.gardenSizeMax = {
|
||||
[Op.gte]: gardenSize
|
||||
};
|
||||
} else if (realEstateTypeObject.hasGardenSize) {
|
||||
checkForIncompleteWanted = true;
|
||||
//we check what to include in query based on real estate type object
|
||||
let gardenSizeQuery = {};
|
||||
if (realEstateTypeObject.hasGardenSize) {
|
||||
if (gardenSize != null) {
|
||||
gardenSizeQuery = {
|
||||
[Op.and]: [
|
||||
{
|
||||
gardenSizeMin: {
|
||||
[Op.lte]: gardenSize
|
||||
}
|
||||
},
|
||||
{
|
||||
gardenSizeMax: {
|
||||
[Op.gte]: gardenSize
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
} else {
|
||||
checkForIncompleteWanted = true;
|
||||
}
|
||||
}
|
||||
|
||||
if (numberOfRooms) {
|
||||
query.numberOfRoomsMin = {
|
||||
[Op.lte]: numberOfRooms
|
||||
};
|
||||
query.numberOfRoomsMax = {
|
||||
[Op.gte]: numberOfRooms
|
||||
};
|
||||
} else if (realEstateTypeObject.hasNumberOfRoom) {
|
||||
checkForIncompleteWanted = true;
|
||||
let numberOfRoomsQuery = {};
|
||||
if (realEstateTypeObject.hasNumberOfRoom) {
|
||||
if (numberOfRooms != null) {
|
||||
//If real estate has defined number of rooms ex. 3 it returns req
|
||||
// that accepts 3 rooms or ones that don't have defined number - null
|
||||
//Ex. they didnt choose advanced filters at all
|
||||
numberOfRoomsQuery = {
|
||||
[Op.and]: [
|
||||
{
|
||||
numberOfRoomsMin: {
|
||||
[Op.or]: {
|
||||
[Op.lte]: numberOfRooms,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
numberOfRoomsMax: {
|
||||
[Op.or]: {
|
||||
[Op.gte]: numberOfRooms,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
} else {
|
||||
// If real estate dont have defined number of rooms ex. null
|
||||
//It returns requests that didn't choose number of rooms - also null
|
||||
//Or ones that picked some values but also picked to includeIncomplete ads (or default)
|
||||
numberOfRoomsQuery = {
|
||||
[Op.or]: [
|
||||
{
|
||||
[Op.and]: [
|
||||
{
|
||||
numberOfRoomsMin: {
|
||||
[Op.is]: null
|
||||
}
|
||||
},
|
||||
{
|
||||
numberOfRoomsMax: {
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
includeIncompleteAds: {
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
}
|
||||
}
|
||||
//Same logic for number of Floors and floors
|
||||
let numberOfFloorsQuery = {};
|
||||
if (realEstateTypeObject.hasNumberOfFloors) {
|
||||
if (numberOfFloors != null) {
|
||||
numberOfFloorsQuery = {
|
||||
[Op.and]: [
|
||||
{
|
||||
numberOfFloorsMin: {
|
||||
[Op.or]: {
|
||||
[Op.lte]: numberOfFloors,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
numberOfFloorsMax: {
|
||||
[Op.or]: {
|
||||
[Op.gte]: numberOfFloors,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
} else {
|
||||
numberOfFloorsQuery = {
|
||||
[Op.or]: [
|
||||
{
|
||||
[Op.and]: [
|
||||
{
|
||||
numberOfFloorsMin: {
|
||||
[Op.is]: null
|
||||
}
|
||||
},
|
||||
{
|
||||
numberOfFloorsMax: {
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
includeIncompleteAds: {
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
}
|
||||
}
|
||||
let floorQuery = {};
|
||||
if (realEstateTypeObject.hasFloorProp) {
|
||||
if (floor != null) {
|
||||
floorQuery = {
|
||||
[Op.and]: [
|
||||
{
|
||||
floorMin: {
|
||||
[Op.or]: {
|
||||
[Op.lte]: floor,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
},
|
||||
{
|
||||
floorMax: {
|
||||
[Op.or]: {
|
||||
[Op.gte]: floor,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
} else {
|
||||
floorQuery = {
|
||||
[Op.or]: [
|
||||
{
|
||||
[Op.and]: [
|
||||
{
|
||||
floorMin: {
|
||||
[Op.is]: null
|
||||
}
|
||||
},
|
||||
{
|
||||
floorMax: {
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
includeIncompleteAds: {
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
};
|
||||
}
|
||||
}
|
||||
|
||||
if (numberOfFloors) {
|
||||
query.numberOfFloorsMin = {
|
||||
[Op.lte]: numberOfFloors
|
||||
};
|
||||
query.numberOfFloorsMax = {
|
||||
[Op.gte]: numberOfFloors
|
||||
};
|
||||
} else if (realEstateTypeObject.hasNumberOfFloors) {
|
||||
checkForIncompleteWanted = true;
|
||||
//Logic for balcony, newBuilding and elevator
|
||||
//If user dont check checkbox for ex. elevator it does not mean he only wants no elevator
|
||||
//If real estate characteristic =true find all req, one that wants charachertistic or dont care - dont need query
|
||||
//If real estate characteristic = false, find all req exept for ones that wants characteristic to be true
|
||||
//If real estate characteristic = null, dont know if true or false, find req that dont care or want char and want incomplete ads
|
||||
let balconyQuery = {};
|
||||
if (realEstateTypeObject.hasBalconyProp && balcony !== true) {
|
||||
if (balcony === false) {
|
||||
balconyQuery = {
|
||||
balcony: {
|
||||
[Op.ne]: true
|
||||
}
|
||||
};
|
||||
} else if (balcony === null) {
|
||||
balconyQuery = {
|
||||
[Op.or]: [
|
||||
{
|
||||
balcony: {
|
||||
[Op.ne]: true
|
||||
}
|
||||
},
|
||||
{
|
||||
[Op.and]: [
|
||||
{
|
||||
balcony: {
|
||||
[Op.eq]: true
|
||||
}
|
||||
},
|
||||
{
|
||||
includeIncompleteAds: {
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
};
|
||||
}
|
||||
}
|
||||
|
||||
if (floor) {
|
||||
query.floorMin = {
|
||||
[Op.lte]: floor
|
||||
};
|
||||
query.floorMax = {
|
||||
[Op.gte]: floor
|
||||
};
|
||||
} else if (realEstateTypeObject.hasFloorProp) {
|
||||
checkForIncompleteWanted = true;
|
||||
let newBuildingQuery = {};
|
||||
if (realEstateTypeObject.hasNewBuildingProp && newBuilding !== true) {
|
||||
if (newBuilding === false) {
|
||||
newBuildingQuery = {
|
||||
newBuilding: {
|
||||
[Op.ne]: true
|
||||
}
|
||||
};
|
||||
} else if (newBuilding === null) {
|
||||
newBuildingQuery = {
|
||||
[Op.or]: [
|
||||
{
|
||||
newBuilding: {
|
||||
[Op.ne]: true
|
||||
}
|
||||
},
|
||||
{
|
||||
[Op.and]: [
|
||||
{
|
||||
newBuilding: {
|
||||
[Op.eq]: true
|
||||
}
|
||||
},
|
||||
{
|
||||
includeIncompleteAds: {
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
};
|
||||
}
|
||||
}
|
||||
let elevatorQuery = {};
|
||||
if (realEstateTypeObject.hasElevatorProp && elevator !== true) {
|
||||
if (elevator === false) {
|
||||
elevatorQuery = {
|
||||
elevator: {
|
||||
[Op.ne]: true
|
||||
}
|
||||
};
|
||||
} else if (elevator === null) {
|
||||
elevatorQuery = {
|
||||
[Op.or]: [
|
||||
{
|
||||
elevator: {
|
||||
[Op.ne]: true
|
||||
}
|
||||
},
|
||||
{
|
||||
[Op.and]: [
|
||||
{
|
||||
elevator: {
|
||||
[Op.eq]: true
|
||||
}
|
||||
},
|
||||
{
|
||||
includeIncompleteAds: {
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
};
|
||||
}
|
||||
}
|
||||
//General query consists of each individual query
|
||||
const query = {
|
||||
adType,
|
||||
realEstateType,
|
||||
subscribed: true,
|
||||
[Op.and]: [
|
||||
geoSearchQueryPart,
|
||||
priceQuery,
|
||||
areaQuery,
|
||||
gardenSizeQuery,
|
||||
numberOfRoomsQuery,
|
||||
numberOfFloorsQuery,
|
||||
floorQuery,
|
||||
balconyQuery,
|
||||
newBuildingQuery,
|
||||
elevatorQuery
|
||||
]
|
||||
};
|
||||
|
||||
if (accessRoadType) {
|
||||
//AccessRoadType is defined - should exists for each ad and estate type
|
||||
if (accessRoadType != null) {
|
||||
query.accessRoadType = {
|
||||
[Op.or]: {
|
||||
[Op.eq]: "ANY",
|
||||
[Op.like]: "ANY",
|
||||
[Op.eq]: accessRoadType
|
||||
}
|
||||
};
|
||||
} else if (realEstateTypeObject.hasAccesRoadType) {
|
||||
checkForIncompleteWanted = true;
|
||||
}
|
||||
|
||||
if (balcony) {
|
||||
query.balcony = {
|
||||
[Op.eq]: balcony
|
||||
} else {
|
||||
//Null values are returned for user request that wanted ANY acces road type
|
||||
query.accessRoadType = {
|
||||
[Op.eq]: "ANY"
|
||||
};
|
||||
} else if (realEstateTypeObject.hasBalconyProp) {
|
||||
checkForIncompleteWanted = true;
|
||||
}
|
||||
|
||||
if (newBuilding) {
|
||||
query.newBuilding = {
|
||||
[Op.eq]: newBuilding
|
||||
};
|
||||
} else if (realEstateTypeObject.hasNewBuildingProp) {
|
||||
checkForIncompleteWanted = true;
|
||||
}
|
||||
|
||||
if (elevator) {
|
||||
query.elevator = {
|
||||
[Op.eq]: elevator
|
||||
};
|
||||
} else if (realEstateTypeObject.hasElevatorProp) {
|
||||
checkForIncompleteWanted = true;
|
||||
}
|
||||
|
||||
//If one of the attributes that exists for property type is null
|
||||
//we include in query to check if incomplete real estates are accepted
|
||||
//Tag to check if incomplete ads are accepted in query which is default
|
||||
if (checkForIncompleteWanted) {
|
||||
query.includeIncompleteAds = {
|
||||
[Op.eq]: true
|
||||
[Op.or]: {
|
||||
[Op.eq]: true,
|
||||
[Op.is]: null
|
||||
}
|
||||
};
|
||||
}
|
||||
return await db.SearchRequest.findAll({ where: query });
|
||||
|
||||
return await db.SearchRequest.findAll({
|
||||
where: query
|
||||
});
|
||||
};
|
||||
|
||||
module.exports = {
|
||||
getSearchRequest,
|
||||
createSearchRequest,
|
||||
findSearchRequestsForRealEstate
|
||||
findSearchRequestsForRealEstate,
|
||||
findAllRequestsForCheckUp
|
||||
};
|
||||
|
||||
@@ -1,5 +1,7 @@
|
||||
"use strict";
|
||||
const db = require("../../models/index");
|
||||
const sequelize = require("sequelize");
|
||||
const Op = sequelize.Op;
|
||||
|
||||
const findRealEstatesForSearchRequest = async searchRequestId => {
|
||||
const query = {
|
||||
|
||||
@@ -1,8 +1,16 @@
|
||||
"use strict";
|
||||
|
||||
const { MAX_REAL_ESTATES_IN_EMAIL, APP_URL } = require("../config/appConfig");
|
||||
const {
|
||||
MAX_REAL_ESTATES_IN_EMAIL,
|
||||
APP_URL,
|
||||
STAGING
|
||||
} = require("../config/appConfig");
|
||||
const { AD_CATEGORY, AD_TYPE, EMAIL_FREQUENCY } = require("../common/enums");
|
||||
|
||||
//Tag to recognize staging from development
|
||||
const stagingTag = STAGING ? "[STAGING] " : "";
|
||||
const wordOfMouthRequest = `Molimo vas <strong>recite svojim prijateljima</strong> za Kivi - što više korisnika budemo imali, moći ćemo više agencija uključiti i više nekretnina imati u bazi. Hvala!`
|
||||
|
||||
const generateEmailFooter = (searchRequestId, emailFrequencyTitle) => {
|
||||
return ` <div>Trenutno ste prijavljeni da obavještenja o novim nekretninama primate <strong>${emailFrequencyTitle.toLowerCase()} </strong>.</div>
|
||||
<div>Ako želite prestati dobijati obavještenja za ovu pretragu, <a href="${APP_URL}/odjava/${searchRequestId}">odjavite ovdje</a></div>
|
||||
@@ -54,7 +62,7 @@ const generateNotificationEmail = (
|
||||
|
||||
const messageBody = dailyNotification ? dailyMessageBody : asapMessageBody;
|
||||
|
||||
return `<h3>Zdravo</h3>
|
||||
return `<h3>${stagingTag}Zdravo</h3>
|
||||
<h4>${messageBody}</h4>
|
||||
<div>
|
||||
${realEstateLinks}
|
||||
@@ -62,6 +70,9 @@ const generateNotificationEmail = (
|
||||
${moreRealEstates}
|
||||
</div>
|
||||
<br/>
|
||||
${wordOfMouthRequest}
|
||||
<br/>
|
||||
<br/>
|
||||
${emailFooter}`;
|
||||
};
|
||||
|
||||
@@ -113,7 +124,7 @@ const generateNewSearchRequestEmail = (searchRequest, matchingRealEstates) => {
|
||||
|
||||
const emailFooter = generateEmailFooter(id, emailFrequencyTitle);
|
||||
|
||||
return `<h3>Zdravo</h3>
|
||||
return `<h3>${stagingTag}Zdravo</h3>
|
||||
<div>Naručili ste da Vam javimo ako se nekretnina sa navedenim uslovima pojavi u oglasima:</div>
|
||||
<br/>
|
||||
<div>
|
||||
@@ -125,12 +136,16 @@ const generateNewSearchRequestEmail = (searchRequest, matchingRealEstates) => {
|
||||
</div>
|
||||
${matchingRealEstates.length > 0 ? instantRealEstatesText : ""}
|
||||
<br/>
|
||||
<br/>
|
||||
${wordOfMouthRequest}
|
||||
<br/>
|
||||
<br/>
|
||||
${emailFooter}`;
|
||||
};
|
||||
|
||||
const generateEmailSubject = (numberOfRealEstates, singleRealEstateTitle) => {
|
||||
if (numberOfRealEstates === 1) {
|
||||
return `Kivi: ${singleRealEstateTitle}`;
|
||||
return `${stagingTag}Kivi: ${singleRealEstateTitle}`;
|
||||
}
|
||||
|
||||
const leastSignificantDigit = numberOfRealEstates % 10;
|
||||
@@ -138,7 +153,7 @@ const generateEmailSubject = (numberOfRealEstates, singleRealEstateTitle) => {
|
||||
const secondLeastSignificantDigit = numberWithoutLastDigit % 10;
|
||||
|
||||
if (leastSignificantDigit === 1 && secondLeastSignificantDigit !== 1) {
|
||||
return `Kivi : ${numberOfRealEstates} nova nekretnina`;
|
||||
return `${stagingTag}Kivi : ${numberOfRealEstates} nova nekretnina`;
|
||||
}
|
||||
|
||||
if (
|
||||
@@ -146,14 +161,58 @@ const generateEmailSubject = (numberOfRealEstates, singleRealEstateTitle) => {
|
||||
leastSignificantDigit <= 4 &&
|
||||
secondLeastSignificantDigit !== 1
|
||||
) {
|
||||
return `Kivi: ${numberOfRealEstates} nove nekretnine`;
|
||||
return `${stagingTag}Kivi: ${numberOfRealEstates} nove nekretnine`;
|
||||
}
|
||||
|
||||
return `Kivi: ${numberOfRealEstates} novih nekretnina`;
|
||||
return `${stagingTag}Kivi: ${numberOfRealEstates} novih nekretnina`;
|
||||
};
|
||||
|
||||
const generateCheckUpEmail = searchRequest => {
|
||||
const realEstateType = AD_CATEGORY[searchRequest.realEstateType];
|
||||
const {
|
||||
id,
|
||||
gardenSizeMin,
|
||||
gardenSizeMax,
|
||||
sizeMin,
|
||||
sizeMax,
|
||||
priceMin,
|
||||
priceMax
|
||||
} = searchRequest;
|
||||
|
||||
let emailFrequencyTitle;
|
||||
switch (searchRequest.emailFrequency) {
|
||||
case EMAIL_FREQUENCY.ASAP.stringId:
|
||||
emailFrequencyTitle = EMAIL_FREQUENCY.ASAP.title;
|
||||
break;
|
||||
case EMAIL_FREQUENCY.DAILY.stringId:
|
||||
emailFrequencyTitle = EMAIL_FREQUENCY.DAILY.title;
|
||||
break;
|
||||
}
|
||||
|
||||
const gardenSize = realEstateType.hasGardenSize
|
||||
? `<div><strong>Kvadratura okućnice: Od ${gardenSizeMin} do ${gardenSizeMax} m2</strong></div>`
|
||||
: ``;
|
||||
|
||||
const emailFooter = generateEmailFooter(id, emailFrequencyTitle);
|
||||
|
||||
return `<h3>${stagingTag}Zdravo</h3>
|
||||
<div><strong>Kivi tim traži nekretnine za Vas i kada to ne vidite.</strong></div>
|
||||
<br />
|
||||
<div>Vaša trenutno aktivna pretraga je:</div>
|
||||
<br/>
|
||||
<div>
|
||||
<div><strong>Tip nekretnine: </strong>${realEstateType.title}</div>
|
||||
<div><strong>Kvadratura nekretnine:</strong> Od ${sizeMin} do ${sizeMax} m2</div>
|
||||
${gardenSize}
|
||||
<div><strong>Cijena:</strong> ${priceMin} do ${priceMax} KM</div>
|
||||
</div>
|
||||
<br/>
|
||||
${emailFooter}`;
|
||||
};
|
||||
|
||||
module.exports = {
|
||||
generateNotificationEmail,
|
||||
generateNewSearchRequestEmail,
|
||||
generateEmailSubject
|
||||
generateEmailSubject,
|
||||
generateCheckUpEmail
|
||||
};
|
||||
|
||||
58
app/helpers/fetchWrapper.js
Normal file
58
app/helpers/fetchWrapper.js
Normal file
@@ -0,0 +1,58 @@
|
||||
const nodeFetch = require("node-fetch");
|
||||
const AbortController = require('abort-controller');
|
||||
const FetchCache = require('@sozialhelden/fetch-cache').default;
|
||||
|
||||
console.log("Fc ", FetchCache)
|
||||
|
||||
const {
|
||||
USER_AGENT,
|
||||
USE_SCRAPER_API,
|
||||
SCRAPER_API_KEY,
|
||||
SCRAPER_API_BASE_URL,
|
||||
NODE_FETCH_TIMEOUT_MS
|
||||
} = require("../config/appConfig");
|
||||
|
||||
const timeout = (ms) => {
|
||||
return new Promise(resolve => setTimeout(resolve, ms));
|
||||
}
|
||||
|
||||
const fetchCache = new FetchCache({
|
||||
fetch: nodeFetch,
|
||||
cacheOptions: {
|
||||
// Don't save more than 100 responses in the cache. Allows infinite responses by default
|
||||
maximalItemCount: 10000,
|
||||
// When should the cache evict responses when its full?
|
||||
evictExceedingItemsBy: 'age', // Valid values: 'lru' or 'age'
|
||||
defaultTTL: 6 * 60 * 60 * 1000 // 6 hours
|
||||
// ...see https://github.com/sozialhelden/hamster-cache for all possible options
|
||||
},
|
||||
});
|
||||
|
||||
|
||||
const fetch = async (url, options = {}, useCache = true) => {
|
||||
const controller = new AbortController();
|
||||
|
||||
const newOptions = Object.assign({}, options);
|
||||
if (!newOptions["headers"]) {
|
||||
newOptions["headers"] = {};
|
||||
}
|
||||
|
||||
newOptions.signal = controller.signal;
|
||||
|
||||
// newOptions["headers"]["User-Agent"] = USER_AGENT;
|
||||
|
||||
let urlToFetchThroughAPI = Buffer.from(url).toString('base64');
|
||||
if (SCRAPER_API_BASE_URL.includes('scraperapi')) {
|
||||
urlToFetchThroughAPI = url;
|
||||
}
|
||||
|
||||
const urlAdaptedForScraping = USE_SCRAPER_API
|
||||
? `${SCRAPER_API_BASE_URL}?api_key=${SCRAPER_API_KEY}&url=${urlToFetchThroughAPI}`
|
||||
: url;
|
||||
const result = useCache ? fetchCache.fetch(urlAdaptedForScraping, newOptions) : nodeFetch(urlAdaptedForScraping, newOptions);
|
||||
|
||||
const timeoutId = setTimeout(() => controller.abort(), NODE_FETCH_TIMEOUT_MS);
|
||||
return result;
|
||||
};
|
||||
|
||||
module.exports = fetch;
|
||||
13
app/helpers/log.js
Normal file
13
app/helpers/log.js
Normal file
@@ -0,0 +1,13 @@
|
||||
const {
|
||||
PRINT_CRAWLER_DEBUG
|
||||
} = require("../config/appConfig");
|
||||
|
||||
const logDebug = (...args) => {
|
||||
if (PRINT_CRAWLER_DEBUG) {
|
||||
console.log(...args);
|
||||
}
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
logDebug
|
||||
};
|
||||
@@ -7,6 +7,26 @@ const currentSearchRequest = async req => {
|
||||
|
||||
return await getSearchRequest(searchRequestId);
|
||||
};
|
||||
module.exports = {
|
||||
currentSearchRequest
|
||||
|
||||
const getUrlParams = function (url) {
|
||||
if (typeof url === 'string' && url.length > 0){
|
||||
const params = {};
|
||||
const questionMarkIndex = url.indexOf('?');
|
||||
if (questionMarkIndex === -1) {
|
||||
return undefined;
|
||||
}
|
||||
const query = url.substring(questionMarkIndex+1);
|
||||
const vars = query.split('&');
|
||||
for (let i = 0; i < vars.length; i++) {
|
||||
const pair = vars[i].split('=');
|
||||
params[pair[0]] = decodeURIComponent(pair[1]);
|
||||
}
|
||||
return params;
|
||||
}
|
||||
return undefined;
|
||||
};
|
||||
|
||||
module.exports = {
|
||||
currentSearchRequest,
|
||||
getUrlParams
|
||||
};
|
||||
|
||||
42
app/migrations/20200121000524-add-priceHistory-table.js
Normal file
42
app/migrations/20200121000524-add-priceHistory-table.js
Normal file
@@ -0,0 +1,42 @@
|
||||
"use strict";
|
||||
|
||||
module.exports = {
|
||||
up: (queryInterface, Sequelize) => {
|
||||
const tableFields = {
|
||||
id: {
|
||||
type: Sequelize.BIGINT,
|
||||
autoIncrement: true,
|
||||
allowNull: false,
|
||||
primaryKey: true
|
||||
},
|
||||
realEstateId: {
|
||||
type: Sequelize.BIGINT,
|
||||
allowNull: false,
|
||||
unique: "uniquePriceRealEstate",
|
||||
references: {
|
||||
model: "RealEstates",
|
||||
key: "id"
|
||||
},
|
||||
onUpdate: "CASCADE",
|
||||
onDelete: "SET NULL"
|
||||
},
|
||||
price: {
|
||||
type: Sequelize.REAL,
|
||||
unique: "uniquePriceRealEstate"
|
||||
},
|
||||
createdAt: {
|
||||
type: Sequelize.DATE,
|
||||
defaultValue: Sequelize.literal("NOW()")
|
||||
},
|
||||
updatedAt: {
|
||||
type: Sequelize.DATE,
|
||||
defaultValue: Sequelize.literal("NOW()")
|
||||
}
|
||||
};
|
||||
return queryInterface.createTable("PriceHistory", tableFields);
|
||||
},
|
||||
|
||||
down: queryInterface => {
|
||||
return queryInterface.dropTable("PriceHistory", {});
|
||||
}
|
||||
};
|
||||
10
app/migrations/20200121094500-add-constraint-priceHistory.js
Normal file
10
app/migrations/20200121094500-add-constraint-priceHistory.js
Normal file
@@ -0,0 +1,10 @@
|
||||
"use strict";
|
||||
module.exports = {
|
||||
up: (queryInterface, Sequelize) =>
|
||||
queryInterface.addConstraint("PriceHistory", ["realEstateId", "price"], {
|
||||
type: "unique",
|
||||
name: "uniquePriceRealEstate"
|
||||
}),
|
||||
down: queryInterface =>
|
||||
queryInterface.removeConstraint("PriceHistory", "uniquePriceRealEstate")
|
||||
};
|
||||
@@ -0,0 +1,14 @@
|
||||
"use strict";
|
||||
|
||||
module.exports = {
|
||||
up: (queryInterface, Sequelize) => {
|
||||
return queryInterface.addColumn("SearchRequests", "includeWithoutPrice", {
|
||||
type: Sequelize.BOOLEAN,
|
||||
defaultValue: true
|
||||
});
|
||||
},
|
||||
|
||||
down: (queryInterface, Sequelize) => {
|
||||
return queryInterface.removeColumn("SearchRequests", "includeWithoutPrice");
|
||||
}
|
||||
};
|
||||
@@ -0,0 +1,14 @@
|
||||
"use strict";
|
||||
|
||||
module.exports = {
|
||||
up: (queryInterface, Sequelize) => {
|
||||
return queryInterface.addColumn("SearchRequests", "notifiedAt", {
|
||||
type: Sequelize.DATE,
|
||||
defaultValue: new Date()
|
||||
});
|
||||
},
|
||||
|
||||
down: (queryInterface, Sequelize) => {
|
||||
return queryInterface.removeColumn("SearchRequests", "notifiedAt");
|
||||
}
|
||||
};
|
||||
@@ -16,7 +16,7 @@ config.logging = parseInt(process.env.SEQUELIZE_LOGGING) ? console.log : false;
|
||||
|
||||
let sequelize;
|
||||
if (config.use_env_variable) {
|
||||
sequelize = new Sequelize(process.env[config.use_env_variable], config);
|
||||
sequelize = new Sequelize(process.env[config.use_env_variable] + "?ssl=true", config);
|
||||
} else {
|
||||
sequelize = new Sequelize(
|
||||
config.database,
|
||||
|
||||
44
app/models/priceHistory.js
Normal file
44
app/models/priceHistory.js
Normal file
@@ -0,0 +1,44 @@
|
||||
"use strict";
|
||||
|
||||
module.exports = (sequalize, DataTypes) => {
|
||||
const PriceHistory = sequalize.define(
|
||||
"PriceHistory",
|
||||
{
|
||||
id: {
|
||||
type: DataTypes.BIGINT,
|
||||
autoIncrement: true,
|
||||
primaryKey: true,
|
||||
allowNull: false
|
||||
},
|
||||
realEstateId: {
|
||||
type: DataTypes.BIGINT,
|
||||
allowNull: false,
|
||||
unique: "uniquePriceRealEstate",
|
||||
references: {
|
||||
model: "RealEstates",
|
||||
key: "id"
|
||||
},
|
||||
onUpdate: "CASCADE",
|
||||
onDelete: "SET NULL"
|
||||
},
|
||||
price: {
|
||||
type: DataTypes.REAL,
|
||||
unique: "uniquePriceRealEstate"
|
||||
}
|
||||
},
|
||||
{
|
||||
freezeTableName: true
|
||||
}
|
||||
);
|
||||
|
||||
PriceHistory.associate = models => {
|
||||
PriceHistory.hasMany(models.RealEstate, {
|
||||
foreignKey: "id",
|
||||
sourceKey: "realEstateId",
|
||||
targetKey: "id",
|
||||
as: "realEstates"
|
||||
});
|
||||
};
|
||||
|
||||
return PriceHistory;
|
||||
};
|
||||
@@ -71,6 +71,7 @@ module.exports = (sequelize, DataTypes) => {
|
||||
type: DataTypes.TEXT
|
||||
},
|
||||
includeIncompleteAds: DataTypes.BOOLEAN,
|
||||
includeWithoutPrice: DataTypes.BOOLEAN,
|
||||
balcony: DataTypes.BOOLEAN,
|
||||
elevator: DataTypes.BOOLEAN,
|
||||
newBuilding: DataTypes.BOOLEAN,
|
||||
@@ -81,7 +82,11 @@ module.exports = (sequelize, DataTypes) => {
|
||||
floorMin: DataTypes.INTEGER,
|
||||
floorMax: DataTypes.INTEGER,
|
||||
accessRoadType: DataTypes.TEXT,
|
||||
heatingType: DataTypes.TEXT
|
||||
heatingType: DataTypes.TEXT,
|
||||
notifiedAt: {
|
||||
type: DataTypes.DATE,
|
||||
defaultValue: new Date()
|
||||
}
|
||||
});
|
||||
|
||||
return SearchRequest;
|
||||
|
||||
6
app/npmScripts/npmCheckUpNotify.js
Normal file
6
app/npmScripts/npmCheckUpNotify.js
Normal file
@@ -0,0 +1,6 @@
|
||||
"use strict";
|
||||
const { checkUpNotify } = require("../services/notificationService");
|
||||
//For testing pursposes
|
||||
(async () => {
|
||||
await checkUpNotify();
|
||||
})();
|
||||
34
app/seeders/20200121150443-initial-price-history.js
Normal file
34
app/seeders/20200121150443-initial-price-history.js
Normal file
@@ -0,0 +1,34 @@
|
||||
"use strict";
|
||||
|
||||
const { RealEstate } = require("../models");
|
||||
|
||||
module.exports = {
|
||||
async up(queryInterface, Sequelize) {
|
||||
//Reading initial data from RealEstate table in db
|
||||
const realEstateInitialData = await RealEstate.findAll();
|
||||
//Extruding data for table PriceHistory
|
||||
const priceHistoryInitialData = realEstateInitialData.map(realEstate => {
|
||||
//Null values canot be recognized by ignore duplicates in sequalize
|
||||
//Value price = 0 indicates 'cijena na upit'
|
||||
const priceTmp =
|
||||
realEstate.dataValues.price === null ? 0 : realEstate.dataValues.price;
|
||||
|
||||
return {
|
||||
realEstateId: realEstate.dataValues.id,
|
||||
price: priceTmp,
|
||||
createdAt: realEstate.dataValues.createdAt,
|
||||
updatedAt: realEstate.dataValues.updatedAt
|
||||
};
|
||||
});
|
||||
|
||||
return queryInterface.bulkInsert(
|
||||
"PriceHistory",
|
||||
priceHistoryInitialData,
|
||||
{}
|
||||
);
|
||||
},
|
||||
|
||||
async down(queryInterface, Sequelize) {
|
||||
return queryInterface.bulkDelete("PriceHistory", null, {});
|
||||
}
|
||||
};
|
||||
@@ -1,4 +1,8 @@
|
||||
"use strict";
|
||||
const { STAGING } = require("../config/appConfig");
|
||||
|
||||
const stagingTag = STAGING ? "[STAGING] " : "";
|
||||
|
||||
const {
|
||||
matchRealEstates,
|
||||
matchSearchRequest
|
||||
@@ -6,12 +10,15 @@ const {
|
||||
const {
|
||||
generateNotificationEmail,
|
||||
generateNewSearchRequestEmail,
|
||||
generateEmailSubject
|
||||
generateEmailSubject,
|
||||
generateCheckUpEmail
|
||||
} = require("../helpers/emailContentGenerator");
|
||||
const {
|
||||
findNotNotifiedMatches,
|
||||
findRealEstatesForSearchRequest
|
||||
} = require("../helpers/db/searchRequestMatch");
|
||||
const { findAllRequestsForCheckUp } = require("../helpers/db/searchRequest");
|
||||
|
||||
const { sendEmail } = require("../services/emailService");
|
||||
|
||||
const notifyForNewRealEstates = async newRealEstates => {
|
||||
@@ -24,13 +31,17 @@ const notifyForNewSearchRequest = async searchRequest => {
|
||||
|
||||
const searchRequestId = searchRequest.id;
|
||||
const matchingRealEstates = matches[searchRequestId].realEstates;
|
||||
|
||||
const emailContent = generateNewSearchRequestEmail(
|
||||
searchRequest,
|
||||
matchingRealEstates
|
||||
);
|
||||
const { email } = searchRequest;
|
||||
await sendEmail(email, "Kivi - novi zahtjev za pretragu", emailContent);
|
||||
//In case of the new search req, notifiedAt column is populated with default value - now (moment of creation)
|
||||
await sendEmail(
|
||||
email,
|
||||
`${stagingTag} Kivi - novi zahtjev za pretragu`,
|
||||
emailContent
|
||||
);
|
||||
};
|
||||
|
||||
const notifyMatches = async (matches, dailyNotification = false) => {
|
||||
@@ -66,6 +77,10 @@ const notifyMatches = async (matches, dailyNotification = false) => {
|
||||
sendEmailPromise.catch(err =>
|
||||
console.log("[Email Sending Failed]", err)
|
||||
);
|
||||
|
||||
//Change time of notified At for searchReq
|
||||
searchRequest.notifiedAt = new Date();
|
||||
searchRequest.save();
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -120,8 +135,30 @@ const notifyRequestsWithDailyOption = async () => {
|
||||
await notifyMatches(matches, true);
|
||||
};
|
||||
|
||||
const checkUpNotify = async () => {
|
||||
/* const searchRequestsForCheckUp = await findAllRequestsForCheckUp();
|
||||
|
||||
const asyncSendEmailActions = [];
|
||||
|
||||
for (const searchRequest of searchRequestsForCheckUp) {
|
||||
const { email } = searchRequest.dataValues;
|
||||
const emailSubject = `${stagingTag}Kivi: Mi tražimo nekretnine za vas!`;
|
||||
const emailContent = generateCheckUpEmail(searchRequest.dataValues);
|
||||
|
||||
const sendEmailPromise = sendEmail(email, emailSubject, emailContent);
|
||||
asyncSendEmailActions.push(sendEmailPromise);
|
||||
sendEmailPromise.catch(err => console.log("[Email Sending Failed]", err));
|
||||
|
||||
//Change time of notified At for searchReq
|
||||
searchRequest.notifiedAt = new Date();
|
||||
searchRequest.save();
|
||||
}
|
||||
await Promise.all(asyncSendEmailActions);*/
|
||||
};
|
||||
|
||||
module.exports = {
|
||||
notifyForNewRealEstates,
|
||||
notifyForNewSearchRequest,
|
||||
notifyRequestsWithDailyOption
|
||||
notifyRequestsWithDailyOption,
|
||||
checkUpNotify
|
||||
};
|
||||
|
||||
@@ -61,9 +61,8 @@
|
||||
<p class="distinguished">
|
||||
<label class="checkbox-label">
|
||||
<input type="checkbox" class="filled-in" name="includeIncompleteAds"
|
||||
<% if (includeIncompleteAds) { %>
|
||||
checked
|
||||
<% } %>>
|
||||
>
|
||||
<span>Uključi i oglase bez potpunih informacija</span>
|
||||
</label>
|
||||
</p>
|
||||
|
||||
@@ -18,6 +18,15 @@
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<br />
|
||||
<p class="distinguished">
|
||||
<label class="checkbox-label">
|
||||
<input type="checkbox" class="filled-in" name="includeWithoutPrice"
|
||||
checked
|
||||
>
|
||||
<span>Uključi i oglase bez cijene</span>
|
||||
</label>
|
||||
</p>
|
||||
<br />
|
||||
|
||||
<div class="row center-align">
|
||||
|
||||
@@ -8,15 +8,25 @@ SEQUELIZE_LOGGING=0- no sequelize logging, 1- log to the console
|
||||
PORT=Port for the app, defaults to 5000
|
||||
APP_BASE_URL=base url for the app
|
||||
|
||||
ENVIRONMENT=Variable to denote development, staging and production
|
||||
|
||||
USER_AGENT=User agent header to send in fetch requests
|
||||
|
||||
MAX_REAL_ESTATES_IN_EMAIL=Max number of real estates that will be shown in email, others will be truncated and URL with full list will be shwon
|
||||
MAX_REAL_ESTATES_IN_FIRST_EMAIL=Max number of real estates that will be shown in first (welcome) email
|
||||
|
||||
CHECK_UP_DAYS=Check up email is sent after this number of days without notification
|
||||
#=============== GOOGLE ANALYTICS =============#
|
||||
GA_ID=Google Analytics ID
|
||||
|
||||
#=============== GOOGLE MAPS =============#
|
||||
API_MAP_KEY=(your-key-here)
|
||||
|
||||
#=============== SCRAPER API SUPORT =============#
|
||||
USE_SCRAPER_API= To turn it on (1) or off (0)
|
||||
SCRAPER_API_KEY= Key for Scraper api
|
||||
SCRAPER_API_BASE_URL= Base url without question mark (example: http://sabur.kivi.ba:1337)
|
||||
|
||||
#=============== AWS SDK EMAIL SETTINGS =======#
|
||||
AWS_KEY_ID=(your-key-here)
|
||||
AWS_SECRET_ACCESS_KEY=(your-key-here)
|
||||
@@ -61,3 +71,9 @@ AKTIDO_CRAWLER_AD_CATEGORIES=comma separated list of enum names of categories to
|
||||
AKTIDO_IGNORED_USERNAMES=!!! This is not used for aktido crawler !!!
|
||||
AKTIDO_DELAY_BETWEEN_PAGES=time in miliseconds to wait before indexing next page
|
||||
AKTIDO_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without stopping when known real estate is found
|
||||
#==SALJIC NEKRETNINE==
|
||||
SALJIC_MAX_PAGES=Restrict crawler to this number of pages
|
||||
SALJIC_MAX_RESULTS_PER_PAGE=For Saljic crawler, this represents how many ads are crawled at once
|
||||
SALJIC_CRAWLER_AD_TYPE=enum name of what type of ads should be crawled, check common/enums.js file for valid values
|
||||
SALJIC_CRAWLER_AD_CATEGORIES=comma separated list of enum names of categories to be included, check common/enums.js file for valid values
|
||||
SALJIC_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without stopping when known real estate is found
|
||||
|
||||
13
index.js
13
index.js
@@ -4,6 +4,7 @@ const bodyParser = require("body-parser");
|
||||
const layout = require("express-layout");
|
||||
const compression = require("compression");
|
||||
const forceSSL = require("./app/helpers/forceSSL");
|
||||
const { logDebug } = require("./app/helpers/log");
|
||||
|
||||
const {
|
||||
APP_PORT,
|
||||
@@ -12,6 +13,7 @@ const {
|
||||
} = require("./app/config/appConfig");
|
||||
const routes = require("./app/routes");
|
||||
const { crawlAll } = require("./app/crawler/crawl");
|
||||
const { checkUpNotify } = require("./app/services/notificationService");
|
||||
const {
|
||||
notifyForNewRealEstates
|
||||
} = require("./app/services/notificationService");
|
||||
@@ -37,12 +39,21 @@ app.listen(APP_PORT, () =>
|
||||
|
||||
let crawlerRunning = STOP_CRAWLER;
|
||||
const crawl = () => {
|
||||
logDebug("Crawl start. crawlerRunning: ", crawlerRunning);
|
||||
if (!crawlerRunning) {
|
||||
crawlerRunning = true;
|
||||
crawlAll().then(newRealEstates => {
|
||||
crawlerRunning = false;
|
||||
logDebug("crawlAll done, new real estate len: ", newRealEstates.length);
|
||||
notifyForNewRealEstates(newRealEstates);
|
||||
}).catch(e => {
|
||||
console.error('Error happened: ', e);
|
||||
}).finally(()=> {
|
||||
crawlerRunning = false;
|
||||
logDebug('Finally done!');
|
||||
});
|
||||
}
|
||||
};
|
||||
|
||||
setInterval(crawl, CRAWLER_INTERVAL * 1000);
|
||||
|
||||
setInterval(checkUpNotify, 1000 * 60 * 60 * 24);
|
||||
|
||||
39
package-lock.json
generated
39
package-lock.json
generated
@@ -40,6 +40,32 @@
|
||||
"@sendgrid/helpers": "^6.3.0"
|
||||
}
|
||||
},
|
||||
"@sozialhelden/fetch-cache": {
|
||||
"version": "2.0.1",
|
||||
"resolved": "https://registry.npmjs.org/@sozialhelden/fetch-cache/-/fetch-cache-2.0.1.tgz",
|
||||
"integrity": "sha512-vMlsdT5JQCGjx1fcFxmMNh7ZKppjjsfUAeZEhhNwhEL7GaqbZXsD1OXEyx2IcRa25ZuZtvJSV6Q3rE77VRdLvg==",
|
||||
"requires": {
|
||||
"@sozialhelden/hamster-cache": "^1.0.0"
|
||||
}
|
||||
},
|
||||
"@sozialhelden/hamster-cache": {
|
||||
"version": "1.0.0",
|
||||
"resolved": "https://registry.npmjs.org/@sozialhelden/hamster-cache/-/hamster-cache-1.0.0.tgz",
|
||||
"integrity": "sha512-/TEGA8mdMawZp4Yq/GrkL+72YL5EGuSeVXC3pKW12YY1t3C+zCN/HZ0HRp4zWF/e67svXcxuz/B0AEQxEdvi7A=="
|
||||
},
|
||||
"@supercharge/goodies": {
|
||||
"version": "1.4.0",
|
||||
"resolved": "https://registry.npmjs.org/@supercharge/goodies/-/goodies-1.4.0.tgz",
|
||||
"integrity": "sha512-Np6u2qjRwiA3wTgzz4n2yduydIjSXqtJWP5cOnNqjdlCR/EUAK86LAOhEcU+YW211D1ksugns3GqpARJDoXQ7g=="
|
||||
},
|
||||
"@supercharge/promise-pool": {
|
||||
"version": "1.3.0",
|
||||
"resolved": "https://registry.npmjs.org/@supercharge/promise-pool/-/promise-pool-1.3.0.tgz",
|
||||
"integrity": "sha512-9/EVrJevSPEqI4i/gRH8Dt7C+FQT65wRRYuu0MDaGmSLZ2aTel0jOGu8Ae84fPiQ+Ah0B80RPFUxk+K+Cz48DA==",
|
||||
"requires": {
|
||||
"@supercharge/goodies": "~1.4.0"
|
||||
}
|
||||
},
|
||||
"@types/caseless": {
|
||||
"version": "0.12.2",
|
||||
"resolved": "https://registry.npmjs.org/@types/caseless/-/caseless-0.12.2.tgz",
|
||||
@@ -79,6 +105,14 @@
|
||||
"resolved": "https://registry.npmjs.org/abbrev/-/abbrev-1.1.1.tgz",
|
||||
"integrity": "sha512-nne9/IiQ/hzIhY6pdDnbBtz7DjPTKrY00P/zvPSm5pOFkl6xuGrGnXn/VtTNNfNtAfZ9/1RtehkszU9qcTii0Q=="
|
||||
},
|
||||
"abort-controller": {
|
||||
"version": "3.0.0",
|
||||
"resolved": "https://registry.npmjs.org/abort-controller/-/abort-controller-3.0.0.tgz",
|
||||
"integrity": "sha512-h8lQ8tacZYnR3vNQTgibj+tODHI5/+l06Au2Pcriv/Gmet0eaj4TwWH41sO9wnHDiQsEj19q0drzdWdeAHtweg==",
|
||||
"requires": {
|
||||
"event-target-shim": "^5.0.0"
|
||||
}
|
||||
},
|
||||
"accepts": {
|
||||
"version": "1.3.5",
|
||||
"resolved": "https://registry.npmjs.org/accepts/-/accepts-1.3.5.tgz",
|
||||
@@ -1074,6 +1108,11 @@
|
||||
"es5-ext": "~0.10.14"
|
||||
}
|
||||
},
|
||||
"event-target-shim": {
|
||||
"version": "5.0.1",
|
||||
"resolved": "https://registry.npmjs.org/event-target-shim/-/event-target-shim-5.0.1.tgz",
|
||||
"integrity": "sha512-i/2XbnSz/uxRCU6+NdVJgKWDTM427+MqYbkQzD321DuCQJUqOuJKIA0IM2+W2xtYHdKOmZ4dR6fExsd4SXL+WQ=="
|
||||
},
|
||||
"events": {
|
||||
"version": "1.1.1",
|
||||
"resolved": "https://registry.npmjs.org/events/-/events-1.1.1.tgz",
|
||||
|
||||
@@ -8,14 +8,17 @@
|
||||
"start": "node ./index.js",
|
||||
"start-mon": "nodemon ./index.js",
|
||||
"migrate": "cd app && npx sequelize db:migrate",
|
||||
"seed": "cd app && npx sequelize db:seed:all",
|
||||
"setup": "docker build -t marketalerts . && docker run -e POSTGRES_USER=docker -e POSTGRES_PASSWORD=docker -e POSTGRES_DB=marketalerts --name pg_marketalerts -d -p 5432:5432 marketalerts && sleep 10 && npm run migrate",
|
||||
"docker-start": "docker start pg_marketalerts",
|
||||
"docker-stop": "docker stop pg_marketalerts",
|
||||
"crawl": "cd app/crawler && node npmCrawl.js",
|
||||
"daily-notify": "cd app/npmScripts && node npmDailyNotify.js",
|
||||
"checkup-notify": "cd app/npmScripts && node npmCheckUpNotify.js",
|
||||
"test-search": "cd test && node searchTest.js",
|
||||
"test-olx-scraper": "cd test && node olxScrapeTest.js",
|
||||
"test-rental-scraper": "cd test && node rentalScrapeTest.js"
|
||||
"test-rental-scraper": "cd test && node rentalScrapeTest.js",
|
||||
"test-saljic-scraper": "cd test && node saljicScrapeTest.js"
|
||||
},
|
||||
"repository": {
|
||||
"type": "git",
|
||||
@@ -29,6 +32,9 @@
|
||||
"dependencies": {
|
||||
"2checkout-node": "0.0.1",
|
||||
"@sendgrid/mail": "^6.3.1",
|
||||
"@sozialhelden/fetch-cache": "^2.0.1",
|
||||
"@supercharge/promise-pool": "^1.3.0",
|
||||
"abort-controller": "^3.0.0",
|
||||
"aws-sdk": "^2.422.0",
|
||||
"bluebird": "^3.5.5",
|
||||
"cheerio": "^1.0.0-rc.2",
|
||||
|
||||
@@ -9,7 +9,7 @@ if (urlToScrape) {
|
||||
|
||||
(async () => {
|
||||
const data = await crawler.scrapeAd(urlToScrape);
|
||||
console.log(data);
|
||||
console.log("Scraped data:", data);
|
||||
})();
|
||||
} else {
|
||||
console.log("No URL to scrape. Use like this : ");
|
||||
|
||||
17
test/saljicScrapeTest.js
Normal file
17
test/saljicScrapeTest.js
Normal file
@@ -0,0 +1,17 @@
|
||||
"use strict";
|
||||
|
||||
const saljicCrawler = require("../app/crawler/specificCrawlers/saljic");
|
||||
|
||||
const urlToScrape = process.argv[2] || undefined;
|
||||
|
||||
if (urlToScrape) {
|
||||
const crawler = new saljicCrawler();
|
||||
|
||||
(async () => {
|
||||
const data = await crawler.scrapeAd(urlToScrape);
|
||||
console.log("Scraped data:", data);
|
||||
})();
|
||||
} else {
|
||||
console.log("No URL to scrape. Use like this : ");
|
||||
console.log("npm run test-saljic-scraper -- URL_TO_SCRAPE");
|
||||
}
|
||||
Reference in New Issue
Block a user