Compare commits

...

94 Commits

Author SHA1 Message Date
Naida Vatric
131536d9fb Olx added preflight check of available concurrent req. 2020-03-06 14:25:02 +01:00
Naida Vatric
824414adad Log for debugging of concurrent req. 2020-03-05 18:05:37 +01:00
Naida Vatric
41c926b5bb Debugging log olx - no normal price. 2020-03-04 23:24:15 +01:00
Naida Vatric
b3708cf842 Fixed olx scraper api bug. 2020-03-04 22:56:06 +01:00
Naida Vatric
f5f8fa276c Created non-concurrent request groups. 2020-03-03 15:21:51 +01:00
Naida Vatric
ccea5fe2aa Changed delay between ads. 2020-03-01 00:44:48 +01:00
Naida Vatric
e1651306eb Changed log for olx. 2020-02-29 19:09:50 +01:00
Naida Vatric
97c09a6da1 Added delay between ads. 2020-02-29 18:52:36 +01:00
Naida Vatric
034106d87a Olx sraper scrape twice. 2020-02-28 22:28:39 +01:00
Naida Vatric
df5e38092d Decomment and saljic smal fix. 2020-02-28 17:12:55 +01:00
Naida Vatric
feb2d04ed6 Olx debugg log. 2020-02-28 15:43:35 +01:00
Naida Vatric
90e171d07d Olx scraper debugging. 2020-02-28 15:41:20 +01:00
Naida Vatric
747f56941a Logged olx price scraping. 2020-02-28 14:21:53 +01:00
Naida Vatric
441f905b29 Olx price parsing changed. 2020-02-27 23:35:18 +01:00
Naida Vatric
edca7f91af Commented checkup out and Prostor fetch changed. 2020-02-27 19:01:11 +01:00
Naida Vatric
44402a9cc4 Merge branch 'saljic-crawler-fix' into 'master'
Saljic crawler changed substring call.

See merge request saburly/marketalarm/web!95
2020-02-26 15:09:22 +00:00
Naida Vatric
b913daa1f7 Merge branch 'checkup-email-bug-fix' into 'master'
Changed checkup email logic.

See merge request saburly/marketalarm/web!97
2020-02-25 15:13:20 +00:00
Naida Vatric
a508f72d7c Merge branch 'master' into 'checkup-email-bug-fix'
# Conflicts:
#   app/services/notificationService.js
2020-02-25 15:12:47 +00:00
Naida Vatric
08ad9edfe1 Merge branch 'scraper-api-support' into 'master'
Added Scraper API option.

See merge request saburly/marketalarm/web!100
2020-02-25 15:10:39 +00:00
Naida Vatric
ce857ddce9 Renamed var. 2020-02-23 23:11:21 +01:00
Naida Vatric
148b2ea863 Changed default. 2020-02-23 16:38:40 +01:00
Naida Vatric
d436d4a37b Added Scraper API option. 2020-02-22 22:15:27 +01:00
Bilal Catic
6791a509d0 make user agent header configurable through env variable 2020-02-20 21:07:16 +01:00
Bilal Catic
edc6e2bbf7 Merge branch 'create-fetch-wrapper-with-user-agent' into 'master'
Create fetch wrapper with user agent

See merge request saburly/marketalarm/web!98
2020-02-20 19:58:32 +00:00
Bilal Catic
4f230020d7 use fetch wrapper instead of node-fetch 2020-02-20 19:49:29 +01:00
Bilal Catic
f62a7200c7 create fetch wrapper with mandatory user agent header 2020-02-20 19:47:30 +01:00
Bilal Catic
cff7cc2e9c apply prettier 2020-02-20 19:46:39 +01:00
Naida Vatric
f56cd5b549 More elegant scrape of lat and long. 2020-02-17 21:55:24 +01:00
Naida Vatric
bc7ce9d708 Changed checkup query to miliseconds. 2020-02-17 21:22:45 +01:00
Naida Vatric
df2a962d0f Merge branch 'prostor-vip-ads-fix' into 'master'
Prostor VIP ads fixed.

See merge request saburly/marketalarm/web!94
2020-02-17 14:44:58 +00:00
Naida Vatric
be4508ebea Merge branch 'include-incomplete-ads-inverse' into 'master'
Default true for include incomplete ads.

See merge request saburly/marketalarm/web!96
2020-02-17 14:44:35 +00:00
Naida Vatric
22bffc126d For staging - checkup fix. 2020-02-15 02:39:38 +01:00
Naida Vatric
06f80296f3 Changed checkup email logic. 2020-02-15 02:30:02 +01:00
Naida Vatric
81fa3f046d Default true for include incomplete ads. 2020-02-15 00:52:06 +01:00
Naida Vatric
addd8c1344 Saljic crawler changed substring call. 2020-02-14 23:42:19 +01:00
Naida Vatric
5bdc8e149a Prostor VIP ads fixed. 2020-02-14 22:41:51 +01:00
Senad Uka
fc7fe3c0b3 Notificaton service disabled 2020-02-14 15:07:42 +01:00
Naida Vatric
b3007123a5 Merge branch 'rename-settings-var' into 'master'
Rename settings var

See merge request saburly/marketalarm/web!93
2020-02-10 20:17:08 +00:00
Naida Vatric
f7d4a9cd07 Renamed settings var to describe purpose. 2020-02-10 21:15:28 +01:00
Naida Vatric
ab6812889a Merge branch 'fixing-saljic-bugs' into 'master'
Fixing saljic bugs

See merge request saburly/marketalarm/web!92
2020-02-09 18:11:00 +00:00
Naida Vatric
b82134e280 Fixed saljic bug for heroku. 2020-02-09 19:09:00 +01:00
Naida Vatric
be378883c8 Just another fix try. 2020-02-08 00:47:00 +01:00
Naida Vatric
8a87b9e253 Another fix. 2020-02-08 00:27:26 +01:00
Naida Vatric
43bc23b164 Another fix. Defined more var. 2020-02-07 22:27:01 +01:00
Naida Vatric
fc6351af46 Added columns and logs for types. 2020-02-07 22:12:53 +01:00
Naida Vatric
6267b2cab4 Merge branch 'staging-tag-to-checkup-email' into 'master'
Added staging tag to checkup email. Email footer bug fixed.

See merge request saburly/marketalarm/web!91
2020-02-06 21:43:56 +00:00
Naida Vatric
97724a47a1 Removed debugg logs. Smal fixes. 2020-02-06 22:40:56 +01:00
Naida Vatric
91a1c6a91e Added more logs for debugging. 2020-02-06 14:12:35 +01:00
Naida Vatric
eb4ab2e341 Changed misspeling. 2020-02-05 23:02:14 +01:00
Naida Vatric
2d0a00b967 Debugging- noOfRealEstates and staging tag. 2020-02-05 21:58:45 +01:00
Naida Vatric
74def9c059 Added staging tag to checkup email. Email footer bug fixed. 2020-02-05 21:35:18 +01:00
Naida Vatric
d29b3eb1b3 Merge branch 'crawler-saljicnekretnine' into 'master'
Crawler saljicnekretnine

See merge request saburly/marketalarm/web!90
2020-02-04 13:09:10 +00:00
Naida Vatric
41b59e8c7c Merge branch 'tag-staging-email' into 'master'
Tag staging email

See merge request saburly/marketalarm/web!85
2020-02-04 13:07:53 +00:00
Naida Vatric
b933fa96d4 Merge branch 'master' into 'tag-staging-email'
# Conflicts:
#   app/config/appConfig.js
2020-02-04 13:07:43 +00:00
Naida Vatric
824db4fbc3 Merge branch 'checkup-email' into 'master'
Added check up email that everything works.

See merge request saburly/marketalarm/web!89
2020-01-31 21:59:35 +00:00
Naida Vatric
0f91841c43 Merge branch 'master' into 'checkup-email'
# Conflicts:
#   app/config/appConfig.js
#   app/services/notificationService.js
2020-01-31 21:59:26 +00:00
Naida Vatric
12b4a8f6ec Merge branch 'ads-without-price' into 'master'
Ads without price

See merge request saburly/marketalarm/web!88
2020-01-31 21:53:36 +00:00
Naida Vatric
17621ad310 Merge branch 'price-history-log' into 'master'
Price history log

See merge request saburly/marketalarm/web!87
2020-01-31 21:53:18 +00:00
Naida Vatric
c461525959 Merge branch 'sliders-formating' into 'master'
Changed sliders for different real estate types.

See merge request saburly/marketalarm/web!86
2020-01-31 21:53:04 +00:00
Naida Vatric
aec9c1e1d5 Merge branch 'more-details-email' into 'master'
More details email

See merge request saburly/marketalarm/web!84
2020-01-31 21:52:33 +00:00
Naida Vatric
2d672f4660 Merge branch 'prostor-vip-ads' into 'master'
Prostor vip ads

See merge request saburly/marketalarm/web!83
2020-01-31 21:52:13 +00:00
Naida Vatric
bc15cf65a5 Merge branch 'results-link' into 'master'
Results link

See merge request saburly/marketalarm/web!82
2020-01-31 21:51:38 +00:00
Naida Vatric
ce11d57ab2 Merge branch 'no-all-results-email' into 'master'
No all results email

See merge request saburly/marketalarm/web!81
2020-01-31 21:51:20 +00:00
Naida Vatric
4a6bcf262e Merge branch 'add-even-more-filters' into 'master'
Add even more filters

See merge request saburly/marketalarm/web!77
2020-01-31 21:50:54 +00:00
Naida Vatric
712cde1632 Changed not to use NODE_ENV variable. 2020-01-31 22:47:47 +01:00
Naida Vatric
1ba7cf8531 Added crawler for Saljic nekretnine. 2020-01-31 22:03:39 +01:00
Naida Vatric
7a7aecb3ee WIP Scraped no of rooms, floors etc. 2020-01-31 00:55:24 +01:00
Naida Vatric
78c4054cde WIP Scraped title, price and location. 2020-01-30 16:24:34 +01:00
Naida Vatric
94ffc2d6d2 WIP Started saljic crawler. 2020-01-29 23:22:39 +01:00
Naida Vatric
b11f18696f Prepared config files. 2020-01-29 01:09:53 +01:00
Naida Vatric
fa46f75dd3 Changed default to switched on. 2020-01-28 22:28:52 +01:00
Naida Vatric
470f53d29b Changed to send email only to some users. 2020-01-24 16:04:28 +01:00
Naida Vatric
40509d2836 Changed logic for checkup.For testing. 2020-01-24 01:01:10 +01:00
Naida Vatric
b2c102bc1a Added check up email that everything works. 2020-01-23 15:48:48 +01:00
Naida Vatric
98263364c7 Added option to include-exclude ads without price 2020-01-23 11:13:53 +01:00
Naida Vatric
5b3491fdba Added migration and model change for searchReq table. 2020-01-23 10:12:56 +01:00
Naida Vatric
d117383802 Tested both ways for realestate and search req filters. 2020-01-17 22:58:22 +01:00
Naida Vatric
870b71a3c7 WIP Changed all logic for searchRequest. 2020-01-17 01:54:06 +01:00
Naida Vatric
e6725355a0 Changed sliders for different realestate types. 2020-01-15 01:33:37 +01:00
Naida Vatric
4fd4018bf6 Merge branch 'sliders-formating' of gitlab.com:saburly/marketalarm/web into add-even-more-filters 2020-01-14 23:19:51 +01:00
Naida Vatric
b9122f8f00 Added tag to email to denote staging from production. 2020-01-14 15:59:17 +01:00
Naida Vatric
fc33c1210a Add more detail to the email 2020-01-13 14:58:09 +01:00
Naida Vatric
511b290096 Login to prostor.ba befoure crawl. 2020-01-13 12:05:33 +01:00
Naida Vatric
ba43fa0713 WIP Changed cookies. 2020-01-13 11:02:26 +01:00
Naida Vatric
e70901d369 WIP Changed login to crawler. 2020-01-13 09:12:03 +01:00
Naida Vatric
8505282670 WiP Login of crawler prostor. 2020-01-12 01:22:50 +01:00
Naida Vatric
64e4835899 Changed redirecting for VIP ads. 2020-01-10 22:52:50 +01:00
Naida Vatric
1658325c4b WIP Fake vip ads. 2020-01-10 19:20:26 +01:00
Naida Vatric
49161c1b60 WIP Changed redirecting for VIP ads. 2020-01-09 12:19:19 +01:00
Naida Vatric
d23ddf849f Results title text made into link. 2020-01-07 01:06:22 +01:00
Naida Vatric
38bd0343f5 Merge branch 'results-link' of gitlab.com:saburly/marketalarm/web into no-all-results-email 2020-01-07 01:01:57 +01:00
Naida Vatric
fa4e0d64de Changed email content to show number of all matching real estates. 2020-01-06 23:59:56 +01:00
Naida Vatric
d5d3a1f306 Changed accesRoadType logic 2019-12-26 23:30:05 +01:00
Naida Vatric
42ff1f762f Changed to avoid falsy values and not defined realestate parametrs. 2019-12-21 02:20:26 +01:00
36 changed files with 2095 additions and 240 deletions

1
.prettierignore Normal file
View File

@@ -0,0 +1 @@
*.ejs

View File

@@ -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
}
@@ -216,14 +295,16 @@ const AD_STATUS = {
STATUS_DELETED: 4,
STATUS_URGENT: 5,
STATUS_DISCOUNTED: 6,
STATUS_RENTED: 7
STATUS_RENTED: 7,
STATUS_VIP: 8
};
const AD_AGENCY = {
OLX: "OLX",
RENTAL: "RENTAL",
PROSTOR: "PROSTOR",
AKTIDO: "AKTIDO"
AKTIDO: "AKTIDO",
SALJIC: "SALJIC"
};
const CRAWLER_AD_TYPE = {

View File

@@ -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: {
@@ -32,6 +36,20 @@ const PRINT_CRAWLER_DEBUG = process.env.PRINT_CRAWLER_DEBUG_INFO || 0;
const API_MAP_KEY = process.env.API_MAP_KEY || "";
const PROSTOR_LOGIN = {
EMAIL: process.env.PROSTOR_LOGIN_EMAIL,
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 || 1; //Default to use
const SCRAPER_API_KEY = process.env.SCRAPER_API_KEY || "";
const NUMBER_OF_CONCURRENT_REQ_SCRAPER_API =
parseInt(process.env.NUMBER_OF_CONCURRENT_REQ_SCRAPER_API) || 10;
module.exports = {
APP_PORT,
APP_URL,
@@ -42,5 +60,12 @@ module.exports = {
MAX_REAL_ESTATES_IN_EMAIL,
MAX_REAL_ESTATES_IN_FIRST_EMAIL,
PRINT_CRAWLER_DEBUG,
API_MAP_KEY
API_MAP_KEY,
STAGING,
CHECK_UP_DAYS,
PROSTOR_LOGIN,
USER_AGENT,
USE_SCRAPER_API,
SCRAPER_API_KEY,
NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
};

View File

@@ -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;

View File

@@ -2,13 +2,14 @@
const {
findRealEstatesForSearchRequest
} = require("../helpers/db/searchRequestMatch");
const { AD_STATUS } = require("../common/enums");
const getRealEstates = async (req, res) => {
const searchRequestId = req.params["searchRequestId"] || "";
const realEstates = await findRealEstatesForSearchRequest(searchRequestId);
const title = "Nekretnine koje odgovaraju Vašim uslovima pretrage";
res.render("realEstates", { realEstates, title });
res.render("realEstates", { realEstates, title, AD_STATUS });
};
module.exports = {

View File

@@ -1,9 +1,11 @@
const { getRealEstateById } = require("../helpers/db/realEstate");
const { AD_STATUS } = require("../common/enums");
const getRedirect = async (req, res) => {
const id = req.params.id || null;
let error = false;
let redirectUrl = undefined;
let vipAd = undefined;
if (!id) {
error = true;
} else {
@@ -13,6 +15,7 @@ const getRedirect = async (req, res) => {
error = true;
} else {
redirectUrl = realEstate.url;
vipAd = realEstate.adStatus === AD_STATUS.STATUS_VIP;
}
} catch (e) {
error = true;
@@ -24,7 +27,7 @@ const getRedirect = async (req, res) => {
res.render("notFound", { title });
} else {
const title = "Preusmjeravanje";
res.render("redirect", { title, redirectUrl });
res.render("redirect", { title, redirectUrl, vipAd });
}
};

View File

@@ -9,12 +9,14 @@ 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 {
OLX_CONFIG,
RENTAL_CONFIG,
PROSTOR_CONFIG,
AKTIDO_CONFIG
AKTIDO_CONFIG,
SALJIC_CONFIG
} = require("./crawlerConfig");
const PostgresSaver = require("./savers/postgres");
@@ -57,6 +59,15 @@ 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
)
];

View File

@@ -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
};

View 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)
};

View File

@@ -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");

View File

@@ -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");
@@ -18,7 +18,9 @@ const {
const {
DEFAULT_TIMEZONE,
PRINT_CRAWLER_DEBUG
PRINT_CRAWLER_DEBUG,
NUMBER_OF_CONCURRENT_REQ_SCRAPER_API,
SCRAPER_API_KEY
} = require("../../config/appConfig");
const OLX_ENUMS = {
@@ -44,6 +46,8 @@ const OLX_ENUMS = {
const { OLX_FORCE_CRAWL } = require("../specificConfigs/olx");
const scraperapiClient = require("scraperapi-sdk")(SCRAPER_API_KEY);
class OlxCrawler {
constructor(
savers = [],
@@ -190,12 +194,40 @@ class OlxCrawler {
let actualNoOfResults =
hrefs.length <= maxResultsPerPage ? hrefs.length : maxResultsPerPage;
const asyncScraping = [];
for (let i = 0; i < actualNoOfResults; i++) {
asyncScraping.push(this.scrapeAd(hrefs[i]));
const scrapedData = [];
for (
let i = 0;
i <= actualNoOfResults;
i = i + NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
) {
const concurrentUrlsToScrape = hrefs.slice(
i,
i + NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
);
//Before it send n req to scraperAPI it send preflight request to check if we have enough concurrent req availabe
//It does not send "real" req until approven internaly
let availableConcurrentReqSlots = false;
do {
availableConcurrentReqSlots = await this.checkAvailableConcurrentReqSlots(
concurrentUrlsToScrape.length
);
} while (availableConcurrentReqSlots !== true);
//
console.log(
`OLX - Sending requests from ${i} to ${i +
NUMBER_OF_CONCURRENT_REQ_SCRAPER_API}.`
);
console.log(`OLX - Urls sent to scrape: `, concurrentUrlsToScrape);
//
const concurrentReqScraperApi = concurrentUrlsToScrape.map(url =>
this.scrapeAd(url)
);
const concurrentReqData = await Promise.all(concurrentReqScraperApi);
concurrentReqData.forEach(reqData => scrapedData.push(reqData));
}
const scrapedData = await Promise.all(asyncScraping);
const filteredScrapedData = scrapedData.filter(adData => !!adData);
return filteredScrapedData;
} catch (e) {
@@ -206,6 +238,7 @@ class OlxCrawler {
async scrapeAd(url) {
// console.log("Scraping : ", url);
try {
const adPageSource = await fetch(url);
const body = await adPageSource.text();
@@ -238,15 +271,28 @@ class OlxCrawler {
//====== PRICE DETECTION AND EXTRACTION =====
let price = null;
const normalPriceValue = $("#pc > p:nth-child(2)").text();
let normalPrice = null;
let urgentPrice = null;
const normalPriceValue = $("#pc > p:nth-child(2)")
.text()
.trim();
const urgentPriceValue = $(
"#artikal_glavni_div > div.artikal_lijevo > div:nth-child(5) > p"
)
.text()
.trim();
//For cases where price is given in discount manner - different from default parsing
const discountPriceValue = $(
"#artikal_glavni_div > div.artikal_lijevo > div.op.pop > p"
)
.text()
.trim();
if (normalPriceValue && normalPriceValue.length > 0) {
price = normalPriceValue;
normalPrice = normalPriceValue
.replace(/\r\n|\n|\r/gm, "")
.replace("KM", "")
.trim();
if (
$("#pc > p.n")
.text()
@@ -256,21 +302,35 @@ class OlxCrawler {
} else {
status = AD_STATUS.STATUS_NORMAL;
}
} else if (urgentPriceValue && urgentPriceValue.length > 0) {
const priceValues = urgentPriceValue.split("KM");
} else if (discountPriceValue && discountPriceValue.length > 0) {
status = AD_STATUS.STATUS_URGENT;
const priceValues = discountPriceValue.split("KM");
normalPrice = priceValues[0].trim();
} else {
console.log("Body:", body);
throw { message: "Can't find normal price" };
}
if (urgentPriceValue && urgentPriceValue.length > 0) {
const priceValues = urgentPriceValue.replace("Cijena", "").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;
if (priceValues.length > 0) {
if (priceValues[0].trim().indexOf("Hitno") != -1) {
urgentPrice = priceValues[0].replace("Hitno", "").trim();
status = AD_STATUS.STATUS_URGENT;
} else {
urgentPrice = priceValues[0].trim();
}
} else if (discountPriceValue && discountPriceValue.length > 0) {
status = AD_STATUS.STATUS_URGENT;
const priceValues = discountPriceValue.split("KM");
urgentPrice = priceValues[1].trim();
} else {
throw { message: "Can't find urgent price" };
}
} else {
throw {
message: "Can't find price (it is not normal nor urgent price ?)"
};
}
price = status === AD_STATUS.STATUS_URGENT ? urgentPrice : normalPrice;
//====== OTHER AD INFORMATION ===============
let adType = null;
let olxId = null;
@@ -278,7 +338,7 @@ 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`
)
@@ -655,6 +715,7 @@ class OlxCrawler {
} catch (e) {
console.error("Exception caught: " + e.message, "\r\nURL:", url);
}
return null;
}
@@ -768,6 +829,9 @@ class OlxCrawler {
if (!priceText) {
return NaN;
}
if (priceText === "Po dogovoru") {
return null;
}
const formattedPriceText = priceText.replace(".", "").replace(",", ".");
return parseFloat(formattedPriceText);
}
@@ -867,8 +931,28 @@ class OlxCrawler {
console.log("sprat = NEPOZNATO [", floorText, "]");
return null;
}
async checkAvailableConcurrentReqSlots(numberOfNeededConcurrentReq) {
try {
const scraperApiAccountInfo = await scraperapiClient.account();
const numberOfUsedConcurrentReq =
scraperApiAccountInfo.concurrentRequests;
const limitOfConcurrentReq = scraperApiAccountInfo.concurrencyLimit;
//Buffer of requests to prevent errors with prefligh requests
const bufferNumberOfReq = 3;
const numberOfAvailableConcurrentReq =
limitOfConcurrentReq - bufferNumberOfReq - numberOfUsedConcurrentReq;
if (numberOfNeededConcurrentReq <= numberOfAvailableConcurrentReq) {
return true;
} else {
return false;
}
} catch (err) {
return false;
}
}
async sleep(ms) {
// console.log("Sleep for:", ms);
return new Promise(resolve => setTimeout(resolve, ms));
}

View File

@@ -1,8 +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,
@@ -16,7 +18,8 @@ const {
const {
PRINT_CRAWLER_DEBUG,
DEFAULT_TIMEZONE
DEFAULT_TIMEZONE,
PROSTOR_LOGIN
} = require("../../config/appConfig");
const { PROSTOR_FORCE_CRAWL } = require("../specificConfigs/prostor");
@@ -60,13 +63,16 @@ class ProstorCrawler {
async crawl() {
const crawlAdCategories = this.crawlerAdCategories;
//We need session cookie to use login privileges
const prostorCookie = await this.getCookies();
//New tag to check if crawler loged in
const login = await this.loginForScraping(PROSTOR_LOGIN, prostorCookie);
const newRealEstates = [];
if (crawlAdCategories) {
//Crawl only if login was successful
if (crawlAdCategories && login) {
const indexGenerators = [];
for (const adCategory of crawlAdCategories) {
indexGenerators.push(this.categoryIndexer(adCategory));
indexGenerators.push(this.categoryIndexer(adCategory, prostorCookie));
}
let done = false;
@@ -119,13 +125,14 @@ class ProstorCrawler {
return newRealEstates;
}
async *categoryIndexer(adCategory) {
async *categoryIndexer(adCategory, prostorCookie) {
const urlAdTypePart = PROSTOR_ENUMS.PROSTOR_AD_TYPE[this.crawlerAdTypes];
const urlCategoryPart = PROSTOR_ENUMS.PROSTOR_AD_CATEGORY[adCategory];
if (urlAdTypePart !== undefined && urlCategoryPart !== undefined) {
const urlPageToCrawl = `${this.baseUrl}?remove_sold=0${urlAdTypePart}${urlCategoryPart}`;
const listOfAllRealEstates = await this.extractRealEstates(
urlPageToCrawl
urlPageToCrawl,
prostorCookie
);
let elementToStartIndexFrom = 0;
@@ -139,7 +146,8 @@ class ProstorCrawler {
elementToStartIndexFrom += realEstatesForSinglePage.length;
const singlePageResults = await this.indexSinglePage(
realEstatesForSinglePage
realEstatesForSinglePage,
prostorCookie
);
const filteredSinglePageResults = singlePageResults.filter(
@@ -163,10 +171,10 @@ class ProstorCrawler {
}
}
async indexSinglePage(realEstatesList) {
async indexSinglePage(realEstatesList, prostorCookie) {
const asyncActions = [];
for (const realEstate of realEstatesList) {
asyncActions.push(this.scrapeAd(realEstate));
asyncActions.push(this.scrapeAd(realEstate, prostorCookie));
}
try {
@@ -180,12 +188,19 @@ class ProstorCrawler {
}
}
async scrapeAd(realEstate) {
async scrapeAd(realEstate, prostorCookie) {
const { lat, lng, property_name, price, size, link, status } = realEstate;
//Status information is given already in realestate list
const adStatus = ProstorCrawler.getStatusId(status);
const url = `https://prostor.ba${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);
@@ -330,7 +345,6 @@ class ProstorCrawler {
furnishingType = FURNISHING_TYPE.NOT_FURNISHED.id;
}
const adStatus = ProstorCrawler.getStatusId(status);
const title = property_name;
const parsedPrice = parseFloat(price.replace(/\./g, "")) || null;
const parsedArea = parseFloat(size);
@@ -408,13 +422,15 @@ class ProstorCrawler {
}
}
async extractRealEstates(url) {
async extractRealEstates(url, prostorCookie) {
if (PRINT_CRAWLER_DEBUG) {
console.log("[PROSTOR] Index page : ", url);
}
try {
const res = await fetch(url);
const res = await nodeFetch(url, {
headers: { Cookie: prostorCookie }
});
const body = await res.text();
const $ = cheerio.load(body);
@@ -548,6 +564,8 @@ class ProstorCrawler {
return AD_STATUS.STATUS_SOLD;
case "Iznajmljeno":
return AD_STATUS.STATUS_RENTED;
case "VIP ponuda":
return AD_STATUS.STATUS_VIP;
default:
console.log("[PROSTOR] Unknown AD_STATUS : [", statusText, "]");
return AD_STATUS.STATUS_NORMAL;
@@ -569,6 +587,54 @@ class ProstorCrawler {
return savers[0].save(results);
//so that we can use some sequelize options and information when data is inserted
}
async loginForScraping(PROSTOR_LOGIN, prostorCookie) {
let formData = new FormData();
formData.append("email", PROSTOR_LOGIN.EMAIL);
formData.append("password", PROSTOR_LOGIN.PASSWORD);
return nodeFetch("https://prostor.ba/moj-prostor/prijava", {
method: "POST",
body: formData,
headers: { Cookie: prostorCookie }
})
.then(page => {
return page.text();
})
.then(resp => {
const $ = cheerio.load(resp);
if (
$("h1")
.text()
.indexOf("Dobrodošli") !== -1
) {
console.log("[PROSTOR]: Crawler loged in!");
return true;
} else {
console.log("[PROSTOR]: Crawler login failed - wrong credentials!");
return false;
}
})
.catch(err => {
console.log("[PROSTOR]: Crawler login error ", err);
});
}
async getCookies() {
const getResponse = await nodeFetch(
"https://prostor.ba/moj-prostor/prijava",
{
headers: { Cookie: "" }
}
);
const raw = getResponse.headers.raw()["set-cookie"];
const cookie = raw
.map(datastring => {
const data = datastring.split(";");
const cookieData = data[0];
return cookieData;
})
.join(";");
return cookie;
}
}
module.exports = ProstorCrawler;

View File

@@ -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");
@@ -399,7 +399,9 @@ class RentalCrawler {
);
if (!publishedDateMoment.isValid()) {
throw {
message: `Invalid published date : ${extractedData["re_realEstates_inserted"]}`
message: `Invalid published date : ${
extractedData["re_realEstates_inserted"]
}`
};
}
@@ -410,7 +412,9 @@ class RentalCrawler {
);
if (!renewedDateMoment.isValid()) {
throw {
message: `Invalid renewed date : ${extractedData["re_realEstates_edited"]}`
message: `Invalid renewed date : ${
extractedData["re_realEstates_edited"]
}`
};
}

View File

@@ -0,0 +1,670 @@
"use strict";
const fetch = require("../../helpers/fetchWrapper");
const cheerio = require("cheerio");
const moment = require("moment-timezone");
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,
NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
} = 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 = 1000
) {
this.savers = savers;
this.baseUrl = "https://www.saljicnekretnine.ba/v2/nekretnine_search";
this.crawlerAdTypes = crawlerAdTypes;
this.crawlerAdCategories = crawlerAdCategories;
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);
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 scrapedData = [];
for (
let i = 0;
i <= actualNoOfResults;
i = i + NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
) {
const concurrentUrlsToScrape = hrefsAbs.slice(
i,
i + NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
);
const concurrentAdTypesOfReq = adTypes.slice(
i,
i + NUMBER_OF_CONCURRENT_REQ_SCRAPER_API
);
const concurrentReqScraperApi = concurrentUrlsToScrape.map(
(url, index) => this.scrapeAd(url, concurrentAdTypesOfReq[index])
);
const concurrentReqData = await Promise.all(concurrentReqScraperApi);
concurrentReqData.forEach(reqData => scrapedData.push(reqData));
}
const filteredScrapedData = scrapedData.filter(adData => !!adData);
return filteredScrapedData;
} catch (e) {
console.error("[SALJIC] Exception caught:" + e);
return [];
}
}
async scrapeAd(url, adTypeAttribute) {
//console.log("[SALJIC] Scraping : ", url);
try {
const adPageSource = await fetch(url);
const body = await adPageSource.text();
const $ = cheerio.load(body);
//Throws error if req to Scraper API proxy wasn't succesful and responds with error
if (body.indexOf("<html>") === -1) {
throw { message: "Scraper API server error." };
}
// 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;
//Extracting main properties
const propertySelectors = {
title:
"div.content-wrap > div.container.clearfix.wpc > div.col-md-8.nobottommargin > div.single-post.nobottommargin > div.entry.clearfix > 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:
"div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-content.topmargin > div.gmap.bottommargin > 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 = $(propertySelectors.latAndLong).attr("src");
let tmpLatLong;
let latText;
let longText;
if (latAndLongSrc && latAndLongSrc.indexOf("openstreetmap") !== -1) {
tmpLatLong = latAndLongSrc.split("marker=")[1];
latText = tmpLatLong.split("%2C")[0];
longText = tmpLatLong.split("%2C")[1];
}
const locationLat = parseFloat(latText) || null;
const locationLong = parseFloat(longText) || null;
//====== DETAIL INFORMATION FIELDS ==========
let area = null,
gardenSize = null,
numberOfRooms = null,
numberOfFloors = null,
floor = null,
accessRoadType = null,
heatingType = null,
furnishingType = null,
balcony = null,
newBuilding = null,
elevator = null,
water = null,
electricity = null,
drainageSystem = null,
registeredInZkBooks = null,
recentlyAdapted = null,
parking = null,
garage = null,
gas = null,
antiTheftDoor = null,
airCondition = null,
phoneConnection = null,
cableTV = null,
internet = null,
basementAttic = null,
storeRoom = null,
videoSurveillance = null,
alarm = null,
suitableForStudents = null,
includingBills = null,
animalsAllowed = null,
pool = null,
exchange = null,
urbanPlanPermit = null,
buildingPermit = null,
utilityConnection = null,
distanceToRiver = null;
let publishedDate = null;
let renewedDate = null;
let realEstateType;
let numberOfViewsAgency = null;
let numberOfViewsKivi = null;
let streetNumber = 0;
let adStatus = status;
let adType = adTypeAttribute;
let shortDescription = descriptions
? descriptions.substring(0, descriptions.indexOf("."))
: "";
let longDescription = descriptions || "";
//Extracting data - Glavne karakteristike
let mainFieldIndex = 1;
do {
const mainFieldSelector = `div.content-wrap > div.container > div.col-md-8.nobottommargin > div.single-post > div.entry > div.entry-content.topmargin > div.col-md-12.bottommargin > ul > li.list-group-item:nth-child(${mainFieldIndex})`;
const mainField = $(mainFieldSelector)
.text()
.replace(/[\n\r\t]/gm, "")
.trim();
const mainFieldTitle = mainField
? mainField.substring(0, mainField.indexOf(" "))
: "";
const mainFieldValue = mainField
? mainField
.substring(mainField.indexOf(" "), mainField.length)
.trim()
: "";
switch (mainFieldTitle) {
case "Površina":
area = parseFloat(
mainFieldValue.substring(0, mainFieldValue.indexOf(" "))
);
break;
case "Okućnica":
gardenSize = parseFloat(
mainFieldValue.substring(0, mainFieldValue.indexOf(" "))
);
break;
case "Broj soba":
numberOfRooms = parseInt(mainFieldValue);
break;
case "Broj spratova":
numberOfFloors = parseInt(mainFieldValue);
break;
case "Sprat":
floor = parseInt(mainFieldValue);
break;
case "Godina renoviranja":
recentlyAdapted = true;
break;
case "Broj parking mjesta":
parking = true;
break;
case "Dostupno od":
const day = mainFieldValue.substring(0, 2);
const month = mainFieldValue.substring(3, 5);
const year = mainFieldValue.substring(6, mainFieldValue.length);
publishedDate = new Date(`${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);
} 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 = "";
//Throws error if realEstateType is null - not read. Still dont know why?
if (realEstateType === null) {
console.log("Body:", body);
throw { message: "Couldn't read real estate type." };
}
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("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;
}
}
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;

View File

@@ -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
});

View File

@@ -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
};

View File

@@ -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 = {

View File

@@ -1,10 +1,18 @@
"use strict";
const { MAX_REAL_ESTATES_IN_EMAIL, APP_URL } = require("../config/appConfig");
const { AD_CATEGORY } = require("../common/enums");
const {
MAX_REAL_ESTATES_IN_EMAIL,
APP_URL,
STAGING
} = require("../config/appConfig");
const { AD_CATEGORY, AD_TYPE, EMAIL_FREQUENCY } = require("../common/enums");
const generateEmailFooter = searchRequestId => {
return `<div>Ako želite prestati dobijati obavještenja za ovu pretragu, <a href="${APP_URL}/odjava/${searchRequestId}">odjavite ovdje</a></div>
//Tag to recognize staging from development
const stagingTag = STAGING ? "[STAGING] " : "";
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/${searchRequestId}">odjavite ovdje</a></div>
<div>Ako želite pogledati ili promijeniti uslove za ovu pretragu, <a href="${APP_URL}/pregled/${searchRequestId}">pogledajte ovdje</a></div>
<br/>
<strong>Vaš,<br/>Kivi tim</strong>`;
@@ -23,17 +31,24 @@ const generateRealEstateLinks = realEstates => {
const generateNotificationEmail = (
realEstates,
searchRequestId,
noAllRealEstates,
dailyNotification = false
) => {
const truncateList = realEstates.length > MAX_REAL_ESTATES_IN_EMAIL;
const realEstatesToShow = truncateList
? realEstates.slice(0, MAX_REAL_ESTATES_IN_EMAIL)
: realEstates;
const allRealEstatesLink = `${APP_URL}/nekretnine/${searchRequestId}`;
const emailFrequencyTitle = dailyNotification
? EMAIL_FREQUENCY.DAILY.title
: EMAIL_FREQUENCY.ASAP.title;
const realEstateLinks = generateRealEstateLinks(realEstatesToShow);
const moreRealEstates = `<div>Kompletan spisak nekretnina možete pogledati na <a href="${allRealEstatesLink}">listi nekretnina</a><div>`;
const emailFooter = generateEmailFooter(searchRequestId);
const moreRealEstates = `<div>Kompletan spisak nekretnina (${noAllRealEstates}) možete pogledati na <a href="${allRealEstatesLink}">listi nekretnina</a><div>`;
const emailFooter = generateEmailFooter(searchRequestId, emailFrequencyTitle);
const asapMessageBody =
realEstates.length > 1
? "Pronašli smo nekretnine koje odgovaraju Vašoj pretrazi"
@@ -46,7 +61,7 @@ const generateNotificationEmail = (
const messageBody = dailyNotification ? dailyMessageBody : asapMessageBody;
return `<h3>Zdravo</h3>
return `<h3>${stagingTag}Zdravo</h3>
<h4>${messageBody}</h4>
<div>
${realEstateLinks}
@@ -59,6 +74,28 @@ const generateNotificationEmail = (
const generateNewSearchRequestEmail = (searchRequest, matchingRealEstates) => {
const realEstateType = AD_CATEGORY[searchRequest.realEstateType];
let adTypeTitle = "";
switch (searchRequest.adType) {
case AD_TYPE.AD_TYPE_SALE.stringId:
adTypeTitle = AD_TYPE.AD_TYPE_SALE.title;
break;
case AD_TYPE.AD_TYPE_RENT.stringId:
adTypeTitle = AD_TYPE.AD_TYPE_RENT.title;
break;
default:
adTypeTitle = "-";
break;
}
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 {
id,
gardenSizeMin,
@@ -70,6 +107,7 @@ const generateNewSearchRequestEmail = (searchRequest, matchingRealEstates) => {
} = searchRequest;
const realEstateLinks = generateRealEstateLinks(matchingRealEstates);
const instantRealEstatesText = `<br/>
<div>
U međuvremenu pogledajte neke od nedavno objavljenih nekretnina koje odgovaraju Vašim uslovima pretrage :<br/>
@@ -80,13 +118,14 @@ const generateNewSearchRequestEmail = (searchRequest, matchingRealEstates) => {
? `<div><strong>Kvadratura okućnice: Od ${gardenSizeMin} do ${gardenSizeMax} m2</strong></div>`
: ``;
const emailFooter = generateEmailFooter(id);
const emailFooter = generateEmailFooter(id, emailFrequencyTitle);
return `<h3>Zdravo</h3>
return `<h3>${stagingTag}Zdravo</h3>
<div>Naručili ste da Vam javimo ako se nekretnina sa navedenim uslovima pojavi u oglasima:</div>
<br/>
<div>
<div><strong>Tip nekretnine: </strong>${realEstateType.title}</div>
<div><strong>Vrsta oglasa: </strong>${adTypeTitle}</div>
<div><strong>Kvadratura nekretnine:</strong> Od ${sizeMin} do ${sizeMax} m2</div>
${gardenSize}
<div><strong>Cijena:</strong> ${priceMin} do ${priceMax} KM</div>
@@ -98,7 +137,7 @@ const generateNewSearchRequestEmail = (searchRequest, matchingRealEstates) => {
const generateEmailSubject = (numberOfRealEstates, singleRealEstateTitle) => {
if (numberOfRealEstates === 1) {
return `Kivi: ${singleRealEstateTitle}`;
return `${stagingTag}Kivi: ${singleRealEstateTitle}`;
}
const leastSignificantDigit = numberOfRealEstates % 10;
@@ -106,7 +145,7 @@ 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 (
@@ -114,14 +153,58 @@ 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
};

View File

@@ -0,0 +1,24 @@
const nodeFetch = require("node-fetch");
const {
USER_AGENT,
USE_SCRAPER_API,
SCRAPER_API_KEY
} = require("../config/appConfig");
const fetch = async (url, options = {}) => {
const newOptions = Object.assign({}, options);
if (!newOptions["headers"]) {
newOptions["headers"] = {};
}
newOptions["headers"]["User-Agent"] = USER_AGENT;
const urlAdaptedForScraping = USE_SCRAPER_API
? `http://api.scraperapi.com/?api_key=${SCRAPER_API_KEY}&url=${url}`
: url;
//
// console.log("Url for scraping:", urlAdaptedForScraping);
return nodeFetch(urlAdaptedForScraping, newOptions);
};
module.exports = fetch;

View File

@@ -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");
}
};

View File

@@ -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");
}
};

View File

@@ -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;

View File

@@ -0,0 +1,6 @@
"use strict";
const { checkUpNotify } = require("../services/notificationService");
//For testing pursposes
(async () => {
await checkUpNotify();
})();

View File

@@ -154,3 +154,7 @@ h3 {
margin-top: 2rem;
margin-bottom: 1rem;
}
.estates-link {
color: rgba(0, 0, 0, 0.87);
}

View File

@@ -1,4 +1,8 @@
"use strict";
const { STAGING } = require("../config/appConfig");
const stagingTag = STAGING ? "[STAGING] " : "";
const {
matchRealEstates,
matchSearchRequest
@@ -6,9 +10,15 @@ const {
const {
generateNotificationEmail,
generateNewSearchRequestEmail,
generateEmailSubject
generateEmailSubject,
generateCheckUpEmail
} = require("../helpers/emailContentGenerator");
const { findNotNotifiedMatches } = require("../helpers/db/searchRequestMatch");
const {
findNotNotifiedMatches,
findRealEstatesForSearchRequest
} = require("../helpers/db/searchRequestMatch");
const { findAllRequestsForCheckUp } = require("../helpers/db/searchRequest");
const { sendEmail } = require("../services/emailService");
const notifyForNewRealEstates = async newRealEstates => {
@@ -21,13 +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,
`${stagingTag} Kivi - novi zahtjev za pretragu`,
emailContent
);
};
const notifyMatches = async (matches, dailyNotification = false) => {
@@ -39,10 +53,18 @@ const notifyMatches = async (matches, dailyNotification = false) => {
const { email, subscribed } = searchRequest;
if (notifyNow && subscribed) {
const allMatchingRealEstates = matches[id].realEstates || [];
//Variable allMatchingRealEstates are real estates that are "new" on the market
//the ones that we notify user in this moment, not all that already exists in db
//New variable allRealEstates are all real estates that exists in db for search req
const allRealEstates = await findRealEstatesForSearchRequest(id);
const noAllRealEstates = allRealEstates.length;
if (allMatchingRealEstates.length > 0) {
const emailContent = generateNotificationEmail(
allMatchingRealEstates,
id,
noAllRealEstates,
dailyNotification
);
const emailSubject = generateEmailSubject(
@@ -55,6 +77,10 @@ const notifyMatches = async (matches, dailyNotification = false) => {
sendEmailPromise.catch(err =>
console.log("[Email Sending Failed]", err)
);
//Change time of notified At for searchReq
searchRequest.notifiedAt = new Date();
searchRequest.save();
}
}
}
@@ -109,8 +135,30 @@ const notifyRequestsWithDailyOption = async () => {
await notifyMatches(matches, true);
};
const checkUpNotify = async () => {
/* const searchRequestsForCheckUp = await findAllRequestsForCheckUp();
const asyncSendEmailActions = [];
for (const searchRequest of searchRequestsForCheckUp) {
const { email } = searchRequest.dataValues;
const emailSubject = `${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
};

View File

@@ -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>

View File

@@ -1,13 +1,29 @@
<div class="row center-align">
<ul class="collection with-header">
<% for(const realEstate of realEstates) { %>
<li class="collection-item">
<div><%= realEstate.title %>
<a href="<%= realEstate.url %>" class="kivi-color secondary-content">
<ul class="collection with-header">
<% for(const realEstate of realEstates) { %>
<li class="collection-item">
<% if(realEstate.adStatus === AD_STATUS.STATUS_VIP) {%>
<div>
<% //This needs to do redirecting instead of direct link to realestate
%>
<a href="/redirect/<%= realEstate.id %>" class="estates-link">
<%= realEstate.title %>
<div class="kivi-color secondary-content">
<i class="material-icons">send</i>
</a>
</div>
</li>
<% } %>
</ul>
</div>
</div>
</a>
</div>
<%} else { %>
<div>
<a href="<%= realEstate.url %>" class="estates-link">
<%= realEstate.title %>
<div class="kivi-color secondary-content">
<i class="material-icons">send</i>
</div>
</a>
</div>
<% }%>
</li>
<% } %>
</ul>
</div>

View File

@@ -1,26 +1,49 @@
<br><br>
<br /><br />
<div class="center">
<div class="preloader-wrapper big active center">
<div class="kivi-spinner-color spinner-layer spinner-green-only">
<div class="circle-clipper left">
<div class="circle"></div>
</div><div class="gap-patch">
<div class="circle"></div>
</div><div class="circle-clipper right">
<div class="circle"></div>
</div>
</div>
<div class="preloader-wrapper big active center">
<div class="kivi-spinner-color spinner-layer spinner-green-only">
<div class="circle-clipper left">
<div class="circle"></div>
</div>
<div class="gap-patch">
<div class="circle"></div>
</div>
<div class="circle-clipper right">
<div class="circle"></div>
</div>
</div>
</div>
</div>
<br>
<br />
<% if(vipAd) { %>
<div class="center">
<h6>
<a href="<%= redirectUrl %>" rel="noreferrer" id="realEstateUrl">Kliknite ovdje ako Vas web preglednik ne preusmjeri automatski</a>
</h6>
<h6>
Ovaj oglas zahtijeva da budete član
<a href="https://prostor.ba/" rel="noreferrer">Prostor.ba</a>.
<br />
<br />
<a href="https://prostor.ba/moj-prostor/prijava" rel="noreferrer"
>Ulogujte se</a
>
ili napravite
<a href="https://prostor.ba/moj-prostor/registracija" rel="noreferrer"
>novi račun</a
>, a potom otvorite <a href="<%= redirectUrl %>" rel="noreferrer">oglas</a>.
</h6>
</div>
<% } else { %>
<div class="center">
<h6>
<a href="<%= redirectUrl %>" rel="noreferrer" id="realEstateUrl"
>Kliknite ovdje ako Vas web preglednik ne preusmjeri automatski</a
>
</h6>
</div>
<% }%>
<script>
window.onload = function() {
document.getElementById('realEstateUrl').click();
}
window.onload = function() {
document.getElementById("realEstateUrl").click();
};
</script>

View File

@@ -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">

View File

@@ -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
NUMBER_OF_CONCURRENT_REQ_SCRAPER_API= Number of requests to send concurrently to Srcaper API proxy
#=============== AWS SDK EMAIL SETTINGS =======#
AWS_KEY_ID=(your-key-here)
AWS_SECRET_ACCESS_KEY=(your-key-here)
@@ -27,6 +37,7 @@ SOURCE_EMAIL=info@saburly.com
CRAWLER_INTERVAL=Interval to run cralwer(s), in seconds
STOP_CRAWLER=Non-zero value will skip crawler execution
PRINT_CRAWLER_DEBUG_INFO=Non-zero value will print crawler debugging info to the server console
#==OLX==
OLX_MAX_PAGES=Restrict crawler to this number of pages
OLX_MAX_RESULTS_PER_PAGE=Only this number or less results from one page will be scraped and saved
@@ -35,6 +46,7 @@ OLX_CRAWLER_AD_CATEGORIES=comma separated list of enum names of categories to be
OLX_IGNORED_USERNAMES=comma separated list of usernames to ignore
OLX_DELAY_BETWEEN_PAGES=time in miliseconds to wait before indexing next page
OLX_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without stopping when known real estate is found
#==RENTAL==
RENTAL_MAX_PAGES=Restrict crawler to this number of pages
RENTAL_MAX_RESULTS_PER_PAGE=Only this number or less results from one page will be scraped and saved
@@ -51,6 +63,8 @@ PROSTOR_CRAWLER_AD_CATEGORIES=comma separated list of enum names of categories t
PROSTOR_IGNORED_USERNAMES=!!! This is not used for prostor crawler !!!
PROSTOR_DELAY_BETWEEN_PAGES=!!! This is not used for prostor crawler !!!
PROSTOR_FORCE_CRAWL=Non-zero value will force crawler to crawl all pages without stopping when known real estate is found
PROSTOR_LOGIN_EMAIL=Email of valid Prostor.ba account for crawling purposes
PROSTOR_LOGIN_PASS=Password of valid Prostor.ba account for crawling purposes
#==AKTIDO==
AKTIDO_MAX_PAGES=Restrict crawler to this number of pages
AKTIDO_MAX_RESULTS_PER_PAGE=Only this number or less results from one page will be scraped and saved
@@ -59,3 +73,8 @@ 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_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

View File

@@ -12,6 +12,7 @@ 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");
@@ -45,4 +46,7 @@ const crawl = () => {
});
}
};
setInterval(crawl, CRAWLER_INTERVAL * 1000);
setInterval(checkUpNotify, 1000 * 60 * 60 * 24);

217
package-lock.json generated
View File

@@ -147,6 +147,14 @@
}
}
},
"argparse": {
"version": "1.0.10",
"resolved": "https://registry.npmjs.org/argparse/-/argparse-1.0.10.tgz",
"integrity": "sha512-o5Roy6tNG4SL/FOkCAN6RzjiakZS25RLYFrcMttJqbdd8BWrnA+fGz57iN5Pb06pvBGvl5gQ0B48dJlslXvoTg==",
"requires": {
"sprintf-js": "~1.0.2"
}
},
"arr-diff": {
"version": "4.0.0",
"resolved": "https://registry.npmjs.org/arr-diff/-/arr-diff-4.0.0.tgz",
@@ -195,6 +203,21 @@
"integrity": "sha1-WWZ/QfrdTyDMvCu5a41Pf3jsA2c=",
"dev": true
},
"async": {
"version": "2.6.3",
"resolved": "https://registry.npmjs.org/async/-/async-2.6.3.tgz",
"integrity": "sha512-zflvls11DCy+dQWzTW2dzuilv8Z5X/pjfmZOWba6TNIVDm+2UDaJmXSOXlasHKfNBs8oo3M0aT50fDEWfKZjXg==",
"requires": {
"lodash": "^4.17.14"
},
"dependencies": {
"lodash": {
"version": "4.17.15",
"resolved": "https://registry.npmjs.org/lodash/-/lodash-4.17.15.tgz",
"integrity": "sha512-8xOcRHvCjnocdS5cpwXQXVzmmh5e5+saE2QGoeQmbKmRS6J3VQppPOIt0MnmE+4xlZoumy0GPG0D0MVIQbNA1A=="
}
}
},
"async-each": {
"version": "1.0.3",
"resolved": "https://registry.npmjs.org/async-each/-/async-each-1.0.3.tgz",
@@ -625,6 +648,11 @@
"resolved": "https://registry.npmjs.org/color-name/-/color-name-1.1.3.tgz",
"integrity": "sha1-p9BVi9icQveV3UIyj3QIMcpTvCU="
},
"colors": {
"version": "1.0.3",
"resolved": "https://registry.npmjs.org/colors/-/colors-1.0.3.tgz",
"integrity": "sha1-BDP0TYCWgP3rYO0mDxsMJi6CpAs="
},
"combined-stream": {
"version": "1.0.7",
"resolved": "https://registry.npmjs.org/combined-stream/-/combined-stream-1.0.7.tgz",
@@ -730,6 +758,25 @@
"resolved": "https://registry.npmjs.org/core-util-is/-/core-util-is-1.0.2.tgz",
"integrity": "sha1-tf1UIgqivFq1eqtxQMlAdUUDwac="
},
"coveralls": {
"version": "3.0.9",
"resolved": "https://registry.npmjs.org/coveralls/-/coveralls-3.0.9.tgz",
"integrity": "sha512-nNBg3B1+4iDox5A5zqHKzUTiwl2ey4k2o0NEcVZYvl+GOSJdKBj4AJGKLv6h3SvWch7tABHePAQOSZWM9E2hMg==",
"requires": {
"js-yaml": "^3.13.1",
"lcov-parse": "^1.0.0",
"log-driver": "^1.2.7",
"minimist": "^1.2.0",
"request": "^2.88.0"
},
"dependencies": {
"minimist": {
"version": "1.2.0",
"resolved": "https://registry.npmjs.org/minimist/-/minimist-1.2.0.tgz",
"integrity": "sha1-o1AIsg9BOD7sH7kU9M1d95omQoQ="
}
}
},
"create-error-class": {
"version": "3.0.2",
"resolved": "https://registry.npmjs.org/create-error-class/-/create-error-class-3.0.2.tgz",
@@ -782,6 +829,11 @@
"resolved": "https://registry.npmjs.org/css-what/-/css-what-2.1.3.tgz",
"integrity": "sha512-a+EPoD+uZiNfh+5fxw2nO9QwFa6nJe2Or35fGY6Ipw1R3R4AGz1d1TEZrCegvw2YTmZ0jXirGYlzxxpYSHwpEg=="
},
"cycle": {
"version": "1.0.3",
"resolved": "https://registry.npmjs.org/cycle/-/cycle-1.0.3.tgz",
"integrity": "sha1-IegLK+hYD5i0aPN5QwZisEbDStI="
},
"d": {
"version": "1.0.1",
"resolved": "https://registry.npmjs.org/d/-/d-1.0.1.tgz",
@@ -1060,6 +1112,11 @@
"prettier-linter-helpers": "^1.0.0"
}
},
"esprima": {
"version": "4.0.1",
"resolved": "https://registry.npmjs.org/esprima/-/esprima-4.0.1.tgz",
"integrity": "sha512-eGuFFw7Upda+g4p+QHvnW0RyTX/SVeJBDM/gCtMARO0cLuT2HcEKnTPvhjV6aGeqrCB/sbNop0Kszm0jsaWU4A=="
},
"etag": {
"version": "1.8.1",
"resolved": "https://registry.npmjs.org/etag/-/etag-1.8.1.tgz",
@@ -1274,6 +1331,11 @@
"resolved": "https://registry.npmjs.org/extsprintf/-/extsprintf-1.3.0.tgz",
"integrity": "sha1-lpGEQOMEGnpBT4xS48V06zw+HgU="
},
"eyes": {
"version": "0.1.8",
"resolved": "https://registry.npmjs.org/eyes/-/eyes-0.1.8.tgz",
"integrity": "sha1-Ys8SAjTGg3hdkCNIqADvPgzCC8A="
},
"fast-deep-equal": {
"version": "2.0.1",
"resolved": "https://registry.npmjs.org/fast-deep-equal/-/fast-deep-equal-2.0.1.tgz",
@@ -1346,13 +1408,23 @@
"integrity": "sha1-+8cfDEGt6zf5bFd60e1C2P2sypE="
},
"form-data": {
"version": "2.3.3",
"resolved": "https://registry.npmjs.org/form-data/-/form-data-2.3.3.tgz",
"integrity": "sha512-1lLKB2Mu3aGP1Q/2eCOx0fNbRMe7XdwktwOruhfqqd0rIJWwN4Dh+E3hrPSlDCXnSR7UtZ1N38rVXm+6+MEhJQ==",
"version": "3.0.0",
"resolved": "https://registry.npmjs.org/form-data/-/form-data-3.0.0.tgz",
"integrity": "sha512-CKMFDglpbMi6PyN+brwB9Q/GOw0eAnsrEZDgcsH5Krhz5Od/haKHAX0NmQfha2zPPz0JpWzA7GJHGSnvCRLWsg==",
"requires": {
"asynckit": "^0.4.0",
"combined-stream": "^1.0.6",
"combined-stream": "^1.0.8",
"mime-types": "^2.1.12"
},
"dependencies": {
"combined-stream": {
"version": "1.0.8",
"resolved": "https://registry.npmjs.org/combined-stream/-/combined-stream-1.0.8.tgz",
"integrity": "sha512-FQN4MRfuJeHf7cBbBMJFXhKSDq+2kAArBlmRBvcvFE5BB1HZKXtSFASDhdlz9zOYwxh8lDdnvmMOe/+5cdoEdg==",
"requires": {
"delayed-stream": "~1.0.0"
}
}
}
},
"forwarded": {
@@ -2465,6 +2537,15 @@
"nopt": "~4.0.1"
}
},
"js-yaml": {
"version": "3.13.1",
"resolved": "https://registry.npmjs.org/js-yaml/-/js-yaml-3.13.1.tgz",
"integrity": "sha512-YfbcO7jXDdyj0DGxYVSlSeQNHbD7XPWvrVWeVUujrQEoZzWJIRrCPoyk6kL6IAjAG2IolMK4T0hNUe0HOUs5Jw==",
"requires": {
"argparse": "^1.0.7",
"esprima": "^4.0.0"
}
},
"jsbn": {
"version": "0.1.1",
"resolved": "https://registry.npmjs.org/jsbn/-/jsbn-0.1.1.tgz",
@@ -2527,6 +2608,11 @@
"invert-kv": "^2.0.0"
}
},
"lcov-parse": {
"version": "1.0.0",
"resolved": "https://registry.npmjs.org/lcov-parse/-/lcov-parse-1.0.0.tgz",
"integrity": "sha1-6w1GtUER68VhrLTECO+TY73I9+A="
},
"locate-path": {
"version": "3.0.0",
"resolved": "https://registry.npmjs.org/locate-path/-/locate-path-3.0.0.tgz",
@@ -2541,6 +2627,11 @@
"resolved": "https://registry.npmjs.org/lodash/-/lodash-4.17.11.tgz",
"integrity": "sha512-cQKh8igo5QUhZ7lg38DYWAxMvjSAKG0A8wGSVimP07SIUEK2UO+arSRKbRZWtelMtN5V0Hkwh5ryOto/SshYIg=="
},
"log-driver": {
"version": "1.2.7",
"resolved": "https://registry.npmjs.org/log-driver/-/log-driver-1.2.7.tgz",
"integrity": "sha512-U7KCmLdqsGHBLeWqYlFA0V0Sl6P08EE1ZrmA9cxjUE0WVqT9qnyVDPz1kzpFEP0jdJuFnasWIfSd7fsaNXkpbg=="
},
"long-timeout": {
"version": "0.1.1",
"resolved": "https://registry.npmjs.org/long-timeout/-/long-timeout-0.1.1.tgz",
@@ -3211,6 +3302,20 @@
"integrity": "sha512-MtEC1TqN0EU5nephaJ4rAtThHtC86dNN9qCuEhtshvpVBkAW5ZO7BASN9REnF9eoXGcRub+pFuKEpOHE+HbEMw==",
"dev": true
},
"promise-request-retry": {
"version": "1.0.2",
"resolved": "https://registry.npmjs.org/promise-request-retry/-/promise-request-retry-1.0.2.tgz",
"integrity": "sha512-zZmu19chRtC6TYeAZaELF8s+Zotl48M6bRnIVjcUrObEjpI4wk+2VpGVRaRgCG6isOqsK4c5IMY7t59Ff2ia0A==",
"requires": {
"async": "^2.6.0",
"bluebird": "^3.5.1",
"coveralls": "^3.0.0",
"req-cwd": "^2.0.0",
"request": "^2.85.0",
"request-promise": "^4.2.2",
"winston": "^2.4.0"
}
},
"proto-list": {
"version": "1.2.4",
"resolved": "https://registry.npmjs.org/proto-list/-/proto-list-1.2.4.tgz",
@@ -3405,6 +3510,22 @@
"integrity": "sha1-jcrkcOHIirwtYA//Sndihtp15jc=",
"dev": true
},
"req-cwd": {
"version": "2.0.0",
"resolved": "https://registry.npmjs.org/req-cwd/-/req-cwd-2.0.0.tgz",
"integrity": "sha1-1AgrTURZgDZkD7c93qAe1T20nrw=",
"requires": {
"req-from": "^2.0.0"
}
},
"req-from": {
"version": "2.0.0",
"resolved": "https://registry.npmjs.org/req-from/-/req-from-2.0.0.tgz",
"integrity": "sha1-10GI5H+TeW9Kpx327jWuaJ8+DnA=",
"requires": {
"resolve-from": "^3.0.0"
}
},
"request": {
"version": "2.88.0",
"resolved": "https://registry.npmjs.org/request/-/request-2.88.0.tgz",
@@ -3430,6 +3551,44 @@
"tough-cookie": "~2.4.3",
"tunnel-agent": "^0.6.0",
"uuid": "^3.3.2"
},
"dependencies": {
"form-data": {
"version": "2.3.3",
"resolved": "https://registry.npmjs.org/form-data/-/form-data-2.3.3.tgz",
"integrity": "sha512-1lLKB2Mu3aGP1Q/2eCOx0fNbRMe7XdwktwOruhfqqd0rIJWwN4Dh+E3hrPSlDCXnSR7UtZ1N38rVXm+6+MEhJQ==",
"requires": {
"asynckit": "^0.4.0",
"combined-stream": "^1.0.6",
"mime-types": "^2.1.12"
}
}
}
},
"request-promise": {
"version": "4.2.5",
"resolved": "https://registry.npmjs.org/request-promise/-/request-promise-4.2.5.tgz",
"integrity": "sha512-ZgnepCykFdmpq86fKGwqntyTiUrHycALuGggpyCZwMvGaZWgxW6yagT0FHkgo5LzYvOaCNvxYwWYIjevSH1EDg==",
"requires": {
"bluebird": "^3.5.0",
"request-promise-core": "1.1.3",
"stealthy-require": "^1.1.1",
"tough-cookie": "^2.3.3"
}
},
"request-promise-core": {
"version": "1.1.3",
"resolved": "https://registry.npmjs.org/request-promise-core/-/request-promise-core-1.1.3.tgz",
"integrity": "sha512-QIs2+ArIGQVp5ZYbWD5ZLCY29D5CfWizP8eWnm8FoGD1TX61veauETVQbrV60662V0oFBkrDOuaBI8XgtuyYAQ==",
"requires": {
"lodash": "^4.17.15"
},
"dependencies": {
"lodash": {
"version": "4.17.15",
"resolved": "https://registry.npmjs.org/lodash/-/lodash-4.17.15.tgz",
"integrity": "sha512-8xOcRHvCjnocdS5cpwXQXVzmmh5e5+saE2QGoeQmbKmRS6J3VQppPOIt0MnmE+4xlZoumy0GPG0D0MVIQbNA1A=="
}
}
},
"require-directory": {
@@ -3450,6 +3609,11 @@
"path-parse": "^1.0.6"
}
},
"resolve-from": {
"version": "3.0.0",
"resolved": "https://registry.npmjs.org/resolve-from/-/resolve-from-3.0.0.tgz",
"integrity": "sha1-six699nWiBvItuZTM17rywoYh0g="
},
"resolve-url": {
"version": "0.2.1",
"resolved": "https://registry.npmjs.org/resolve-url/-/resolve-url-0.2.1.tgz",
@@ -3494,6 +3658,16 @@
"resolved": "https://registry.npmjs.org/sax/-/sax-1.2.1.tgz",
"integrity": "sha1-e45lYZCyKOgaZq6nSEgNgozS03o="
},
"scraperapi-sdk": {
"version": "1.0.3",
"resolved": "https://registry.npmjs.org/scraperapi-sdk/-/scraperapi-sdk-1.0.3.tgz",
"integrity": "sha512-wFzdVptJHAA13HWMxR6DxsesA95cx0eBvylh2CHH9UmzBYor7N54jxgL473IW1VZEferSCNpwlW2R/B3zTPDsQ==",
"requires": {
"promise-request-retry": "^1.0.2",
"request": "^2.88.0",
"request-promise": "^4.2.5"
}
},
"semver": {
"version": "5.6.0",
"resolved": "https://registry.npmjs.org/semver/-/semver-5.6.0.tgz",
@@ -3816,6 +3990,11 @@
"extend-shallow": "^3.0.0"
}
},
"sprintf-js": {
"version": "1.0.3",
"resolved": "https://registry.npmjs.org/sprintf-js/-/sprintf-js-1.0.3.tgz",
"integrity": "sha1-BOaSb2YolTVPPdAVIDYzuFcpfiw="
},
"sshpk": {
"version": "1.16.1",
"resolved": "https://registry.npmjs.org/sshpk/-/sshpk-1.16.1.tgz",
@@ -3832,6 +4011,11 @@
"tweetnacl": "~0.14.0"
}
},
"stack-trace": {
"version": "0.0.10",
"resolved": "https://registry.npmjs.org/stack-trace/-/stack-trace-0.0.10.tgz",
"integrity": "sha1-VHxws0fo0ytOEI6hoqFZ5f3eGcA="
},
"static-extend": {
"version": "0.1.2",
"resolved": "https://registry.npmjs.org/static-extend/-/static-extend-0.1.2.tgz",
@@ -3858,6 +4042,11 @@
"resolved": "https://registry.npmjs.org/statuses/-/statuses-1.4.0.tgz",
"integrity": "sha512-zhSCtt8v2NDrRlPQpCNtw/heZLtfUDqxBM1udqikb/Hbk52LK4nQSwr10u77iopCW5LsyHpuXS0GnEc48mLeew=="
},
"stealthy-require": {
"version": "1.1.1",
"resolved": "https://registry.npmjs.org/stealthy-require/-/stealthy-require-1.1.1.tgz",
"integrity": "sha1-NbCYdbT/SfJqd35QmzCQoyJr8ks="
},
"string-width": {
"version": "2.1.1",
"resolved": "https://registry.npmjs.org/string-width/-/string-width-2.1.1.tgz",
@@ -4329,6 +4518,26 @@
"string-width": "^2.1.1"
}
},
"winston": {
"version": "2.4.4",
"resolved": "https://registry.npmjs.org/winston/-/winston-2.4.4.tgz",
"integrity": "sha512-NBo2Pepn4hK4V01UfcWcDlmiVTs7VTB1h7bgnB0rgP146bYhMxX0ypCz3lBOfNxCO4Zuek7yeT+y/zM1OfMw4Q==",
"requires": {
"async": "~1.0.0",
"colors": "1.0.x",
"cycle": "1.0.x",
"eyes": "0.1.x",
"isstream": "0.1.x",
"stack-trace": "0.0.x"
},
"dependencies": {
"async": {
"version": "1.0.0",
"resolved": "https://registry.npmjs.org/async/-/async-1.0.0.tgz",
"integrity": "sha1-+PwEyjoTeErenhZBr5hXjPvWR6k="
}
}
},
"wkx": {
"version": "0.4.8",
"resolved": "https://registry.npmjs.org/wkx/-/wkx-0.4.8.tgz",

View File

@@ -14,9 +14,12 @@
"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-saljic-scraper": "cd test && node saljicScrapeTest.js",
"test-rental-scraper": "cd test && node rentalScrapeTest.js",
"test-scraper-api": "cd test && node scraperAPITest.js"
},
"repository": {
"type": "git",
@@ -40,6 +43,7 @@
"express": "^4.16.4",
"express-ejs-layouts": "^2.5.0",
"express-layout": "^0.1.0",
"form-data": "^3.0.0",
"html-to-text": "^5.1.1",
"moment": "^2.24.0",
"moment-timezone": "^0.5.26",
@@ -48,6 +52,7 @@
"pg": "^7.10.0",
"prettier": "^1.19.1",
"react-step-wizard": "^5.1.0",
"scraperapi-sdk": "^1.0.3",
"sequelize": "^5.18.4",
"sequelize-cli": "^5.5.0"
},

View File

@@ -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
View 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");
}

19
test/scraperAPITest.js Normal file
View File

@@ -0,0 +1,19 @@
const { SCRAPER_API_KEY } = require("../app/config/appConfig");
const scraperapiClient = require("scraperapi-sdk")(SCRAPER_API_KEY);
async function logUsedConcurrentReq() {
try {
const response = await scraperapiClient.account();
const dateOfLog = new Date().toLocaleString();
console.log(
dateOfLog,
" Number of concurrent requests: ",
response.concurrentRequests
);
} catch (err) {
console.log(err.message);
}
}
setInterval(logUsedConcurrentReq, 1000);