move specific crawler config to the separated files
This commit is contained in:
@@ -5,8 +5,8 @@
|
||||
All environment specific configuration is read here and
|
||||
passed to the crawlers and savers.
|
||||
*/
|
||||
const OlxCrawler = require("./specific/olx");
|
||||
const RentalCrawler = require("./specific/rental");
|
||||
const OlxCrawler = require("./specificCrawlers/olx");
|
||||
const RentalCrawler = require("./specificCrawlers/rental");
|
||||
|
||||
const { OLX_CONFIG, RENTAL_CONFIG } = require("./crawlerConfig");
|
||||
const PostgresSaver = require("./savers/postgres");
|
||||
|
||||
@@ -1,72 +1,8 @@
|
||||
"use strict";
|
||||
require("dotenv").config({ path: __dirname + "/./../../.env" });
|
||||
const { CRAWLER_AD_TYPE, AD_CATEGORY } = require("../common/enums");
|
||||
|
||||
const olxCrawlerAdType =
|
||||
process.env.OLX_CRAWLER_AD_TYPE !== undefined
|
||||
? CRAWLER_AD_TYPE[process.env.OLX_CRAWLER_AD_TYPE]
|
||||
: null;
|
||||
|
||||
const rentalCrawlerAdType =
|
||||
process.env.RENTAL_CRAWLER_AD_TYPE !== undefined
|
||||
? CRAWLER_AD_TYPE[process.env.RENTAL_CRAWLER_AD_TYPE]
|
||||
: null;
|
||||
|
||||
const olxParsedCrawlerAdCategories =
|
||||
process.env.OLX_CRAWLER_AD_CATEGORIES !== undefined
|
||||
? process.env.OLX_CRAWLER_AD_CATEGORIES.split(",").map(category =>
|
||||
category.trim()
|
||||
)
|
||||
: ["FLAT", "HOUSE"];
|
||||
|
||||
const rentalParsedCrawlerAdCategories =
|
||||
process.env.RENTAL_CRAWLER_AD_CATEGORIES !== undefined
|
||||
? process.env.RENTAL_CRAWLER_AD_CATEGORIES.split(",").map(category =>
|
||||
category.trim()
|
||||
)
|
||||
: ["FLAT", "HOUSE"];
|
||||
|
||||
const olxIgnoredUsernames =
|
||||
process.env.OLX_IGNORED_USERNAMES !== undefined
|
||||
? process.env.OLX_IGNORED_USERNAMES.split(",").map(username =>
|
||||
username.trim()
|
||||
)
|
||||
: [];
|
||||
|
||||
const rentalIgnoredUsernames = [];
|
||||
|
||||
const transformedOlxCrawlerAdCategories = olxParsedCrawlerAdCategories
|
||||
.map(categoryName =>
|
||||
AD_CATEGORY[categoryName] ? AD_CATEGORY[categoryName].id : undefined
|
||||
)
|
||||
.filter(category => !!category);
|
||||
|
||||
const transformedRentalCrawlerAdCategories = rentalParsedCrawlerAdCategories
|
||||
.map(categoryName =>
|
||||
AD_CATEGORY[categoryName] ? AD_CATEGORY[categoryName].id : undefined
|
||||
)
|
||||
.filter(category => !!category);
|
||||
|
||||
const OLX_CONFIG = {
|
||||
OLX_MAX_PAGES: parseInt(process.env.OLX_MAX_PAGES) || 500,
|
||||
OLX_MAX_RESULTS_PER_PAGE:
|
||||
parseInt(process.env.OLX_MAX_RESULTS_PER_PAGE) || 50,
|
||||
OLX_CRAWLER_AD_TYPE: olxCrawlerAdType || CRAWLER_AD_TYPE.NONE,
|
||||
OLX_CRAWLER_AD_CATEGORIES: transformedOlxCrawlerAdCategories,
|
||||
OLX_IGNORED_USERNAMES: olxIgnoredUsernames || [],
|
||||
OLX_DELAY_BETWEEN_PAGES: parseInt(process.env.OLX_DELAY_BETWEEN_PAGES) || 1000
|
||||
};
|
||||
|
||||
const RENTAL_CONFIG = {
|
||||
RENTAL_MAX_PAGES: parseInt(process.env.RENTAL_MAX_PAGES) || 500,
|
||||
RENTAL_MAX_RESULTS_PER_PAGE:
|
||||
parseInt(process.env.RENTAL_MAX_RESULTS_PER_PAGE) || 50,
|
||||
RENTAL_CRAWLER_AD_TYPE: rentalCrawlerAdType || CRAWLER_AD_TYPE.NONE,
|
||||
RENTAL_CRAWLER_AD_CATEGORIES: transformedRentalCrawlerAdCategories,
|
||||
RENTAL_IGNORED_USERNAMES: rentalIgnoredUsernames || [],
|
||||
RENTAL_DELAY_BETWEEN_PAGES:
|
||||
parseInt(process.env.RENTAL_DELAY_BETWEEN_PAGES) || 1000
|
||||
};
|
||||
const OLX_CONFIG = require("./specificConfigs/olx");
|
||||
const RENTAL_CONFIG = require("./specificConfigs/rental");
|
||||
|
||||
module.exports = {
|
||||
OLX_CONFIG,
|
||||
|
||||
37
app/crawler/specificConfigs/olx.js
Normal file
37
app/crawler/specificConfigs/olx.js
Normal file
@@ -0,0 +1,37 @@
|
||||
"use strict";
|
||||
const { CRAWLER_AD_TYPE, AD_CATEGORY } = require("../../common/enums");
|
||||
|
||||
const olxCrawlerAdType =
|
||||
process.env.OLX_CRAWLER_AD_TYPE !== undefined
|
||||
? CRAWLER_AD_TYPE[process.env.OLX_CRAWLER_AD_TYPE]
|
||||
: null;
|
||||
|
||||
const olxParsedCrawlerAdCategories =
|
||||
process.env.OLX_CRAWLER_AD_CATEGORIES !== undefined
|
||||
? process.env.OLX_CRAWLER_AD_CATEGORIES.split(",").map(category =>
|
||||
category.trim()
|
||||
)
|
||||
: ["FLAT", "HOUSE"];
|
||||
|
||||
const olxIgnoredUsernames =
|
||||
process.env.OLX_IGNORED_USERNAMES !== undefined
|
||||
? process.env.OLX_IGNORED_USERNAMES.split(",").map(username =>
|
||||
username.trim()
|
||||
)
|
||||
: [];
|
||||
|
||||
const transformedOlxCrawlerAdCategories = olxParsedCrawlerAdCategories
|
||||
.map(categoryName =>
|
||||
AD_CATEGORY[categoryName] ? AD_CATEGORY[categoryName].id : undefined
|
||||
)
|
||||
.filter(category => !!category);
|
||||
|
||||
module.exports = {
|
||||
OLX_MAX_PAGES: parseInt(process.env.OLX_MAX_PAGES) || 500,
|
||||
OLX_MAX_RESULTS_PER_PAGE:
|
||||
parseInt(process.env.OLX_MAX_RESULTS_PER_PAGE) || 50,
|
||||
OLX_CRAWLER_AD_TYPE: olxCrawlerAdType || CRAWLER_AD_TYPE.NONE,
|
||||
OLX_CRAWLER_AD_CATEGORIES: transformedOlxCrawlerAdCategories,
|
||||
OLX_IGNORED_USERNAMES: olxIgnoredUsernames || [],
|
||||
OLX_DELAY_BETWEEN_PAGES: parseInt(process.env.OLX_DELAY_BETWEEN_PAGES) || 1000
|
||||
};
|
||||
33
app/crawler/specificConfigs/rental.js
Normal file
33
app/crawler/specificConfigs/rental.js
Normal file
@@ -0,0 +1,33 @@
|
||||
"use strict";
|
||||
const { CRAWLER_AD_TYPE, AD_CATEGORY } = require("../../common/enums");
|
||||
|
||||
const rentalCrawlerAdType =
|
||||
process.env.RENTAL_CRAWLER_AD_TYPE !== undefined
|
||||
? CRAWLER_AD_TYPE[process.env.RENTAL_CRAWLER_AD_TYPE]
|
||||
: null;
|
||||
|
||||
const rentalParsedCrawlerAdCategories =
|
||||
process.env.RENTAL_CRAWLER_AD_CATEGORIES !== undefined
|
||||
? process.env.RENTAL_CRAWLER_AD_CATEGORIES.split(",").map(category =>
|
||||
category.trim()
|
||||
)
|
||||
: ["FLAT", "HOUSE"];
|
||||
|
||||
const rentalIgnoredUsernames = [];
|
||||
|
||||
const transformedRentalCrawlerAdCategories = rentalParsedCrawlerAdCategories
|
||||
.map(categoryName =>
|
||||
AD_CATEGORY[categoryName] ? AD_CATEGORY[categoryName].id : undefined
|
||||
)
|
||||
.filter(category => !!category);
|
||||
|
||||
module.exports = {
|
||||
RENTAL_MAX_PAGES: parseInt(process.env.RENTAL_MAX_PAGES) || 500,
|
||||
RENTAL_MAX_RESULTS_PER_PAGE:
|
||||
parseInt(process.env.RENTAL_MAX_RESULTS_PER_PAGE) || 50,
|
||||
RENTAL_CRAWLER_AD_TYPE: rentalCrawlerAdType || CRAWLER_AD_TYPE.NONE,
|
||||
RENTAL_CRAWLER_AD_CATEGORIES: transformedRentalCrawlerAdCategories,
|
||||
RENTAL_IGNORED_USERNAMES: rentalIgnoredUsernames || [],
|
||||
RENTAL_DELAY_BETWEEN_PAGES:
|
||||
parseInt(process.env.RENTAL_DELAY_BETWEEN_PAGES) || 1000
|
||||
};
|
||||
@@ -1,6 +1,6 @@
|
||||
"use strict";
|
||||
|
||||
const olxCrawler = require("../app/crawler/specific/olx");
|
||||
const olxCrawler = require("../app/crawler/specificCrawlers/olx");
|
||||
|
||||
const urlToScrape = process.argv[2] || undefined;
|
||||
|
||||
|
||||
Reference in New Issue
Block a user