Adding new Admin UI. Updating Fredy to V3.0.0 as it has been a large rewrite. Thanks for all contributions and help on the way!
This commit is contained in:
Christian Kellner
2021-01-21 16:09:23 +01:00
committed by GitHub
parent 8185bfe818
commit b2847f6834
124 changed files with 9768 additions and 1495 deletions

View File

@@ -20,7 +20,6 @@ function applyBlacklist(o) {
}
const config = {
enabled: null,
url: null,
crawlContainer: '.tabelle',
crawlFields: {
@@ -29,20 +28,23 @@ const config = {
size: '.tabelle .inner_object_data .data_boxes div:nth-child(1)',
rooms: '.tabelle .inner_object_data .data_boxes div:nth-child(2)',
title: '.tabelle .inner_object_data .tabelle_inhalt_titel_black | removeNewline | trim',
description: '.tabelle .inner_object_data .objekt_beschreibung | removeNewline | trim'
description: '.tabelle .inner_object_data .objekt_beschreibung | removeNewline | trim',
},
paginate: '.pagination_blocks div:last a@href',
normalize: normalize,
filter: applyBlacklist
filter: applyBlacklist,
};
exports.init = (sourceConfig, blacklist) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
appliedBlackList = blacklist || [];
};
//must match the id of the source given in the config!
exports.id = () => 'einsAImmobilien';
exports.metaInformation = {
name: '1a Immobilien',
baseUrl: 'https://www.1a-immobilienmarkt.de/',
id: __filename.slice(__dirname.length + 1, -3),
};
exports.config = config;

View File

@@ -20,7 +20,6 @@ function applyBlacklist(o) {
}
const config = {
enabled: null,
url: null,
crawlContainer: '#result-list-stage .item',
crawlFields: {
@@ -29,20 +28,23 @@ const config = {
size: 'div[id*="selArea_"] | trim',
title: '.item a img@title',
link: 'a[id*="lnkImgToDetails_"]@href',
address: '.item .box-25 .ellipsis .text-100 | removeNewline | trim'
address: '.item .box-25 .ellipsis .text-100 | removeNewline | trim',
},
paginate: '#idResultList .margin-bottom-6.margin-bottom-sm-12 .panel a.pull-right@href',
normalize: normalize,
filter: applyBlacklist
filter: applyBlacklist,
};
exports.init = (sourceConfig, blacklist) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
appliedBlackList = blacklist || [];
};
//must match the id of the source given in the config!
exports.id = () => 'immonet';
exports.metaInformation = {
name: 'Immonet',
baseUrl: 'https://www.immonet.de/',
id: __filename.slice(__dirname.length + 1, -3),
};
exports.config = config;

View File

@@ -1,42 +0,0 @@
const utils = require('../utils');
let appliedBlackList = [];
function normalize(o) {
const title = o.title.replace('NEU', '');
const address = (o.address || '').replace(/\(.*\),.*$/, '').trim();
return Object.assign(o, { title, address });
}
function applyBlacklist(o) {
return !utils.isOneOf(o.title, appliedBlackList);
}
const config = {
enabled: null,
url: null,
crawlContainer: '#resultListItems li.result-list__listing',
crawlFields: {
id: '.result-list-entry@data-obid | int',
price: '.result-list-entry .result-list-entry__criteria .grid-item:first-child dd | removeNewline | trim',
size: '.result-list-entry .result-list-entry__criteria .grid-item:nth-child(2) dd | removeNewline | trim',
title: '.result-list-entry .result-list-entry__brand-title-container h5 | removeNewline | trim',
link: '.result-list-entry .result-list-entry__brand-title-container@href',
address: '.result-list-entry .result-list-entry__map-link'
},
paginate: '#pager .align-right a@href',
normalize: normalize,
filter: applyBlacklist
};
exports.init = (sourceConfig, blacklist) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
};
//must match the id of the source given in the config!
exports.id = () => 'immoscout';
exports.config = config;

View File

@@ -17,7 +17,6 @@ function applyBlacklist(o) {
}
const config = {
enabled: null,
url: null,
crawlContainer: '.immoliste .js-object.listitem_wrap ',
crawlFields: {
@@ -26,20 +25,22 @@ const config = {
size: '.js-object.listitem_wrap .hardfacts_3 div:nth-child(2)| removeNewline | trim',
title: '.listcontent.clear h2',
link: 'a@href',
address: '.listcontent .details .listlocation| removeNewline | trim'
address: '.listcontent .details .listlocation| removeNewline | trim',
},
paginate: '#pnlPaging #nlbPlus@href',
normalize: normalize,
filter: applyBlacklist
filter: applyBlacklist,
};
exports.init = (sourceConfig, blacklist) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
appliedBlackList = blacklist || [];
};
exports.metaInformation = {
name: 'Immowelt',
baseUrl: 'https://www.immowelt.de/',
id: __filename.slice(__dirname.length + 1, -3),
};
//must match the id of the source given in the config!
exports.id = () => 'immowelt';
exports.config = config;

View File

@@ -1,56 +0,0 @@
const utils = require('../utils');
let appliedBlackList = [];
let appliedBlacklistedDistricts = [];
function normalize(o) {
const id = o.id
.split('/')
.filter(Boolean)
.reverse()[0];
const price = o.price == null ? 'unknown' : o.price.trim().replace('Preis', '');
let size = o.size == null ? 'unknown' : o.size.replace('Wohnfläche: ', '').replace('ca. ', '');
size += ' / ' + o.rooms;
const address = '---';
return Object.assign(o, { id, price, size, address });
}
function applyBlacklist(o) {
const titleNotBlacklisted = !utils.isOneOf(o.title, appliedBlackList);
const descNotBlacklisted = !utils.isOneOf(o.description, appliedBlackList);
const isBlacklistedDistrict =
appliedBlacklistedDistricts.length === 0 ? false : utils.isOneOf(o.title, appliedBlacklistedDistricts);
return !isBlacklistedDistrict && titleNotBlacklisted && descNotBlacklisted;
}
const config = {
enabled: null,
url: null,
crawlContainer: '#resultList .resultitem-content-container',
crawlFields: {
id: '.resultitem-content-container a@href',
price: '.description .rent | removeNewline | trim',
title: '.resultitem-content-container a@title',
link: '.resultitem-content-container a@href',
rooms: '.resultitem-content-container .no-of-rooms | removeNewline | trim',
size: '.resultitem-content-container .living-area | removeNewline | trim'
},
paginate: '.markt_pagination_pageLinkNext .markt_pagination_link@href',
normalize: normalize,
filter: applyBlacklist
};
exports.init = (sourceConfig, blacklist, blacklistedDistricts) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
appliedBlacklistedDistricts = blacklistedDistricts;
};
//must match the id of the source given in the config!
exports.id = () => 'kalaydo';
exports.config = config;

View File

@@ -13,13 +13,12 @@ function applyBlacklist(o) {
const titleNotBlacklisted = !utils.isOneOf(o.title, appliedBlackList);
const descNotBlacklisted = !utils.isOneOf(o.description, appliedBlackList);
const isBlacklistedDistrict =
appliedBlacklistedDistricts.length === 0 ? false : utils.isOneOf(o.description, appliedBlacklistedDistricts);
appliedBlacklistedDistricts.length === 0 ? false : utils.isOneOf(o.description, appliedBlacklistedDistricts);
return !isBlacklistedDistrict && titleNotBlacklisted && descNotBlacklisted;
}
const config = {
enabled: null,
url: null,
crawlContainer: '#srchrslt-adtable .ad-listitem',
crawlFields: {
@@ -29,22 +28,24 @@ const config = {
title: '.aditem-main .text-module-begin a | removeNewline | trim',
link: '.aditem-main .text-module-begin a@href | removeNewline | trim',
description: '.aditem-main p:not(.text-module-end) | removeNewline | trim',
address: '.aditem-details | trim | removeNewline'
address: '.aditem-details | trim | removeNewline',
},
paginate: '#srchrslt-pagination .pagination-next@href',
normalize: normalize,
filter: applyBlacklist
filter: applyBlacklist,
};
exports.metaInformation = {
name: 'Ebay Kleinanzeigen',
baseUrl: 'https://www.ebay-kleinanzeigen.de/',
id: __filename.slice(__dirname.length + 1, -3),
};
exports.init = (sourceConfig, blacklist, blacklistedDistricts) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlacklistedDistricts = blacklistedDistricts;
appliedBlackList = blacklist;
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlacklistedDistricts = blacklistedDistricts || [];
appliedBlackList = blacklist || [];
};
//must match the id of the source given in the config!
exports.id = () => 'kleinanzeigen';
exports.config = config;

View File

@@ -11,7 +11,6 @@ function applyBlacklist(o) {
}
const config = {
enabled: null,
url: null,
crawlContainer: '.nbk-container >div article',
crawlFields: {
@@ -28,10 +27,13 @@ const config = {
exports.init = (sourceConfig, blacklist) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
appliedBlackList = blacklist || [];
};
//must match the id of the source given in the config!
exports.id = () => 'neubauKompass';
exports.metaInformation = {
name: 'Neubau Kompass',
baseUrl: 'https://www.neubaukompass.de/',
id: __filename.slice(__dirname.length + 1, -3),
};
exports.config = config;

View File

@@ -14,7 +14,6 @@ function applyBlacklist(o) {
}
const config = {
enabled: null,
url: null,
crawlContainer: '#main_column .wgg_card',
crawlFields: {
@@ -23,20 +22,23 @@ const config = {
price: '.middle .col-xs-3 |removeNewline |trim',
size: '.middle .text-right |removeNewline |trim',
title: '.truncate_title a |removeNewline |trim',
link: '.truncate_title a@href'
link: '.truncate_title a@href',
},
paginate: '.pagination-sm:first a:last@href',
normalize: normalize,
filter: applyBlacklist
filter: applyBlacklist,
};
exports.init = (sourceConfig, blacklist) => {
config.enabled = sourceConfig.enabled;
config.url = sourceConfig.url;
appliedBlackList = blacklist;
appliedBlackList = blacklist || [];
};
//must match the id of the source given in the config!
exports.id = () => 'wgGesucht';
exports.metaInformation = {
name: 'Wg gesucht',
baseUrl: 'https://www.wg-gesucht.de/',
id: __filename.slice(__dirname.length + 1, -3),
};
exports.config = config;