forked from KEMT/zpwiki
		
	
		
			
				
	
	
		
			143 lines
		
	
	
		
			4.0 KiB
		
	
	
	
		
			JavaScript
		
	
	
	
	
	
			
		
		
	
	
			143 lines
		
	
	
		
			4.0 KiB
		
	
	
	
		
			JavaScript
		
	
	
	
	
	
//Jan Holp, DP 2021
 | 
						|
 | 
						|
 | 
						|
//client2 = cassandra
 | 
						|
//client1  = elasticsearch 
 | 
						|
//-----------------------------------------------------------------
 | 
						|
 | 
						|
//require for PageRank 
 | 
						|
var  Pagerank = require('../lib/pagerank')
 | 
						|
var fs = require('fs')
 | 
						|
var path = require('path')
 | 
						|
 | 
						|
//require the Elasticsearch librray
 | 
						|
const elasticsearch = require('elasticsearch');
 | 
						|
const client1 = new elasticsearch.Client({
 | 
						|
   hosts: [ 'localhost:9200']
 | 
						|
});
 | 
						|
client1.ping({
 | 
						|
     requestTimeout: 30000,
 | 
						|
 }, function(error) {
 | 
						|
 // at this point, eastic search is down, please check your Elasticsearch service
 | 
						|
     if (error) {
 | 
						|
         console.error('Elasticsearch cluster is down!');
 | 
						|
     } else {
 | 
						|
         console.log('ELasticSearch is ok');
 | 
						|
     }
 | 
						|
 });
 | 
						|
 | 
						|
//create new index - skweb2
 | 
						|
client1.indices.create({
 | 
						|
    index: 'skweb2'
 | 
						|
}, function(error, response, status) {
 | 
						|
    if (error) {
 | 
						|
        console.log(error);
 | 
						|
    } else {
 | 
						|
        console.log("created a new index", response);
 | 
						|
    }
 | 
						|
});
 | 
						|
 | 
						|
//indexing method
 | 
						|
const bulkIndex = function bulkIndex(index, type, data) {
 | 
						|
	let bulkBody = [];
 | 
						|
	//id = 1;
 | 
						|
const errorCount = 0;
 | 
						|
	data.forEach(item => {
 | 
						|
		bulkBody.push({
 | 
						|
			index: {
 | 
						|
				_index: index,
 | 
						|
				_type:  type,
 | 
						|
				_id :   item.target_link,  // documents id is url
 | 
						|
			}
 | 
						|
		});
 | 
						|
		bulkBody.push(item);
 | 
						|
	});
 | 
						|
        console.log(bulkBody);
 | 
						|
        //console.log(object_list.id);
 | 
						|
 | 
						|
	client1.bulk({body: bulkBody})
 | 
						|
		.then(response => {
 | 
						|
 | 
						|
			response.items.forEach(item => {
 | 
						|
				if (item.index && item.index.error) {
 | 
						|
					console.log(++errorCount, item.index.error);
 | 
						|
				}
 | 
						|
			});
 | 
						|
			console.log(
 | 
						|
				`Successfully indexed ${data.length - errorCount}
 | 
						|
				out of ${data.length} items`
 | 
						|
			);
 | 
						|
		})
 | 
						|
		.catch(console.err);
 | 
						|
};
 | 
						|
 | 
						|
 | 
						|
const cassandra = require('cassandra-driver');
 | 
						|
const client2 = new cassandra.Client({ contactPoints: ['localhost:9042'], localDataCenter: 'datacenter1', keyspace: 'websucker' });
 | 
						|
const query1 = 'SELECT domain_name FROM websucker.domain_quality WHERE good_count > 0 ALLOW FILTERING';
 | 
						|
//const query2 = 'SELECT * from websucker.content WHERE domain_name = ' + domain_name[i] + 'ALLOW FILTERING'; // body_size > 0
 | 
						|
 | 
						|
//-------------------------------------------------------------------------
 | 
						|
 | 
						|
var domain_name = [];         // pole domain name 
 | 
						|
var object_list = [];         // pole ktore obsahuje vsetky dokumenty pre jednotilive domain name
 | 
						|
const linkProb = 0.85;    // high numbers are more stable (Pagerank)
 | 
						|
const tolerance = 0.0001;  // sensitivity for accuracy of convergence
 | 
						|
 | 
						|
 | 
						|
client2.execute(query1)                             // vyselektujeme vsetky domenove mena a ulozime do pola
 | 
						|
  .then(result => {
 | 
						|
  		let pole = result.rows.map(r => {
 | 
						|
  			domain_name.push(r.domain_name)
 | 
						|
  		});
 | 
						|
  		console.log("Vsetky domenove mena : " , domain_name);
 | 
						|
 	domain_name.forEach(name => {										// pre kazde domenove meno spustime select nizsie, kt. vyberie vsetky clanky ktore niesu prazdne
 | 
						|
			let query = 'SELECT * from websucker.content WHERE domain_name = ' + "'" + name + "'" + ' and body_size > 0 ALLOW FILTERING';
 | 
						|
			client2.execute(query).then( res => {
 | 
						|
				object_list = res.rows.map(rr => {
 | 
						|
				return {
 | 
						|
					 domain_name: rr.domain_name,
 | 
						|
					 title: rr.title,
 | 
						|
					 body: rr.body,
 | 
						|
					 links: rr.links,
 | 
						|
					 target_link: rr.target_link,
 | 
						|
					// pagerank: Pagerank(rr.links ,linkProb,tolerance, function (err, res) {
 | 
						|
					// 	return res;
 | 
						|
					// })
 | 
						|
				}
 | 
						|
 | 
						|
				
 | 
						|
 | 
						|
			});
 | 
						|
			//console.log(object_list);
 | 
						|
			bulkIndex('skweb2', 'web_page', object_list);
 | 
						|
			}).catch(error => console.log(error));
 | 
						|
})
 | 
						|
 | 
						|
}).catch(err => console.log(err));
 | 
						|
 | 
						|
//volanie funkcie pre vypocet Pageranku a definovane premenne
 | 
						|
 | 
						|
//Larger numbers (0.85) //var linkProb = 0.85;    
 | 
						|
//accuracy at which we terminate 
 | 
						|
//--------------Pagerank 
 | 
						|
//	const linkProb = 0.85;    // high numbers are more stable (Pagerank)
 | 
						|
//	const tolerance = 0.0001;  // sensitivity for accuracy of convergence
 | 
						|
   /*
 | 
						|
	var nodeMatrix = [
 | 
						|
        [object_list.links]
 | 
						|
    ];
 | 
						|
*/
 | 
						|
/*
 | 
						|
 | 
						|
const PR = function PR(nodeMatrix,linkProb,tolerance){
 | 
						|
 | 
						|
Pagerank(nodeMatrix, linkProb, tolerance, function (err, res) {
 | 
						|
 | 
						|
 | 
						|
    return res;
 | 
						|
    //console.log(res);
 | 
						|
           
 | 
						|
});
 | 
						|
}
 | 
						|
*/ |