[Explore Service | Services]:

1. extractUrlsFromSearch.ts: changed results parsing (used landing parsing instead of searh) to call method "checkIfAllowed()" (no index filtering) | Added more numbers for statistics.
2. package.json: Upgrade node version.
This commit is contained in:
Konstantina Galouni 2021-08-06 14:19:58 +03:00
parent 9f48768fe3
commit f5cb4c2acd
2 changed files with 124 additions and 9 deletions

View File

@ -7,6 +7,9 @@ import {ResultPreview} from "../../explore/src/app/openaireLibrary/utils/result-
import {Identifier} from "../../explore/src/app/openaireLibrary/utils/string-utils.class";
import {SearchFields} from "../../explore/src/app/openaireLibrary/utils/properties/searchFields";
import {ContextsService} from "../../explore/src/app/openaireLibrary/claims/claim-utils/service/contexts.service";
import {ResultLandingComponent} from "../../explore/src/app/openaireLibrary/landingPages/result/resultLanding.component";
import {ResultLandingService} from "../../explore/src/app/openaireLibrary/landingPages/result/resultLanding.service";
import {ResultLandingInfo} from "../../explore/src/app/openaireLibrary/utils/entities/resultLandingInfo";
const request = require('superagent');
@ -27,6 +30,7 @@ function get(resultsPerUrl) {
let promiseArray = [];
for (let key of keys) {
// comment out for communities query check
if(key == "community") {
await communitiesPromise;
}
@ -48,7 +52,8 @@ function get(resultsPerUrl) {
valueId = idArray[0];
}
}
if(!valueId || !publicCommunities.includes(valueId)) {
// if(!valueId) {
if(!valueId || !publicCommunities.includes(valueId)) { // comment out for communities query check
console.log("hidden community: "+valueId);
continue;
}
@ -64,7 +69,7 @@ function get(resultsPerUrl) {
reject(err);
} else {
parseAllUrls(response, allUrls);
resolve();
resolve(value => value);
}
})
}).catch(error => {
@ -79,6 +84,9 @@ function get(resultsPerUrl) {
await Promise.all(promiseArray);
console.log("\nDuplicate urls: "+alreadyin + " vs unique urls: "+notin);
console.log("\nNo indexed urls: "+noIndexedUrls + " vs final urls: "+finalUrls);
console.log("\nPublications: "+publications + " - Datasets: "+datasets +
" - Software: "+software + " - Other: "+other + " --- urls with pid: "+urlsWithPid);
fs.appendFile("./" + fileName, "\n</urlset>", function (err) {
if (err) {
@ -92,7 +100,7 @@ function get(resultsPerUrl) {
}
// });
function parseAllUrls(response: any, allUrls: any) {
function parseAllUrls_old(response: any, allUrls: any) {
// let allUrls: any = [];
let responses: any = response.body['results'];
@ -123,7 +131,7 @@ function parseAllUrls(response: any, allUrls: any) {
}
if(searchResults.length == 0) {
fs.appendFileSync("./"+errorFileName, response.statusCode+" "+response.request.url+"/n");
fs.appendFileSync("./"+errorFileName, response.statusCode+" "+response.request.url+"\n");
}
for(let j=0; j<searchResults.length; j++) {
let resultPreview: any = ResultPreview.searchResultConvert(searchResults[j], searchResults[j].entityType);
@ -149,6 +157,93 @@ function parseAllUrls(response: any, allUrls: any) {
}
function parseAllUrls(response: any, allUrls: any) {
// let allUrls: any = [];
let responses: any = response.body['results'];
let length = responses ? (Array.isArray(responses) ? responses.length : 1) : 0;
if(length < 100 && length > 0) {
console.log("num of results: "+length + " " + response.request.url);
}
if(length == 0) {
fs.appendFileSync("./"+errorFileName, response.statusCode+" "+response.request.url+"\n");
}
for (let i = 0; i < length; i++) {
let curResponse = Array.isArray(responses) ? responses[i] : responses;
let resData = curResponse['result']['metadata']['oaf:entity'];
let param = [resData['oaf:result'], resData['oaf:result']['title'], resData['oaf:result']['rels']['rel'],
resData['oaf:result']['children'], resData['oaf:result']['pid'], resData['oaf:result']['journal'],
resData['oaf:result']['language'], resData['oaf:result']['subject'], resData['oaf:result']['context'],
resData['oaf:result']['creator'], resData['oaf:result']['country'], resData['oaf:result']['programmingLanguage'],
(resData['extraInfo'] !== undefined && resData['extraInfo']['citations'] !== undefined)
? resData['extraInfo']['citations']['citation'] : null,
curResponse['result']['header']['dri:status'], curResponse];
let resultLandingService = new ResultLandingService(null);
let resultLandingInfo: ResultLandingInfo = resultLandingService.parseResultLandingInfo(param, null, properties);
let resultPreview: any = ResultPreview.resultLandingInfoConvert(resultLandingInfo, resultLandingInfo.resultType);
let pid: any = Identifier.getResultPIDFromIdentifiers(resultPreview.identifiers);
let url;
if(pid && pid.id) {
url = getUrlByType(resultPreview.resultType, pid, encodeURIComponent(pid.id));
} else {
url = getUrlByType(resultPreview.resultType, null, resultLandingInfo.relcanId);
}
if(allUrls.has(url)) {
alreadyin++;
} else {
let resultLandingComponent = new ResultLandingComponent(null, null, null,
null, null, null, null, null, null,
null, null, null, null, null);
if(!resultLandingComponent.checkIfAllowed(resultLandingInfo)) {
noIndexedUrls++;
fs.appendFileSync("./"+noIndexFileName, url+"\n");
} else {
finalUrls++;
let urlPre = "\n<url><loc>";
let urlSuf = "</loc></url>";
fs.appendFileSync("./"+fileName, urlPre + url + urlSuf);
}
allUrls.add(url);
notin++;
}
// .pipe(map(res => [res['result']['header']['dri:status'], res['result']['metadata']['oaf:entity'], res]))
// .pipe(map(res => [
// res[1]['oaf:result'], // 0
// res[1]['oaf:result']['title'], // 1
// res[1]['oaf:result']['rels']['rel'], // 2
// res[1]['oaf:result']['children'], // 3
// res[1]['oaf:result']['pid'], // 4
// res[1]['oaf:result']['journal'], // 5
// res[1]['oaf:result']['language'], // 6
// res[1]['oaf:result']['subject'], // 7
// res[1]['oaf:result']['context'], // 8
// res[1]['oaf:result']['creator'], // 9
// res[1]['oaf:result']['country'] , // 10
// res[1]['oaf:result']['programmingLanguage'], // 11 - software
// //res[1]['oaf:result']['resulttype'],
// (res[1]['extraInfo'] !== undefined && res[1]['extraInfo']['citations'] !== undefined)
// ? res[1]['extraInfo']['citations']['citation'] : null, // 12
// res[0], // 13
// res[2] // 14
// ]))
// .pipe(map(res => this.parseResultLandingInfo(res, provenanceActionVocabulary, properties)));
}
return allUrls;
}
// function parseAllUrls1(response) {
// let allUrls = [];
//
@ -186,22 +281,28 @@ function getUrlByType(type: any, pid: any, id: any) {
let parameter: any = "";
if (type === "publication") {
publications++;
parameter = "articleId";
} else if (type === "dataset") {
datasets++;
parameter = "datasetId";
} else if (type === "software") {
software++;
parameter = "softwareId";
} else if (type === "other") {
other++;
parameter = "orpId";
} else {
parameter = "id";
}
if(pid) {
urlsWithPid++;
parameter = "pid";
}
return landingPrefix+type+"?"+parameter+"="+id;
}
// comment out for communities query check
function getCommunities() {
communitiesPromise = new Promise((resolve, reject) => {
request.get(contextUrl, async function (err: any, communitiesResponse: any) {
@ -210,7 +311,7 @@ function getCommunities() {
} else {
const contextsService = new ContextsService();
publicCommunities = contextsService.parseCommunities(communitiesResponse.body, false).map(value => value.id);
resolve();
resolve(value => value);
}
})
}).catch(error => console.error("Error getting communities ", error));
@ -222,6 +323,7 @@ function buildSiteMap(resultsPerUrl) {
let date = new Date();
fileName = "sitemap_"+date.getFullYear()+"_"+(date.getMonth()+1)+"_"+date.getDate()+".xml";//+"_"+date.getTime();
errorFileName = "error_"+date.getFullYear()+"_"+(date.getMonth()+1)+"_"+date.getDate()+".txt";//+"_"+date.getTime();
noIndexFileName = "noIndex_"+date.getFullYear()+"_"+(date.getMonth()+1)+"_"+date.getDate()+".txt";//+"_"+date.getTime();
console.log("Buiding sitemap in file: "+fileName+"\n");
let sitemap = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" +
@ -233,24 +335,37 @@ function buildSiteMap(resultsPerUrl) {
}
});
getCommunities();
getCommunities(); // comment out for communities query check
get(resultsPerUrl);
}
let fileName;
let errorFileName;
let noIndexFileName;
const fs = require('fs');
let alreadyin = 0; // duplicate urls
let notin= 0;
let finalUrls = 0;
let noIndexedUrls = 0;
let publications = 0;
let datasets = 0;
let software = 0;
let other = 0;
let urlsWithPid = 0;
// comment out for communities query check
let communitiesPromise;
let publicCommunities = [];
const refineUrl = "https://services.openaire.eu/search/v2/api/resources2/?format=json&refine=true&fields=resultbestaccessright&fields=relfunder&fields=instancetypename&fields=resultlanguagename&fields=community&fields=resulthostingdatasource&fields=country&type=results&page=0&size=0";
const resultsUrlPrefix = "https://services.openaire.eu/search/v2/api/resources2/?format=json";
const landingPrefix = "https://explore.openaire.eu/search/";
// const refineUrl = "https://services.openaire.eu/search/v2/api/resources2/?format=json&refine=true&fields=resultbestaccessright&fields=relfunder&fields=instancetypename&fields=resultlanguagename&fields=resulthostingdatasource&fields=country&type=results&page=0&size=0";
// comment out for communities query check
const refineUrl = "https://services.openaire.eu/search/v2/api/resources2/?format=json&refine=true&fields=resultbestaccessright&fields=relfunder&fields=instancetypename&fields=resultlanguagename&fields=community&fields=resulthostingdatasource&fields=country&type=results&page=0&size=0";
const contextUrl = "https://services.openaire.eu/openaire/contexts/";
buildSiteMap(100);
buildSiteMap(150);

View File

@ -14,7 +14,7 @@
"@types/node": "^8.0.30"
},
"engines": {
"node": "8.1.x"
"node": "16.3.0"
},
"author": "Konstantina Galouni <kgalouni@di.uoa.gr>",
"license": "NKUA"