Algorithm selection

git-svn-id: http://svn.research-infrastructures.eu/public/d4science/gcube/trunk/data-analysis/nlphub@165247 82a268e6-3cf1-43bd-a215-b396298e98cf
This commit is contained in:
Erik Perrone 2018-03-19 15:31:23 +00:00
parent e2729e357b
commit 43a834505d
9 changed files with 267 additions and 80 deletions

View File

@ -65,6 +65,7 @@ public class NLPHub extends HttpServlet {
RunnerCommander commander = new RunnerCommander(algs, request.getParameter("plink"), request.getParameter("annotations"), token,
response);
commander.setSleepTime(100);
commander.setMaxWaitingTime(90*1000);
commander.startProcess();
/*
if (algs.length >= 1) {

View File

@ -79,26 +79,15 @@ public class NLPMapper extends HttpServlet {
}
String annotations = request.getParameter("annotations");
//System.out.println("annotations\n: " + annotations);
String language = request.getParameter("lang");
PrintWriter writer = response.getWriter();
//System.out.println("language\n: " + language);
//System.out.println("tokens length: " + tokens.length);
// for(int u=0; u<tokens.length; u++) {
// System.out.println("[" + u + "] " + tokens[u]);
// }
try {
String text = getDocument(documentLink);
text = text.replaceAll("\n", "\\\\n");;
text = text.replaceAll("\r", "\\\\r");
text = text.replaceAll("\t", "\\\\t");
text = text.replaceAll("\"", "\\\\\"");
// byte[] b = text.getBytes();
// System.out.println(Constants.hexDump(b));
String out = "{";
out += "\"text\":\"" + text + "\",";
@ -109,11 +98,12 @@ public class NLPMapper extends HttpServlet {
for (int i = 0; i < tokens.length; i++) {
String token = tokens[i];
String[] t = token.split(":::");
//System.out.println(t[0]);
//System.out.println(t[1]);
if(t[1].equals(Constants.ERROR_ID)) {
logger.error("Algorithm " + t[0] + " in error. Bypass...");
continue;
}
try {
String json = ((JsonMapper) (getMapper(t[0]).newInstance())).getJson(t[0], t[1]);
//System.out.println("" + json);
out += json;
if (i < tokens.length - 1)
out += ",";
@ -123,7 +113,6 @@ public class NLPMapper extends HttpServlet {
}
out += "]}";
String resultLink = saveResult(out);
String outResult = "{";
outResult += "\"link\":\"" + resultLink + "\",";
@ -172,6 +161,8 @@ public class NLPMapper extends HttpServlet {
}
String link = ws.getPublicLink(fileName, token);
logger.info("Output json [" + fileName + "] created in " + (System.currentTimeMillis() - now) + " millisec.");
return link;
}

View File

@ -2,6 +2,9 @@ package org.gcube.nlphub.legacy;
import javax.servlet.http.HttpServletRequest;
public class Constants {
//public static String DATAMINER_URL = "http://dataminer0-proto.d4science.org";
//public static String DATAMINER_URL = "http://dataminer2-proto.d4science.org";
public static String DATAMINER_URL = "http://dataminer-prototypes.d4science.org";
public static String DEFAULT_DESCRIPTION = "NlpHub upload";
public static String TOKEN_PARAMETER = "gcube-token";
public static String TEST_TOKEN = "df2cc5f5-63ee-48c1-b2a6-1210030c57b8-843339462";

View File

@ -54,7 +54,8 @@ public class NLpLanguageRecognizer extends DataminerClient {
public static void run(String sentence, String token, String publicLink, HttpServletResponse response) throws NlpHubException {
try {
String urlService = "http://dataminer-prototypes.d4science.org/wps/WebProcessingService?request=Execute&service=WPS&Version=1.0.0";
//String urlService = "http://dataminer-prototypes.d4science.org/wps/WebProcessingService?request=Execute&service=WPS&Version=1.0.0";
String urlService = Constants.DATAMINER_URL + "/wps/WebProcessingService?request=Execute&service=WPS&Version=1.0.0";
urlService += "&gcube-token=" + token;
urlService += "&lang=en-US";
urlService += "&Identifier=" + RECOGNIZER_ID;

View File

@ -20,8 +20,8 @@ import org.w3c.dom.NodeList;
import org.xml.sax.InputSource;
public class NlpAsyncNerRunner extends AsyncHttpRequest {
public final static String WPS_EXECUTE_URL = "http://dataminer-prototypes.d4science.org/wps/WebProcessingService?request=Execute&service=WPS&Version=1.0.0";
public final static String WPS_DESCRIBE_PROCESS_URL = "http://dataminer-prototypes.d4science.org/wps/WebProcessingService?request=DescribeProcess&service=WPS&Version=1.0.0";
public final static String WPS_EXECUTE_URL = Constants.DATAMINER_URL + "/wps/WebProcessingService?request=Execute&service=WPS&Version=1.0.0";
public final static String WPS_DESCRIBE_PROCESS_URL = Constants.DATAMINER_URL + "/wps/WebProcessingService?request=DescribeProcess&service=WPS&Version=1.0.0";
private String identifier, token, httpMethod, annotations, publicLink, language;
private Logger logger = Logger.getLogger(NlpAsyncNerRunner.class.getSimpleName());
@ -38,7 +38,6 @@ public class NlpAsyncNerRunner extends AsyncHttpRequest {
serviceUrl += "&" + setUrl(params);
super.setBaseUrl(serviceUrl);
super.setMethod(httpMethod);
//System.out.println(serviceUrl);
}
public NlpAsyncNerRunner(String baseUrl, String method) {

View File

@ -103,7 +103,7 @@ pre {
float: left;
margin: 10px;
color: white;
background-color: #4CAF50 !important;
background-color: #4CAF50; /*!important;*/
padding: 0.6rem;
text-transform: uppercase;
vertical-align: middle;
@ -189,6 +189,7 @@ pre {
#logo-image {
width: 200px;
height: auto;
cursor: pointer;
}
.ajax-file-upload-container {
@ -239,6 +240,12 @@ select {
border: 1px solid silver;
height: 60vh;
margin: 2px;
overflow-y: auto;
}
#ner-result-accordion {
height: 60vh;
margin: 2px;
}
.left-side-half {
@ -280,6 +287,10 @@ select {
}
}
#algorithm-param_div {
font-size: 12px !important;
}
#result-text-div {
overflow-y: auto;
}
@ -288,6 +299,16 @@ select {
overflow-y: auto;
}
.collapsible-body {
padding-left: 2px !important;
padding-right: 2px !important;
}
.collapsible-header {
color: white;
background-color: #4CAF50;
}
/*
input[type="checkbox"]:not(:checked){
position: unset !important;
@ -306,6 +327,11 @@ input[type="checkbox"]:checked {
pointer-events: auto;
}
.collapsible {
-webkit-box-shadow: none !important;
-moz-box-shadow: none !important;;
box-shadow: none !important;;
}
/*
#result-params-div input[type="checkbox"]:not(:checked){
position: unset !important;

View File

@ -65,8 +65,6 @@
</div>
<div class="column margin-left-10px">
<div class="centered full-width" id="fu-container">
<!-- <div class="waves-effect waves-light btn green darken-1"
id="upload-button">Upload text file</div>-->
<div class="waves-effect waves-light darken-1"
id="upload-button">Upload text file</div>
</div>
@ -101,7 +99,18 @@
</div>
<div></div>
<div class="left-side" id="result-text-div"></div>
<div class="right-side" id="result-params-div"></div>
<div class="right-side">
<ul class="collapsible" data-collapsible="accordion" id="ner-result-accordion">
<li>
<div class="collapsible-header">Annotations</div>
<div class="collapsible-body" id="result-params-div"></div>
</li>
<li>
<div class="collapsible-header">Algorithms</div>
<div class="collapsible-body" id="algorithm-params-div"></div>
</li>
</ul>
</div>
<div style="text-align: left;">
<div class="green-button float-right-div" id="back-ner-ui-button">back</div>
</div>

View File

@ -11,6 +11,8 @@ var named = null;
var resultText = "";
var hexLetters = '0123456789ABCDEF';
var txtFlag = true;
var checkedAnnotation = "";
var checkedAlgs = [];
// ------------------------------------------------------------------------------------------------------
// Starting point: get the csv file with algorithm parameters and set the page
@ -19,6 +21,7 @@ var txtFlag = true;
$(document).ready(function() {
disableExecuteButton();
$('.collapsible').collapsible();
$.ajax({
url : "https://data.d4science.org/bnp4UDNyb21lSURkQXdDUnlSS0JkVWgzWk1KMUxWVGZHbWJQNStIS0N6Yz0",
type : "GET",
@ -33,13 +36,14 @@ $(document).ready(function() {
buildAnnotations();
resizeTable();
resizeLogo();
enableAjaxFileUpload();
$("#logo-image").click(function() {
backHandler();
});
$("#back-ner-ui-button").click(function() {
$("#ner-result-container").hide();
$("#ner-ui").show();
$("#input-textarea").val("");
publicLink = null;
jsonContent = null;
backHandler();
});
disableExecuteButton();
@ -72,8 +76,7 @@ $(document).ready(function() {
.append(
"<div id=\"reset-upload\" class=\"centered\">Cancel</div>");
$("#reset-upload").hide();
$("#reset-upload").click(
function() {
$("#reset-upload").click(function() {
resetExecuteButton();
$("#file-info").empty();
$("#reset-upload").css(
@ -94,6 +97,9 @@ $(document).ready(function() {
txtFlag = false;
alert("The application supports text file only (.TXT)");
}
publicLink = null;
textAreaEnable(false);
$("#input-textarea").val("");
},
onSubmit: function(files) {
var submitFlag = txtFlag;
@ -107,7 +113,6 @@ $(document).ready(function() {
selectAnnotationsByLanguage();
$("#reset-upload").css("display",
"inline");
//console.log("files: " + files);
$("#file-info").empty();
$("#file-info")
.append(
@ -134,6 +139,42 @@ $(document).ready(function() {
});
});
enableAjaxFileUpload= function() {
$(".ajax-file-upload").attr("enabled", "true");
$(".ajax-file-upload").css("background", pageGreen);
try {
document.getElementById("ajax-file-upload-id").removeEventListener("click", listenAjaxFileUpload);
}catch(err) {
console.log(err.message);
}
}
disableAjaxFileUpload= function() {
$(".ajax-file-upload").attr("enabled", "false");
$(".ajax-file-upload").css("background", pageGray);
try {
document.getElementById("ajax-file-upload-id").removeEventListener("click", listenAjaxFileUpload);
document.getElementById("ajax-file-upload-id").addEventListener("click", listenAjaxFileUpload);
}catch(err) {
document.getElementById("ajax-file-upload-id").addEventListener("click", listenAjaxFileUpload);
}
}
listenAjaxFileUpload = function(event) {
event.preventDefault();
}
backHandler = function() {
if($("#ner-ui").is(":visible"))
return;
$("#ner-result-container").hide();
$("#ner-ui").show();
if(publicLink != null) {
enableExecuteButton();
}
}
/*
* Utility function processing the language indication returned by the language recognition service
@ -286,16 +327,22 @@ isEnabled = function() {
}
enableDisableTextArea = function() {
if ($("#input-textarea").val() == "")
if ($("#input-textarea").val() == "") {
disableExecuteButton();
else
enableAjaxFileUpload();
}
else {
enableExecuteButton();
disableAjaxFileUpload();
}
}
/*
* Set the listeners on the text area and the execute button
*/
setEventListeners = function() {
$(".ajax-file-upload").attr("id", "ajax-file-upload-id");
$("#input-textarea").on("keyup", function() {
enableDisableTextArea();
});
@ -303,7 +350,7 @@ setEventListeners = function() {
$("#input-textarea").change(function() {
enableDisableTextArea();
});
window.addEventListener('contextmenu', function(e) {
//alert("You've tried to open context menu"); //here you draw your own menu
//e.preventDefault();
@ -421,6 +468,7 @@ launchAlgorithm = function() {
}
if (algList.length == 0) {
alert("Warning. No algorithm matching with selected language.");
hideProgress();
return;
}
algList = algList.substring(0, algList.length - 1);
@ -436,8 +484,6 @@ launchAlgorithm = function() {
async : true,
success : function(data, stato) {
textAreaEnable(true);
$("#file-info").empty();
$("#reset-upload").css("display", "none");
if (typeof (data.response) != 'undefined') {
var jsonOut = getOutputJson(data.message);
} else if (typeof (data.error) != 'undefined') {
@ -447,13 +493,13 @@ launchAlgorithm = function() {
hideProgress();
alert("Unexpected response");
}
resetExecuteButton();
//resetExecuteButton();
},
error : function(richiesta, stato, errori) {
hideProgress();
textAreaEnable(true);
alert("Unexpected Error. Status: " + richiesta.status);
resetExecuteButton();
//resetExecuteButton();
}
});
}
@ -464,7 +510,13 @@ launchAlgorithm = function() {
getOutputJson = function(message) {
var tobemap = "";
for (var i = 0; i < message.length; i++) {
tobemap += message[i] + "|";
var splitted = message[i].split(":::");
if(splitted[1].toLowerCase().trim() == "error") {
alert("Some error occurrend running algorithm " + splitted[0] + ". Trying to get other results...");
}
else {
tobemap += message[i] + "|";
}
}
tobemap = tobemap.substring(0, tobemap.length - 1);
@ -486,9 +538,7 @@ getOutputJson = function(message) {
&& (data.response.trim().toUpperCase() == "ERROR")) {
alert("ERROR\n" + data.message);
} else {
$("#reset-upload").hide();
savedAnnotations = "";
publicLink = null;
showResult(data);
}
},
@ -496,7 +546,7 @@ getOutputJson = function(message) {
hideProgress();
$("#reset-upload").hide();
savedAnnotations = "";
publicLink = null;
//publicLink = null;
alert("Unexpected Error. Status: " + richiesta.status);
}
});
@ -526,10 +576,14 @@ showResult = function(data) {
$("#result-text-div").empty();
$("#result-text-div").append("<p>" + localText + "</p>");
showAnnotationList(jsonContent.output.annotations);
buildAlgortihmList();
buildAnnotationList(jsonContent.output.annotations);
}
showAnnotationList = function(list) {
/*
* Build the list of radio-button controls with the annotations
*/
buildAnnotationList = function(list) {
var colorDisabled = "CCCCCC";
var color;
var colors = [];
@ -542,24 +596,93 @@ showAnnotationList = function(list) {
}
$("#result-params-div").append("<form id='colored-annotations'></form>");
var firstRadio = true;
for (var i = 0; i < annotations.length; i++) {
var cb = "<input type='radio' name='foo' ";
var cb = "<input type='radio' name='foo-ner-annotations' ";
if (!checkAnnotation(annotations[i])) {
cb += "disabled ";
colors[i] = colorDisabled;
}
if(firstRadio) {
if(cb.indexOf("disabled") < 0) {
firstRadio = false;
cb += "checked ";
}
}
cb += "named='" + annotations[i] + "' value='" + colors[i]
+ "' id='color-" + i + "' class='filled-in'>";
cb += "<label for='color-" + i
+ "'><span style='font-weight:bold; color:#" + colors[i] + "'>"
+ annotations[i] + "</span></label><br>";
$("#colored-annotations").append(cb);
}
$("#colored-annotations :radio").change(function() {
if (this.checked) {
rewriteText(this.getAttribute("named"), "#" + this.value);
checkedAnnotation = this.getAttribute("named");
rewriteText("#" + this.value);
}
});
showAnnotationsList();
$("#colored-annotations :radio").change();
}
/*
* Utility
*/
showAnnotationsList = function() {
$('.collapsible').collapsible('open', 0);
}
/*
* build the list of check-box controls for algorithms
*/
buildAlgortihmList = function() {
$("#algorithm-params-div").empty();
for (var i = 0; i < jsonContent.output.result.length; i++) {
var cb = "<p><input type='checkbox' name='foo-ner-algs' checked named='"
+ jsonContent.output.result[i].algorithm + "' id='algorithm" + i + "'>";
cb += "<label for='algorithm" + i + "'>" + findNameByAlgorithmId(jsonContent.output.result[i].algorithm) +"</label></p>";
$("#algorithm-params-div").append(cb);
}
buildCheckedAlgs();
$("#algorithm-params-div :checkbox").change(function () {
buildCheckedAlgs();
rewriteText($("#colored-annotations :checked")[0].value);
});
}
/*
* build the internal global variable checkedAlgs used to filter results
*/
buildCheckedAlgs = function() {
checkedAlgs = [];
for (var i = 0; i < jsonContent.output.result.length; i++) {
if($("#algorithm" + i).is(":checked")) {
checkedAlgs.push($("#algorithm" + i).attr("named"));
}
}
}
/*
* Utility function
*/
findNameByAlgorithmId = function(id) {
for(i in algorithms) {
if(algorithms[i].id.substr(algorithms[i].id.lastIndexOf(".")+1) == id)
return algorithms[i].name;
}
return id;
}
/*
@ -597,7 +720,7 @@ enhanceColor = function(color) {
/*
* Write the html paragraph <p> containing the text highlighted on annotation value
*/
rewriteText = function(annotation, color) {
rewriteText = function(color) {
$("#result-text-div").empty();
var complemetar = 0xFFFFFF - eval("0x" + color.substring(1));
@ -609,42 +732,47 @@ rewriteText = function(annotation, color) {
showProgress();
$(".hidden-div").show();
window.setTimeout(function() {
var indices = getIndices(annotation);
var indices = getIndices();
$(".hidden-div").hide();
$("#result-header-right").empty();
$("#result-header-right").append("<span style='color:" + color + ";'>" + annotation + "</span> occurs " + indices.length + " times.");
var indexedText = "";
$("#result-header-right").append("<span style='color:" + color + ";'>" + checkedAnnotation + "</span> occurs " + indices.length + " times.");
if ((typeof (indices) == 'undefined') || (indices.length == 0)) {
indexedText = resultText;
if(indices.length != 0) {
var indexedText = "";
if ((typeof (indices) == 'undefined') || (indices.length == 0)) {
indexedText = resultText;
indexedText = indexedText.replace(/\n/g, "<br>");
indexedText = indexedText.replace(/</g, "&lt;");
indexedText = indexedText.replace(/>/g, "&gt;");
$("#result-text-div").append("<p>" + indexedText + "</p>");
return;
}
var t = 0;
var offset = 0;
for (var i = 0; i < indices.length; i++) {
var index = indices[i];
var start = index[0];
var end = index[1];
indexedText += resultText.substring(t, start);
var colored = "<span style='color:" + color + "; background:"
+ complement + "; font-weight:bold;'>"
+ resultText.substring(start, end) + "</span>";
indexedText += colored;
t = end;
}
if (t < resultText.length)
indexedText += resultText.substring(t);
indexedText = indexedText.replace(/\n/g, "<br>");
indexedText = indexedText.replace(/</g, "&lt;");
indexedText = indexedText.replace(/>/g, "&gt;");
$("#result-text-div").append("<p>" + indexedText + "</p>");
return;
}
var t = 0;
var offset = 0;
for (var i = 0; i < indices.length; i++) {
var index = indices[i];
var start = index[0];
var end = index[1];
indexedText += resultText.substring(t, start);
var colored = "<span style='color:" + color + "; background:"
+ complement + "; font-weight:bold;'>"
+ resultText.substring(start, end) + "</span>";
indexedText += colored;
t = end;
else {
$("#result-text-div").append("<p>" + resultText + "</p>");
}
if (t < resultText.length)
indexedText += resultText.substring(t);
indexedText = indexedText.replace(/\n/g, "<br>");
$("#result-text-div").append("<p>" + indexedText + "</p>");
}, 50);
}
@ -668,13 +796,25 @@ checkAnnotation = function(annotation) {
/*
* Retrieve the annotation indices from the json
*/
getIndices = function(annotation) {
getIndices = function() {
var indices = [];
// get indices
for (var i = 0; i < jsonContent.output.result.length; i++) {
// filter on algorithm
var found = false;
for(var j=0; j<checkedAlgs.length; j++) {
if(checkedAlgs[j] == jsonContent.output.result[i].algorithm) {
found = true;
break;
}
}
if(!found)
continue;
var entities = jsonContent.output.result[i].entities;
for (var j = 0; j < entities.length; j++) {
a = entities[j][annotation];
// filter on annotation
a = entities[j][checkedAnnotation];
if (typeof a != 'undefined') {
indices[i] = [];
for (var k = 0; k < a.length; k++) {
@ -685,6 +825,9 @@ getIndices = function(annotation) {
}
}
if(indices.length == 0)
return indices;
var indices2 = [];
for (var i = 0; i < indices.length; i++) {
if (typeof (indices[i]) != 'undefined')
@ -699,7 +842,6 @@ getIndices = function(annotation) {
*/
mergeIndices = function(indices) {
var newIndices = [];
//console.log("--mergeIndices: indices.length=" + indices.length);
if (indices.length == 1)
newIndices = indices[0];
else
@ -801,7 +943,16 @@ $(".ajax-file-upload-abort").on("DOMAttrModified", function(event) {
*/
buildLanguageList = function() {
// build the language selection input control
var defaultLanguage = null;
for (i in languages) {
languages[i] = languages[i].trim();
languages[i] = languages[i].replace(/\w\S*/g,
function(txt) {
return txt.charAt(0).toUpperCase() + txt.substr(1).toLowerCase();
});
if(languages[i] == "English")
defaultLanguage = languages[i];
var opt = "<option value=\"" + languages[i]
+ "\" selected=\"selected\">" + languages[i] + "</option>";
if (i > 1)
@ -809,9 +960,15 @@ buildLanguageList = function() {
+ "</option>";
$("#language-select").append(opt);
}
if(defaultLanguage != null)
$("#language-select").val(defaultLanguage);
$("#language-select").on("change", function() {
selectAnnotationsByLanguage();
});
$("#language-select").change();
}
/*

View File

@ -73,7 +73,7 @@ mergeAll = function(indices) {
// [1] 'compareSegment'.
// when two segment are equals or included compareSegment returns 0
// when two segment are intersecting compareSegment returns -1
// when two segment have are external (no intersection) compareSegment returns 1
// when two segment are external (no intersection) compareSegment returns 1
//
// [2] 'mergeSegment'
// returns the "union" of two segments