enhanced merge segments algorithm.
git-svn-id: http://svn.research-infrastructures.eu/public/d4science/gcube/trunk/data-analysis/nlphub@165167 82a268e6-3cf1-43bd-a215-b396298e98cf
This commit is contained in:
Erik Perrone
parent
b2c98580da
commit
3c689932ed
|
|
@ -0,0 +1,8 @@
|
|||
log4j.rootLogger=INFO,stdout
|
||||
log4j.logger.com.endeca=INFO
|
||||
# Logger for crawl metrics
|
||||
log4j.logger.com.endeca.itl.web.metrics=INFO
|
||||
|
||||
log4j.appender.stdout=org.apache.log4j.ConsoleAppender
|
||||
log4j.appender.stdout.layout=org.apache.log4j.PatternLayout
|
||||
log4j.appender.stdout.layout.ConversionPattern=%p\t%d{ISO8601}\t%r\t%c\t[%t]\t%m%n
|
||||
|
|
@ -102,12 +102,12 @@ $(document).ready(function() {
|
|||
},
|
||||
onSuccess : function(files, data, xhr) {
|
||||
hideProgress();
|
||||
console.log('' + data.language);
|
||||
//console.log('' + data.language);
|
||||
checkLanguage(data.language);
|
||||
selectAnnotationsByLanguage();
|
||||
$("#reset-upload").css("display",
|
||||
"inline");
|
||||
console.log("files: " + files);
|
||||
//console.log("files: " + files);
|
||||
$("#file-info").empty();
|
||||
$("#file-info")
|
||||
.append(
|
||||
|
|
@ -373,6 +373,7 @@ showProgress = function() {
|
|||
$(".progress-circular-div").css("left", left);
|
||||
$(".progress-circular-div").css("top", top);
|
||||
$(".hidden-div").css("display", "block");
|
||||
$(".hidden-div").show();
|
||||
}
|
||||
|
||||
/*
|
||||
|
|
@ -380,6 +381,7 @@ showProgress = function() {
|
|||
*/
|
||||
hideProgress = function() {
|
||||
$(".hidden-div").css("display", "none");
|
||||
$(".hidden-div").hide();
|
||||
}
|
||||
|
||||
/*
|
||||
|
|
@ -471,7 +473,7 @@ getOutputJson = function(message) {
|
|||
parameters += "&plink=" + encodeURI(publicLink);
|
||||
parameters += "&lang=" + $("#language-select").val();
|
||||
parameters += "&tobemap=" + encodeURI(tobemap);
|
||||
console.log(parameters);
|
||||
//console.log(parameters);
|
||||
|
||||
$.ajax({
|
||||
url : "/nlphub/nlphub-mapper-servlet?" + parameters,
|
||||
|
|
@ -479,7 +481,7 @@ getOutputJson = function(message) {
|
|||
async : true,
|
||||
success : function(data, stato) {
|
||||
hideProgress();
|
||||
console.log(data);
|
||||
//console.log(data);
|
||||
if ((typeof (data.response) != "undefined")
|
||||
&& (data.response.trim().toUpperCase() == "ERROR")) {
|
||||
alert("ERROR\n" + data.message);
|
||||
|
|
@ -604,49 +606,46 @@ rewriteText = function(annotation, color) {
|
|||
var G = enhanceColor(complement.substring(2,4));
|
||||
var B = enhanceColor(complement.substring(4));
|
||||
complement = "#" + R + G + B;
|
||||
console.log("-getIndices: start");
|
||||
var indices = getIndices(annotation);
|
||||
console.log("-getIndices: end");
|
||||
$("#result-header-right").empty();
|
||||
$("#result-header-right").append("<span style='color:" + color + ";'>" + annotation + "</span> occurs " + indices.length + " times.");
|
||||
var indexedText = "";
|
||||
showProgress();
|
||||
$(".hidden-div").show();
|
||||
window.setTimeout(function() {
|
||||
var indices = getIndices(annotation);
|
||||
$(".hidden-div").hide();
|
||||
|
||||
$("#result-header-right").empty();
|
||||
$("#result-header-right").append("<span style='color:" + color + ";'>" + annotation + "</span> occurs " + indices.length + " times.");
|
||||
var indexedText = "";
|
||||
|
||||
if ((typeof (indices) == 'undefined') || (indices.length == 0)) {
|
||||
indexedText = resultText;
|
||||
indexedText = indexedText.replace(/\n/g, "<br>");
|
||||
indexedText = indexedText.replace(/</g, "<");
|
||||
indexedText = indexedText.replace(/>/g, ">");
|
||||
$("#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);
|
||||
|
||||
if ((typeof (indices) == 'undefined') || (indices.length == 0)) {
|
||||
indexedText = resultText;
|
||||
indexedText = indexedText.replace(/\n/g, "<br>");
|
||||
indexedText = indexedText.replace(/</g, "<");
|
||||
indexedText = indexedText.replace(/>/g, ">");
|
||||
$("#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);
|
||||
/*
|
||||
offset += countSubstringOccurrencies(resultText.substring(t, start),
|
||||
"\n");
|
||||
offset += countSubstringOccurrencies(resultText.substring(t, start),
|
||||
"\r");
|
||||
offset += countSubstringOccurrencies(resultText.substring(t, start),
|
||||
"\t");
|
||||
start += offset;
|
||||
end += offset;*/
|
||||
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>");
|
||||
$("#result-text-div").append("<p>" + indexedText + "</p>");
|
||||
|
||||
}, 50);
|
||||
}
|
||||
|
||||
/*
|
||||
|
|
@ -699,14 +698,12 @@ getIndices = function(annotation) {
|
|||
* Merge the indices
|
||||
*/
|
||||
mergeIndices = function(indices) {
|
||||
console.log("--mergeIndices start");
|
||||
var newIndices = [];
|
||||
//console.log("--mergeIndices: indices.length=" + indices.length);
|
||||
if (indices.length == 1)
|
||||
newIndices = indices[0];
|
||||
else
|
||||
newIndices = mergeAll(indices);
|
||||
console.log("--mergeIndices end");
|
||||
return newIndices;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -43,43 +43,6 @@ compareSegments = function(s1, s2) {
|
|||
return (s1[0] - s2[0]);
|
||||
}
|
||||
|
||||
cleanSegmentList = function(list) {
|
||||
var cleaned = [];
|
||||
for(var i=0; i<list.length; i++) {
|
||||
if(cleaned.length == 0)
|
||||
cleaned[cleaned.length] = list[i];
|
||||
else {
|
||||
var found = false;
|
||||
for(var j=0; j<cleaned.length; j++) {
|
||||
if(compareSegment(cleaned[j], list[i]) == 0) {
|
||||
if(cleaned[j][0] > list[i][0])
|
||||
cleaned[j] = list[i];
|
||||
found = true;
|
||||
break;
|
||||
}
|
||||
if(compareSegment(cleaned[j], list[i]) < 0) {
|
||||
if(cleaned[j][0] <= list[i][0])
|
||||
cleaned[j] = [cleaned[j][0], list[i][1]];
|
||||
else
|
||||
cleaned[j] = [list[i][0], cleaned[j][1]];
|
||||
found = true;
|
||||
}
|
||||
}
|
||||
if(!found) cleaned[cleaned.length] = list[i];
|
||||
}
|
||||
}
|
||||
return cleaned;
|
||||
}
|
||||
|
||||
compareSegmentList = function(list1, list2) {
|
||||
if(list1.length != list2.length)
|
||||
return false;
|
||||
for(var i=0; i<list1.length; i++)
|
||||
if((list1[i][0] != list2[i][0]) || (list1[i][1] != list2[i][1]))
|
||||
return false;
|
||||
return true;
|
||||
}
|
||||
|
||||
/**
|
||||
* mergeAll: merge indices
|
||||
* parameters: indices = Array of Array of indices (index = [start, end])
|
||||
|
|
@ -87,42 +50,46 @@ compareSegmentList = function(list1, list2) {
|
|||
*/
|
||||
|
||||
mergeAll = function(indices) {
|
||||
console.log("---mergeAll indices.length=" + indices.length);
|
||||
var m = [];
|
||||
for(var i=0; i<indices[0].length; i++) {
|
||||
for(var j=1; j<indices.length; j++) {
|
||||
for(var k=0; k<indices[j].length; k++) {
|
||||
var res = mergeSegment(indices[0][i], indices[j][k]);
|
||||
for(var u=0; u<res.length; u++)
|
||||
m[m.length] = res[u];
|
||||
}
|
||||
}
|
||||
}
|
||||
/*
|
||||
var end = false;
|
||||
var counter = 1;
|
||||
while(!end) {
|
||||
var old = m.slice();
|
||||
m = cleanSegmentList(m);
|
||||
//console.log(m);
|
||||
end = compareSegmentList(old, m);
|
||||
counter++;
|
||||
if(counter == 100)
|
||||
end = true;
|
||||
}
|
||||
return m.sort(compareSegments);*/
|
||||
|
||||
// first of all: creates a 1-dimension array with all data
|
||||
for(var i=0; i<indices.length; i++) {
|
||||
m = m.concat(indices[i]);
|
||||
}
|
||||
|
||||
//
|
||||
// second step: sort the array
|
||||
// for our purposes a segment is 'lower than' another segment if the left value of the segment
|
||||
// is lower than the left value of the other segment. In other words:
|
||||
// [a, b] < [c, d] if a < c
|
||||
//
|
||||
m = m.sort(compareSegments);
|
||||
|
||||
var m2 = [];
|
||||
|
||||
//
|
||||
// merging procedure:
|
||||
// the procedure uses the functions:
|
||||
// [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
|
||||
//
|
||||
// [2] 'mergeSegment'
|
||||
// returns the "union" of two segments
|
||||
//
|
||||
var current = m[0];
|
||||
for(var i=0; i<m.length; i++) {
|
||||
var cfr = compareSegment(current, m[i]);
|
||||
switch(cfr) {
|
||||
case 0:
|
||||
case -1:
|
||||
// if segments are the same or intersected the result is the merged segment
|
||||
current = mergeSegment(current, m[i])[0];
|
||||
break;
|
||||
default:
|
||||
// if segments are external mergeSegment produce two segments: the first is ready to be stored in the output vector
|
||||
// the second is to be compared with others
|
||||
var s = mergeSegment(current, m[i]);
|
||||
m2[m2.length] = s[0];
|
||||
current = s[1];
|
||||
|
|
@ -130,7 +97,10 @@ mergeAll = function(indices) {
|
|||
}
|
||||
}
|
||||
|
||||
if((current[0] != m2[m2.length-1][0]) || (current[1] != m2[m2.length-1][1]))
|
||||
if(m2.length == 0) {
|
||||
m2[0] = current;
|
||||
}
|
||||
else if((current[0] != m2[m2.length-1][0]) || (current[1] != m2[m2.length-1][1]))
|
||||
m2[m2.length] = current;
|
||||
return m2;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -18,33 +18,33 @@ var indices = [
|
|||
//console.log(newIndices);
|
||||
// answer is: [0,19], [21,23]
|
||||
|
||||
indices = [
|
||||
indices1 = [
|
||||
[[7,19], [21,25]],
|
||||
[[1,5], [18,22]]
|
||||
];
|
||||
//answer is: [1,5], [7,25]
|
||||
|
||||
indices = [
|
||||
indices2 = [
|
||||
[[7,16], [21,25]],
|
||||
[[1,5], [18,22]]
|
||||
];
|
||||
//answer is: [1,5], [7,16], [18,25]
|
||||
|
||||
indices = [
|
||||
indices3 = [
|
||||
[[7,16], [21,25]],
|
||||
[[1,5], [18,22]],
|
||||
[[2,30]]
|
||||
];
|
||||
//answer is: [1,30]
|
||||
|
||||
indices = [
|
||||
indices4 = [
|
||||
[[1,4], [5, 8], [21,25]],
|
||||
[[1,5], [18,22]],
|
||||
[[0,2], [11,15], [27,31]]
|
||||
];
|
||||
//answer is: [0,8],[11,15],[18,25],[27,31]
|
||||
|
||||
|
||||
var indices = indices4;
|
||||
var mm = mergeAll(indices);
|
||||
|
||||
console.log(mm);
|
||||
|
|
|
|||
Loading…
Reference in New Issue