Skip to content

Commit

Permalink
Fixed several tokenized and whitespace trimming bugs. Added a multi-r…
Browse files Browse the repository at this point in the history 
…eference xml file to the demo folder.
  • Loading branch information
nmadnani committed Apr 9, 2013
1 parent 7acb3fd commit f08f773
Show file tree
Hide file tree
Showing 6 changed files with 522 additions and 90 deletions.
78 changes: 40 additions & 38 deletions bleu.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -214,16 +214,17 @@
// and when there are two systems, then between the the hyps
if (tstSets[1] != undefined) {
var lcslist;
var tokhyp1 = tstSets[0].documents[docid][segnum]
var tokhyp2 = tstSets[1].documents[docid][segnum];
var tokhyp1 = $('#tokenizeCheckBox').attr('checked') ? tstSets[0].untokenizedDocuments[docid][segnum] : tstSets[0].documents[docid][segnum];
var tokhyp2 = $('#tokenizeCheckBox').attr('checked') ? tstSets[1].untokenizedDocuments[docid][segnum] : tstSets[1].documents[docid][segnum];
lcslist = lcs(tokhyp1.split(' '), tokhyp2.split(' '));
}
else {
var hypLCSList = [];
var refLCSLists = [];
var tokhyp = tstSets[0].documents[docid][segnum];
// var tokhyp = tstSets[0].documents[docid][segnum];
var tokhyp = $('#tokenizeCheckBox').attr('checked') ? tstSets[0].untokenizedDocuments[docid][segnum] : tstSets[0].documents[docid][segnum];
for (var i=0; i<refSets.length; i++) {
var tokref = refSets[i].documents[docid][segnum];
var tokref = $('#tokenizeCheckBox').attr('checked') ? refSets[i].untokenizedDocuments[docid][segnum] : refSets[i].documents[docid][segnum];
var lcslist = lcs(tokhyp.split(' '), tokref.split(' '));
hypLCSList.push.apply(hypLCSList, lcslist.xindices);
refLCSLists.push(lcslist.yindices);
Expand Down Expand Up @@ -259,23 +260,23 @@

// Add all the references
for(var i=0; i < refSets.length; i++) {
if (tstSets[1] != undefined) {
var ref = refSets[i].untokenizedDocuments[docid][segnum];
ref = $('#caseCheckBox').attr('checked') ? ref : ref.toLowerCase();
tablehtml += '<tr>' +
'<td align="center" id="segDetailHeader">Reference<br/>(' + refSets[i].refid + ')</td>' +
'<td id="segDetailRef' + i + '">' + ref + '</td>' +
'</tr>';
}
else {
var tokref = refSets[i].documents[docid][segnum];
tokref = $('#caseCheckBox').attr('checked') ? tokref : tokref.toLowerCase();
ref_html = lcsColorize(tokref, refLCSLists[i], '#CC3333');
tablehtml += '<tr>' +
'<td align="center" id="segDetailHeader">Reference<br/>(' + refSets[i].refid + ')</td>' +
'<td id="segDetailRef' + i + '">' + ref_html + '</td>' +
'</tr>';
}
if (tstSets[1] != undefined) {
var ref = $('#tokenizeCheckBox').attr('checked') ? refSets[i].untokenizedDocuments[docid][segnum] : refSets[i].documents[docid][segnum];
ref = $('#caseCheckBox').attr('checked') ? ref : ref.toLowerCase();
tablehtml += '<tr>' +
'<td align="center" id="segDetailHeader">Reference<br/>(' + refSets[i].refid + ')</td>' +
'<td id="segDetailRef' + i + '">' + ref + '</td>' +
'</tr>';
}
else {
var tokref = $('#tokenizeCheckBox').attr('checked') ? refSets[i].untokenizedDocuments[docid][segnum] : refSets[i].documents[docid][segnum];
tokref = $('#caseCheckBox').attr('checked') ? tokref : tokref.toLowerCase();
ref_html = lcsColorize(tokref, refLCSLists[i], '#CC3333');
tablehtml += '<tr>' +
'<td align="center" id="segDetailHeader">Reference<br/>(' + refSets[i].refid + ')</td>' +
'<td id="segDetailRef' + i + '">' + ref_html + '</td>' +
'</tr>';
}
}

// Add the hypothesis or hypotheses as the case may be
Expand All @@ -296,7 +297,7 @@
'</tr>';
}
else {
var hyp = tstSets[0].untokenizedDocuments[docid][segnum];
var hyp = $('#tokenizeCheckBox').attr('checked') ? tstSets[0].untokenizedDocuments[docid][segnum] : tstSets[0].documents[docid][segnum];
tokhyp = $('#caseCheckBox').attr('checked') ? tokhyp : tokhyp.toLowerCase();
hyp_html = lcsColorize(tokhyp, hypLCSList, '#CC3333');
tablehtml += '<tr>' +
Expand Down Expand Up @@ -698,7 +699,7 @@
var segs = [];
$(this).find("seg").each(function (seg) {
//var segid = $(this).attr("id");
segs.push($(this).text());
segs.push($(this).text().trim());
// segs.push(tokenize($(this).text(), $('#caseCheckBox').attr('checked')));
});
srcSet.documents[docid] = segs;
Expand All @@ -718,7 +719,7 @@
var segs = [];
var originalSegs = txt.split('\n').slice(0, -1);
for (var i=0; i < originalSegs.length; i++) {
segs.push(originalSegs[i]);
segs.push(originalSegs[i].trim());
}
srcSet.documents[docid] = segs;
}
Expand Down Expand Up @@ -760,12 +761,12 @@
$(this).find("seg").each(function (seg) {
//var segid = $(this).attr("id");
if ($('#tokenizeCheckBox').attr('checked')) {
segs.push($(this).text());
segs.push($(this).text().trim());
}
else {
segs.push(tokenize($(this).text(), $('#caseCheckBox').attr('checked')));
segs.push(tokenize($(this).text().trim(), $('#caseCheckBox').attr('checked')));
}
originalSegs.push($(this).text());
originalSegs.push($(this).text().trim());
});
tstSet.documents[docid] = segs;
tstSet.untokenizedDocuments[docid] = originalSegs;
Expand All @@ -783,10 +784,10 @@
var originalSegs = txt.split('\n').slice(0, -1);
for (var i=0; i < originalSegs.length; i++) {
if ($('#tokenizeCheckBox').attr('checked')) {
segs.push(originalSegs[i]);
segs.push(originalSegs[i].trim());
}
else {
segs.push(tokenize(originalSegs[i], $('#caseCheckBox').attr('checked')));
segs.push(tokenize(originalSegs[i].trim(), $('#caseCheckBox').attr('checked')));
}
}
tstSet.documents[docid] = segs;
Expand Down Expand Up @@ -822,11 +823,12 @@
$(this).find("seg").each(function (seg) {
//var segid = $(this).attr("id");
if ($('#tokenizeCheckBox').attr('checked')) {
segs.push(tokenize($(this).text(), $('#caseCheckBox').attr('checked')));
segs.push($(this).text().trim());
}
else {
segs.push($(this).text());
segs.push(tokenize($(this).text().trim(), $('#caseCheckBox').attr('checked')));
}
originalSegs.push($(this).text().trim());
});
tstSet.documents[docid] = segs;
tstSet.untokenizedDocuments[docid] = originalSegs;
Expand All @@ -844,10 +846,10 @@
var originalSegs = txt.split('\n').slice(0, -1);
for (var i=0; i < originalSegs.length; i++) {
if ($('#tokenizeCheckBox').attr('checked')) {
segs.push(originalSegs[i]);
segs.push(originalSegs[i].trim());
}
else {
segs.push(tokenize(originalSegs[i], $('#caseCheckBox').attr('checked')));
segs.push(tokenize(originalSegs[i].trim(), $('#caseCheckBox').attr('checked')));
}
}
tstSet.documents[docid] = segs;
Expand Down Expand Up @@ -890,12 +892,12 @@
$(this).find("seg").each(function(seg) {
//var segid = $(this).attr("id");
if ($('#tokenizeCheckBox').attr('checked')) {
segs.push($(this).text());
segs.push($(this).text().trim());
}
else {
segs.push(tokenize($(this).text(), $('#caseCheckBox').attr('checked')));
segs.push(tokenize($(this).text().trim(), $('#caseCheckBox').attr('checked')));
}
originalSegs.push($(this).text());
originalSegs.push($(this).text().trim());
});
rset.documents[docid] = segs;
rset.untokenizedDocuments[docid] = originalSegs;
Expand All @@ -915,10 +917,10 @@
var originalSegs = txt.split('\n').slice(0, -1);
for (var i=0; i < originalSegs.length; i++) {
if ($('#tokenizeCheckBox').attr('checked')) {
segs.push(originalSegs[i]);
segs.push(originalSegs[i].trim());
}
else {
segs.push(tokenize(originalSegs[i], $('#caseCheckBox').attr('checked')));
segs.push(tokenize(originalSegs[i].trim(), $('#caseCheckBox').attr('checked')));
}
}
rset.documents[docid] = segs;
Expand Down
Loading

0 comments on commit f08f773

Please sign in to comment.