client/textanalysis: sanitize quotes, fixes #293

2 files changed, 16 insertions, 10 deletions

diff --git a/res/client/css/style.css b/res/client/css/style.css
index 531b15c6..63893106 100644
--- a/res/client/css/style.css
+++ b/res/client/css/style.css
@@ -49,9 +49,6 @@ a:hover {
 a:active {
   color: #919095; }
 
-button {
-  border: none; }
-
 input {
   background-color: white;
   color: #363636;
@@ -439,7 +436,7 @@ body.debug .debug-ui-right {
   fill: #8b3ab0; }
 
 .circle.project {
-  fill: #40C200; }
+  fill: #40c200; }
 
 .circle.third-internship-proposal {
   fill: #33c2e0; }
@@ -680,7 +677,7 @@ body.debug .debug-ui-right {
   border-top: 6px solid #363636; }
 
 .info.type-project {
-  border-top: 6px solid #40C200; }
+  border-top: 6px solid #40c200; }
 
 .info .info-card-attr {
   display: block;
@@ -760,8 +757,8 @@ body.debug .debug-ui-right {
   width: 100%;
   min-height: 48px;
   text-align: center;
-  background-color: #D3F2F8;
-  color: #1D6E7F;
+  background-color: #d3f2f8;
+  color: #1d6e7f;
   border-top: 1px solid #8eb6bf;
   border-bottom: 1px solid #8eb6bf; }
 
@@ -986,5 +983,3 @@ p {
 
 svg text::selection {
   background: none; }
-
-/*# sourceMappingURL=style.css.map */
diff --git a/src/client/textanalysis.js b/src/client/textanalysis.js
index e0b57dbc..9a4cd412 100644
--- a/src/client/textanalysis.js
+++ b/src/client/textanalysis.js
@@ -28,6 +28,17 @@ function selectedType()
     return nodetypes[typeindex];
 }
 
+/*
+ * Place to do general cleanup of the text. Right now just replacing all quotation types
+ * with a single type, separately for double quotes and single quotes.
+ */
+function cleanup(text)
+{
+    return text
+        .replace(/[״”“„‟″‶]/, '"')   // Convert typographic double quotes
+        .replace(/[`׳’‘‚‛′‵]/, "'"); // Convert typographic single quotes
+}
+
 /**
  * Tokenizer for input.
  *
@@ -251,7 +262,7 @@ var textAnalyser = function (spec) {
 
     //Sentence Sequencing
     //Build the words and cuts the main elements
-    tokens = tokenize(newtext, '#', '"');
+    tokens = tokenize(cleanup(newtext), '#', '"');
     sentence = tokens.map(function (d) { return d.token; });
 
     // build new node,link arrays in order of appearance