KNL-1253 Improved support to add an option to better handle interpret…

…ing html into the intended plain text git-svn-id: https://source.sakaiproject.org/svn/kernel/trunk@309575 66ffb92e-73f9-0310-93c1-f5514f145a0a
yuyilin · May 16, 2014 · de33331 · de33331
1 parent b0e59e7
commit de33331
Show file tree

Hide file tree

Showing 5 changed files with 28 additions and 13 deletions.
diff --git a/kernel/api/src/main/java/org/sakaiproject/util/api/FormattedText.java b/kernel/api/src/main/java/org/sakaiproject/util/api/FormattedText.java
@@ -392,8 +392,12 @@ public String processFormattedText(final String strFromBrowser, StringBuilder er
      * Strips html/xml tags from a string and returns the cleaned version
      *
      * @param text any text (if this is null or empty then the input text is returned unchanged)
+     * @param smartSpacing if true then try to make the text represent the intent of the html,
+     *                     trims out duplicate spaces, converts block type html into a space, etc.,
+     *                     else just removes html tags and leaves all other parts of the string intact,
+     *                     NOTE: false is also slightly faster
      * @return the cleaned string
      */
-    public String stripHtmlFromText(String text);
+    public String stripHtmlFromText(String text, boolean smartSpacing);
 
 }
diff --git a/kernel/api/src/main/java/org/sakaiproject/util/api/MockFormattedText.java b/kernel/api/src/main/java/org/sakaiproject/util/api/MockFormattedText.java
@@ -183,7 +183,7 @@ public String sanitizeHrefURL(String urlToSanitize) {
     }
 
     @Override
-    public String stripHtmlFromText(String text) {
+    public String stripHtmlFromText(String text, boolean smartSpacing) {
         log.warn(WARNING);
         return text;
     }

diff --git a/kernel/kernel-impl/src/main/java/org/sakaiproject/util/impl/FormattedTextImpl.java b/kernel/kernel-impl/src/main/java/org/sakaiproject/util/impl/FormattedTextImpl.java
@@ -1161,10 +1161,17 @@ public String sanitizeHrefURL(String urlToSanitize) {
     }
 
     @Override
-    public String stripHtmlFromText(String text) {
+    public String stripHtmlFromText(String text, boolean smartSpacing) {
         // KNL-1253 use Jsoup
         if (text != null && !"".equals(text)) {
-            text = org.jsoup.Jsoup.clean(text, org.jsoup.safety.Whitelist.none());
+            if (smartSpacing) {
+                // replace block level html with an extra space (to try to preserve the intent)
+                text = text.replaceAll("/br>", "/br> ").replaceAll("/p>", "/p> ").replaceAll("/tr>", "/tr> ");
+            }
+            text = org.jsoup.Jsoup.clean(text, "", org.jsoup.safety.Whitelist.none(), new org.jsoup.nodes.Document.OutputSettings().prettyPrint(false).outline(false));
+            if (smartSpacing) {
+                text = text.replaceAll("\\s+", " ").trim(); // eliminate extra whitespaces
+            }
         }
         return text;
     }

diff --git a/kernel/kernel-impl/src/test/java/org/sakaiproject/util/impl/FormattedTextTest.java b/kernel/kernel-impl/src/test/java/org/sakaiproject/util/impl/FormattedTextTest.java
@@ -842,28 +842,32 @@ public void testKNL_1253() {
 
         // no change
         text = null;
-        result = formattedText.stripHtmlFromText(text);
+        result = formattedText.stripHtmlFromText(text,true);
         assertEquals(text, result);
 
         text = "";
-        result = formattedText.stripHtmlFromText(text);
+        result = formattedText.stripHtmlFromText(text,true);
         assertEquals(text, result);
 
         text = "azeckoski";
-        result = formattedText.stripHtmlFromText(text);
+        result = formattedText.stripHtmlFromText(text,true);
         assertEquals(text, result);
 
         // changed
         text = "<b>azeckoski</b>";
-        result = formattedText.stripHtmlFromText(text);
+        result = formattedText.stripHtmlFromText(text,true);
         assertEquals("azeckoski", result);
 
         text = "<a href='www.vt.edu'><b>azeckoski</b> is AZ</a>";
-        result = formattedText.stripHtmlFromText(text);
+        result = formattedText.stripHtmlFromText(text,true);
         assertEquals("azeckoski is AZ", result);
 
         text = "<table><tr><th>Column1</th></tr><tr><td>Row1</td></tr></table>";
-        result = formattedText.stripHtmlFromText(text);
+        result = formattedText.stripHtmlFromText(text,true);
+        assertEquals("Column1 Row1", result);
+
+        text = "<table><tr><th>Column1</th></tr><tr><td>Row1</td></tr></table>";
+        result = formattedText.stripHtmlFromText(text,false);
         assertEquals("Column1Row1", result);
     }
 

diff --git a/kernel/kernel-util/src/main/java/org/sakaiproject/util/FormattedText.java b/kernel/kernel-util/src/main/java/org/sakaiproject/util/FormattedText.java
@@ -176,10 +176,10 @@ public static String sanitizeHrefURL(String urlToSanitize) {
     }
 
     /**
-     * @see org.sakaiproject.util.api.FormattedText#stripHtmlFromText(String)
+     * @see org.sakaiproject.util.api.FormattedText#stripHtmlFromText(String,boolean)
      */
-    public String stripHtmlFromText(String text) {
-        return getFormattedText().stripHtmlFromText(text);
+    public String stripHtmlFromText(String text, boolean smartSpacing) {
+        return getFormattedText().stripHtmlFromText(text, smartSpacing);
     }
 
 }