Changeset 35732 for other-projects

Timestamp:

2021-11-11T17:02:12+13:00 (2 years ago)

Author:

cstephen

Message:

Improve DI return to account for newlines in JSON.
Fix DI ignoring line breaks.

Location:

other-projects/the-macronizer/trunk/src/java

Files:

• monogram/restorer/TxtRestorer.java (modified) (1 diff)
• web/servlets/DirectInput.java (modified) (2 diffs)

other-projects/the-macronizer/trunk/src/java/monogram/restorer/TxtRestorer.java

@@ -25 +25 @@
 
     public TxtRestorer() {
-        pattern = Pattern.compile("([a-zA-Zā-ūĀ-Ū0-9]+|\\p{Punct})|(.)");
+        pattern = Pattern.compile("([a-zA-Zā-ūĀ-Ū0-9]+|\\p{Punct})|(.)", Pattern.DOTALL);
     }
 

other-projects/the-macronizer/trunk/src/java/web/servlets/DirectInput.java

@@ -10 +10 @@
 import java.io.OutputStreamWriter;
 import java.io.UnsupportedEncodingException;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 import javax.servlet.RequestDispatcher;
@@ -108 +110 @@
                 JsonWriter writer = gsonInstance.newJsonWriter(response.getWriter());
                 writer.beginArray();
-
-                for (String element : restoredFragment.split("\\ "))
+                
+                final Pattern pattern = Pattern.compile("([^\\s]+)|([\\r\\n]+)");// Pattern.compile("([a-zA-Zā-ūĀ-Ū0-9</>]+|\\p{Punct})|([\\r?\\n]+)", Pattern.DOTALL);
+                final Matcher matcher = pattern.matcher(restoredFragment);
+
+                while (matcher.find())
                 {
+                    String word = matcher.group(1);
+                    final String lineBreak = matcher.group(2);
+                    
+                    if (word == null && lineBreak == null)
+                    {
+                        continue;
+                    }
+
                     writer.beginObject();
 
-                    if (element.contains("<mark>"))
+                    if (word != null)
                     {
-                        element = element.replace("<mark>", "").replace("</mark>", "");
-                        writer.name("macronised");
-                        writer.value(true);
+                        if (word.contains("<mark>"))
+                        {
+                            word = word.replace("<mark>", "").replace("</mark>", "");
+                            writer.name("macronised");
+                            writer.value(true);
+                        }
+
+                        writer.name("w");
+                        writer.value(word);
                     }
 
-                    writer.name("w");
-                    writer.value(element);
+                    if (lineBreak != null)
+                    {
+                        writer.name("linebreak");
+                        writer.value(lineBreak);
+                    }
 
                     writer.endObject();