Small fix in the VCF parser (now a bit quicker) and --help (--noutfa

[jvcard.git] / src / be / nikiroo / jvcard / tui / StringUtils.java

diff --git a/src/be/nikiroo/jvcard/tui/StringUtils.java b/src/be/nikiroo/jvcard/tui/StringUtils.java
index e9353c83ccde43c2c77128359fe329cfc11dec17..382922605260260c8f4998da741757d277ba9f4d 100644 (file)
--- a/src/be/nikiroo/jvcard/tui/StringUtils.java
+++ b/src/be/nikiroo/jvcard/tui/StringUtils.java
@@ -1,14 +1,48 @@
 package be.nikiroo.jvcard.tui;
 
+import java.text.Normalizer;
+import java.text.Normalizer.Form;
+import java.util.regex.Pattern;
+
 import com.googlecode.lanterna.gui2.LinearLayout.Alignment;
 
 public class StringUtils {
+       static private Pattern marks = Pattern
+                       .compile("[\\p{InCombiningDiacriticalMarks}\\p{IsLm}\\p{IsSk}]+");
 
+       /**
+        * Fix the size of the given {@link String} either with space-padding or by
+        * shortening it.
+        * 
+        * @param text
+        *            the {@link String} to fix
+        * @param width
+        *            the size of the resulting {@link String} if the text fits or
+        *            if cut is TRUE
+        * 
+        * @return the resulting {@link String} of size <i>size</i>
+        */
        static public String padString(String text, int width) {
                return padString(text, width, true, Alignment.Beginning);
        }
 
-       // TODO: doc it, width of -1 == no change to text
+       /**
+        * Fix the size of the given {@link String} either with space-padding or by
+        * optionally shortening it.
+        * 
+        * @param text
+        *            the {@link String} to fix
+        * @param width
+        *            the size of the resulting {@link String} if the text fits or
+        *            if cut is TRUE
+        * @param cut
+        *            cut the {@link String} shorter if needed
+        * @param align
+        *            align the {@link String} in this position if we have enough
+        *            space
+        * 
+        * @return the resulting {@link String} of size <i>size</i> minimum
+        */
        static public String padString(String text, int width, boolean cut,
                        Alignment align) {
 
@@ -47,4 +81,58 @@ public class StringUtils {
                return text;
        }
 
+       /**
+        * Sanitise the given input to make it more Terminal-friendly by removing
+        * combining characters.
+        * 
+        * @param input
+        *            the input to sanitise
+        * @param allowUnicode
+        *            allow Unicode or only allow ASCII Latin characters
+        * 
+        * @return the sanitised {@link String}
+        */
+       static public String sanitize(String input, boolean allowUnicode) {
+               return sanitize(input, allowUnicode, !allowUnicode);
+       }
+
+       /**
+        * Sanitise the given input to make it more Terminal-friendly by removing
+        * combining characters.
+        * 
+        * @param input
+        *            the input to sanitise
+        * @param allowUnicode
+        *            allow Unicode or only allow ASCII Latin characters
+        * @param removeAllAccents
+        *            TRUE to replace all accentuated characters by their non
+        *            accentuated counter-parts
+        * 
+        * @return the sanitised {@link String}
+        */
+       static public String sanitize(String input, boolean allowUnicode,
+                       boolean removeAllAccents) {
+
+               if (removeAllAccents) {
+                       input = Normalizer.normalize(input, Form.NFKD);
+                       input = marks.matcher(input).replaceAll("");
+               }
+
+               input = Normalizer.normalize(input, Form.NFKC);
+
+               if (!allowUnicode) {
+                       StringBuilder builder = new StringBuilder();
+                       for (int index = 0; index < input.length(); index++) {
+                               char car = input.charAt(index);
+                               // displayable chars in ASCII are in the range 32<->255,
+                               // except DEL (127)
+                               if (car >= 32 && car <= 255 && car != 127) {
+                                       builder.append(car);
+                               }
+                       }
+                       input = builder.toString();
+               }
+
+               return input;
+       }
 }