Merge branch 'develop' into alpha/JAL-3362_Jalview_212_alpha

[jalview.git] / src / jalview / util / StringUtils.java

diff --git a/src/jalview/util/StringUtils.java b/src/jalview/util/StringUtils.java
index 2e8ace8..1f33522 100644 (file)
--- a/src/jalview/util/StringUtils.java
+++ b/src/jalview/util/StringUtils.java
@@ -20,6 +20,8 @@
  */
 package jalview.util;
 
+import java.io.UnsupportedEncodingException;
+import java.net.URLEncoder;
 import java.util.ArrayList;
 import java.util.List;
 import java.util.regex.Pattern;
@@ -29,8 +31,16 @@ public class StringUtils
   private static final Pattern DELIMITERS_PATTERN = Pattern
           .compile(".*='[^']*(?!')");
 
+  private static final char PERCENT = '%';
+
   private static final boolean DEBUG = false;
 
+  /*
+   * URL encoded characters, indexed by char value
+   * e.g. urlEncodings['='] = urlEncodings[61] = "%3D"
+   */
+  private static String[] urlEncodings = new String[255];
+
   /**
    * Returns a new character array, after inserting characters into the given
    * character array.
@@ -107,29 +117,6 @@ public class StringUtils
   }
 
   /**
-   * Returns the last part of 'input' after the last occurrence of 'token'. For
-   * example to extract only the filename from a full path or URL.
-   * 
-   * @param input
-   * @param token
-   *          a delimiter which must be in regular expression format
-   * @return
-   */
-  public static String getLastToken(String input, String token)
-  {
-    if (input == null)
-    {
-      return null;
-    }
-    if (token == null)
-    {
-      return input;
-    }
-    String[] st = input.split(token);
-    return st[st.length - 1];
-  }
-
-  /**
    * Parses the input string into components separated by the delimiter. Unlike
    * String.split(), this method will ignore occurrences of the delimiter which
    * are nested within single quotes in name-value pair values, e.g. a='b,c'.
@@ -146,7 +133,7 @@ public class StringUtils
     {
       return null;
     }
-    List<String> jv = new ArrayList<String>();
+    List<String> jv = new ArrayList<>();
     int cp = 0, pos, escape;
     boolean wasescaped = false, wasquoted = false;
     String lstitem = null;
@@ -444,4 +431,118 @@ public class StringUtils
     }
     return text;
   }
+
+  /**
+   * Answers the input string with any occurrences of the 'encodeable' characters
+   * replaced by their URL encoding
+   * 
+   * @param s
+   * @param encodable
+   * @return
+   */
+  public static String urlEncode(String s, String encodable)
+  {
+    if (s == null || s.isEmpty())
+    {
+      return s;
+    }
+
+    /*
+     * do % encoding first, as otherwise it may double-encode!
+     */
+    if (encodable.indexOf(PERCENT) != -1)
+    {
+      s = urlEncode(s, PERCENT);
+    }
+
+    for (char c : encodable.toCharArray())
+    {
+      if (c != PERCENT)
+      {
+        s = urlEncode(s, c);
+      }
+    }
+    return s;
+  }
+
+  /**
+   * Answers the input string with any occurrences of {@code c} replaced with
+   * their url encoding. Answers the input string if it is unchanged.
+   * 
+   * @param s
+   * @param c
+   * @return
+   */
+  static String urlEncode(String s, char c)
+  {
+    String decoded = String.valueOf(c);
+    if (s.indexOf(decoded) != -1)
+    {
+      String encoded = getUrlEncoding(c);
+      if (!encoded.equals(decoded))
+      {
+        s = s.replace(decoded, encoded);
+      }
+    }
+    return s;
+  }
+
+  /**
+   * Answers the input string with any occurrences of the specified (unencoded)
+   * characters replaced by their URL decoding.
+   * <p>
+   * Example: {@code urlDecode("a%3Db%3Bc", "-;=,")} should answer
+   * {@code "a=b;c"}.
+   * 
+   * @param s
+   * @param encodable
+   * @return
+   */
+  public static String urlDecode(String s, String encodable)
+  {
+    if (s == null || s.isEmpty())
+    {
+      return s;
+    }
+
+    for (char c : encodable.toCharArray())
+    {
+      String encoded = getUrlEncoding(c);
+      if (s.indexOf(encoded) != -1)
+      {
+        String decoded = String.valueOf(c);
+        s = s.replace(encoded, decoded);
+      }
+    }
+    return s;
+  }
+
+  /**
+   * Does a lazy lookup of the url encoding of the given character, saving the
+   * value for repeat lookups
+   * 
+   * @param c
+   * @return
+   */
+  private static String getUrlEncoding(char c)
+  {
+    if (c < 0 || c >= urlEncodings.length)
+    {
+      return String.valueOf(c);
+    }
+
+    String enc = urlEncodings[c];
+    if (enc == null)
+    {
+      try
+      {
+        enc = urlEncodings[c] = URLEncoder.encode(String.valueOf(c),
+                "UTF-8");
+      } catch (UnsupportedEncodingException e)
+      {
+        enc = urlEncodings[c] = String.valueOf(c);
+      }
+    }
+    return enc;
+  }
 }