1 files changed, 59 insertions, 8 deletions

diff --git a/java/src/com/android/inputmethod/latin/StringUtils.java b/java/src/com/android/inputmethod/latin/StringUtils.java
index 90c3fcdd2..11ef60dc9 100644
--- a/java/src/com/android/inputmethod/latin/StringUtils.java
+++ b/java/src/com/android/inputmethod/latin/StringUtils.java
@@ -22,6 +22,10 @@ import java.util.ArrayList;
 import java.util.Locale;
 
 public final class StringUtils {
+    public static final int CAPITALIZE_NONE = 0;  // No caps, or mixed case
+    public static final int CAPITALIZE_FIRST = 1; // First only
+    public static final int CAPITALIZE_ALL = 2;   // All caps
+
     private StringUtils() {
         // This utility class is not publicly instantiable.
     }
@@ -102,20 +106,30 @@ public final class StringUtils {
         }
     }
 
-    public static String toTitleCase(final String s, final Locale locale) {
+    public static String capitalizeFirstCodePoint(final String s, final Locale locale) {
+        if (s.length() <= 1) {
+            return s.toUpperCase(locale);
+        }
+        // Please refer to the comment below in
+        // {@link #capitalizeFirstAndDowncaseRest(String,Locale)} as this has the same shortcomings
+        final int cutoff = s.offsetByCodePoints(0, 1);
+        return s.substring(0, cutoff).toUpperCase(locale) + s.substring(cutoff);
+    }
+
+    public static String capitalizeFirstAndDowncaseRest(final String s, final Locale locale) {
         if (s.length() <= 1) {
-            // TODO: is this really correct? Shouldn't this be s.toUpperCase()?
-            return s;
+            return s.toUpperCase(locale);
         }
         // TODO: fix the bugs below
         // - This does not work for Greek, because it returns upper case instead of title case.
         // - It does not work for Serbian, because it fails to account for the "lj" character,
         // which should be "Lj" in title case and "LJ" in upper case.
-        // - It does not work for Dutch, because it fails to account for the "ij" digraph, which
-        // are two different characters but both should be capitalized as "IJ" as if they were
-        // a single letter.
-        // - It also does not work with unicode surrogate code points.
-        return s.toUpperCase(locale).charAt(0) + s.substring(1);
+        // - It does not work for Dutch, because it fails to account for the "ij" digraph when it's
+        // written as two separate code points. They are two different characters but both should
+        // be capitalized as "IJ" as if they were a single letter in most words (not all). If the
+        // unicode char for the ligature is used however, it works.
+        final int cutoff = s.offsetByCodePoints(0, 1);
+        return s.substring(0, cutoff).toUpperCase(locale) + s.substring(cutoff).toLowerCase(locale);
     }
 
     private static final int[] EMPTY_CODEPOINTS = {};
@@ -171,4 +185,41 @@ public final class StringUtils {
         }
         return list.toArray(new String[list.size()]);
     }
+
+    // This method assumes the text is not null. For the empty string, it returns CAPITALIZE_NONE.
+    public static int getCapitalizationType(final String text) {
+        // If the first char is not uppercase, then the word is either all lower case or
+        // camel case, and in either case we return CAPITALIZE_NONE.
+        final int len = text.length();
+        int index = 0;
+        for (; index < len; index = text.offsetByCodePoints(index, 1)) {
+            if (Character.isLetter(text.codePointAt(index))) {
+                break;
+            }
+        }
+        if (index == len) return CAPITALIZE_NONE;
+        if (!Character.isUpperCase(text.codePointAt(index))) {
+            return CAPITALIZE_NONE;
+        }
+        int capsCount = 1;
+        int letterCount = 1;
+        for (index = text.offsetByCodePoints(index, 1); index < len;
+                index = text.offsetByCodePoints(index, 1)) {
+            if (1 != capsCount && letterCount != capsCount) break;
+            final int codePoint = text.codePointAt(index);
+            if (Character.isUpperCase(codePoint)) {
+                ++capsCount;
+                ++letterCount;
+            } else if (Character.isLetter(codePoint)) {
+                // We need to discount non-letters since they may not be upper-case, but may
+                // still be part of a word (e.g. single quote or dash, as in "IT'S" or "FULL-TIME")
+                ++letterCount;
+            }
+        }
+        // We know the first char is upper case. So we want to test if either every letter other
+        // than the first is lower case, or if they are all upper case. If the string is exactly
+        // one char long, then we will arrive here with letterCount 1, and this is correct, too.
+        if (1 == capsCount) return CAPITALIZE_FIRST;
+        return (letterCount == capsCount ? CAPITALIZE_ALL : CAPITALIZE_NONE);
+    }
 }