[Jython-checkins] SF.net SVN: jython:[5555] trunk/jython

[<pj...@us...>]

Revision: 5555
          http://jython.svn.sourceforge.net/jython/?rev=5555&view=rev
Author:   pjenvey
Date:     2008-11-07 08:20:22 +0000 (Fri, 07 Nov 2008)

Log Message:
-----------
add a simple unicodeobject::PyUnicode_EncodeDecimal equivalent for unicode
number conversion. fixes test_unicode.test_codecs_errors

Modified Paths:
--------------
    trunk/jython/Lib/test/test_unicode_jy.py
    trunk/jython/src/org/python/core/PyUnicode.java

Modified: trunk/jython/Lib/test/test_unicode_jy.py
===================================================================
--- trunk/jython/Lib/test/test_unicode_jy.py	2008-11-07 05:27:30 UTC (rev 5554)
+++ trunk/jython/Lib/test/test_unicode_jy.py	2008-11-07 08:20:22 UTC (rev 5555)
@@ -50,7 +50,15 @@
         self.assertEqual(ord(bar[2]), 92)
         self.assertEqual(ord(bar[3]), 110)
 
+    def test_encode_decimal(self):
+        self.assertEqual(int(u'\u0039\u0032'), 92)
+        self.assertEqual(int(u'\u0660'), 0)
+        self.assertEqual(int(u' \u001F\u0966\u096F\u0039'), 99)
+        self.assertEqual(long(u'\u0663'), 3)
+        self.assertEqual(float(u'\u0663.\u0661'), 3.1)
+        self.assertEqual(complex(u'\u0663.\u0661'), 3.1+0j)
 
+
 def test_main():
     test_support.run_unittest(UnicodeTestCase)
 

Modified: trunk/jython/src/org/python/core/PyUnicode.java
===================================================================
--- trunk/jython/src/org/python/core/PyUnicode.java	2008-11-07 05:27:30 UTC (rev 5554)
+++ trunk/jython/src/org/python/core/PyUnicode.java	2008-11-07 08:20:22 UTC (rev 5555)
@@ -1311,6 +1311,93 @@
         return newSubsequenceIterator();
     }
 
+    @Override
+    public PyComplex __complex__() {
+        return new PyString(encodeDecimal()).__complex__();
+    }
+
+    @Override
+    public int atoi(int base) {
+        return new PyString(encodeDecimal()).atoi(base);
+    }
+
+    @Override
+    public PyLong atol(int base) {
+        return new PyString(encodeDecimal()).atol(base);
+    }
+
+    @Override
+    public double atof() {
+        return new PyString(encodeDecimal()).atof();
+    }
+
+    /**
+     * Encode unicode into a valid decimal String. Throws a UnicodeEncodeError on invalid
+     * characters.
+     *
+     * @return a valid decimal as an encoded String
+     */
+    private String encodeDecimal() {
+        if (isBasicPlane()) {
+            return encodeDecimalBasic();
+        }
+
+        int digit;
+        StringBuilder sb = new StringBuilder();
+        int i = 0;
+        for (Iterator<Integer> iter = newSubsequenceIterator(); iter.hasNext(); i++) {
+            int codePoint = iter.next();
+            if (Character.isWhitespace(codePoint)) {
+                sb.append(' ');
+                continue;
+            }
+            digit = Character.digit(codePoint, 10);
+            if (digit >= 0) {
+                sb.append(digit);
+                continue;
+            }
+            if (0 < codePoint && codePoint < 256) {
+                sb.appendCodePoint(codePoint);
+                continue;
+            }
+            // All other characters are considered unencodable
+            codecs.encoding_error("strict", "decimal", string, i, i + 1,
+                                  "invalid decimal Unicode string");
+        }
+        return sb.toString();
+    }
+
+    /**
+     * Encode unicode in the basic plane into a valid decimal String. Throws a
+     * UnicodeEncodeError on invalid characters.
+     *
+     * @return a valid decimal as an encoded String
+     */
+    private String encodeDecimalBasic() {
+        int digit;
+        StringBuilder sb = new StringBuilder();
+        for (int i = 0; i < string.length(); i++) {
+            char ch = string.charAt(i);
+            if (Character.isWhitespace(ch)) {
+                sb.append(' ');
+                continue;
+            }
+            digit = Character.digit(ch, 10);
+            if (digit >= 0) {
+                sb.append(digit);
+                continue;
+            }
+            if (0 < ch && ch < 256) {
+                sb.append(ch);
+                continue;
+            }
+            // All other characters are considered unencodable
+            codecs.encoding_error("strict", "decimal", string, i, i + 1,
+                                  "invalid decimal Unicode string");
+        }
+        return sb.toString();
+    }
+
     @ExposedMethod
     final String unicode_toString() {
         return toString();


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.