charactertranslationtest.java

html解析包 可以很方便的解析html 纯java 实现

    }

    public void check (byte[] reference, byte[] result)
        throws
            IOException
    {
        InputStream ref;
        InputStream in;
        int i;
        int i1;
        int i2;

        ref = new ByteArrayInputStream (reference);
        in = new ByteArrayInputStream (result);
        i = 0;
        do
        {
            i1 = ref.read ();
            i2 = in.read ();
            if (i1 != i2)
                fail ("byte difference detected at offset " + i + " expected " + i1 + ", actual " + i2);
            i++;
        }
        while (-1 != i1);
        ref.close ();
        in.close ();
    }

    public void testHexNumericEncoding ()
        throws
            IOException
    {
        try
        {
            Translate.ENCODE_HEXADECIMAL = true;
            assertEquals (
                "hex value incorrect",
                "֫ is a non-existant character.",
                Translate.encode ("\u05AB is a non-existant character."));
        }
        finally
        {
            Translate.ENCODE_HEXADECIMAL = false;
        }
    }

    public void testLastCharacterEntityReference ()
        throws
            IOException
    {
        assertEquals (
            "poorly terminated numeric character reference doesn't work",
            "The character entity reference\u200cshould be decoded",
            Translate.decode ("The character entity reference&zwnjshould be decoded"));
    }

    public void testEncodeDecodePage () throws IOException
    {
        URL url;
        URLConnection connection;
        InputStream in;
        ByteArrayOutputStream out;
        byte[] bytes;
        byte[] result;
        int c;

        // get some bytes
        url = new URL ("http://sourceforge.net/projects/htmlparser");
        connection = url.openConnection ();
        in = connection.getInputStream ();
        out = new ByteArrayOutputStream ();
        while (-1 != (c = in.read ()))
            out.write (c);
        in.close ();
        out.close ();
        bytes = out.toByteArray ();

        // run it through
        result = encodedecode (bytes);
        
        // check
        check (bytes, result);
    }

    /**
     * Check all references read in from the w3.org site.
     * If this test fails but the others pass, suspect that the list of
     * entity references has been augmented. The updated list is in the
     * CharacterEntityReferenceList.java file in your home directory.
     */
    public void testEncodeDecodeAll ()
    {
        CharacterReference[] list;
        StringBuffer stimulus;
        StringBuffer response;
        CharacterReference ref;
        String string;

        list = getReferences ();
        stimulus = new StringBuffer ();
        response = new StringBuffer ();
        for (int i = 0; i < list.length; i++)
        {
            ref = list[i];
            stimulus.append ((char)ref.getCharacter ());
            response.append ("&");
            response.append (ref.getKernel ());
            response.append (";");
        }
        string = Translate.encode (stimulus.toString ());
        if (!string.equals (response.toString ()))
            fail ("encoding incorrect, expected \n\"" + response.toString () + "\", encoded \n\"" + string + "\""); 
        string = Translate.decode (string);
        if (!string.equals (stimulus.toString ()))
            fail ("decoding incorrect, expected \n\"" + stimulus.toString () + "\", decoded \n\"" + string + "\", encoded \n\"" + response.toString () + "\""); 
    }

    public void testEncodeDecodeRandom ()
    {
        Random random;
        CharacterReference[] list;
        StringBuffer stimulus;
        StringBuffer response;
        char character;
        CharacterReference ref;
        String string;

        random = new Random ();
        list = getReferences ();
        stimulus = new StringBuffer ();
        response = new StringBuffer ();
        for (int i = 0; i < 1000; i++)
        {
            for (int j = 0; j < 10; j++)
            {
                // some random characters
                for (int k = 0; k < 10; k++)
                {
                    character = (char)random.nextInt (127);
                    if (character >= ' ')
                    {
                        if ('&' == character)
                        {
                            stimulus.append (character);
                            response.append ("&amp;");
                        }
                        else if ('"' == character)
                        {
                            stimulus.append (character);
                            response.append ("&quot;");
                        }
                        else if ('<' == character)
                        {
                            stimulus.append (character);
                            response.append ("&lt;");
                        }
                        else if ('>' == character)
                        {
                            stimulus.append (character);
                            response.append ("&gt;");
                        }
                        else
                        {
                            stimulus.append (character);
                            response.append (character);
                        }
                    }
                }
                ref = list[random.nextInt (list.length)];
                stimulus.append ((char)ref.getCharacter ());
                response.append ("&");
                response.append (ref.getKernel ());
                response.append (";");
                // some more random characters
                for (int k = 0; k < 10; k++)
                {
                    character = (char)random.nextInt (127);
                    if (character >= ' ')
                    {
                        if ('&' == character)
                        {
                            stimulus.append (character);
                            response.append ("&amp;");
                        }
                        else if ('"' == character)
                        {
                            stimulus.append (character);
                            response.append ("&quot;");
                        }
                        else if ('<' == character)
                        {
                            stimulus.append (character);
                            response.append ("&lt;");
                        }
                        else if ('>' == character)
                        {
                            stimulus.append (character);
                            response.append ("&gt;");
                        }
                        else
                        {
                            stimulus.append (character);
                            response.append (character);
                        }
                    }
                }
            }
            string = Translate.encode (stimulus.toString ());
            if (!string.equals (response.toString ()))
                fail ("encoding incorrect, expected \n\"" + response.toString () + "\", encoded \n\"" + string + "\""); 
            string = Translate.decode (string);
            if (!string.equals (stimulus.toString ()))
                fail ("decoding incorrect, expected \n\"" + stimulus.toString () + "\", decoded \n\"" + string + "\", encoded \n\"" + response.toString () + "\""); 
            stimulus.setLength (0);
            response.setLength (0);
        }   
        
    }

    public void testEncodeDecodeRandomNoSemi ()
    {
        Random random;
        CharacterReference[] list;
        StringBuffer stimulus;
        StringBuffer response;
        char character;
        int index;
        CharacterReference ref;
        String kernel;
        ArrayList forbidden;
        String string;

        random = new Random ();
        list = getReferences ();
        stimulus = new StringBuffer ();
        response = new StringBuffer ();
        for (int i = 0; i < 1000; i++)
        {
            for (int j = 0; j < 10; j++)
            {
                // some random characters
                for (int k = 0; k < 10; k++)
                {
                    character = (char)random.nextInt (127);
                    if (character >= ' ')
                    {
                        if ('&' == character)
                        {
                            stimulus.append (character);
                            response.append ("&amp;");
                        }
                        else if ('"' == character)
                        {
                            stimulus.append (character);
                            response.append ("&quot;");
                        }
                        else if ('<' == character)
                        {
                            stimulus.append (character);
                            response.append ("&lt;");
                        }
                        else if ('>' == character)
                        {
                            stimulus.append (character);
                            response.append ("&gt;");
                        }
                        else
                        {
                            stimulus.append (character);
                            response.append (character);
                        }
                    }
                }
                index = random.nextInt (list.length);
                ref = list[index];
                kernel = ref.getKernel ();
                stimulus.append ((char)ref.getCharacter ());
                response.append ("&");
                response.append (kernel);
                // to be fair, we ensure that the next character isn't valid
                // for a different reference, i.e. &sup shouldn't be followed
                // by a 1, 2, 3 or e
                forbidden = new ArrayList ();
                for (int k = index + 1; k < list.length; k++)
                    if (list[k].getKernel ().regionMatches (
                        0,
                        kernel,
                        0,
                        kernel.length ()))
                        forbidden.add (new Character (list[k].getKernel ().charAt (kernel.length ())));
                    else
                        break;
                do
                {
                    character = (char)random.nextInt (127);
                    if (   (' ' <= character)
                        && ('&' != character)
                        && ('"' != character)
                        && ('<' != character)
                        && ('>' != character)
                        && (';' != character)
                        && !(forbidden.contains (new Character (character))))
                    {
                        stimulus.append (character);
                        response.append (character);
                        character = 0;
                    }
                    else
                        character = ' ';
                        
                }
                while (0 != character);
                // some more random characters
                for (int k = 0; k < 10; k++)
                {
                    character = (char)random.nextInt (127);
                    if (character >= ' ')
                    {
                        if ('&' == character)
                        {
                            stimulus.append (character);
                            response.append ("&amp;");
                        }
                        else if ('"' == character)
                        {
                            stimulus.append (character);
                            response.append ("&quot;");
                        }
                        else if ('<' == character)
                        {
                            stimulus.append (character);
                            response.append ("&lt;");
                        }
                        else if ('>' == character)
                        {
                            stimulus.append (character);
                            response.append ("&gt;");
                        }
                        else
                        {
                            stimulus.append (character);
                            response.append (character);
                        }
                    }
                }
            }
            string = Translate.decode (response.toString ());
            if (!string.equals (stimulus.toString ()))
                fail ("decoding incorrect:\nexpected \"" + stimulus.toString () + "\"\n decoded \"" + string + "\"\n encoded \"" + response.toString () + "\""); 
            stimulus.setLength (0);
            response.setLength (0);
        }   
    }
}