Name: bsC130419 Date: 05/25/2001


java version "1.3.0"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.3.0)
Java HotSpot(TM) Client VM (build 1.3.0, mixed mode)


When -encoding option is specified to javadoc, Java source codes are converted
correctly, but package.html aren't.

For example, an open-source project manage java source codes and related
documents includes "package.html" in Shift-JIS and share them from various
environments (e.g. Japanese Solaris which character encoding is Japanese-EUC).
On solaris, package.html is converted to '???' sequences.

This is a simple javadoc's bug.

j2sdk1_3_0-src-sol/src/share/classes/com/sun/tools/javadoc/DocImpl.java has the
following method which reads "package.html":

    String readHTMLDocumentation(java.io.InputStream input, String filename)
throws java.io.IOException {
        int filesize = input.available();
        byte[] filecontents = new byte[filesize];
        input.read(filecontents, 0, filesize);
        input.close();
        String rawDoc = new String(filecontents);

In fact, "new String(filecontents)" is the same as "new String(filecontents,
..., ..., ByteToCharConverter.getDefault())". Therefore, package.html always is
converted from platform's native character encoding, not character encoding
specified by -encoding option.

  To fix this bug, add an encoding argument to it like "new String(filecontents,
encoding)"; it is saved in an Env object and acquired by getCharacterEncoding().
(Review ID: 124196)
======================================================================