simple emacs tags generator for java source - only indexes classes and interfaces.

Gen.java

import java.io.BufferedOutputStream;
import java.io.ByteArrayOutputStream;
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.File;
import java.io.Reader;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.io.StreamTokenizer;
import java.io.Writer;
import java.util.ArrayList;
import java.util.List;

// Usage:
// java Gen <full-path-to-source-tree>
//
// That recursively traverses it, looks for *.java
// and attempts to generate a tags file that ONLY
// indexes classes and interfaces (including inner ones.)

public class Gen
{
    public static void main(String args[])
        throws IOException
    {
        File root = new File(args[0]);
        File tags = new File("tags");
        if (tags.canRead()) {
            throw new IOException
                ("Won't overwrite existing tags file - "+
                 "please move or remove it first.");
        }
        BufferedOutputStream bout = new BufferedOutputStream
            (new FileOutputStream(tags));
        process(root, bout);
        bout.close();
    }

    private final static void process(File root, OutputStream out)
        throws IOException
    {
        if (root.isDirectory()) {
            File[] children = root.listFiles();
            if (children != null) {
                for (int i=0; i<children.length; i++) {
                    process(children[i], out);
                }
            }
            return;
        }
        String path = root.getPath();
        if (!path.endsWith(".java")) { return; }
        analyzeFile(root, out);
    }

    // crude technique [ahem] way to parse the file.  I just want
    // class names; and this parser fits into the "works for me"
    // category.
    private final static void analyzeFile
        (File root, OutputStream out)
        throws IOException
    {
        CountingReader br = null;
        try {
            br = new CountingReader(new FileReader(root));
            StreamTokenizer stok = new StreamTokenizer(br);
            stok.resetSyntax();
            stok.slashStarComments(true);
            stok.slashSlashComments(true);
            stok.wordChars('a', 'z');
            stok.wordChars('A', 'Z');
            stok.wordChars('_', '_');
            stok.wordChars('.', '.'); // takes care of MyClass.class
            stok.wordChars('0', '9');

            stok.wordChars(128+32, 255);
            stok.whitespaceChars(0, ' ');
            stok.quoteChar('"');

            Locations loc = new Locations(root.getPath());
            stok.nextToken();
            parse(stok, br, loc, "");
            dumpTag(out, loc);
            int a4444 = 444;
        }
        finally {
            if (br != null) {
                try { br.close(); }
                catch (IOException ign) {}
            }
        }
    }

    // always enter with stok.ttype pointing to the
    // token to be parsed.
    private final static void parse
        (StreamTokenizer stok, CountingReader br, Locations loc, String pfx)
        throws IOException
    {
        while (true) {
            switch (stok.ttype) {

            case StreamTokenizer.TT_EOF:
                return;

            case '{':
                // eat and push a level down
                stok.nextToken();
                parse(stok, br, loc, pfx);
                break;

            case '}':
                // eat and pop up a level
                stok.nextToken();
                return;

            case StreamTokenizer.TT_WORD:
                if ("class".equals(stok.sval) ||
                    "interface".equals(stok.sval)) {
                    parseLocationBlock(stok, br, loc, pfx);
                }
                else {
                    // fuggetaboudit
                    stok.nextToken();
                }
                break;

            default:
                // eat and keep going.
                stok.nextToken();
                break;
            }
        }
    }

    private final static void parseLocationBlock
        (StreamTokenizer stok, CountingReader br, Locations loc, String pfx)
        throws IOException
    {
        // This is a bit fiddly; because the tokenizer keep going
        // till it goes one token past the name. But by that time,
        // we may have hit a newline.
        String cn = expectName(stok);
        int curoff = br.m_lcount - 1;
        String lpfx = br.m_lb.toString();
        if (lpfx.length() == 0) {
            lpfx = br.m_prev.toString();
            curoff -= lpfx.length();
        }

        StringBuilder sb = new StringBuilder();
        sb.append(lpfx);
        sb.append("\u007f"); // ^?
        sb.append(pfx);
        sb.append(cn);
        sb.append("\u0001"); // ^A
        sb.append(String.valueOf(stok.lineno()));
        sb.append(",");
        sb.append(String.valueOf(curoff));
        loc.m_refs.add(sb.toString());

        while (stok.nextToken() != '{') {
            if (stok.ttype == StreamTokenizer.TT_EOF) {
                throw new IOException("Missing { at "+stok.lineno());
            }
        }
        // leave cursor at start of next token.
        stok.nextToken();
        parse(stok, br, loc, pfx+cn+".");
    }

    private final static String expectName(StreamTokenizer stok)
        throws IOException
    {
        if (stok.nextToken() != StreamTokenizer.TT_WORD) {
            throw new IOException("Unexpected - non-word at "+ stok.lineno());
        }
        return stok.sval;
    }

    private final static void dumpTag(OutputStream out, Locations loc)
        throws IOException
    {
        // header ^L^J
        out.write(0x0c);
        out.write(0x0a);

        // I need the length of the section.
        StringBuilder sb = new StringBuilder();
        for (String ref: loc.m_refs) {
            sb.append(ref);
            sb.append("\n"); // ^J
        }
        byte[] def = asBytes(sb.toString());

        out.write(asBytes(loc.m_path));
        out.write(asBytes(","));
        out.write(asBytes(String.valueOf(def.length)));
        out.write(0x0a);
        out.write(def);
    }

    private final static byte[] asBytes(String s)
        throws IOException
    { return s.getBytes("utf-8"); }

    private final static class Locations
    {
        private Locations(String path)
        { m_path = path; }

        private final String m_path;
        private final List<String> m_refs = new ArrayList<String>();
    }

    private final static class CountingReader
        extends BufferedReader
    {
        public CountingReader(Reader in)
        { super(in); }

        @Override
        public int read()
            throws IOException
        {
            m_count++;
            int ret = super.read();
            if (ret == '\n') {
                m_prev = m_lb;
                m_lb = new StringBuilder();
                m_lcount = m_count;
            }
            else {
                m_lb.append((char) ret);
            }
            return ret;
        }
        private int m_count = 0;
        private int m_lcount = 0;
        private StringBuilder m_lb = new StringBuilder();
        private StringBuilder m_prev = null;
    }
}

Frustrated with the tag generators for java [massively complex, or unusable ones] for my simple needs [I just want to quickly jump around the android source via class names] here's a single file that walks through a source tree and does the trick for me.

$ javac Gen.java
$ java Gen ~/android-sdk-macosx/sources/android-19
$ ls -lh tags
-rw-r--r--  1 kbs  staff   1.6M Nov 24 13:46 tags
$