jocaml/Jocaml.java

import java.io.File;
import java.io.FileInputStream;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.channels.Channels;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

public class Jocaml {
    private static String TRAILER_MAGIC = "Caml1999X028";
    private static long MARSHAL_MAGIC_SMALL = 0x8495A6BE;
    private static long MARSHAL_MAGIC_BIG = 0x8495A6BF;

    private static enum CamlInst {
        ACC0(0, 0),
        ACC1(1, 0),
        ACC2(2, 0),
        ACC3(3, 0),
        ACC4(4, 0),
        ACC5(5, 0),
        ACC6(6, 0),
        ACC7(7, 0),
        ACC(8, 1),
        PUSH(9, 0),
        PUSHACC0(10, 0),
        PUSHACC1(11, 0),
        PUSHACC2(12, 0),
        PUSHACC3(13, 0),
        PUSHACC4(14, 0),
        PUSHACC5(15, 0),
        PUSHACC6(16, 0),
        PUSHACC7(17, 0),
        PUSHACC(18, 1),
        POP(19, 1),
        ASSIGN(20, 1),
        ENVACC1(21, 0),
        ENVACC2(22, 0),
        ENVACC3(23, 0),
        ENVACC4(24, 0),
        ENVACC(25, 1),
        PUSHENVACC1(26, 0),
        PUSHENVACC2(27, 0),
        PUSHENVACC3(28, 0),
        PUSHENVACC4(29, 0),
        PUSHENVACC(30, 1),
        PUSH_RETADDR(31, 1),
        APPLY(32, 1),
        APPLY1(33, 0),
        APPLY2(34, 0),
        APPLY3(35, 0),
        APPTERM(36, 2),
        APPTERM1(37, 1),
        APPTERM2(38, 1),
        APPTERM3(39, 1),
        RETURN(40, 1),
        RESTART(41, 0),
        GRAB(42, 1),
        CLOSURE(43, 2),
        CLOSUREREC(44, 2),
        OFFSETCLOSUREM2(45, 0),
        OFFSETCLOSURE0(46, 0),
        OFFSETCLOSURE2(47, 0),
        OFFSETCLOSURE(48, 1),
        PUSHOFFSETCLOSUREM2(49, 0),
        PUSHOFFSETCLOSURE0(50, 0),
        PUSHOFFSETCLOSURE2(51, 0),
        PUSHOFFSETCLOSURE(52, 1),
        GETGLOBAL(53, 1),
        PUSHGETGLOBAL(54, 1),
        GETGLOBALFIELD(55, 2),
        PUSHGETGLOBALFIELD(56, 2),
        SETGLOBAL(57, 1),
        ATOM0(58, 0),
        ATOM(59, 1),
        PUSHATOM0(60, 0),
        PUSHATOM(61, 1),
        MAKEBLOCK(62, 2),
        MAKEBLOCK1(63, 1),
        MAKEBLOCK2(64, 1),
        MAKEBLOCK3(65, 1),
        MAKEFLOATBLOCK(66, 1),
        GETFIELD0(67, 0),
        GETFIELD1(68, 0),
        GETFIELD2(69, 0),
        GETFIELD3(70, 0),
        GETFIELD(71, 1),
        GETFLOATFIELD(72, 1),
        SETFIELD0(73, 0),
        SETFIELD1(74, 0),
        SETFIELD2(75, 0),
        SETFIELD3(76, 0),
        SETFIELD(77, 1),
        SETFLOATFIELD(78, 1),
        VECTLENGTH(79, 0),
        GETVECTITEM(80, 0),
        SETVECTITEM(81, 0),
        GETBYTESCHAR(82, 0),
        SETBYTESCHAR(83, 0),
        BRANCH(84, 1),
        BRANCHIF(85, 1),
        BRANCHIFNOT(86, 1),
        SWITCH(87, 2),
        BOOTNOT(88, 0),
        PUSHTRAP(89, 1),
        POPTRAP(90, 0),
        RAISE(91, 0),
        CHECK_SIGNALS(92, 0),
        C_CALL1(93, 1),
        C_CALL2(94, 1),
        C_CALL3(95, 1),
        C_CALL4(96, 1),
        C_CALL5(97, 1),
        C_CALLN(98, 2),
        CONST0(99, 0),
        CONST1(100, 0),
        CONST2(101, 0),
        CONST3(102, 0),
        CONSTINT(103, 1),
        PUSHCONST0(104, 0),
        PUSHCONST1(105, 0),
        PUSHCONST2(106, 0),
        PUSHCONST3(107, 0),
        PUSHCONSTINT(108, 1),
        NEGINT(109, 0),
        ADDINT(110, 0),
        SUBINT(111, 0),
        MULINT(112, 0),
        DIVINT(113, 0),
        MODINT(114, 0),
        ANDINT(115, 0),
        ORINT(116, 0),
        XORINT(117, 0),
        LSLINT(118, 0),
        LSRINT(119, 0),
        ASRINT(120, 0),
        EQ(121, 0),
        NEQ(122, 0),
        LTINT(123, 0),
        LEINT(124, 0),
        GTINT(125, 0),
        GEINT(126, 0),
        OFFSETINT(127, 1),
        OFFSETREF(128, 1),
        ISINT(129, 0),
        GETMETHOD(130, 0),
        BEQ(131, 2),
        BNEQ(132, 2),
        BLTINT(133, 2),
        BLEINT(134, 2),
        BGTINT(135, 2),
        BGEINT(136, 2),
        ULTINT(137, 0),
        UGEINT(138, 0),
        BULTINT(139, 2),
        BUGEINT(140, 2),
        GETPUBMET(141, 2),
        GETDYNMET(142, 0),
        STOP(143, 0),
        EVENT(144, 0),
        BREAK(145, 0),

        RERAISE(146, 0), // ???
        RAISE_NOTRACE(147, 0),
        GETSTRINGCHAR(148, 0);

        public final int code;
        public final int numArgs;
        CamlInst(int code, int numArgs) {
            this.code = code;
            this.numArgs = numArgs;
        }

        static {
            for (int i = 0; i < CamlInst.values().length; i++) {
                if (CamlInst.values()[i].code != i) {
                    throw new RuntimeException("what");
                }
            }
        }
    }

    private static class CodeOffset {
        public final int offset;

        public CodeOffset(int offset) {
            this.offset = offset;
        }

        public String toString() {
            return "CodeOffset<" + offset + ">";
        }

        public boolean equals(Object other) {
            return other instanceof CodeOffset && ((CodeOffset) other).offset == this.offset;
        }

        public int hashCode() {
            return offset * 13;
        }
    }

    private static class Block {
        public final int tag;
        public final List<Object> contents;

        public Block(int tag, List<Object> contents) {
            this.tag = tag;
            this.contents = contents;
        }

        public String toString() {
            return "Block<" + tag + ": " + contents.toString() + ">";
        }
    }

    private static class Section {
        String name;
        ByteBuffer data;

        public Section(String name, ByteBuffer data) {
            this.name = name;
            this.data = data;
        }

        public String toString() {
            return "Section<name=" + name + ", length=" + data.capacity() + ">";
        }

        public List<String> unstring() {
            List<String> strings = new ArrayList<>();
            for (int i = 0; i < this.data.capacity();) {
                StringBuffer buf = new StringBuffer();
                while (this.data.get(i) != 0) {
                    buf.append((char) this.data.get(i));
                    i += 1;
                }
                strings.add(buf.toString());
                i += 1;
            }
            return strings;
        }

        private Object parseNextObject(List<Object> shared) throws Exception {
            int nextTag = this.data.get() & (short) 0xff;
            switch (nextTag) {
                // integers
                case 0x00: {
                    byte value = this.data.get();
                    return Long.valueOf(value);
                }
                case 0x01: {
                    short value = this.data.getShort();
                    return Long.valueOf(value);
                }
                case 0x02: {
                    int value = this.data.getInt();
                    return Long.valueOf(value);
                }
                case 0x03: {
                    long value = this.data.getLong();
                    return Long.valueOf(value);
                }

                // shared elements
                case 0x04: {
                    long offset = this.data.get() & (long) 0xff;
                    return shared.get(shared.size() - 1 - (int) offset);
                }
                case 0x05: {
                    long offset = this.data.getShort() & (long) 0xff;
                    return shared.get(shared.size() - 1 - (int) offset);
                }
                case 0x06: {
                    long offset = this.data.getInt() & (long) 0xff;
                    return shared.get(shared.size() - 1 - (int) offset);
                }

                // blocks
                case 0x08: {
                    long header = this.data.getInt() & (long) 0xff;
                    int tag = (int) header & 0xff;
                    long size = header >> 10;
                    List<Object> elems = new ArrayList<>();
                    Block block = new Block(tag, elems);
                    if (size > 0) {
                        shared.add(block);
                        for (int i = 0; i < size; i++) {
                            elems.add(parseNextObject(shared));
                        }
                    }
                    return block;
                }
                case 0x13: {
                    throw new Exception("long block not supported");
                }

                // string
                case 0x09: {
                    int len = this.data.get() & (short) 0xff;
                    byte[] buf = new byte[len];
                    this.data.get(buf);
                    String s = new String(buf);
                    shared.add(s);
                    return s;
                }
                case 0x0a: {
                    long len = this.data.getInt() & (long) 0xff;
                    byte[] buf = new byte[(int)len];
                    this.data.get(buf);
                    String s = new String(buf);
                    shared.add(s);
                    return s;
                }

                // float
                case 0x0b:
                case 0x0c: {
                    Double obj = Double.valueOf(this.data.getDouble());
                    shared.add(obj);
                    return obj;
                }
                case 0x0D:
                case 0x0E: {
                    throw new Exception("double array not supported");
                }
                case 0x07:
                case 0x0f: {
                    throw new Exception("double array not supported");
                }

                // misc
                case 0x10: {
                    throw new Exception("code pointer with checksum");
                }
                case 0x11: {
                    throw new Exception("code pointer with closure");
                }
                case 0x12:
                case 0x18:
                case 0x19: {
                    StringBuffer buf = new StringBuffer();
                    while (true) {
                        int next = this.data.get() & (short) 0xff;
                        if (next == 0) {
                            break;
                        }
                        buf.append((char) next);
                    }
                    String name = buf.toString();
                    switch (name) {
                        // int64
                        case "_j": {
                            long v = Long.valueOf(this.data.getLong());
                            shared.add(v);
                            return v;
                        }
                        default:
                            throw new Exception("unknown custom op: " + name);
                    }
                }
                default: {
                    if (nextTag < 0x20) {
                        throw new Exception("invalid tag " + nextTag);
                    } else if (nextTag < 0x40) {
                        int length = nextTag & 0x1f;
                        byte[] buf = new byte[(int)length];
                        this.data.get(buf);
                        String s = new String(buf);
                        shared.add(s);
                        return s;
                    } else if (nextTag < 0x80) {
                        Long value = Long.valueOf(nextTag & 0x3f);
                        return value;
                    } else {
                        long size = (nextTag >> 4) & 0x07;
                        int tag = nextTag & 0x0f;
                        List<Object> elems = new ArrayList<>();
                        Block block = new Block(tag, elems);
                        if (size > 0) {
                            shared.add(block);
                            for (int i = 0; i < size; i++) {
                                elems.add(parseNextObject(shared));
                            }
                        }
                        return block;
                    }
                }
            }
        }

        public List<Object> unmarshal() throws Exception {
            this.data.order(ByteOrder.BIG_ENDIAN);
            int magic = this.data.getInt(0);
            if (magic != MARSHAL_MAGIC_SMALL) {
                throw new Exception("bad marshal magic");
            }

            // int byteLength = this.data.getInt(4);
            int numObjs = this.data.getInt(8);
            System.out.println(numObjs);

            List<Object> shared = new ArrayList<>();
            List<Object> objs = new ArrayList<>();

            this.data.position(20);

            while (this.data.position() < this.data.capacity()) {
                Object next = parseNextObject(shared);
                objs.add(next);
            }

            System.out.println(shared.size());
            return objs;
        }

        public void disassemble(List<String> prim) throws Exception {
            this.data.order(ByteOrder.LITTLE_ENDIAN);
            this.data.position(0);
            while (this.data.position() < this.data.capacity()) {
                int pc = this.data.position() / 4;
                int op = this.data.getInt();
                String rep = null;
                try {
                    CamlInst inst = CamlInst.values()[op];
                    rep = inst.name();
                    if (inst == CamlInst.C_CALL1 || inst == CamlInst.C_CALL2
                            || inst == CamlInst.C_CALL3 || inst == CamlInst.C_CALL4
                            || inst == CamlInst.C_CALL5 || inst == CamlInst.C_CALLN) {
                        int p = this.data.getInt();
                        rep += " " + prim.get(p);
                        if (inst == CamlInst.C_CALLN) {
                            int n = this.data.getInt();
                            rep += " " + n;
                        }
                    } else if (inst == CamlInst.CLOSUREREC) {
                        int nfuncs = this.data.getInt();
                        int nargs = this.data.getInt();
                        rep += " " + nfuncs + " " + nargs;
                        for (int i = 0; i < nfuncs; i++) {
                            rep += " " + this.data.getInt();
                        }
                    } else {
                        for (int i = 0; i < inst.numArgs; i++) {
                            rep += " " + this.data.getInt();
                        }
                    }
                } catch (ArrayIndexOutOfBoundsException e) {
                    rep = "<UNKNOWN " + op + ">";
                }
                System.out.printf("%05d: ", pc);
                System.out.println(rep);
            }
        }
    }

    public static Section findSection(Section[] sections, String name) {
        for (Section section : sections) {
            if (section.name.equals(name)) {
                return section;
            }
        }
        return null;
    }

    public static void main(String[] args) throws Exception {
        ByteBuffer buf;
        int len;
        {
            FileInputStream in = new FileInputStream(new File(args[0]));
            len = in.available();
            buf = ByteBuffer.allocate(len);
            Channels.newChannel(in).read(buf);
        }

        for (int i = 0; i < 12; i++) {
            if (buf.get(len - 12 + i) != (byte) TRAILER_MAGIC.charAt(i)) {
                throw new Exception("invalid trailer");
            }
        }

        int numSections = buf.getInt(len - 16);
        Section[] sections = new Section[numSections];

        for (int i = numSections - 1, sectionBase = len - 16 - (numSections * 8); i >= 0; i--) {
            int base = len - 16 - (numSections * 8) + (i * 8);
            char[] name = new char[4];
            name[0] = (char) buf.get(base + 0);
            name[1] = (char) buf.get(base + 1);
            name[2] = (char) buf.get(base + 2);
            name[3] = (char) buf.get(base + 3);
            int sectionLength = buf.getInt(base + 4);
            sectionBase -= sectionLength;
            byte[] data = new byte[sectionLength];
            buf.position(sectionBase);
            buf.get(data);
            sections[i] = new Section(new String(name), ByteBuffer.wrap(data));
        }

        System.out.println(Arrays.toString(sections));
        List<Object> symb = findSection(sections, "SYMB").unmarshal();
        System.out.println(symb);
        List<Object> crcs = findSection(sections, "CRCS").unmarshal();
        System.out.println(crcs);
        List<Object> data = findSection(sections, "DATA").unmarshal();
        System.out.println(data);
        List<String> prim = findSection(sections, "PRIM").unstring();
        findSection(sections, "CODE").disassemble(prim);
    }
}