noderl/dist/bert.js

445 lines
14 KiB
JavaScript
Raw Normal View History

2024-01-28 11:32:24 +00:00
export const Types = {
BERT_START: 131,
SMALL_ATOM: 115,
ATOM: 100,
BINARY: 109,
SMALL_INTEGER: 97,
INTEGER: 98,
SMALL_BIG: 110,
LARGE_BIG: 111,
FLOAT: 99,
STRING: 107,
LIST: 108,
SMALL_TUPLE: 104,
LARGE_TUPLE: 105, // TODO: Implement.
MAP: 116,
NIL: 106,
NEW_FLOAT: 70,
ZERO: 0,
};
export const Lang = {
ELIXIR: 0,
ERLANG: 1,
};
export class Bert {
allBinariesAsString;
mapKeyAsAtom;
decodeUndefinedValues;
convention;
outputBuffer;
constructor(allBinariesAsString = false, mapKeyAsAtom = true, decodeUndefinedValues = true, convention = Lang.ELIXIR) {
this.allBinariesAsString = allBinariesAsString;
this.mapKeyAsAtom = mapKeyAsAtom;
this.decodeUndefinedValues = decodeUndefinedValues;
this.convention = convention;
this.outputBuffer = Buffer.alloc(10000000);
this.outputBuffer[0] = Types.BERT_START;
}
toAtom = toAtom;
toTuple = toTuple;
#encode = (obj, buffer) => this[`encode_${typeof obj}`](obj, buffer);
2024-01-28 13:23:50 +00:00
encode = (obj, copy = true) => {
2024-01-28 13:58:59 +00:00
const tailBuffer = this.#encode(obj, this.outputBuffer.subarray(1));
process.stderr.write(`tailbuffer length: ${tailBuffer.length}\n`);
2024-01-28 11:32:24 +00:00
if (tailBuffer.length === 0) {
throw new Error("Bert encode a too big term, encoding buffer overflow");
}
2024-01-28 13:23:50 +00:00
else if (copy) {
2024-01-28 13:58:59 +00:00
const ret = Buffer.alloc(tailBuffer.length + 1);
this.outputBuffer.copy(ret, 0, 0, ret.length + 1);
2024-01-28 13:23:50 +00:00
return ret;
2024-01-28 11:32:24 +00:00
}
else {
2024-01-28 13:58:59 +00:00
return this.outputBuffer.subarray(0, tailBuffer.length + 1);
2024-01-28 11:32:24 +00:00
}
};
#decode = (buffer) => {
const t = buffer[0];
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(1);
2024-01-28 11:32:24 +00:00
switch (t) {
case Types.SMALL_ATOM:
return this.decode_atom(buffer, 1);
case Types.ATOM:
return this.decode_atom(buffer, 2);
case Types.BINARY:
return this.decode_binary(buffer);
case Types.SMALL_INTEGER:
return this.decode_integer(buffer, 1, true);
case Types.INTEGER:
return this.decode_integer(buffer, 4);
case Types.SMALL_BIG:
return this.decode_big(buffer, 1);
case Types.LARGE_BIG:
return this.decode_big(buffer, 4);
case Types.FLOAT:
return this.decode_float(buffer);
case Types.NEW_FLOAT:
return this.decode_new_float(buffer);
case Types.STRING:
return this.decode_string(buffer);
case Types.LIST:
return this.decode_list(buffer);
case Types.SMALL_TUPLE:
return this.decode_tuple(buffer, 1);
// case Types.LARGE_TUPLE:
// return this.decode_large_tuple(buffer, 4)
case Types.NIL:
return this.decode_nil(buffer);
case Types.MAP:
return this.decode_map(buffer);
default:
throw new Error(`Unexpected BERT type: ${t}`);
}
};
decode = (buffer) => {
if (buffer[0] !== Types.BERT_START) {
2024-01-28 16:57:08 +00:00
throw new Error("Invalid BERT start magic");
2024-01-28 11:32:24 +00:00
}
2024-01-28 14:02:33 +00:00
const obj = this.#decode(buffer.subarray(1));
2024-01-28 11:32:24 +00:00
if (obj.rest.length !== 0) {
2024-01-28 16:57:08 +00:00
throw new Error(`Invalid BERT, remainder was: ${obj.rest.length}`);
2024-01-28 11:32:24 +00:00
}
return obj.value;
};
encode_string = (obj, buffer) => {
if (this.convention === Lang.ELIXIR) {
2024-01-28 13:58:59 +00:00
process.stderr.write(`encode string as elixir\n`);
2024-01-28 11:32:24 +00:00
return this.encode_binary(Buffer.from(obj), buffer);
}
else {
buffer[0] = Types.STRING;
2024-01-28 13:23:50 +00:00
buffer.writeUInt16BE(Buffer.byteLength(obj, "utf-8"), 1);
2024-01-28 11:32:24 +00:00
const len = buffer.write(obj, 3);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 3 + len);
2024-01-28 11:32:24 +00:00
}
};
encode_boolean = (obj, buffer) => {
if (obj) {
return this.#encode(this.toAtom("true"), buffer);
}
else {
return this.#encode(this.toAtom("false"), buffer);
}
};
encode_number = (obj, buffer) => {
const isInteger = obj % 1 === 0;
// Handle floats...
if (!isInteger) {
return this.encode_float(obj, buffer);
}
// Small int...
if (isInteger && obj >= 0 && obj < 256) {
buffer[0] = Types.SMALL_INTEGER;
buffer.writeUInt8(obj, 1);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 2);
2024-01-28 11:32:24 +00:00
}
// 4 byte int...
if (isInteger && obj >= -134217728 && obj <= 134217727) {
buffer[0] = Types.INTEGER;
buffer.writeInt32BE(obj, 1);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 5);
2024-01-28 11:32:24 +00:00
}
// Bignum...
const numBuffer = Buffer.alloc(buffer.length);
if (obj < 0) {
obj *= -1;
numBuffer[0] = 1;
}
else {
numBuffer[0] = 0;
}
let offset = 1;
while (obj !== 0) {
numBuffer[offset] = obj % 256;
obj = Math.floor(obj / 256);
offset++;
}
if (offset < 256) {
buffer[0] = Types.SMALL_BIG;
buffer.writeUInt8(offset - 1, 1);
numBuffer.copy(buffer, 2, 0, offset);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 2 + offset);
2024-01-28 11:32:24 +00:00
}
else {
buffer[0] = Types.LARGE_BIG;
buffer.writeUInt32BE(offset - 1, 1);
numBuffer.copy(buffer, 5, 0, offset);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 5 + offset);
2024-01-28 11:32:24 +00:00
}
};
encode_float = (obj, buffer) => {
// float...
buffer[0] = Types.NEW_FLOAT;
buffer.writeDoubleBE(obj, 1);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 9);
2024-01-28 11:32:24 +00:00
};
encode_object = (obj, buffer) => {
// Check if it's an atom, binary, or tuple...
if (obj === null) {
const undefinedAtom = this.convention === Lang.ELIXIR ? "nil" : "undefined";
return this.#encode(this.toAtom(undefinedAtom), buffer);
}
if (Buffer.isBuffer(obj)) {
return this.encode_binary(obj, buffer);
}
if (Array.isArray(obj)) {
return this.encode_array(obj, buffer);
}
if (obj.type === "Atom") {
return this.encode_atom(obj, buffer);
}
if (obj.type === "Tuple") {
return this.encode_tuple(obj, buffer);
}
// Treat the object as an associative array...
return this.encode_map(obj, buffer);
};
encode_atom = (obj, buffer) => {
buffer[0] = Types.ATOM;
buffer.writeUInt16BE(obj.value.length, 1);
const len = buffer.write(obj.value, 3);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 3 + len);
2024-01-28 11:32:24 +00:00
};
encode_binary = (obj, buffer) => {
2024-01-28 13:58:59 +00:00
process.stderr.write(`encode binary\n`);
2024-01-28 11:32:24 +00:00
buffer[0] = Types.BINARY;
buffer.writeUInt32BE(obj.length, 1);
2024-01-28 13:23:50 +00:00
const len = obj.copy(buffer, 5);
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 5 + len);
2024-01-28 11:32:24 +00:00
};
encode_undefined = (_obj, buffer) => {
return this.#encode(null, buffer);
};
encode_tuple = (obj, buffer) => {
if (obj.length < 256) {
buffer[0] = Types.SMALL_TUPLE;
buffer.writeUInt8(obj.length, 1);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(0, 2);
2024-01-28 11:32:24 +00:00
}
else {
buffer[0] = Types.LARGE_TUPLE;
buffer.writeUInt32BE(obj.length, 1);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(0, 5);
2024-01-28 11:32:24 +00:00
}
for (let i = 0; i < obj.length; ++i) {
buffer = this.#encode(obj[i], buffer);
}
return buffer;
};
encode_array = (obj, buffer) => {
if (obj.length === 0) {
buffer[0] = Types.NIL;
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 1);
2024-01-28 11:32:24 +00:00
}
buffer[0] = Types.LIST;
buffer.writeUInt32BE(obj.length, 1);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(0, 5);
2024-01-28 11:32:24 +00:00
for (let i = 0; i < obj.length; ++i) {
buffer = this.#encode(obj[i], buffer);
}
buffer[0] = Types.NIL;
2024-01-28 13:58:59 +00:00
return buffer.subarray(0, 1);
2024-01-28 11:32:24 +00:00
};
encode_map = (obj, buffer) => {
const keys = Object.keys(obj);
buffer[0] = Types.MAP;
buffer.writeUInt32BE(keys.length, 1);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(0, 5);
2024-01-28 11:32:24 +00:00
for (let i = 0; i < keys.length; ++i) {
const key = this.mapKeyAsAtom ? this.toAtom(keys[i]) : keys[i];
buffer = this.#encode(key, buffer);
buffer = this.#encode(obj[keys[i]], buffer);
}
return buffer;
};
decode_atom = (buffer, count) => {
const size = this.bytesToInt(buffer, count, true);
buffer = Buffer.from(buffer, count);
let value = buffer.toString("utf8", 0, size);
if (value === "true") {
value = true;
}
else if (value === "false") {
value = false;
}
else if (this.decodeUndefinedValues &&
this.convention === Lang.ELIXIR &&
value === "nil") {
value = null;
}
else if (this.decodeUndefinedValues &&
this.convention === Lang.ERLANG &&
value === "undefined") {
value = null;
}
else {
value = this.toAtom(value);
}
return {
value,
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(size),
2024-01-28 11:32:24 +00:00
};
};
decode_binary = (buffer) => {
const size = this.bytesToInt(buffer, 4, true);
buffer = Buffer.from(buffer, 4);
const bin = Buffer.alloc(size);
buffer.copy(bin, 0, 0, size);
return {
value: this.convention === Lang.ELIXIR && this.allBinariesAsString
? bin.toString()
: bin,
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(size),
2024-01-28 11:32:24 +00:00
};
};
decode_integer = (buffer, count, unsigned = false) => {
return {
value: this.bytesToInt(buffer, count, unsigned),
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(count),
2024-01-28 11:32:24 +00:00
};
};
decode_big = (buffer, count) => {
const size = this.bytesToInt(buffer, count, false);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(count);
2024-01-28 11:32:24 +00:00
let num = 0;
const isNegative = buffer[0] === 1;
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(1);
2024-01-28 11:32:24 +00:00
for (let i = size - 1; i >= 0; --i) {
const n = buffer[i];
if (num === 0) {
num = n;
}
else {
num = num * 256 + n;
}
}
if (isNegative) {
num = num * -1;
}
return {
value: num,
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(size),
2024-01-28 11:32:24 +00:00
};
};
decode_float = (buffer) => {
const size = 31;
return {
value: parseFloat(buffer.toString("utf8", 0, size)),
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(size),
2024-01-28 11:32:24 +00:00
};
};
decode_new_float = (buffer) => {
return {
value: buffer.readDoubleBE(0),
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(8),
2024-01-28 11:32:24 +00:00
};
};
decode_string = (buffer) => {
2024-01-28 17:19:00 +00:00
const sizeLen = this.convention === Lang.ELIXIR ? 4 : 2;
2024-01-28 17:12:26 +00:00
const size = this.bytesToInt(buffer, sizeLen, true);
buffer = buffer.subarray(sizeLen);
2024-01-28 11:32:24 +00:00
return {
value: buffer.toString("utf8", 0, size),
2024-01-28 13:58:59 +00:00
rest: buffer.subarray(size),
2024-01-28 11:32:24 +00:00
};
};
decode_list = (buffer) => {
const arr = [];
const size = this.bytesToInt(buffer, 4, true);
2024-01-28 17:12:26 +00:00
buffer = buffer.subarray(4);
2024-01-28 11:32:24 +00:00
for (let i = 0; i < size; ++i) {
const el = this.#decode(buffer);
arr.push(el.value);
buffer = el.rest;
}
const lastChar = buffer[0];
if (lastChar !== Types.NIL) {
throw new Error("List does not end with NIL");
}
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(1);
2024-01-28 11:32:24 +00:00
return {
value: arr,
rest: buffer,
};
};
decode_map = (buffer) => {
const map = {};
const size = this.bytesToInt(buffer, 4, true);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(4);
2024-01-28 11:32:24 +00:00
for (let i = 0; i < size; ++i) {
let el = this.#decode(buffer);
const key = el.value;
el = this.#decode(el.rest);
const value = el.value;
map[key] = value;
buffer = el.rest;
}
return {
value: map,
rest: buffer,
};
};
decode_tuple = (buffer, count) => {
const arr = [];
const size = this.bytesToInt(buffer, count, true);
2024-01-28 13:58:59 +00:00
buffer = buffer.subarray(count);
2024-01-28 11:32:24 +00:00
for (let i = 0; i < size; ++i) {
const el = this.#decode(buffer);
arr.push(el.value);
buffer = el.rest;
}
return {
value: this.toTuple(arr),
rest: buffer,
};
};
decode_nil = (buffer) => {
// nil is an empty list
return {
value: [],
rest: buffer,
};
};
bytesToInt = (buffer, length, unsigned) => {
switch (length) {
case 1:
return unsigned ? buffer.readUInt8(0) : buffer.readInt8(0);
case 2:
return unsigned ? buffer.readUInt16BE(0) : buffer.readInt16BE(0);
case 4:
return unsigned ? buffer.readUInt32BE(0) : buffer.readInt32BE(0);
}
};
binary_to_list = (str) => {
const ret = [];
for (let i = 0; i < str.length; ++i)
ret.push(str[i]);
return ret;
};
}
/**
* Convert object to atom.
*
*/
export const toAtom = (str) => ({
type: "Atom",
value: str,
toString: () => str,
});
/**
* Convert array of items to tuple.
*
*/
export const toTuple = (arr) => ({
...arr,
type: "Tuple",
length: arr.length,
value: arr,
toString: () => "{" +
arr
.filter((i) => i !== "")
.map((i) => i.toString())
.join(", ") +
"}",
});