How to implement unsigned right shift for BigInt in JavaScript? - javascript

I tried this sort of implementation, but it doesn't appear to be working.
function urs32(n, amount) {
const mask = (1 << (32 - amount)) - 1
return (n >> amount) & mask
}
function flip32(n) {
const mask = (1 << 32) - 1
return ~n & mask
}
log(~0b10101010 >>> 0, urs32(~0b10101010, 0))
log(~0b10101010 >>> 0, flip32(0b10101010))
function log(a, b) {
console.log(a.toString(2), b.toString(2))
}
I would expect for a to equal b in both cases, if done right. Basically I am trying to flip 32-bits (so 1's become 0s, 0's become 1s). I see that 1 << 32 === 0, so to get the value, I do 2 ** 32, but still doesn't work.
How do you implement the equivalent of ~n >>> 0 on a BigInt?
Basically what I am trying to do is create the countLeadingOnes functions (out of the countLeadingZeroes functions), like so:
const LEADING_ZERO_BIT_TABLE = makeLeadingZeroTable()
function makeLeadingZeroTable() {
let i = 0
const table = new Uint8Array(256).fill(0)
while (i < 256) {
let count = 8
let index = i
while (index > 0) {
index = (index / 2) | 0
count--
}
table[i] = count
i++
}
return table
}
function countLeadingZeroes32JS(n)
{
let accum = LEADING_ZERO_BIT_TABLE[n >>> 24];
if (accum === 8) {
accum += LEADING_ZERO_BIT_TABLE[(n >>> 16)]
}
if (accum === 16) {
accum += LEADING_ZERO_BIT_TABLE[(n >>> 8)]
}
if (accum === 24) {
accum += LEADING_ZERO_BIT_TABLE[ n ]
}
return accum;
}
function countLeadingZeroes16JS(n)
{
let accum = LEADING_ZERO_BIT_TABLE[n >>> 8]
if (accum === 8) {
accum += LEADING_ZERO_BIT_TABLE[n]
}
return accum;
}
function countLeadingZeroes8JS(n)
{
return LEADING_ZERO_BIT_TABLE[n]
}
console.log('countLeadingZeroes32JS', countLeadingZeroes32JS(0b10100010001000100010001000100010))
console.log('countLeadingZeroes32JS', countLeadingZeroes32JS(0b00100010001000100010001000100010))
console.log('countLeadingZeroes32JS', countLeadingZeroes32JS(0b00000010001000100010001000100010))
console.log('countLeadingZeroes16JS', countLeadingZeroes16JS(0b1010001000100010))
console.log('countLeadingZeroes16JS', countLeadingZeroes16JS(0b0010001000100010))
console.log('countLeadingZeroes16JS', countLeadingZeroes16JS(0b0000001000100010))
console.log('countLeadingZeroes16JS', countLeadingZeroes16JS(0b0000000000100010))
console.log('countLeadingZeroes8JS', countLeadingZeroes8JS(0b10100010))
console.log('countLeadingZeroes8JS', countLeadingZeroes8JS(0b00100010))
console.log('countLeadingZeroes8JS', countLeadingZeroes8JS(0b00000010))
function countLeadingOnes32JS(n) {
return countLeadingZeroes32JS(~n >>> 0)
}
function countLeadingOnes16JS(n) {
return countLeadingZeroes16JS(~n >>> 0)
}
function countLeadingOnes8JS(n) {
return countLeadingZeroes8JS(~n >>> 0)
}
console.log('countLeadingOnes32JS', countLeadingZeroes32JS(0b00100010001000100010001000100010))
console.log('countLeadingOnes32JS', countLeadingZeroes32JS(0b11100010001000100010001000100010))
console.log('countLeadingOnes32JS', countLeadingZeroes32JS(0b11111100001000100010001000100010))
console.log('countLeadingOnes16JS', countLeadingOnes16JS(0b0100001000100010))
console.log('countLeadingOnes16JS', countLeadingOnes16JS(0b1111110000100010))
console.log('countLeadingOnes16JS', countLeadingOnes16JS(0b1111111111000010))
console.log('countLeadingOnes8JS', countLeadingOnes8JS(0b01000010))
console.log('countLeadingOnes8JS', countLeadingOnes8JS(0b11000010))
console.log('countLeadingOnes8JS', countLeadingOnes8JS(0b11111100))
But it appears that ~n >>> 0 doesn't work on 32-bit integers. How to get this working properly?

How to implement unsigned right shift for BigInt in JavaScript?
Unsigned right-shift is difficult to define meaningfully for arbitrary-size integers, so before you (or anyone) can implement it, you'll have to decide how you want it to behave.
That said, considering the rest of this question, I don't see why you would even need this.
I would expect for a to equal b in both cases
Why would it? Unsigned right-shift and bit flipping are different operations and produce different results.
I see that 1 << 32 === 0
Nope, 1 << 32 === 1. JavaScript (like x86 CPUs) performs an implicit &31 on the shift amount, so since 32 & 31 === 0, ... << 32 is the same as ... << 0.
How do you implement the equivalent of ~n >>> 0 on a BigInt?
The equivalent of ~n is ~n. (That's not a typo. It's literally the same thing.)
The equivalent of ... >>> 0 is BigInt.asUintN(32, ...). (Note that neither the Number version nor the BigInt version shifts anything, so this doesn't answer your headline question "how to implement USR for BigInt".)
it appears that ~n >>> 0 doesn't work on 32-bit integers.
It sure does work. In fact, it only works on 32-bit integers.
The >>> 0 part is completely unnecessary though, you could just drop it.
The reason why this line:
console.log('countLeadingOnes32JS', countLeadingZeroes32JS(0b00100010001000100010001000100010))
isn't producing the number of leading ones is because the function it's calling is ...Zeroes...; an apparent copy-paste bug.
The reason why countLeadingOnes16JS isn't working correctly is because ~ in JavaScript always flips 32 bits. Since a 16-bit number's 32-bit representation has (at least) 16 leading zeros, those all become ones after flipping, and countLeadingZeroes16JS gets an input that's far bigger than it can handle: LEADING_ZERO_BIT_TABLE[n >>> 8] looks up an element that doesn't exist in the table, because the result of n >>> 8 is a 24-bit number in this case, not an 8-bit number. The solution is to use a mask after flipping; a valid implementation of clo16 might be:
function countLeadingOnes16(n) {
return countLeadingZeroes16(~n & 0xFFFF);
}
No BigInts and no >>> 0 required.
countLeadingOnes8 is similar.
You may want to read https://en.wikipedia.org/wiki/Two%27s_complement (or some other description of that concept) to understand what's going on with bitwise operations on negative numbers.
You may also want to learn how to debug your own code. There's a range of techniques: for example, you could have:
inserted console.log statements for intermediate results,
or stepped through execution in a debugger,
or simply evaluated small snippets in the console,
any of which would have made it very easy for you to see what's happening on the path from input number to end result.
For anyone else reading this: there's Math.clz32, which is highly efficient because it gets compiled to a machine instruction, so implementing countLeadingZeros by hand is unnecessary and wasteful. For smaller widths, just subtract: function clz8(n) { return Math.clz32(n) - 24; }

Related

The meaning of this `long` function (two's complement and bit-shifting)

I have encountered this function:
const LIMIT32 = 2147483648; // The limit at which a 32-bit number switches signs == 2 ^ 31
function long(v) {
// Two's complement
if (v >= LIMIT32) {
v = -(2 * LIMIT32 - v);
}
return [(v >> 24) & 0xFF, (v >> 16) & 0xFF, (v >> 8) & 0xFF, v & 0xFF];
}
// e.g.
[-3, -2, -1, 0, 1,
-2147483649,-2147483648,-2147483647,
2147483647,2147483648,2147483649].forEach(x =>
console.log(`${x}: ${long(x)}`)
);
I'm wondering generally what this function is doing (why it's returning an array, and what the array elements are).
Then I'm wondering why it takes the v and does what looks like a sign flip and some multiplication.
Finally, the meaning of the bitshift and & operations for each item, why it's as multiples of 8, and why they chose 0xFF.
I'm wondering generally what this function is doing (why it's returning an array, and what the array elements are).
It returns an array of the 4 bytes that make up a int32 value. Why someone wrote the code to do that? I don't know.
Then I'm wondering why it takes the v and does what looks like a sign flip and some multiplication.
Because that's how int32 works: 0x7FFFFFFF + 1 === -0x80000000.
Although it is unnecessary in this code, the bit operations will take care of everything.
Finally, the meaning of the bitshift and & operations for each item, why it's as multiples of 8, and why they chose 0xFF.
Getting the distinct bytes of the int32, each one 8 bit long.

Decomposing a value into results of powers of two

Is it possible to get the integers that, being results of powers of two, forms a value?
Example:
129 resolves [1, 128]
77 resolves [1, 4, 8, 64]
I already thought about using Math.log and doing also a foreach with a bitwise comparator. Is any other more beautiful solution?
The easiest way is to use a single bit value, starting with 1 and shift that bit 'left' until its value is greater than the value to check, comparing each bit step bitwise with the value. The bits that are set can be stored in an array.
function GetBits(value) {
var b = 1;
var res = [];
while (b <= value) {
if (b & value) res.push(b);
b <<= 1;
}
return res;
}
console.log(GetBits(129));
console.log(GetBits(77));
console.log(GetBits(255));
Since shifting the bit can be seen as a power of 2, you can push the current bit value directly into the result array.
Example
You can adapt solutions from other languages to javascript. In this SO question you'll find some ways of solving the problem using Java (you can choose the one you find more elegant).
decomposing a value into powers of two
I adapted one of those answers to javascript and come up with this code:
var powers = [], power = 0, n = 129;// Gives [1,128] as output.
while (n != 0) {
if ((n & 1) != 0) {
powers.push(1 << power);
}
++power;
n >>>= 1;
}
console.log(powers);
Fiddle
Find the largest power of two contained in the number.
Subtract from the original number and Add it to list.
Decrement the exponent and check if new 2's power is less than the number.
If less then subtract it from the original number and add it to list.
Otherwise go to step 3.
Exit when your number comes to 0.
I am thinking of creating a list of all power of 2 numbers <= your number, then use an addition- subtraction algorithm to find out the group of correct numbers.
For example number 77:
the group of factors is { 1,2,4,8,16,32,64} [ 64 is the greatest power of 2 less than or equal 77]
An algorithm that continuously subtract the greatest number less than or equal to your number from the group you just created, until you get zero.
77-64 = 13 ==> [64]
13-8 = 7 ==> [8]
7-4 = 3 ==> [4]
3-2 = 1 ==> [2]
1-1 = 0 ==> [1]
Hope you understand my algorithm, pardo my bad english.
function getBits(val, factor) {
factor = factor || 1;
if(val) {
return (val % 2 ? [factor] : []).concat(getBits(val>>1, factor*2))
}
return [];
}
alert(getBits(77));

How to get value of a UInt64, toString only way?

I have this:
"ctypes.UInt64("7")"
It is returned by this:
var chars = SendMessage(hToolbar, TB_GETBUTTONTEXTW, local_tbb.idCommand, ctypes.voidptr_t(0));
so
console.log('chars=', chars, chars.toString(), uneval(chars));
gives
'chars=' 'UInt64 { }' "7" 'ctypes.UInt64("7")'
So I can get the value by going chars.toString(), but I have to run a parseInt on that, is there anyway to read it like a property? Like chars.UInt64?
The problem with 64-bit integers in js-ctypes is that Javascript lacks a compatible type. All Javascript numbers are IEEE double precision floating point numbers (double), and those can represent 53-bit integers at most. So you shouldn't even be trying to parse the int yourself, unless you know for a fact that the result would fit into a double. E.g. You cannot know this for pointers.
E.g. consider the following:
// 6 * 8-bit = 48 bit; 48 < 53, so this is OK
((parseInt("0xffffffffffff", 16) + 2) == parseInt("0xffffffffffff", 16)) == false
// However, 7 * 8-bit = 56 bit; 56 < 53, so this is not OK
((parseInt("0xffffffffffffff", 16) + 2) == parseInt("0xffffffffffffff", 16)) == true
// Oops, that compared equal, because a double precision floating point
// cannot actual hold the parseInt result, which is still well below 64-bit!
Lets deal with 64-bit integers in JS properly...
If you just want to comparisons, use UInt64.compare()/Int64.compare(), e.g.
// number == another number
ctypes.UInt64.compare(ctypes.UInt64("7"), ctypes.UInt64("7")) == 0
// number != another number
ctypes.UInt64.compare(ctypes.UInt64("7"), ctypes.UInt64("6")) != 0
// number > another number
ctypes.UInt64.compare(ctypes.UInt64("7"), ctypes.UInt64("6")) > 0
// number < another number
ctypes.UInt64.compare(ctypes.UInt64("7"), ctypes.UInt64("8")) < 0
If you need the result, but are not sure it is a 32-bit unsigned integer, you can detect if you're dealing with 32 bit unsigned integers that are just packed into Uint64:
ctypes.UInt64.compare(ctypes.UInt64("7"), ctypes.UInt64("0xffffffff")) < 0
And the analog for 32-bit signed integers in Int64, but you need to compare minimum and maximum:
ctypes.Int64.compare(ctypes.Int64("7"), ctypes.Int64("2147483647")) < 0 &&
ctypes.Int64.compare(ctypes.Int64("7"), ctypes.Int64("-2147483648")) > 0
So, once you know or detected that something will fit into a JS double, it is safe to call parseInt on it.
var number = ...;
if (ctypes.UInt64.compare(number, ctypes.UInt64("0xffffffff")) > 0) {
throw Error("Whoops, unexpectedly large value that our code would not handle correctly");
}
chars = parseInt(chars.toString(), 10);
(For the sake of completeness, there is also UInt64.hi()/Int64.hi() and UInt64.lo()/Int64.lo() to get the high and low 32-bits for real 64-bit integers and do 64-bit integer math yourself (e.g.), but beware of endianess).
PS: The return value of SendMessage is intptr_t not uintptr_t, which is important here because SendMessage(hwnd, TB_GETBUTTONTEXT, ...) may return -1 on failure!
So putting all this together (untested):
var SendMessage = user32.declare(
'SendMessageW',
ctypes.winapi_abi,
ctypes.intptr_t,
ctypes.voidptr_t, // HWND
ctypes.uint32_t, // MSG
ctypes.uintptr_t, // WPARAM
ctypes.intptr_t // LPARAM
);
// ...
var chars = SendMessage(hToolbar, TB_GETBUTTONTEXTW, local_tbb.idCommand, ctypes.voidptr_t(0));
if (ctypes.Int64.compare(chars, ctypes.Int64("0")) < 0) {
throw new Error("TB_GETBUTTONTEXT returned a failure (negative value)");
}
if (ctypes.Int64.comare(chars, ctypes.Int64("32768")) > 0) {
throw new Error("TB_GETBUTTONTEXT returned unreasonably large number > 32KiB");
}
chars = parseInt(chars.toString());

Replace modulus by function

When I try to do 8067 % 80.67 I get 80.66999999999983, instead of 0 beacuse of known floating point javascript behaviour.
So I went and made a function for this, to avoid floating point javascript errors.
function math(a, b) {
var left = Math.abs(a),
times = 1,
abs = a >= 0 ? 1 : -1;
while (Math.abs(a) >= b * times) {
left -= b;
times++;
}
return (a - (b * (times - 1))) * abs;
}
http://jsfiddle.net/s5w3C/
So my question is: is this usefull, ie a good tool to use instead of %? is there cases where this will also give falsy results like the modulus % oprator.
I am looking for a tools to calculate % consistently.
I didn't really inspect the algorithm for correctness, but if you care about efficiency, this is a bad idea. Basically, the larger the input, the slower your code will execute.
I think any fix will only work to a certain level of accuracy and for certain sized numbers. Perhaps something like the following will be sufficient:
function nearlyMod(a, b) {
var precision = ('' + b).split('.').length;
var estimate = (a % b).toFixed(precision);
return estimate == b ? 0 : +estimate;
}
console.log(nearlyMod(8067, 80.66)); // 1
console.log(nearlyMod(8067, 80.67)); // 0
console.log(nearlyMod(8067, 80.68)); // 79.68
It tests if the result is an even divisor within the precision of the original number. If so, it returns 0, otherwise it returns a number to the same precision (which may or may not be what you want).
The result is always a number (the value returned from toFixed is a string, hence +estimate).
A better name might be "roundedMod" or similar.

How do I swap endian-ness (byte order) of a variable in javascript

I am receiving and sending a decimal representation of two little endian numbers. I would like to:
shift one variable 8 bits left
OR them
shift a variable number of bits
create 2 8 bit numbers representing the first and second half of the 16 bit number.
javascript (according to https://developer.mozilla.org/en/JavaScript/Reference/Operators/Bitwise_Operators) uses big endian representation when shifting...
endianness is a bit foreign to me (I am only 90 percent sure that my outlined steps are what i want.) so swapping is a bit dizzying. please help! I only really need to know how to swap the order in an efficient manner. (I can only think of using a for loop on a toString() return value)
function swap16(val) {
return ((val & 0xFF) << 8)
| ((val >> 8) & 0xFF);
}
Explanation:
Let's say that val is, for example, 0xAABB.
Mask val to get the LSB by &ing with 0xFF: result is 0xBB.
Shift that result 8 bits to the left: result is 0xBB00.
Shift val 8 bits to the right: result is 0xAA (the LSB has "dropped off" the right-hand side).
Mask that result to get the LSB by &ing with 0xFF: result is 0xAA.
Combine the results from steps 3 and step 5 by |ing them together:
0xBB00 | 0xAA is 0xBBAA.
function swap32(val) {
return ((val & 0xFF) << 24)
| ((val & 0xFF00) << 8)
| ((val >> 8) & 0xFF00)
| ((val >> 24) & 0xFF);
}
Explanation:
Let's say that val is, for example, 0xAABBCCDD.
Mask val to get the LSB by &ing with 0xFF: result is 0xDD.
Shift that result 24 bits to the left: result is 0xDD000000.
Mask val to get the second byte by &ing with 0xFF00: result is 0xCC00.
Shift that result 8 bits to the left: result is 0xCC0000.
Shift val 8 bits to the right: result is 0xAABBCC (the LSB has "dropped off" the right-hand side).
Mask that result to get the second byte by &ing with 0xFF00: result is 0xBB00.
Shift val 24 bits to the right: result is 0xAA (everything except the MSB has "dropped off" the right-hand side).
Mask that result to get the LSB by &ing with 0xFF: result is 0xAA.
Combine the results from steps 3, 5, 7 and 9 by |ing them together:
0xDD000000 | 0xCC0000 | 0xBB00 | 0xAA is 0xDDCCBBAA.
Such function can be used to change endianness in js:
const changeEndianness = (string) => {
const result = [];
let len = string.length - 2;
while (len >= 0) {
result.push(string.substr(len, 2));
len -= 2;
}
return result.join('');
}
changeEndianness('AA00FF1234'); /// '3412FF00AA'
Use the << (bit shift) operator. Ex: 1 << 2 == 4.
I really think that the underlying implementation of JavaScript will use whatever endianess the platform it is running on is using. Since you cannot directly access memory in JavaScript you won't ever have to worry about how numbers are represented physically in memory. Bit shifting integer values always yield the same result no matter the endianess. You only see a difference when looking at individual bytes in memory using pointers.
Here is a oneliner for arrays to swap between big and little endian (and vise versa). The swapping is done using reverse on byte level. I guess for large arrays, it is more efficient than looping over scalar swap function.
function swapbyte(x) {
return new Float64Array(new Int8Array(x.buffer).reverse().buffer).reverse()
}
// Example
buf = new ArrayBuffer(16); // for 2 float64 numbers
enBig = new Float64Array(buf);
enBig[0] = 3.2073756306779606e-192;
enBig[1] = 2.7604354232023903e+199;
enLittle = swapbyte(enBig)
// two famous numbers are revealed
console.log(enLittle)
// Float64Array [ 6.283185307179586, 2.718281828459045 ]
// swapping again yields the original input
console.log(swapbyte(enLittle))
// Float64Array [ 3.2073756306779606e-192, 2.7604354232023903e+199 ]

Categories