Signed ADLER-32 algorithm implementation in JS (for the browser and nodejs). Emphasis on correctness and performance.
npm install adler-32
In the browser:
The browser exposes a variable ADLER32
ADLER32.buf(byte array or buffer) assumes the argument is a set of 8 bit
unsigned integers (e.g. nodejs
Buffer or simple array of ints)
ADLER32.bstr(binary string) interprets the argument as a binary string where
i-th byte is
ADLER32.str(string) interprets the argument as a standard JS string
make test will run the nodejs-based test. To run the in-browser tests, run a
local server and go to the
ctest directory. To update the browser artifacts,
To generate the bits file, use the
adler32 function from python zlib:
>>> from zlib import adler32>>> x="foo bar baz٪☃🍣">>> adler32(x)1543572022>>> adler32(x+x)-2076896149>>> adler32(x+x+x)2023497376
make perf will run algorithmic performance tests (which should justify certain
decisions in the code).
js-crc has more performance notes
Bit twiddling is much faster than taking the mod on Safari and older Firefoxes.
Instead of taking the literal mod 65521, it is faster to keep it in the integers
65536 ~ 15 mod 65521 so for nonnegative integer
a = (a >>> 16) * 65536 + (a & 65535) [equality]a ~ (a >>> 16) * 15 + (a & 65535) mod 65521
The mod is taken at the very end, since the intermediate result may exceed 65521
The magic numbers were chosen so as to not overflow a 31-bit integer:
F[n_] := Reduce[x*(x + 1)*n/2 + (x + 1)*(65521) < (2^31 - 1) && x > 0, x, Integers]F (* bstr: x \[Element] Integers && 1 <= x <= 3854 *)F (* ascii: x \[Element] Integers && 1 <= x <= 5321 *)
Subtract up to 4 elements for the unicode case.
Please consult the attached LICENSE file for details. All rights not explicitly granted by the Apache 2.0 license are reserved by the Original Author.