DmitrySoshnikov

ES7 String trim functions

String.prototype.trim ( )

Return result of StringTrim abstract operation passing this value as thisArg, and TrimBoth as the type.

String.prototype.trimRight ( )

Return result of StringTrim abstract operation passing this value as thisArg, and TrimRight as the type.

DmitrySoshnikov

/**
 * = LL parser =
 *
 * by Dmitry Soshnikov <dmitry.soshnikov@gmail.com>
 * MIT Style license
 *
 * Often one can see manually written LL parsers implemented as
 * recursive descent. Approach in this diff is a classical parse table
 * state machine.
 *

DmitrySoshnikov

/**
 * = Recursive descent parser =
 *
 * MIT Style License
 * By Dmitry Soshnikov <dmitry.soshnikov@gmail.com>
 *
 * In this short lecture we'll cover the basic (non-predictive, backtracking)
 * recursive descent parsing algorithm.
 *
 * Recursive descent is an LL parser: scan from left to right, doing

DmitrySoshnikov

/**
 * = AST printer =
 *
 * by Dmitry Soshnikov <[email protected]>
 * MIT Style License
 *
 * Once we have finished operating on our AST, we need to generate the
 * code back from that (possibly transformed) AST.
 *
 * We choose a very simple AST format here, which we used in the

DmitrySoshnikov

/**
 * Callable objects in ES2015.
 * by Dmitry Soshnikov <[email protected]>
 * MIT Style License
 */

class Callable extends Function {
  constructor() {
    super('(' + Callable.prototype.__call__ + ')();');
    return this.bind(this);

DmitrySoshnikov

Generated code optimizations and tricks

I've been playing with generated (native) code in order to test how compilers (specific or in general) translate a = a + 1 and a++, and found some other interesting optimizations. I used gcc compiler for this simple program:

int main() {
  int a = 0;
  a = a + 1;

  int b = 0;

DmitrySoshnikov

/**
 * LL(1) parser. Building parsing table, part 1: First and Follow sets.
 *
 * NOTICE: see full implementation in the Syntax tool, here:
 * https://github.com/DmitrySoshnikov/syntax/blob/master/src/sets-generator.js
 *
 * by Dmitry Soshnikov <[email protected]>
 * MIT Style License
 *
 * An LL(1)-parser is a top-down, fast predictive non-recursive parser,

DmitrySoshnikov

/**
 * Building LL(1) parsing table from First and Follow sets.
 *
 * by Dmitry Soshnikov <dmitry.soshnikov@gmail.com>
 * MIT Style License
 *
 * This diff is a continuation of what we started in the previous two diffs:
 *
 * Dependencies:
 *

DmitrySoshnikov

# Python
#
# Closures when they are used as default params, capture bindings
# from the environment one level above the class.
#

x = 1 # global env

def wrap():

DmitrySoshnikov

# Compiled code sharing between
# several functions.

arr = []
for k in [1, 2, 3]:
  arr.append(lambda: k)

print(arr[0].__code__ == arr[1].__code__) # True
print(arr[0].__code__.co_code) # bytecode