Bug 6383 – Unpacking from dynamic array, lazy ranges

This is a spinoff of Issue 6365 (AutoTupleDeclaration). Issue 6365 is about syntax sugar to unpack tuples, but unpacking dynamic arrays or lazy ranges too is handy and commonly useful, because in D most arrays are not fixed-sized, and many functions return a lazy range, like lazy splitting, regular expression matches, etc. The 'lengths don't match' exception is already present in D and I think it's acceptable (when the lengths are statically known there is no need to test lengths at run time, so this exception throwing is not present, so it's nothrow code): int[] array = [1, 2, 3]; int[2] a2 = array[]; ------------------- Some usage examples, assign from a dynamic array, in D2: import std.string; void main() { auto s = " foo bar "; auto ss = s.split(); auto sa = ss[0]; auto sb = ss[1]; } In Python2.6: >>> sa, sb = " foo bar ".split() >>> sa 'foo' >>> sb 'bar' Proposed: import std.string; void main() { auto s = " foo bar "; auto (sa, sb) = s.split(); } ------------------- From lazy range, D2: import std.algorithm, std.conv, std.string; void main() { string s = " 15 27"; // splitter doesn't work yet for this // (here it doesn't strip the leading space) //auto xy = map!(to!int)(splitter(s)); auto xy = map!(to!int)(split(s)); int x = xy[0]; int y = xy[1]; } In Python: >>> x, y = map(int, " 15 27".split()) >>> x 15 >>> y 27 >>> from itertools import imap >>> x, y = imap(int, " 15 27".split()) >>> x 15 >>> y 27 Proposed: import std.algorithm, std.conv, std.string; void main() { string s = " 15 27"; (int x, int y) = map!(to!int)(splitter(s)); } ------------------- From a lazy range, another example. See: http://rosettacode.org/wiki/Range_expansion#D D2 code: import std.stdio, std.regex, std.string, std.conv, std.range; int[] rangeExpand(string txt) { int[] result; foreach (r; std.string.split(txt, ",")) { auto m = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); if (m[2].empty) result ~= to!int(m[1]); else result ~= array(iota(to!int(m[1]), to!int(m[3])+1)); } return result; } Python: import re def rangeexpand(txt): lst = [] for rng in txt.split(','): start,end = re.match('^(-?\d+)(?:-(-?\d+))?$', rng).groups() if end: lst.extend(xrange(int(start),int(end)+1)) else: lst.append(int(start)) return lst print(rangeexpand('-6,-3--1,3-5,7-11,14,15,17-20')) Proposed (just the important line), from dynamic array: auto (start,end) = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); Or better, from the range of regex captures: auto (start,end) = match(r, r"^(-?\d+)(-?(-?\d+))?$").captures; Similar unpacking from Python list (that is a dynamic array) or from lazy iterable is quite common in scripting-style Python code.

Sorry, this is not correct: auto (start,end) = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); auto (start,end) = match(r, r"^(-?\d+)(-?(-?\d+))?$").captures;

(In reply to comment #1) > Sorry, this is not correct: > > auto (start,end) = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); > auto (start,end) = match(r, r"^(-?\d+)(-?(-?\d+))?$").captures; I'm puzzled, in this case captures contain way more then 2 strings: 0 - whole match 1 - (-?\d+) 2 - (-?(-?\d+))? 3 - (-?\d+)

(In reply to comment #2) > (In reply to comment #1) > > Sorry, this is not correct: > > > > auto (start,end) = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); > > auto (start,end) = match(r, r"^(-?\d+)(-?(-?\d+))?$").captures; > > I'm puzzled, in this case captures contain way more then 2 strings: > 0 - whole match > 1 - (-?\d+) > 2 - (-?(-?\d+))? > 3 - (-?\d+) What are you puzzled about? I have already said that code I have written is not correct.

(In reply to comment #3) > (In reply to comment #2) > > (In reply to comment #1) > > > Sorry, this is not correct: > > > > > > auto (start,end) = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); > > > auto (start,end) = match(r, r"^(-?\d+)(-?(-?\d+))?$").captures; > > > > I'm puzzled, in this case captures contain way more then 2 strings: > > 0 - whole match > > 1 - (-?\d+) > > 2 - (-?(-?\d+))? > > 3 - (-?\d+) > > What are you puzzled about? I have already said that code I have written is not > correct. Ah, don't worry, I supposed you pasted a correction, sily me.

I think this would be very useful.(In reply to comment #2) > (In reply to comment #1) > > Sorry, this is not correct: > > > > auto (start,end) = array(match(r, r"^(-?\d+)(-?(-?\d+))?$").captures); > > auto (start,end) = match(r, r"^(-?\d+)(-?(-?\d+))?$").captures; > > I'm puzzled, in this case captures contain way more then 2 strings: > 0 - whole match > 1 - (-?\d+) > 2 - (-?(-?\d+))? > 3 - (-?\d+) So why couldn't "start" and "end" unpack the whole match and (-?\d+) and ignore the last two? I totally want to see unpacking in D2.

Not the place + is the subject of a current DIP.