D: how to remove last char in string? - string

Question

I need to remove last char in string in my case it's comma (","):
foreach(line; fcontent.splitLines)
{
string row = line.split.map!(a=>format("'%s', ", a)).join;
writeln(row.chop.chop);
}
I have found only one way - to call chop two times. First remove \r\n and second remove last char.
Is there any better ways?

Answer 1

import std.array;
if (!row.empty)
row.popBack();

Answer 2

As it usually happens with string processing, it depends on how much Unicode do you care about.
If you only work with ASCII it is very simple:
import std.encoding;
// no "nice" ASCII literals, D really encourages Unicode
auto str1 = cast(AsciiString) "abcde";
str1 = str1[0 .. $-1]; // get slice of everything but last byte
auto str2 = cast(AsciiString) "abcde\n\r";
str2 = str2[0 .. $-3]; // same principle
In "last char" actually means unicode code point (http://unicode.org/glossary/#code_point) it gets a bit more complicated. Easy way is to just rely on D automatic decoding and algorithms:
import std.range, std.stdio;
auto range = "кириллица".retro.drop(1).retro();
writeln(range);
Here retro (http://dlang.org/phobos/std_range.html#.retro) is a lazy reverse iteration function. It takes any range (unicode string is a valid range) and returns wrapper that is capable of iterating it backwards.
drop (http://dlang.org/phobos/std_range.html#.drop) simply pops a single range element and ignores it. Calling retro again will reverse the iteration order back to normal, but now with the last element dropped.
Reason why it is different from ASCII version is because of nature of Unicode (specifically UTF-8 which D defaults to) - it does not allow random access to any code point. You actually need to decode them all one by one to get to any desired index. Fortunately, D takes care of all decoding for you hiding it behind convenient range interface.
For those who want even more Unicode correctness, it should be possible to operate on graphemes (http://unicode.org/glossary/#grapheme):
import std.range, std.uni, std.stdio;
auto range = "abcde".byGrapheme.retro.drop(1).retro();
writeln(range);
Sadly, looks like this specific pattern is not curently supported because of bug in Phobos. I have created an issue about it : https://issues.dlang.org/show_bug.cgi?id=14394

Answer 3

NOTE: Updated my answer to be a bit cleaner and removed the lambda function in 'map!' as it was a little ugly.
import std.algorithm, std.stdio;
import std.string;
void main(){
string fcontent = "I am a test\nFile\nwith some,\nCommas here and\nthere,\n";
auto data = fcontent
.splitLines
.map!(a => a.replaceLast(","))
.join("\n");
writefln("%s", data);
}
auto replaceLast(string line, string toReplace){
auto o = line.lastIndexOf(toReplace);
return o >= 0 ? line[0..o] : line;
}

Answer 4

module main;
import std.stdio : writeln;
import std.string : lineSplitter, join;
import std.algorithm : map, splitter, each;
enum fcontent = "some text\r\nnext line\r\n";
void main()
{
fcontent.lineSplitter.map!(a=>a.splitter(' ')
.map!(b=>"'" ~ b ~ "'")
.join(", "))
.each!writeln;
}

Answer 5

Take a look, I use this extension method to replace any last character or sub-string, for example:
string testStr = "Happy holiday!";<br>
Console.Write(testStr.ReplaceVeryLast("holiday!", "Easter!"));
public static class StringExtensions
{
public static string ReplaceVeryLast(this string sStr, string sSearch, string sReplace = "")
{
int pos = 0;
sStr = sStr.Trim();
do
{
pos = sStr.LastIndexOf(sSearch, StringComparison.CurrentCultureIgnoreCase);
if (pos >= 0 && pos + sSearch.Length == sStr.Length)
sStr = sStr.Substring(0, pos) + sReplace;
} while (pos == (sStr.Length - sSearch.Length + 1));
return sStr;
}
}