Wrong behavior of "runeSubstr" from module "unicode".

[lscrd]

Proc "runeSubstr" from module "unicode" is broken for negative lengths.

Example

import unicode

let s1 = "abcdef"
let s2 = "abcdéf"

echo s1.runeSubstr(0, -1)
echo s2.runeSubstr(0, -1)

Current Output

abcd
abcd�

Expected Output

abcde
abcde

Possible Solution

Replacing the negative indexes with the following works:

echo s1.runeSubstr(0, s1.runeLen - 1)
echo s2.runeSubstr(0, s2.runeLen - 1)

Additional Information

$ nim -v
Nim Compiler Version 1.4.6 [Linux: amd64]
...

Issue also exists in development version.

[ringabout]

runeSubStr is very fragile for negative number, confusing and hard to use(or fix).

A worse example:

import std/unicode

let s4 = "白头如新倾盖如故"
echo s4.runeSubStr(-10, 3)

/usercode/in.nim(4)      in
/playground/nim/lib/pure/unicode.nim(414) runeSubStr
/playground/nim/lib/system/fatal.nim(49) sysFatal
Error: unhandled exception: value out of range: -2 notin 0 .. 9223372036854775807 [RangeDefect]

Not to say

import std/unicode

let s4 = "白头如新倾盖如故"
doAssert s4.runeSubStr(-10, 10) == "白头如新倾盖如故"

works.

Well as documentation said, it is meant to work. But it is really terrible and unnatural.

import std/unicode

let s4 = "白头如新倾盖如故"
echo s4.runeSubStr(-10, -1)

This doesn't work

/usercode/in.nim(4)      in
/playground/nim/lib/pure/unicode.nim(412) runeSubStr
/playground/nim/lib/system/fatal.nim(49) sysFatal
Error: unhandled exception: value out of range: -2 notin 0 .. 9223372036854775807 [RangeDefect]