Crystal: Discuss: Determines monospace(unicode) display width of a string

Created on 22 Oct 2017 · 9Comments · Source: crystal-lang/crystal

Crystal was built-in size and byte size in String, why not add monospace(unicode) display width, it's better in terminal and other usage.

Examples:

hello | size: 5, bytesize: 5, display width: 5
你好 | size: 2, bytesize: 6, display width: 4
ｈｅｌｌｏ | size: 5, bytesize: 15, display width: 10
здравствуйте | size: 12, bytesize: 24, display width: 12
namasté | size: 7, bytesize: 8, display width: 7
márħaban | size: 8, bytesize: 10, display width: 8
여보세요 | size: 4, bytesize: 12, display width: 8
おはよう | size: 4, bytesize: 12, display width: 8
مرحبًا | size: 6, bytesize: 12, display width: 5
здравствуйте | size: 12, bytesize: 24, display width: 12

terminal display width:

|hello|
+-----+
+----+
|你好|
+----+
+----------+
|ｈｅｌｌｏ|
+----------+
+------------+
|здравствуйте|
+------------+
+-------+
|namasté|
+-------+
+--------+
|márħaban|
+--------+
+--------+
|여보세요|
+--------+
+--------+
|おはよう|
+--------+
+-----+
|مرحبًا|
+-----+
+------------+
|здравствуйте|
+------------+

Ruby implementations:

Source

icyleaf

Most helpful comment

However this causes real problem, in crystal tool context for example:

2017-10-23 18 19 38

Japanese (also size != display_width) named variable breaks table column.

MakeNowJust on 23 Oct 2017

👍5

All 9 comments

Sounds like a shard idea to me. Perhaps a bit heavy and obscure for the stdlib.

RX14 on 22 Oct 2017

However this causes real problem, in crystal tool context for example:

2017-10-23 18 19 38

Japanese (also size != display_width) named variable breaks table column.

MakeNowJust on 23 Oct 2017

👍5

It was affect all CJK languages and emoji too

icyleaf on 23 Oct 2017

Couldn't we use libc's wcswidth() function for that?

Sija on 23 Oct 2017

I'm also not sure this belongs to the standard library, despite that fact that some tools won't work well (well, they work well, you just get a display that's not nice, but it's still easy to understand). But if wcswidth is portable we could use it (I don't know if that works well, though, there doesn't seem to be a standard way to know the width of a char).