[BUG] Standard library replace_regex function fails to replace slash characters

[hdwalters]

Describe the bug
The standard library replace_regex function calls sed internally, with command s/.../.../, but if the search or replace string itself contains /, this causes a sed syntax error.

To Reproduce
Reproduce with one liner command:

$ amber -e 'import * from "std/text"; echo replace_regex("/path/to/file.txt", "/", "#", true)'
sed: -e expression #1, char 0: no previous regular expression

Expected behavior
It should behave like this:

$ amber -e 'import * from "std/text"; echo replace_regex("/path/to/file.txt", "/", "#", true)'
#path#to#file.txt

[hdwalters]

I already have a workaround for this, which is to make replace_regex escape slash characters:

pub fun replace_regex(source: Text, search: Text, replace: Text, extended: Bool = false): Text {
    unsafe {
        search = replace(search, "/", "\/")
        replace = replace(replace, "/", "\/")
        ...

[hdwalters]

Note, if slash characters are already escaped, this may have slightly unexpected behaviour:

$ amber -e 'import * from "std/text"; echo replace_regex("/path/to/file.txt", "\/", "#", true)'
/path/to/file.txt

[Mte90]

sed can use any symbol as separator so you can do s#something#any# and it will works.

[hdwalters]

sed can use any symbol as separator so you can do s#something#any# and it will works.

Yes I'm aware of that. Unfortunately we can't guarantee that callers of this stdlib function will not want to use "#" (or any other character we might pick) in the search or replace string; so it doesn't really give us anything.