Re: [PATCH] git-prompt: fix reading files with windows line endings

From: Robert Abel <rabel@robertabel.eu>
Date: Nov 30, 2017, 06:22 UTC
Message-ID: <cacbf41e-3b4a-99e2-a0e0-50bb4cd9e152@robertabel.eu>
In-Reply-To: <alpine.DEB.2.21.1.1711300100320.6482@virtualbox>

Hi Johannes,

On 30 Nov 2017 01:21, Johannes Schindelin wrote:

Show 6 quoted lines

> On Wed, 29 Nov 2017, Robert Abel wrote:
>> This means that it should be okay to just do
>>
>>> test -r "$f" && IFS=" \t\r\n" read "$@" < "$f"
> 
> I am afraid that this won't work: when I call

I managed to trick myself with that one, yes... Apparently I had already converted my HEAD back to Unix line endings.

However, I noticed that the behavior of read is apparently ambiguous for the last (or a single) variable:

From POSIX.1-2008:

Show 7 quoted lines

> If there are fewer vars than fields, the last var shall be set to a
> value comprising the following elements:
> - The field that corresponds to the last var in the normal assignment
>   sequence described above
> - The delimiter(s) that follow the field corresponding to the last var
> - The remaining fields and their delimiters, with trailing IFS white
>   space ignored

I read that last "ignored" as "trailing IFS white space shall not be appended". Apparently, people implementing read read it as "trailing IFS while space shall not be processed further"

Thus, the behavior for trailing IFS white space is different in case of one or two variables:

    printf '123 456\r\n' | while IFS=$' \t\r\n' read foo bar
    do
        printf 'foo: %s' "$foo" | hexdump -C
        printf 'bar: %s' "$bar" | hexdump -C
    done

This works as expected trimming the trailing \r:
    00000000  66 6f 6f 3a 20 31 32 33                           |foo: 123|
    00000008
    00000000  62 61 72 3a 20 34 35 36                           |bar: 456|
    00000008

While doing the same just reading a single variable

    printf '123 456\r\n' | while IFS=$' \t\r\n' read foo
    do
        printf 'foo: %s' "$foo" | hexdump -C
        printf 'bar: %s' "$bar" | hexdump -C
    done

prints

    00000000  66 6f 6f 3a 20 31 32 33  20 34 35 36 0d           |foo:
123 456.|
    0000000d
    00000000  62 61 72 3a 20                                    |bar: |
    00000005

Notice the 0d at the end of foo, which didn't get trimmed.

So reading a dummy variable along with the actual content variable works for git-prompt:

    __git_eread ()
    {
        local f="$1"
        local dummy
        shift
        test -r "$f" && IFS=$'\r\n' read "$@" dummy < "$f"
    }

I feel like this would be the most readable solution thus far.

Regards,

Robert