Remove trailing whitespaces from driver when converting

[joaopluigi]

Hey there,

I had some problems when trying to update my wslgit.sh to the last version 2019-02-08. Debugging it, I found that my variable MOUNTED_DRVFS was getting some trailing whitespaces when splitted on mount_len = split(_mount, mount_list, "\n");. For that reason, the result of, for example, to_unix_path_by_awk function would be something like:

/c <SPACE>/apps/cool_app

Which was leading to error.

I solved by removing the whitespaces using gsub on awk

# Usage: to_unix_path_by_awk "$path"
function to_unix_path_by_awk() {
	.
        .
        .
	for(i = 1; i <= mount_len ; i += 2 ) {
	  gsub(/ /, "", mount_list[i])
	  gsub(/ /, "", mount_list[i+1])
	  if(driver != mount_list[i]) continue;
	    suffix = substr($0, 3); gsub(/\\/, "/", suffix); gsub("//", "/", suffix);
	    print mount_list[i+1] suffix;
        .
        .
        .
}

I'm not sure if this was the correct approach. Also, I don't know why I was getting the trailing spaces.

See my changes on: https://github.com/hangxingliu/wslgit/compare/master...joaopluigi:fix/convertion-whitespace?expand=1

[hangxingliu]

@joaopluigi Thanks for your report. I am also curious why there are extra white characters.

Can you report some information to here too:

1. The output of following commands:

mount -t drvfs
mount -t drvfs | base64 -

2. Your Windows version and OS build code. (you can find them in Settings > System > About)

---

By the way, I may reply later because of my work.

[hangxingliu]

And I guess the problem is caused by function get_mounted_drvfs:

function get_mounted_drvfs() {
	# region need-to-be-replaced-in-unit-test
	#     The previous line is used for mark the following statments
	#     need to be replaced to other implementation for unit test (travis-CI)
	mount -t drvfs | awk '{
		if(split($0, lr, "type drvfs") < 2) next;
		if(split(lr[1], part, "on") < 2) next;

		drive = part[1];     gsub(/^\s/, "", drive);    gsub(/\s$/, "", drive);
		mount_to = part[2];  gsub(/^\s/, "", mount_to); gsub(/\s$/, "", mount_to);
		print toupper(drive) "\n" mount_to;
	}';
	# endregion need-to-be-replaced-in-unit-test
}

Maybe the \s in gsub(/^\s/, "", drive); gsub(/\s$/, "", drive); and gsub(/^\s/, "", mount_to); gsub(/\s$/, "", mount_to); should be replaced to \s+.

[joaopluigi]

Thanks for the answer! These are the info you asked for:

$ mount -t drvfs
C: on /c type drvfs (rw,noatime,uid=1000,gid=1000,metadata,case=off)
$ mount -t drvfs | base64 -
Qzogb24gL2MgdHlwZSBkcnZmcyAocncsbm9hdGltZSx1aWQ9MTAwMCxnaWQ9MTAwMCxtZXRhZGF0
YSxjYXNlPW9mZikK

Windows version: 1809 (I'm pretty sure I was getting this error on 1803 also)
OS build: 17763.316

[hangxingliu]

@joaopluigi And could you report the output of following command to here to, thanks

awk --version

Because I found a Q&A on the StackOverflow:

More generally, you can use [[:space:]] to match a space, a tab or a newline (GNU Awk also supports \s), and [[:blank:]] to match a space or a tab.

https://stackoverflow.com/questions/29512854/how-to-define-a-space-in-a-regular-expression-in-awk

[joaopluigi]

Ok, my awk is pointing to mawk

mawk 1.3.3 Nov 1996, Copyright (C) Michael D. Brennan

compiled limits:
max NF             32767
sprintf buffer      2040

[hangxingliu]

And you can try replace the implement of function get_mounted_drvfs to:

function get_mounted_drvfs() {
	# region need-to-be-replaced-in-unit-test
	#     The previous line is used for mark the following statments
	#     need to be replaced to other implementation for unit test (travis-CI)
	mount -t drvfs | awk '
	function trim(s) { gsub(/^[[:space:]]+/, "", s); gsub(/[[:space:]]+$/, "", s); return s; }
	{
		if(split($0, lr, "type drvfs") < 2) next;
		if(split(lr[1], part, "on") < 2) next;

		drive = trim(part[1]); mount_to = trim(part[2]);
		print toupper(drive) "\n" mount_to;
	}';
	# endregion need-to-be-replaced-in-unit-test
}

And try again, It maybe working fine for you. @joaopluigi

[joaopluigi]

This solution did not work @hangxingliu

[hangxingliu]

@joaopluigi Have you updated that wslgit.sh to your WSL? Because I replaced gawk to awk in the script and working fine. (But I forgot copy wslgit.sh into WSL before test succeed)

---

And the reason why don't replace space characters after get_mount_drvfs is because the wrong path will be generated if mount path contains space char. (Eg. /win disk/c, although basically no one will do like this)

---

P.s. And my next reply will be appear in here after a few hours. I have to take a nap

[joaopluigi]

@joaopluigi Have you updated that wslgit.sh to your WSL? Because I replaced gawk to awk in the script and working fine. (But I forgot copy wslgit.sh into WSL before test succeed)

I did the change directly on /usr/bin/wslgit.sh and tried. It only worked when I added back the gsub workaround

And the reason why don't replace space characters after get_mount_drvfs is because the wrong path will be generated if mount path contains space char. (Eg. /win disk/c, although basically no one will do like this)

You're right. I agree

P.s. And my next reply will be appear in here after a few hours. I have to take a nap

Haha go for it, get some rest

[hangxingliu]

@joaopluigi Sorry for reply late. (I am busying on the work and graduation thesis recently) And i found the the root of problem about this issue in my scripts. and I will fix it in the latest commit.

The root of problem are:

1. the awk binary in get_mount_drvfs is awk but not $AWK. so the result will be converted by gawk but not mawk if you installed gawk
   • This command could explain: readlink "$(readlink "$(which awk)")"
2. mawk doesn't support [[:space]] in regex expression.

I will be grateful if you can give me a feedback of my latest commit for fix this issue.

[joaopluigi]

It worked just fine!

Thanks for that