Re: [PATCH 1/3] scripts: add spelling_sanitizer.sh script

From: Andy Shevchenko <andy.shevchenko@gmail.com>
To: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Nicolas Dichtel <nicolas.dichtel@6wind.com>,
	Joe Perches <joe@perches.com>, Jason Baron <jbaron@akamai.com>,
	Stefani Seibold <stefani@seibold.net>,
	Jacob Keller <jacob.e.keller@intel.com>,
	Thomas Graf <tgraf@suug.ch>,
	Herbert Xu <herbert@gondor.apana.org.au>,
	Jens Axboe <axboe@kernel.dk>, Petr Mladek <pmladek@suse.com>,
	Sergey Senozhatsky <senozhatsky@chromium.org>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
	Rasmus Villemoes <linux@rasmusvillemoes.dk>,
	linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 1/3] scripts: add spelling_sanitizer.sh script
Date: Fri, 11 Jun 2021 10:58:49 +0300	[thread overview]
Message-ID: <CAHp75VfX95GVkd6iJ-aYNp7nO56nLSxgreE4fDXAm3h3p6VEjg@mail.gmail.com> (raw)
In-Reply-To: <20210611071241.16728-2-thunder.leizhen@huawei.com>

On Fri, Jun 11, 2021 at 10:19 AM Zhen Lei <thunder.leizhen@huawei.com> wrote:
>
> The file scripts/spelling.txt recorded a large number of
> "mistake||correction" pairs. These entries are currently maintained in
> order, but the results are not strict. In addition, when someone wants to
> add some new pairs, he either sort them manually or write a script, which
> is clearly a waste of labor. So add this script. It removes the duplicates
> first, then sort by correctly spelled words. Sorting based on misspelled
> words is not chose because it is uncontrollable.

chosen

...

> +#!/bin/sh

If you want to have stricter rules applied, use
#!/bin/sh -efu
in all your shell scripts, it will show you a lot of problems.

Missed SPDX.

> +src=spelling.txt

> +tmp=spelling_mistake_correction_pairs.txt

It will pollute the source tree, so use `mktemp` or utilize O=. In
case there is no O= supplied (or whatever equivalent to describe
output folder) you will get it in the source tree, so it needs to be
Git-ignored.

> +cd `dirname $0`

Useless use of dirname. Check for %, %%, #, and ## substitutions (`man sh`).
IIRC dirname equivalent is ${0%/*}.

> +# Convert the format of 'codespell' to the current
> +sed -r -i 's/ ==> /||/' $src
> +
> +# Move the spelling "mistake||correction" pairs into file $tmp

> +# There are currently 9 lines of comments in $src, so the text starts at line 10
> +sed -n '10,$p' $src > $tmp
> +sed -i '10,$d' $src

This is fragile, use proper comment line detection.

> +# Remove duplicates first, then sort by correctly spelled words
> +sort -u $tmp -o $tmp
> +sort -t '|' -k 3 $tmp -o $tmp

Can be one pipeline

> +# Append sorted results to comments
> +cat $tmp >> $src

I believe it can be done in a better way, but I was not thinking about it.

> +# Delete the temporary file
> +rm -f $tmp

What if the script will be trapped? It's good to handle SIGHUP I
suppose, so we won't leave garbage behind us.

> +cd - > /dev/null

-- 
With Best Regards,
Andy Shevchenko