Mismatch between CLI and JS usage due to echo

[alexandercerutti]

I was using this library via CLI to generate some mock IDs that should match actual IDs generated through our code, that uses NPM integrated.

By following the examples in the README, I did something like this:

echo 'site1.com/podc_article1' | pnpm base32
# OUTCOME: edmq8t9h5thpyv9fe1qp8ruzc5t78ub3dhjk22g

(yes, I'm using pnpm and a monorepo, so the only way to reach base32 without installing it globally, is this one).

However, when I run the same string in the JS code directly, without passing through the CLI, I get a sightly different ID:

base32.encode("site1.com/podc_article1")
// Outcome: edmq8t9h5thpyv9fe1qp8ruzc5t78ub3dhjk2

By using the CLI to decode both IDS, the string matches correctly:

echo 'edmq8t9h5thpyv9fe1qp8ruzc5t78ub3dhjk2' | pnpm base32 -d
echo 'edmq8t9h5thpyv9fe1qp8ruzc5t78ub3dhjk22g' | pnpm base32 -d

# Outcome: site1.com/podc_article1

I was remembering there were some kind of weird behaviors with echo, and a short research confirmed to me: echo prints an additional EOL character.

So, I tried using printf and the output became coherent with the usage via JS.

Checking the examples in the readme, ofc, do the same:

echo "Hello World" | pnpm base32
#> 91jprv3f41bpywkccg50

printf "Hello World" | pnpm base32
#> 91jprv3f41bpywkccg

I'm not sure this is something that can or should be fixed in either the library OR the bin file.
However, this thing caused me an issue, as the id I was generating, were for a NOSQL DB and I was performing the tests. It made me go crazy.

@agnoster can I help in any way with this? I was thinking of editing the README in the meantime.

[lmammino]

If i have to guess, based on a similar problem, this is because echo "something" will not just echo something but it will actually echo something\n. One way to fix this is by using echo -n "something".