Using to curl to do urlencode in bash script

From: Balakrishnan Balasubramanian via curl-users <curl-users_at_lists.haxx.se>
Date: Sat, 24 Dec 2022 23:27:36 -0500

In a bash script, I want to generate a url whose query parameters need
to be url-encoded. I have the following helper function:

     gen_iv () {
         rhash=$1
         url=$2
         python3 <(: <<- EOS
                 from urllib.parse import urlencode
                 query = urlencode({"url": "$url", "rhash": "$rhash"})
                 print(f'https://t.me/iv?{query}')
                 EOS
         )
     }

I am trying to get rid of python dependency. I found solutions online
using awk, perl, jq. But all of them adds a dependency.

Hacked a below curl solution:

     gen_iv_curl () {
         rhash=$1
         url=$2
         (
                 curl -s -w '%{url}\n' --url-query "url=$url"
--url-query "rhash=$rhash" "http://127.0.0.1:0/iv" | sed
's!^http://127.0.0.1:0!https://t.me!'
         ) || true
     }

     ❯ gen_iv_curl feb0cb1f19cbc2
"https://daniel.haxx.se/blog/2022/12/23/the-curl-fragment-trick/"
https://t.me/iv?url=https%3a%2f%2fdaniel.haxx.se%2fblog%2f2022%2f12%2f23%2fthe-curl-fragment-trick%2f&rhash=feb0cb1f19cbc2

Seems to work fine but have some questions:

1. Is there a better/cleaner way to do this without extra dependency?
(No pure bash hack unless readable/bug free)

2. Are there any cases where this may not work?

3. Is it ok to rely on current curl behavior? In particular, does `-w`
guaranteed to work and print the url even if curl command fails? Or
could this change in future curl versions?
------------------------------------------------------------------------

-- 
Unsubscribe: https://lists.haxx.se/listinfo/curl-users
Etiquette:   https://curl.se/mail/etiquette.html

Received on 2022-12-25