How to replace part of multi-line string with sed - sed

Question

I want to replace newlines in a string with <br/> on all places not inside triple backticks statement(```)
String:
This is some requirements just talking about c++ with
`char` types supported by C++:
```
using SecureString = BasicSecureString<char>;
using WSecureString = BasicSecureString<wchar_t>;
using U16SecureString = BasicSecureString<char16_t>;
using U32SecureString = BasicSecureString<char32_t>;
```
And continuing to write
stuff
Expected result:
This is some requirements just talking about c++ with<br/>`char` types supported by C++:<br/>```
using SecureString = BasicSecureString<char>;
using WSecureString = BasicSecureString<wchar_t>;
using U16SecureString = BasicSecureString<char16_t>;
using U32SecureString = BasicSecureString<char32_t>;
```<br/>And continuing to write<br/>stuff
What I currently have is something like:
sed --null-data '/```.*```/!s/\n/<br\/>/g'
But it's only working on inputs which don't include the backticks.
Does someone have any hints?

Answer 1

With perl or awk
perl -0777 -pe 's#```.*?```(*SKIP)(*F)|\n#<br/>#sg'
awk '/```/{f=!f} {ORS = f ? RS : "<br/>"} 1'
The perl solution is similar to what you tried with sed.
-0777 will slurp entire input file, similar to sed -z but -z works by using ASCII NUL as line separator
```.*?```(*SKIP)(*F) to prevent this matching portion from being changed
|\n specified what should be matched
<br/> replacement string
s flag to allow . to match newline characters as well
With awk, the output record separator is changed based on value of f which changes whenever input has triple backticks. The advantage with this approach is that whole input doesn't have to be slurped.
If you don't wish to change the last newline in the file, use
perl -0777 -pe 's#```.*?```(*SKIP)(*F)|\n(?!\z)#<br/>#sg'