I’m currently in a project where the client has a custom, but not entirely consistent or known subset of utf-8.
They want us to keep the form content as it is, but remove the “bad” characters. Our current approach is to just forward everything as it is and wait for someone to complain. How TF am I supposed to remove a character without changing the message?
Yeah I had a backend with poor support for anything that wasn’t ASCII
PHP is like this. Poor Unicode support, but it treats strings as raw bytes so it usually works well enough. It turns out a programming language can take data from a form, save it to a database, then later load and render it, without having to know what those bytes actually mean, as long as the app or browser knows it’s UTF-8, for example through a Content-Type header or meta tag.
The tricky thing is the all the standard string manipulation functions (strlen, substr, etc) don’t handle Unicode properly at all and they deal with number of bytes rather than number of characters. You need to use the “multibyte” (Unicode-ready) equivalents like mb_substr, but a lot of PHP developers forget to do this and end up with string truncation code that cuts UTF-8 characters in half (e.g.if it’s truncating a long title with Emoji in it, it might cut off the title in the middle of the three bytes that represent the Emoji and only leave 1 or 2 of them)
You just need to ensure you validate character by character (NOT byte by byte) and allow characters in the Emoji Unicode ranges (which are well-defined in the Unicode standard). Using a library is a great idea though.
One of my projects was validation for form submission and emojis melted me. I gave up trying to do it from scratch and trusted a library.
I’m currently in a project where the client has a custom, but not entirely consistent or known subset of utf-8.
They want us to keep the form content as it is, but remove the “bad” characters. Our current approach is to just forward everything as it is and wait for someone to complain. How TF am I supposed to remove a character without changing the message?
Removed by mod
PHP is like this. Poor Unicode support, but it treats strings as raw bytes so it usually works well enough. It turns out a programming language can take data from a form, save it to a database, then later load and render it, without having to know what those bytes actually mean, as long as the app or browser knows it’s UTF-8, for example through a Content-Type header or meta tag.
The tricky thing is the all the standard string manipulation functions (
strlen,substr, etc) don’t handle Unicode properly at all and they deal with number of bytes rather than number of characters. You need to use the “multibyte” (Unicode-ready) equivalents likemb_substr, but a lot of PHP developers forget to do this and end up with string truncation code that cuts UTF-8 characters in half (e.g.if it’s truncating a long title with Emoji in it, it might cut off the title in the middle of the three bytes that represent the Emoji and only leave 1 or 2 of them)You just need to ensure you validate character by character (NOT byte by byte) and allow characters in the Emoji Unicode ranges (which are well-defined in the Unicode standard). Using a library is a great idea though.
Why not simply discard them?