strange character replacement on SO

Question

I added a comment to jQuery ID selector not working but something strange is happening. The end code reads "); but SO decided to change that to ";);.

I did not add a ; in front of the ).

To further explain what's happening here, the following:

"http://example.com/." http://example.com/.
"http://example.com." http://example.com.
"http://example.com" http://example.com

renders as:

Answer 1

To explain what happens here:

"http://xyz.com/a/b/c"

is turned into

"http://xyz.com/a/b/c"

The auto-linker sees this as a link to http://xyz.com/a/b/c&quot, so the semicolon is left over, and that's where it suddenly appears from.

Granted, this is isn't exactly the expected result.

Answer 2

I was able to reproduce it

We were trying

$("img[id=Char_" + Character + "]").attr("src", "image.xboxlive.com/global/t.584108a4/tile/0/2800d");

But as jao said in the comments

What i did was, I copied the string from the original post, then pasted it into the comment box (where it looked okay) $("img[id='Char_" + Character + "']").attr("src", "image.xboxlive.com/global/t.584108a4/tile/0/2800d";); but now it does the same thing here. it probably is some funny character replacing thing – jao

The code that jao is copy / pasting is

$("img[id=Char_" + Character + "]").attr("src", "http://image.xboxlive.com/global/t.584108a4/tile/0/2800d");

As balpha demonstrated in an example

"http://xyz.com/a/b/c"

is turned into

"http://xyz.com/a/b/c"

The http:// in the link is the key

Try the code with http:// and see

Answer 3

This behavior is fixed, in that it is working better for all use cases but not perfect in your use case.

As balpha mentioned 7 years ago, the problem is that link extraction happens on HTML and the regular expression that handles it matches http://xyz.com/a/b/c&quot as a URL in http://xyz.com/a/b/c". This is because want trailing semicolons to be treated as punctuation.

I've made things a little smarter now in that:

1. We never break up HTML entities when parsing URLs in rendered HTML.
2. We unserialize trailing entities from matched strings and try re-evaluating them.
3. We exclude trailing semicolons both before and after trailing entities.

Only trailing entities are considered because of other hacks that I was asked not to break.

The combination works nicely for URLs in quotes in lists. In your case, things are better in that the there's not a syntax error in your code, but not perfect because the ) prevents your " from being considered a trailing entity and puts it in the URL.

For code, it's still a best practice to wrap the whole thing in backticks, `, especially since the URL matcher strips off the protocol breaking the code in another way.