Using a Regular Expression to Find XML character references for control characters

Question

I need some help figuring out the regex for XML character references to control characters, in decimal or hex.

These sequences look like the following:







In other words, they are an ampersand, followed by a pound, followed by an optional 'x' to denote hexadecimal mode, followed by 1 to 4 decimal (or hexadecimal) digits, followed by a semicolon.

I'm specifically trying to identify those sequences where they contain (inclusive) numbers from decimal 0 to 31, or hexadecimal 0 to 1F.

Can anyone figure out the regex for this??

Do you want to accept leading zeroes? Such as your second case? If so, how many leading zeroes are acceptable? ``? — carlpett, Sep 15 '11 at 20:34
Yes, leading 0's need to be tolerated; there can be at most 4 digits, including leading 0's. Thus, 3, 03, 003, and 0003 are all valid and refer to the same character. — Ken Mason, Sep 15 '11 at 20:39

Nicolas Webb · Answer 1 · 2011-09-15T21:42:41.750

3

&#(0{0,2}[1-2]\d|000\d|0{0,2}3[01]|x0{0,2}[01][0-9A-Fa-f]);

It's not the most elegant, but it should work.

Verified in RegexBuddy.

results

edited Sep 15 '11 at 21:42

answered Sep 15 '11 at 20:41

Nicolas Webb

1,312
10
22

Both legitimate bugs. Teach me to dash these off in the middle of a release :). – Nicolas Webb Sep 15 '11 at 21:30
Change `0{0,3}[1-2]?\d` to `0{0,2}[1-2]\d|000\d` to correct the extra-leading-0 problem FJ is alluding to. – jpj625 Sep 15 '11 at 21:33
I was hoping to avoid Yet Another Alternation but yeah...that'll work. – Nicolas Webb Sep 15 '11 at 21:40

Justin Morgan - On strike · Accepted Answer · 2011-09-16T14:09:45.750

3

If you use a zero-width lookahead assertion to restrict the number of digits, you can write the rest of the pattern without worrying about the length restriction. Try this:

&#(?=x?[0-9A-Fa-f]{1,4})0*([12]?\d|3[01]|x0*1?[0-9A-Fa-f]);

Explanation:

(?=x?[0-9A-Fa-f]{1,4})  #Restricts the numeric portion to at most four digits, including leading zeroes.
0*                      #Consumes leading zeroes if there is no x.
[12]?\d                 #Allows decimal numbers 0 - 29, inclusive.
3[01]                   #Allows decimal 30 or 31.
x0*1?[0-9A-Fa-f]        #Allows hexadecimal 0 - 1F, inclusive, regardless of case or leading zeroes.

This pattern allows leading zeroes after the x, but the (?=x?[0-9A-Fa-f]{1,4}) part prevents them from occurring before an x.

edited Sep 16 '11 at 14:09

answered Sep 15 '11 at 21:02

Justin Morgan - On strike

30,035
12
80
104

2

A few good answers; this one wins for clarity and great explanation! – Ken Mason Sep 16 '11 at 17:23
@Ken - Glad I could help. BTW, `(?=x?[^;]{1,4})(0*[12]?\d|0*3[01]|x0*1?[0-9A-Fa-f]);` should also work if you find that clearer. – Justin Morgan - On strike Sep 16 '11 at 18:25

Andrew Clark · Answer 3 · 2011-09-15T20:38:18.397

2

I think the following should work:

&#(?:x0{0,2}[01]?[0-9a-fA-F]|0{0,2}(?:[012]?[0-9]|3[01]));

Here is a Rubular:
http://www.rubular.com/r/VEYx25Fdpj

edited Sep 15 '11 at 20:38

answered Sep 15 '11 at 20:30

Andrew Clark

202,379
35
273
306

Using a Regular Expression to Find XML character references for control characters

3 Answers3