Question

I know this question seems stupid, but it isn’t. I mean what is it exactly. I have a fair understanding of the parsing problem. I know BNF/EBNF, I’ve written grammar to parse simple context-free languages in one of my college courses. I just never met regular expressions before! The only thing that I remember about it is that context-free grammar can do all what regular expression can do.

Also, is it useful for a usual coding to parse strings? A simple example would be helpful.

Answer 1

Regular expressions first came around in mathematics and automata theory. A regular expression is simply something which defines a regular language. Without going too much into what “regular” means, think of a language as this way:

1. A language is made up of strings. English is a language, for example, and its made of strings.
2. Those strings are made of symbols – called an alphabet. So a string is just a concatenation of symbols from the alphabet.

So you could have a string (which is, remember, just a concatenation of symbols) which is not part of a given language. Or it could be in the language.

So lets say you have an alphabet made of 2 symbols: “0” and “1”. And lets say you want to create a language using the symbols in that alphabet. You could create the following rule: “In order for a string to be in my language, it must have only 0’s and 1’s in it.”

So these strings are in your language:

• 0
• 1
• 01
• 11001101
• …etc

These would not be in your language:

• 2
• peaches
• 00101105

That’s a pretty simple language. How about this: “In my language, each string [analogous to a valid ‘word’ in English] must being with a 0, and then can be followed by any number of 0’s or 1’s”

These are in the language:

• 0111111
• 0000000
• 0101010110001

These are not:

• 1
• 10000
• 1010
• 2000000

Well rather than defining the language using words – and these languages might get very complex (“1 followed by 2 0’s followed by any combination of 1’s and 0’s ending with a 1”), we came up with this syntax called “regular expressions” to define the language.

The first language would have been:

(0|1)*

(0 or 1, repeated infinitely)

The next: 0(0|1)*

(0, followed by any number of 0’s and 1’s).

So lets think of programming now. When you create a regex, you are saying “Look at this text. Return to me strings which match this pattern.” Which is really saying “I have defined a language. Return to me all strings within this document which are in my language.”

So when you create a “regex”, you are actually defining a regular language, which is a mathematical concept. (In actuality, perl-like regex define “nonregular” languages, but that is a separate issue.)

By learning the syntax of regex, you are learning the ins and outs of how to create a language, so that later you can see if a given string is “in” the language. Thus, commonly, people say that regex are for pattern matching – which is basically what you are doing when you look at a pattern, and see if it “matches” the rules for your language.

(this was long. does it answer your question at all?)