Hi www.google.com (http://www.google.com)
Hi http://google.com
Hi http://www.google.com
All the above lines must be detected and converted to URLs from text. These are the conditions
If I output these lines using *echo* then I should get a clickable output.
Hope this helps
Okay...so start by breaking those strings down. As an example, I'll do the first one for you.
1. Starts with "www."
2. Contains only valid domain name characters (a-z,0-9,-,.) ("google")
3. The domain portion should end with a "." and then 2-4 a-z characters (".com")
3. Possibly followed by a slash "/"
4. Possible followed by a URI (probably allow any characters here except spaces)
5. String should be terminated by a "/", "a-z" or "0-9". This will help us to not capture punctuation at the end of a URL. For example:
"www.google.com,"
GOOD: <a href="www.google.com">www.google.com</a>,
BAD: <a href="www.google.com,">www.google.com,</a>
So, by breaking down how to match the string, we can build our regular expression.
/(www\.[a-z\d-\.]*\.[a-z]{2,4}?\/?(?:[^\s]*[\/a-z\d]))/ig
Now, let's break this down:
/ : start your regex
( : begin capturing (we'll use this later)
www : match the text "www"
\. : match a period *
[ : begin a character set **
a-z : match any letter
\d : match any digit
- : match a hyphen
\. : match a period *
] : end the character set **
* : match multiple characters in the previous set
\. : match a period *
[ : start another character set **
a-z : match any letter
] : end the character set **
{2,4}: match the previous character set 2-4 times
? : perhaps followed by a...
\/ : forward slash
? : perhaps followed by a...
(?: : begin a non-capturing block ***
[ : begin another character set **
^ : matching any characters that are NOT
\s : spaces
] : end the character set **
* : match the previous character set as many times as possible
[ : start another character set **
\/ : matching a forward slash *
a-z : matching any letter
\d : matching any digit
] : end the character set
) : end the non-capturing block ***
) : end our capturing block
/ : end our regex
i : perform a case-insensitive search
g : perform the search globally (ie, match multiple urls)
* Regular expressions have some reserved characters, including "." and "/". They mean other things on their own, so to denote those actual characters, you need to escape them with a backslash.
** A character set matches any character in brackets ([ ]). It is usually followed by a +, *, or a number range {1, 3} to tell how many characters to match.
*** Parentheses indicate a "block". It's a way of organizing your regular expression into smaller chunks. On their own, they also capture the contents and create what is called a "back reference". This allows you to use the contents of your regex later. They are also useful when you want to apply an operator to a whole portion of a regular expression, like in our example. We have a piece that represents the URI -- [^\s]*[\/a-z\d]
That whole piece may or may not be present, so we want to precede it with a ?
?[^\s]*[\/a-z\d]
Problem is, that will only examine the first character set, so we need a block:
?([^\s]*[\/a-z\d])
Except now the problem is that we have created another capturing block that we don't need. To tell the regex engine not to create a back reference to that block, we use the ?: notation
Our back references are created in the order that they are started (although we only have one...) and are available via the variables $1...$n;
So, to use a back reference:
echo preg_replace("/a(b)c/ig", $1, "abc abc");
yields: "b b"
In this example, we've matched the abc twice, and created a back reference to the "b". We then replace all of our matches "abc" with our back reference "b".
So, to convert a url to a link, the basic formula is:
$myRegex = "/(www\.[a-z\d-\.]*\.[a-z]{2,4}?\/?(?:[^\s]*[\/a-z\d]))/ig";
preg_replace($myRegex, "<a href='$1'>$1</a>", $text);
I would recommend writing a separate regex for urls beginning with http...I'll let you go ahead and try your hand. Post what you come up with here, and we can go over it if you're having problems.
Also, check out this site for quick and easy testing: http://www.regextester.com/
Matt
