Question

Possible Duplicate:
R regular expression: http matching

I’m working to capture URLs from a chunk of source code using regex.

The URL’s follow a pattern and are in the following form:

• http://www.google.com/…./1-1,1″
• http://www.google.com/…./1-2,2″
• http://www.google.com/…./1-20,20″

so far I can get to the url using the following code:

pattern = paste("1-", 1:20,",", 1:20, "\"", sep="")

this gives me a vector of:

• 1-1,1
• 1-2,2
• …..
• 1-20,20

then I can use these vectors to give me a position or the URLs inside the soure code .

Let’s say for example that the whole source code is simply: “http://www.google.com/word/1-1,1>”

`regexpr("1-1,1", test1k, TRUE)`

gives me:

[1] 28 attr(,”match.length”) [1] 5

this means that the pattern 1-1,1 starts at length 28. Given this information, how would I select the whole URL starting at “http://ww…” until the end “1-1,1>”.

I guess what I’m asking is, give the position 28, is there a function to select the nearest “http://” string going backwards (this marks the start of the URL). Similarly, given the position 28, is there a way to select the nearest “>” character going forward (this marks the end of the URL).

Answer 1

Rather than creating all possible combinations, just use the \\d character, which will match any digit. For example:

 regexpr("1-\\d+,\\d+", test1k, TRUE)

To select the whole URL, you want to start the regular expression with “http”, and then have it continue until the first time this pattern is matched. One simple way is:

regexpr("http.*?1-\\d+,\\d+", test1k, TRUE)

The .* pattern has three parts. The . matches any character, the * means “any number of that character, and the ? means that it’s not greedy (otherwise, this will take up the entire string from the first http to the last 1-\\d+,\\d+.

For example:

> regexpr("http.*1-?\\d+,\\d+", "The URL that you are looking for is http://www.google.com/word/1-1,1", TRUE)
[1] 37
attr(,"match.length")
[1] 32
attr(,"useBytes")
[1] TRUE