looking for a better regex solution

Question

my input is:
<span question_number="18"> blah blah blah 1</span><span question_number="19"> blah blah blah 2</span>

and I want my regex to match this <span question_number="somenumber">xxxx</span> pattern
and the desired output is 1.somenumber 2.xxxx

I wrote a naive solution which could cover
<span question_number="18"> blah blah blah 1</span>
<span question_number="19"> blah blah blah 2</span>
notice: they are on different lines
the output is : 18, blah blah blah 1 and 19,blah blah blah 2

but when the input is <span question_number="18"> blah blah blah 1</span><span question_number="19"> blah blah blah 2</span>
which is on the same line

my output is 18, blah blah blah 1</span><span question_number="19"> blah blah blah 2

how could I bypass this problem?

Update: regex: /\<span question_number=(?:\")*(\d*)(?:\")*>(.*)<\/span>/ig

testinput:
case1 -> two lines of code
<span question_number="54">often graces doorways tied into ropes called</span>
<span question_number="54">often graces doorways tied into ropes called <i>ristras</i>.</span>
case2 -> one line of code
<span question_number="54">often graces doorways tied into ropes called</span><span question_number="54">often graces doorways tied into ropes called <i>ristras</i>.</span>

Update2:
This is not a dom , it is just a plain text that I want to process.

Update3: so my problem about Regex is solved, now I have a question about comparing the proessing speed between regex or dom operation ? how could implement such a test ?

Answer 1

Although you are not parsing an entire HTML document, your input obviously contains HTML elements.

In either case, Nokogiri is the library of choice:

require 'nokogiri'

input = '<span question_number="18"> blah blah blah 1</span><span question_number="19"> blah blah blah 2</span>'

doc = Nokogiri::HTML.fragment(input)
doc.css('span').map { |s| [s[:question_number], s.text] }
#=> [["18", " blah blah blah 1"], ["19", " blah blah blah 2"]]

Answer 2

If it really isn't HTML (hmm?) you could do it with

<span question_number="(\d+)">(.*?)<\/span>

See it here at regex101.

The problem with your original regex is that it's greedy. The part (.*) will match as many characters it can, making sure the remaining <\/span> still can be matched. So it finds the first <span... and matches up to the last </span>. My attempt at a solution is non-greedy (The ? in (.*?)), thus just matching to the first </span>.

Answer 3

Even though you insist that this isn't HTML, it sure looks and smells like it, and it can, in fact, easily be parsed by an HTML parser:

require 'nokogiri'

doc = Nokogiri::HTML.fragment <<~'HTML'
  <span question_number="54">often graces doorways tied into ropes called</span> 
  <span question_number="54">often graces doorways tied into ropes called <i>ristras</i>.</span>
HTML

doc.xpath('span').map {|span| next span[:question_number].to_i, span.text }
#=> [[54, "often graces doorways tied into ropes called"], [54, "often graces doorways tied into ropes called ristras."]]

It is not quite clear to me why you insist on not using an HTML parser for something that is obviously HTML.

Answer 4

I've looked at this problem as if there was a string involved - not a DOM environment. At the end of the day it's < and > that suddenly make it HTML. If you are in control of that string and you understand what it will contain and the boundaries of it then there are many solutions to a problem if it's specific to your needs.

Anyway, if you are looking for an answer and you know all of your questions absolutely live inside a <span> with an attribute of "question_number" then I guess you could do something like this. No Regex.

This is a simple version demonstrating how you could extract the information from a HTML string. For simplicity I've stuck it inside a textarea so you can see it actually working. You could copy this code and run it.

However, in reality you will probably want to get the innerHTML value of a container that you know contains all of the <span> tags.

I know there would be a number of different ways of solving this as many suggested but this is an answer to your specific need.

<html><body>
    <textarea id='htmlstring'>
        <div>Random HTML Before</div>
        <span question_number="18">blah blah blah 1</span>
        <span question_number="19">blah blah blah 1</span>
        <span question_number="21">blah blah blah 1</span>
        <span question_number="22">blah blah blah 1</span>
        <div>Random HTML After</div>
    </textarea>
    <script type="text/javascript">
        var t = document.getElementById('htmlstring');
        var q = t.value.split("<span question_number=");
        q.shift();
        for(var i in q){
            var d = q[i].split("</span>")[0];
            d = d.replace("\">","|");
            d = d.replace("\"","");
            d = d.split("|");
            alert("num="+d[0]+" val="+d[1]);
        }
    </script>
</body></html>