How to replace tag in an XHTML file with a regular expression replace?

Samir · Apr 07, 2017#12017-04-07T13:39+00:00

Hi everyone

I have a task

Input
»Er ist ein Freund von Stone. Es geht um eine Zimmerrenovierung, kann also spät werden, weil meine anderen Termine jetzt alle nach hinten rutschen.«
Freund von Stone. Es geht um eine Zimmerrenovierung, kann also spät werden, weil meine anderen Termine jetzt alle nach hinten rutschen.«

Output
»Er ist ein Freund von Stone. Es geht um eine Zimmerrenovierung, kann also spät werden, weil meine anderen Termine jetzt alle nach hinten rutschen.«
Freund von Stone. Es geht um eine Zimmerrenovierung, kann also spät werden, weil meine anderen Termine jetzt alle nach hinten rutschen.«

How to replace tag with , with like adobe example

fleggy · Apr 08, 2017#22017-04-08T06:48+00:00

Hi Samir,

try this Perl regex:
(?s)(<(span) class="(?(3)[^"]++|(?|(b)old|(i)talic))">((?>(?:(?!<\2\b)(?!</\2\b).)++|(?1))*+)</\2>)

and replace:
<\3>\4\</\3>

Unfortunately you have to repeate Replace All until nothing is replaced because of nested blocks.

BR, Fleggy

Samir · Apr 08, 2017#32017-04-08T09:06+00:00

Thanks fleggy
This pattern has worked very good
I am not understand properly this pattern, please can you explain this pattern in details....

fleggy · Apr 08, 2017#42017-04-08T11:29+00:00

Hi Samir

hope this will help you. I am not a very good teacher

Code: Select all

(?s)                          -- . matches also CR/LF
(                             -- the 1st group begins. Used in recursion
  <(span) class="             -- match <span class=" and capture span in the 2nd group
  (?(3)                       -- test if the group 3 has been already captured
    [^"]++                    -- YES: we are already in the recursion and whatever can be matched until the "
    |                         -- NO: we are on the top level and only bold or italic can be matched (the very beginning tag must contain bold or italic)
    (?|                       -- the branch reset group begins
      (b)old|(i)talic         -- match bold or italic and capture the first letter in the 3rd group
    )                         -- end of the branch reset group
  )                           -- end of the test
  ">                          -- match the rest of the tag
  (                           -- the 4th group begins. It matches the inner part between <span><\span>
                              -- a recursion will be used to find the correct closing tag
                              -- the inner part must consist either from any text but opening/closing tag or from another <span></span> block or can be empty
    (?>                       -- atomic group for better performance
      (?:                     -- non-capturing group for better performance
        (?!<\2\b)(?!</\2\b).  -- match any character if the current text is not <span or <\span
      )++                     -- and possessively repeat
      |                       -- OR
      (?1)                    -- if the current text is <span or <\span then try to match it recursively
    )*+                       -- this part can repeat 0 or more times possessively
  )                           -- end of the 4th group
  </\2>                       -- match the closing tag
)                             -- end of the 1st group

Apr 10, 2017#52017-04-10T07:58+00:00

Hi Samir,

this modified regex is better:

(?s)(<(span)\b(?(3)[^>]*+|(?: class="(?|(b)old|(i)talic))")>((?>(?:(?!<\2\b)(?!</\2\b).)++|(?1))*+)</\2>)

because now it works with any attributes in the tag in the text between and 
E.G. the previous one fails in this text:

First partnested partfinal part

BR, Fleggy

Apr 10, 2017#62017-04-10T17:07+00:00

And here is a version which does not need a condition. I think this regex is more comprehensible. On the top level it begins only with and the inner recursion begins with any form of tag .

(?s)<(span) class="(?|(b)old|(i)talic)">(((?>(?:(?!<\1\b)(?!</\1\b).)++|<\1\b[^>]*+>(?3))*+)</\1>)

and replace with:
<\2>\4\</\2>

BR, Fleggy

PeM
You can simply modify the beginning part

<(span) class="(?|(b)old|(i)talic)">

to match any other tag and keep the rest of the regex same if the beginning part contains two capturing groups. Otherwise you have to renumber them accordingly.
For example:

<(div) style="padding:(\d+)px">(((?>(?:(?!<\1\b)(?!</\1\b).)++|<\1\b[^>]*+>(?3))*+)</\1>)

or without the second capturing group inside the tag

<(div) style="padding:\d+px">(((?>(?:(?!<\1\b)(?!</\1\b).)++|<\1\b[^>]*+>(?2))*+)</\1>)