Help with regex



  • Hi, I am trying to build a regex for the following:
    I have a file that has sections that are titled as follows:

    =======      MY FIRST SECTION      ===================
    

    I need to extract the words MY DESCRIPTION only
    As in this example, there can be multiple spaces within the section name, and there is no set amount of words - there can be 1,2,3,4 etc.

    I started with the following:

    =+\s+\K.+(?= )
    

    which returns to me the whole description, including the spaces at the end.
    How can I get the words without the spaces?
    (If I put a ? after the \K.+, then it will only find the first word!)

    Anyone know how to do this?
    Thanks,
    David



  • Hello David,

    =+\s+\K.+[^\s](?= )
    

    this seems to work although I don’t know why?

    =+ one or multiple equal signs followed by
    \s+ one or more whitespace chars
    \K reset position to current location
    .+ one or more chars
    [^\s] but not whitespace chars and
    (?= ) match space but consumes zero characters.

    Cheers
    Claudia



  • nope - doesn’t work as expected



  • This one seems to be working

    ^=+\s+\K.+[^\s](?= )
    

    But still haveing problems to understand why.

    Cheers
    Claudia



  • Hi Claudia - Thanks!
    Works perfect, and simple! Why didn’t I think of this! :)
    The reason why it works is as follows:
    After resetting the position with \K, the regex continues:

    • find any characters (with the .) as much as it can find (+ without ?)
    • until and including a non-white_space character
    • followed by (and not including) a space

    although without the [^\s] the .+ finds all words until and including the last letter, by adding [^\s] followed by a non-including space (i.e., (?= )), the regex has to minimize it’s find with .+ by one character in order to fulfill the request of [^\s](?= )

    I hope I explained that clear enough…

    Thanks a lot Claudia!
    All the best
    David



  • Hello David,

    perfect, I now get it.

    I always interpreted [^\s] as don’t find whitespace chars whereas it is
    find everything except whitespace chars.

    Another step on my regex ladder.

    Thank you
    Claudia



  • Hi Claudia, happy to hear that I can be a help to you!
    (After all, I am still so grateful that you pushed me into trying out python… never stopped since! :) - by the way, those lines of equal signs and section names… they’re produced with a python script! :) )

    Although this is not the topic of discussion… for some reason the function list parser is only showing me the first 2 results of this regex.
    I.e., I have 10 of these lines in my files, when I put the regex as a class_name in function list, it only finds the first 2 occurrences! And if I erase the first, then it finds the next 2…

    Do you know what can be the cause of this?

    Thanks,
    David



  • Hi David,

    yes python rules. For me it is one of those languages which suits me. I can read it
    and understand most of it instantly. Nice to see that you get infected to.

    Regarding the function list, to be honest, no, I did a test (borrowed fortan configuration and modified it) with eleven section
    and it is working. See here.

    Could it be that you modified the document without saving it, then it needs a refresh click.
    Otherwise … hmmm how did you do the association? With a language ID? extention or userdefined?
    Mine is using the id.

    Cheers
    Claudia



  • Nevermind… I ‘sort of’ figured it out
    There was obviously some type of overlapping going on (not sure what), but when I changed the regex of the mainexpr a little bit, everything fell into place

    (To explain fully would take a lot of time and writing - and reading :) )



  • good to see you did it ;-)

    Have a nice day
    Claudia



  • Hi Claudia, thanks
    I hadn’t seen your post when I wrote mine!

    Regarding association, I did it by userDefinedLang ID
    (By the way… thanks for that too! I couldn’t live without it! :)
    [still waiting and hoping for version 3!] )

    Thanks
    David

    (P.S. My replies are delayed since my reputation is under 2… :( - :) )



  • Hi, Davey and Claudia,

    I understood why the regex ^=+\s+\K.+[^\s](?= ) is correct and the regex =+\s+\K.+[^\s](?= ), without the ^ character, is NOT correct

    Just imagine the subject text, below :

    =======      MY FIRST SECTION      ===================
    =======      MY SECOND SECTION      ===================
    

    You have to remember that the regex \s is stricly identical to the regex [\t\n\x0B\f\r\x20\xA0]


    So, if you consider the wrong regex =+\s+\K.+[^\s](?= )

    • A first click, on the Find Next button, matches the text MY FIRST SECTION. Nice !

    • A second click, on the Find Next button, matches, wrongly, the string ======= MY SECOND SECTION. Why ?

    Well, after the first click, the cursor is located between the last letter N and the space, in the first line


    So, that wrong regex =+\s+\K.+[^\s](?= ) matches :

    • The =================== string, at the end of the first line, due to the regex =+

    • The EOL characters ( \r\n ) of the first line, which are, both, \s characters !

    • The \K form forget the present regex, so the cursor is reset between the \n of the first line and the first = character of the second line

    • Then, after backtracking, the ======= MY SECOND SECTIO string, due to the .+ part of the regex

    • Finally, the N character, as it’s a NON BLANK character ( [^\s] )

    • The space character, although not part of the final regex, must be present, after the last N character, of the second line : that’s right

    On the contrary, after the first match, when we use the regex ^=+\s+\K.+[^\s](?= ), the cursor location is correctly reset, first, at the beginning of the second line !

    Cheers,

    guy038



  • Hello guy038,

    thank you for the detailed and good explanation.
    When creating a regex I’m still don’t care enough about the current cursor position and the correct meaning of parts
    of the regex like \K. Like you said, it is not only the reset but als the forget the previous matches which is important.
    But I hope this gets done when doing more and more of these. Yesterday I found a nice webpage
    which supports me understanding my regexes ;-)

    Cheers
    Claudia



  • Hi guy038
    Thanks for the extra clarity that you provided - as you always do!

    All the best
    Davey



  • Hello, Claudia and DaveyD,

    I had a look to the webpage, that you found out Claudia. Really interesting, indeed !

    However, in order to get the same matches than the Notepad++ regex engine, we should take care about the following points, when using the site :

    https://regex101.com


    • We must use the default PCRE flavor, on the left part of the window, which seems to have the closest behaviour than the N++ regex engine

    • In the gmixXsuUAJ field, you should, systematically, add the two modifiers gm

      • The g modifier means it will indicate all the matches of the test regex, and not only the first one ! ( just like the Find All button of the Mark tab ! )

      • The m modifier means that the anchors ^ and $ represent the beginning and the end of each line, as the N++ regex engine does ( implicit modifier (?m) )

    • In the gmixXsuUAJ field, you will add the i modifier, if you DON’T check the Match case option OR if your N++ regex begins with the (?i) form

    • In the gmixXsuUAJ field, you will add the s modifier, if you have CHECKED the . matches newline option OR if your N++ regex begins with the (?s) form

    • In the gmixXsuUAJ field, you will add the x modifier, if your N++ regex begins with the form (?x)

    • In the gmixXsuUAJ field, you will NOT indicate the m modifier, if your N++ regex begins with the form (?-m)


    So, giving our last example of the test string, below,

    =======      MY FIRST SECTION      ===================
    =======      MY SECOND SECTION      ===================
    
    • If you enter the regex =+\s+\K.+[^\s](?= ), in the Regular Expression field and gm after the slash, in the modifiers field, you should have the strings MY FIRST SECTION and ======= MY SECOND SECTION, both, highlighted in blue

    • Now, just add the ^ anchor, at the beginning of the regex -> This time, you should have the two strings MY FIRST SECTION and MY SECOND SECTION, both, highlighted in blue !

    Cheers,

    guy038



  • Hi guy038,

    thanky you very much for the infos. I was looking for a boost:regex tester but couldn’t find anyone.
    So with this site and your explanation it makes it easier to find the misunderstandings when using
    regexes.

    Cheers
    Claudia



  • Hi Loreia,
    Just to mention, when I first started, I found a website (similar to the one you mentioned above) that was really helpful, and I liked it mostly because they had a desktop app that included the same features!
    You can take a here http://regexr.com/
    I think the download is somewhere else www.gskinner.com - however, this is from memory
    I’m not saying that this is better than what you found, but just another nice option with a desktop app.

    I think everything guy038 mentioned above applies here too, except that here, the g modifier is enabled by default

    Davey



  • Loreia ?? Did some UDL lately??
    Yeah, https://github.com/gskinner/regexr/
    Worth a try

    Thanks and cheers
    Claudia



  • :) Yes, I did! - Sorry about that confusion!

    Let me know what you think - I thought it was great and it really helped me in the beginning.

    All the best
    Davey


Log in to reply