Re: regular expressions

On 20/11/06, Paul Novitski <paul@juniperwebcraft.com> wrote:

Børge, Here's how I would think this one through: First, I'm having to make several guesses at the nature of your text content: - You use the single word "topic" but I'll assume this can be multiple words and spaces. - Your source string includes a space after "rest of the text " while your marked-up result doesn't. However I will assume that you really do mean the rest of the text until end-of-string. - Your source string also includes a space before the initial <c> but your regexp pattern doesn't. I'll assume that both beginning and ending spaces are unintentional. Your source string:

         "<c> FFFFFF topic <c> 999999 rest of the text"

consists of these parts: 1) [start-of-string] 2) "<c> "

3) "FFFFFF"     (color code 1)

4) " "

5) "topic"      (text 1)

6) " <c> "

7) "999999"     (color code 2)

8) " " 9) "rest of the text" (text 2) 10) [end-of-string] i.e.: 1) [start-of-string] 2) <c> + whitespace 3) color code 1 4) whitespace 5) one or more characters 6) whitespace + <c> + whitespace 7) color code 2 8) whitespace 9) one or more characters 10) [end-of-string] This suggests the regexp pattern: 1) ^ 2) <c>\s 3) ([0-9A-F]{6}) 4) \s 5) (.+) 6) \s<c>\s 7) ([0-9A-F]{6}) 8) \s 9) (.+) 10) $ /^<c>\s([0-9A-F]{6})\s(.+)\s<c>\s([0-9A-F]{6})\s(.+)$/i Everything in the source string that you need to retain needs to be in parentheses so regexp can grab it. In 5) I can let the pattern be greedy, safe in the knowledge that there WILL be a /s<c> to terminate the character-grab. I end with the pattern modifier /i so it will work with lowercase letters in the RGB color codes. PHP: $sText = '<c> FFFFFF topic <c> 999999 rest of the text'; $sPattern = '/^<c>\s([0-9A-F]{6})\s(.+)\s<c>\s([0-9A-F]{6})\s(.+)$/i'; preg_match($sPattern, $sText, $aMatches); print_r($aMatches); result: Array (

     [0] => <c> FFFFFF topic <c> 999999 rest of the text
     [1] => FFFFFF
     [2] => topic
     [3] => 999999
     [4] => rest of the text

) This isolates the four substrings you want in regexp references $1 through $4. Replacement: [Tangentially, I'd like to comment that font tags are passe. I urge you to use spans with styling instead. I normally dislike using inline styles (style details mixed with the HTML), but in this case (as far as I know) you don't have any choice. If you can, I suggest you replace the literal color codes with style names and define the precise colors in your stylesheet, not your database. [What this further suggests is that you ought to have two discrete database fields, topic and description, if you can, rather than combining them into one field that needs to be parsed. Then you can output something like:

         <span class="topic">TOPIC</span> <span class="desc">DESCRIPTION</span>

and leave the RGB color codes out of this layer of your application altogether.] However, working with the data you've been dealt: $sTagBegin = '<span style="color:#'; $sTagEnd = ';">'; $sCloseTag = '</span>'; $sReplacement = $sTagBegin . '$1' . $sTagEnd . '$2' . $sCloseTag .

                 $sTagBegin . '$3' . $sTagEnd . '$4' . $sCloseTag;

echo preg_replace($sPattern, $sReplacement, $sText); result: <span style="color:#FFFFFF;">topic</span> <span style="color:#999999;">rest of the text</span> ____________________________ It's tempting to write the pattern more succinctly to take advantage of the repeating pattern of the source text:

         <c> COLORCODE text

The regexp pattern might be: 1) \s* 2) <c>\s 3) ([0-9A-F]{6}) 4) \s 5) ([^<]+) 1) optional whitespace 2) <c> + whitespace 3) color code 4) whitespace 5) one or more characters until the next < $sText = '<c> FFFFFF topic <c> 999999 rest of the text'; $sPattern = '/\s*<c>\s([0-9A-F]{6})\s([^<]+)/i'; preg_match_all($sPattern, $sText, $aMatches); result: Array (

     [0] => Array
         (
             [0] =>  FFFFFF topic
             [1] =>  999999 rest of the text
         )

     [1] => Array
         (
             [0] => FFFFFF
             [1] => 999999
         )

     [2] => Array
         (
             [0] => topic
             [1] => rest of the text
         )

) In this case, we need to specify the tag pattern only once: $sReplacement = $sTagBegin . '$1' . $sTagEnd . '$2' . $sCloseTag; echo preg_replace($sPattern, $sReplacement, $sText); result: <span style="color:#FF0000;">topic </span> <span style="color:#00FF00;">rest of the text</span> Notice is that this results in whitespace after the topic string. Someone more knowledgeable in regular expressions can probably tell you how to eliminate that, perhaps by using a regexp assertion: http://php.net/manual/en/reference.pcre.pattern.syntax.php#regexp.reference.assertions Regards, Paul __________________________ Paul Novitski Juniper Webcraft Ltd. http://juniperwebcraft.com

Paul, I just got around to reading this thread. The post of yours that I quote above has got to be one of the best posts that I've read in the 5 years that I've been on and off the php list. The way you break that regex down taught me things that have eluded me for half a decade. Although I have nothing to do with the OP, I really want to say thanks for that bit of information. Dotan Cohen http://lyricslist.com/ http://what-is-what.com/