SPIN -- string wild card code

Bobb Fwed

Has anyone written any code that could accept wild cards to match strings against other strings? I'd hate to work on something like this if it has already be done (and likely done better than what I will be able to do).

kuroneko

Does this help [thread=117496]Pattern matching and data extraction in Spin using regular expressions[/thread]? Courtesy of Phil Pilgrim.

Bobb Fwed

kuroneko wrote: »

Does this help [thread=117496]Pattern matching and data extraction in Spin using regular expressions[/thread]? Courtesy of Phil Pilgrim.

More complicated than I want, 628 longs of RAM seems a bit excessive for just wild card functionality, but it is a start. Thanks.

WBA Consulting

Have you seen this thread? It came to mind, but not sure if it can be tweaked for what you need.

Pattern matching and data extraction in Spin using regular expressions by PhiPi
http://forums.parallax.com/showthread.php?117496-Pattern-matching-and-data-extraction-in-Spin-using-regular-expressions.

Bobb Fwed

WBA Consulting wrote: »

Have you seen this thread? It came to mind, but not sure if it can be tweaked for what you need.

Pattern matching and data extraction in Spin using regular expressions by PhiPi
http://forums.parallax.com/showthread.php?117496-Pattern-matching-and-data-extraction-in-Spin-using-regular-expressions.

Nope, I've never seen that before, ever. :-P

Phil Pilgrim (PhiPi)

How about this one?

-Phil

'Sorry Bobb, I tried not to press Post Reply, but the urge was just too strong!

Mike G

I use this in my Spinneret HttpRequest object. I'm sure you can modify the code to handle wild card characters.

PUB MatchPattern(source, pattern, matchCase, startIndex) | i, j, sourceLength, patternlength

  sourceLength := strsize(source)
  patternlength := strsize(pattern)
  result := -1
  ptrEnd := 0
  ptrStart := 0

  i := 0
  j := 0
   
  repeat i from startIndex to sourceLength
    ' Try to match the first char in pattern to the
    ' source 
    if(IsCharMatch(source + i, pattern + j, matchCase) )    
      'Found a match
      'Store the current pointer position and increment the source pointer by one byte

      ptrStart := source + i
      'Lets see if the pattern lines up with the source
      repeat j from 0 to patternLength - 1
        ifnot(IsCharMatch(source + i + j, pattern + j, matchCase))
          j := 0
          ptrEnd := -1
          ptrStart := -1
          quit
          
    if(j > 0)
      ptrEnd := ptrStart + patternLength
      result := i
      quit   

  
PRI IsCharMatch(sourceChar, patternChar, matchCase)
  '' Initilize the return value
  result := false


  if(matchCase)
    if(byte[sourceChar] == byte[patternChar])
      result := true   
  else
    if((byte[sourceChar] == byte[patternChar]) or (byte[sourceChar] == byte[patternChar]+32))
      result := true

Now that I know about Phil's regex code, I'll probably use that.

localroger

I wrote a version of the DIR command in PropCMD which recognized pre*.p* wildcards. I'm not sure if the version in the Obex has that, I was working on a much more elaborate version before I realized that what I really wanted to do simply was not going to fit in Hub RAM and drifted off into expanded memory bufferland. I'll see what's what tomorrow at work but if you haven't ever done so you might want to grab the version of PropCMD I posted and see what's there.

kuroneko

PRI IsCharMatch(sourceChar, patternChar, matchCase)
  '' Initilize the return value
  result := false


  if(matchCase)
    if(byte[sourceChar] == byte[patternChar])
      result := true   
  else
    [COLOR="red"]if((byte[sourceChar] == byte[patternChar]) or (byte[sourceChar] == byte[patternChar]+32))[/COLOR]
      result := true

How does this work for byte[sourceChar] == "A" and byte[patternChar] == "a"? Shouldn't this involve | $20?

WBA Consulting

Wow, apparently it took my post a half hour to show up. When I responded, your thread had just popped up and didn't have any responses. Oh well, got a laugh, so it wasn't completely a waste!

Bobb Fwed

Well, it's not perfect, but I have found a solution. It works exactly as expected (unlimited wild cards), but it's recursive, so there is less control over memory usage, and running this on a pattern with lots of wild cards could use a lot of STACK space (8 longs per call). I'm going to see if I can put an option to remove case sensitivity,

I'm going to put this into my string object(s):

PRI match_pattern (sourceAddr, patternAddr)
'' ported code from http://vijayinterviewquestions.blogspot.com/2007/07/write-c-program-which-does-wildcard.html
'' seems to work good, but it's recursive, so it could use a significant
'' amount of STACK space depending on the number wild cards there are

  REPEAT WHILE (byte[sourceAddr])                                               ' go through entire string
    CASE byte[patternAddr]                                                      ' check for wild card character
      "*":
        REPEAT
          patternAddr++                                                         ' do this at least once, repeat until no more wild card characters
        WHILE byte[patternAddr] == "*"

        IF NOT(byte[patternAddr])                                               ' return true if end of string is reached
          RETURN true

       REPEAT WHILE byte[sourceAddr]
          IF (match_pattern(sourceAddr++, patternAddr))                         ' recurse!
            RETURN true
        RETURN false

      OTHER:
        IF byte[sourceAddr] <> byte[patternAddr]                                ' check for a match
          RETURN false   

    patternAddr++
    sourceAddr++

  REPEAT WHILE (byte[patternAddr] == "*")                                       ' skip repetitive wild card characters
    patternAddr++
  RETURN NOT(byte[patternAddr])

Mike G

@kuroneko, yes it should.