Can the StringReplace function be used with wildcards?

[CHeighlund]

Can the Delphi StringReplace function be used with wildcard values? That is, if I wanted to, for data entry purposes, read in a file with datestamps in it, and put quotes around the stamps, would the StringReplace function allow me to do so without having to know the exact string value?

Ex: Read in value xx/yy/zzzz. Convert to 'xx/yy/zzzz', where the values of xx, yy, and zzzz remain the same. Again, can StringReplace do this, or do I need to look into other methods?

 

[Griffyn]

No, but there are always workarounds. In your example you could simply do

  DateStr := '''' + DateStr + '''';

 

[towerbase]

Possibly more elegant:

DateStr := QuotedStr( DateStr );

Andrew
Hampshire, UK

 

[CHeighlund]

I apologize, I should have made this clearer.

The datestamp isn't the entire string. The values being drawn in are from a csv file, and the (varying) date value is stuck near the center of each record. Because the records before it have variable size, I don't think I can just pluck out the particular substr I need to work with. I apologize for the confusion.

I already know about QuotedStr(), although I thank you for bringing it up. My problem was concerned with the proper method to isolate that portion, although following your suggestion may help me to get around the mono-focus block I'm running into. Thank you.

 

[towerbase]

It looks like you need to use regular expressions. As far as I know, the Delphi VCL does not provide support for regular expressions but you can get third party components. Take a look at

http://www.regular-expressions.info/pcre.html

Regular expressions provide a very powerful tool for search and replace type operations.

Andrew
Hampshire, UK

 

[harebrain]

It looks like you need to use regular expressions.

I beg to differ. It should be rather simple to parse a string for a fixed-length substring with constant delimiters, even though the delimiters are within the substring. It just takes a little arithmetic: minus 2, plus 4.

CHeighlund, don't over-think or over-complicate this.

 

[lespaul]

ok harebrain, the OP said:

The values being drawn in are from a csv file, and the (varying) date value is stuck near the center of each record. Because the records before it have variable size, I don't think I can just pluck out the particular substr I need to work with.

Assuming the two following strings are the text coming from the csv and you needed to extract the date portion, how would you do it?

[tt]
This is the beginning of the string 1 234 5 and here is the 1/15/2007 that I am 77895 trying to extract 58 579
This is the second string 5 624 5 and I need the 2/15/2007 now but it's in a different place 58 710[/tt]

(I'm not being obnoxious or anything, I'd really like to know how you would do something like that!)

Leslie

In an open world there's no need for windows and gates

http://www.ubuntu.com

 

[Griffyn]

Hi Leslie,

The trick is identifying something that is constant across all records. In this case you could say the constant thing will be a slash (/) followed by 2 characters followed by a slash. You could then write some code the searches for a slash in each record, and then checks if the character 3 bytes ahead is also a slash - if it is, you extract a portion around that spot as necessary - that's what harebrain was suggesting.

You could tie it down further of course, by also checking if the characters in between the slashes are 0..9 and also if the 2 prior to the slash are 0..9 and the 4 after the second slash are 0..9. With that sort of rigid checking, you can be sure you've found a date. You could even do a StrToDate check to be sure it's a valid date.

Regular expressions do this sort of thing for you automatically by allowing you to specify a mask such as '00/00/0000' where the 0's represent any number and the slashes are literal. A regex search simply finds any sequence of characters in a string that match the mask. Plus regex searches allow for wildcards so you can have a variable number of characters between literals in the mask.

 

[harebrain]

...here is the 1/15/2007 that I am 77895 trying ...

Note that neither of your examples conforms to the OP's spec. Could be why you're having trouble with this.

But, what Griffyn said: find a slash, back up 2, extract a substring of 10, and see if that's a date. Simple.

REs are wonderful. But, if REs aren't already a familiar tool, the overhead and the learning curve in this case argue against it. Unless the OP has some time on his/her hands, I'd advise a more direct approach using intrinsic string-handling routines.

And when a programmer does have spare time, studying up on REs will pay off better than studying almost anything else.

 

[lespaul]

Note that neither of your examples conforms to the OP's spec.

The only specifications I saw are:

The datestamp isn't the entire string. The values being drawn in are from a csv file, and the (varying) date value is stuck near the center of each record. Because the records before it have variable size, I don't think I can just pluck out the particular substr I need to work with. I apologize for the confusion.

I have a date near the center of each record and variable size of information prior to that date....am I missing something?

Where can I find more on Regular Expressions and how to use them?

 

[harebrain]

Apparently you missed this:

Ex: Read in value xx/yy/zzzz. Convert to 'xx/yy/zzzz',

As for REs, try

http://regexlib.com/CheatSheet.aspx

and

http://www.regular-expressions.info/tools.html,

and don't forget that you can search this site, too.

 

[harebrain]

Just found this in Companion Disk 1 to D7 (can't vouch for earlier/later versions.)

On the CD, look for X:\droopy_eyes_software\faststrings -- the FastStringFuncs.pas unit has a function, StringMatches, that performs the kind of wildcard matching that the OP requested.

Still, I think the approach suggested by Griffyn and myself is safer: using wildcards leaves more chances of finding something other than a date. But we did find a wildcard match routine for you. Enjoy!