Regular Expressions

Thou mewling ill-breeding pignut!

AspDotNetDev19-Jun-12 12:34

AspDotNetDev

19-Jun-12 12:34

Is this C#? I've not seen that method of creating named capture groups. Though, if you are going to name your capture groups, you may as well retrieve them by name rather than by index.

Also, you only check if the line contains the string to find. Nowhere in your code do you actually look for that as a key or value.

And consider that regular expressions are greedy. Your "Value" capture group will capture everything after the first ": ".

Also, according to your regular expression, you can have an empty value ("*" means zero or more). I suspect that is not what you want, but maybe you do.

Finally, I'm not really sure how "SERVOFW: SDLSUPPORTED: 1" fits into the key/value scenario. Maybe that's a typo.

Based on the assumption that values can have spaces, but keys can't have spaces, this may be a more appropriate regular expression:

(?<KEY>(?!:| ).)+: (?<VALUE>((?!:).)+)(?= (((?!:| ).)+:)|$)

Keep in mind that this will find all matches, so you don't watch to just find the first match as you are doing in your example. You want to loop through them all and compare the key/value against the word you are looking for.

modified 19-Jun-12 18:41pm.

Thou mewling ill-breeding pignut!

AspDotNetDev19-Jun-12 12:52

AspDotNetDev

19-Jun-12 12:52

AspDotNetDev wrote:
I'm not really sure how "SERVOFW: SDLSUPPORTED: 1" fits into the key/value scenario

Now that I think about it, "SERVOFW" is probably a key with an empty value. That'd take a bit more time to figure out. I'll leave that as an exercise for the reader. Smile | :)

Kevin Marois19-Jun-12 13:02

Kevin Marois

19-Jun-12 13:02

I'm piecing toget code formv various places becuase it all looks like greek to me. I'm fairly frustrated at this point. This shouldn't be this difficult.

Anyway, here's my code:

if (Line.Contains(WordToFind))
{
    Regex r = new Regex(
            "(?:(?'Key'\\S+): (?'Value'.*?))",
        RegexOptions.RightToLeft
        | RegexOptions.CultureInvariant
        | RegexOptions.Compiled
        );

    Match m = r.Match(WordToFind);

    if (m.Success)
    {
    }
}

I get no matches for the word "Model".

Also, once I get a match, how do I get the Key/Value data???

If it's not broken, fix it until it is

Thou mewling ill-breeding pignut!

AspDotNetDev19-Jun-12 13:35

AspDotNetDev

19-Jun-12 13:35

Why did you edit your response? The code I see in the email notification I was sent for your reply to my message looks more correct than the code I now see.

For one, you're going to need a for loop to iterate over the result of Matches (not Match). Match finds a single result, and Matches finds all results.

For two, r.Match(WordToFind) is searching "WordToFind" rather than "Line".

Kevin Marois wrote:
once I get a match, how do I get the Key/Value data

You can get groups based on the name. Something like this (I don't have my compiler open, so it may vary slightly):

String key = m.Groups["Key"].Value;
String value = m.Groups["Value"].Value;

Kevin Marois wrote:
RegexOptions.RightToLeft

This may drastically change the functionality, but I'm not exactly sure what it does, so you may want to Google this.

Kevin Marois wrote:
I'm fairly frustrated at this point. This shouldn't be this difficult.

Regular expressions are complicated, but powerful. You'll get used to them over time. From what others say, Expresso is a good tool to learn regular expressions, though I use a custom tool I built for myself.

can someone tell me whats wrong with this expression?

ChekGuy29-Apr-12 11:48

ChekGuy

29-Apr-12 11:48

pretty new to regex. this is a great article.

can someone tell me whats wrong with my expression? I'm using this from C#. I'm getting the response from a blog in a malformed xml format. Need to extract entries out of it. In a simple format inpu is similar to the following.

string input = @"<entry><id>tag:myblog.com try</entry><entry><id>tag:myblog.com tryagain</entry><entry><id>tag:myblog.com hello </enty>";

I need to identify the number of entries, and then processing each of them.

Regex blogsRegEx = new Regex(@"<entry><id>tag:myblog.*</entry>");

MatchCollection blogEntries = blogsRegEx.Matches(input);

I always get just 1 entry. It matches the whole thing instead of matching multiple strings in the pattern <entry<id>tag:myblog....</entry>.

Can someone help what am I missing here? do i need to use subexpressions here?

modified 29-Apr-12 19:28pm.

Re: can someone tell me whats wrong with this expression?

egenis29-Apr-12 19:22

egenis

29-Apr-12 19:22

Not sure about the regex (It makes my brain hurt), but you can use linq instead

string findText = @"<entry><id>tag:myblog";
int entriesCount = blogsText.Count(t => t.equals(findText)); // you can use Contains() as well I suppose

Have not tested it but it should give you the correct results by just tweaking your findText variable.

Re: can someone tell me whats wrong with this expression?

Peter_in_278029-Apr-12 19:46

Peter_in_2780

29-Apr-12 19:46

By default, regex matching is greedy. That is, wildcards will match the longest possible chunk of input, so you need to make your .* non-greedy. You do that by putting a ? after it, so your line becomes

Regex blogsRegEx = new Regex(@"<entry><id>tag:myblog.*?</entry>");

If you are going to do ANYTHING nontrivial with regexes, get a copy of Expresso. (See our Free Tools forum for details.)

Cheers,
Peter

Software rusts. Simon Stephenson, ca 1994.

Either start or not a "d"

Bernhard Hiller6-Mar-12 0:50

Bernhard Hiller

6-Mar-12 0:50

I try to get the date number format from a date format string. The date number format is - if present - either a single d or a double d. But the whole string can contain a place holder for the dayname also - ddd or dddd. How can I get the d or dd, if present?
I tried [^d](?<pattern>d{1,2})[^d] which works well with the standard German format (dddd, d. MMMM yyyy). But that fails when the daynumber comes first (say: d. dddd MMMM yyyy; here, ^(?<pattern>d{1,2})[^d] does the trick), and it fails when the daynumber comes last.
Actually I need something like a "either start of the string or not a d", i.e. ^|[^d] - but that does not work. How can that be solved?

Re: Either start or not a "d"

Bernhard Hiller6-Mar-12 2:02

Bernhard Hiller

6-Mar-12 2:02

Solved it with @"\b(?<pattern>d{1,2})\b"

Trying to match non-quoted sections

PIEBALDconsult24-Jan-12 4:30

24-Jan-12 4:30

Given the name of an SQL table, I am looking to add quotes as necessary (with an loose definition of "necessary"). With SQL Server, the parts of the name should be wrapped in brackets ([ and ]), MySQL uses backticks (`). So, for example:

database.schema.table
database.schema.[table]
database.[schema].table
[database].[schema].table

should all be transformed to:

[database].[schema].[table]

What I have working; is capturing Wrapped and Unwrapped sections separately, wrapping the Unwrapped sections, and joining the sections back together. But it occurred to me that if I could Match only the Unwrapped sections, I could use Replace.

However, I have so far been unsuccessful in my attempts (otherwise I wouldn't be posting). Does anyone out there have an idea of how to do this? I'm thinking it may involve Balancing Groups, but I've never used them before so I'm finding them confusing.

This is not urgent. Big Grin | :-D

Edit: I must have been over-thinking it D'Oh! | :doh:

. What I have now is (?<=^|\.)[^\[\]\.]+(?=$|\.)

modified 24-Jan-12 11:00am.

Andreas Gieriet7-Apr-12 4:33

7-Apr-12 4:33

Are you looking for this?

string sql = @"database.schema.table
database.schema.[table]
database.[schema].table
[database].[schema].table
 
should all be transformed to:
 
[database].[schema].[table]
";
string pattern = @"(?:\[?(\w+)\]?)?\.\[?(\w+)\]?";
Func<Match, string> replace = m =>
    (m.Groups[1].Success ? "[" + m.Groups[1].Value + "]" : "") + ".[" + m.Groups[2].Value + "]";
Console.WriteLine("{0}", Regex.Replace(sql, pattern, m=>replace(m)));

PIEBALDconsult7-Apr-12 9:17

7-Apr-12 9:17

Does that work for names that contain SPACEs? my database.my schema.my table

And Excel worksheet names that include a dollar sign ($) at the end?

(I realize those were not listed in the original spec.)

Andreas Gieriet7-Apr-12 11:04

7-Apr-12 11:04

You will run into problems here.
The problem arises with spaces in the name, since the following pattern:

...([\w\s]+)...

matches

delete my database

as well as

delete database

In this case I guess you don't get away without a parser (use the Regex for tokenizing, use the parser to detect all commands and translate the arguments where needed).

Any names without spaces get easily translated, though, e.g.:

string pattern = @"(?:\[?([\w\$]+)\]?)?\.\[?([\w\$]+)\]?";

And if you have optional spaces around "[" and ".", the following a bit more complicted regex will do:

...
string open = @"(?:\[\s*?)?";
string close = @"(?:\s*?\])?";
string ident = @"([\w\$]+)";
string prefix = @"(?:" + open + ident + close + @"\s*?)?";
string suffix = @"(?:" + open + ident + close + @")";
string pattern = prefix + @"\.\s*?" + suffix;
...

PIEBALDconsult9-Apr-12 18:03

9-Apr-12 18:03

Andreas Gieriet wrote:
matches

delete my database

as well
as

delete database

I expect the string to contain only the database, schema, and table names.

Andreas Gieriet10-Apr-12 9:59

10-Apr-12 9:59

The Regex sees a line like
aaa bbb ccc . ddd . eee fff
What part of aaa bbb ccc is the database name? Only ccc or bbb ccc, etc.? You see the problem?
The same for eee fff.

Non-escaped/non-wrapped spaces in names is guess work to make them wrapped into [...].
I.e. to get from aaa bbb ccc .... to aaa [bbb ccc] .... is rather difficult, unless you know what aaa means or you say from outside that bbb ccc is a single name.

Quite a challenge. Wink | ;-)

Cheers

Andi

PIEBALDconsult11-Apr-12 3:10

11-Apr-12 3:10

That should result in

[aaa bbb ccc ].[ ddd ].[ eee fff]

Andreas Gieriet11-Apr-12 13:11

11-Apr-12 13:11

The line
aaa bbb ccc.ddd eee ...
could be
ALTER TABLE dbo.tVersion ADD ...
which in your approach would result in
[ALTER TABLE dbo].[tVersion ADD] ...

Forget about spaces or get as input the individual names (db name, table name, etc.) or make a parser that detects all language constructs and their db, table, etc. positions...

I still think it's not worth the effort with names that contain spaces - too fragile.

Cheers
Andi

PIEBALDconsult11-Apr-12 14:45

11-Apr-12 14:45

No, the string contains only the database, schema, and table name separated by periods as per the original post.

Andreas Gieriet11-Apr-12 21:52

11-Apr-12 21:52

I was confused since I understood (say: assumed...) that you have an SQL script that you want to patch... Never assume anything Wink | ;-)

In that case your initial regex is probably the simplest solution.
Cheers
Andi