jumpfroggy
Member
|
I just installed 2.0.9.806, and tried to search for (.+), which in the old PN would search for an entire line. Now ( searches for a parenthesis instead of grouping the match.
However, when I change it to (.+) it highlights from the cursor to the end of the file, and then the second search highlights the entire file. Same thing happens with ^.+$. Maybe the problem is with the multi-line flag not being set for the regex?
Is there any documentation for the regex used in PN? I’d be glad to help make the wiki page, if the wiki would let me into my account (see my other thread). Thanks.
|
simon
Key Master
|
Hmm, the line-end marker does seem to be broken now. It was definitely working at one point, so I must have broken it again. The bol marker ^ still works fine, but the eol marker ‘$’ seems to be matching only the very last character in the file. I’ll fix it, and have opened this bug:
http://code.google.com/p/pnotepad/issues/detail?id=120
|
simon
Key Master
|
Oh, and you’re right in that PN now uses standards-conformant syntax so grouping is done with plain parens:
Capture group: (…)
Left paren: (
Right paren: )
|
simon
Key Master
|
Ah, I see what’s happening. The reason for the whole file being matched in this case is that regular expressions are by default greedy and the ‘.’ character is matching line breaks too. This means that the regular expression keeps gobbling characters until it can’t match any more. This means that you get the first line start you encounter up to the last line end you encounter.
To make regex match in non-greedy mode you use the ? character after either * or +. Your expression, therefore, would be:
^.+?$
This then correctly matches just one line.
|
simon
Key Master
|
The regular expressions engine can be configured to not match . against newline characters – perhaps that would be a good idea?
|
simon
Key Master
|
I’ve just realised that replacing $ with something doesn’t do what I expected, it simply inserts after the end-of-line.
This is because end-of-line matches are zero-width meaning that they simply match the space between the CRLF and the first character on the next line.
What I’m not sure of is how regular expressions would normally be used to replace empty lines? Replacing ^$ just inserts after empty lines – the behaviour in PN is pretty close to what Visual Studio does.
Anyone any examples of expected behaviour?
|
kwantum
Member
|
Hi guys
I’m using version 2.0.9
Ok, I’m new to this whole search/replace pattern matching stuff. I did read the wiki pages and I now have a pretty good idea of how to use regex. I started with something simple, of course, like the “Fred2XXX” example, but I didn’t get the same results: nothing was replaced.
I’m trying to replace this:
*ORIENTATION 1.20
with this:
<Rotations y = “1.20″ />
Would somebody mind showing me another example?
|
simon
Key Master
|
Find: *ORIENTATION ([0-9.]+)
Replace: <Rotations y = "1" />
Explanation of “Find”:
We start with * because * is a special regular expressions character and putting the in front says that you want the character not the special character.
ORIENTATION matches exactly that, and then we come to the bit that captures the number. The parentheses around the expression () say that you want to use the bit of text that you find later. As it’s the first set of parentheses this will be capture group 1.
The bit inside the parentheses [0-9.]+ is a character set, including digits 0-9 and the . character. This is another special character so you need to excape it again (using the ). This matches any one of those characters, and the plus sign afterwards says to include one or more of these characters.
Explanation of “Replace”:
Most of the replacement is just text, the only special bit is 1 that gets the text from the group 1 that you stored in the find text.
Hope that helps!
|
simon
Key Master
|
To update the conversation above about multi-line regular expressions, I found a bug in the Boost Xpressive regular expressions library and this has now been fixed so we should get some more sane results. You still can’t replace ^$ with a blank line though, and I don’t think that’s a valid use case for regular expressions. In VIM you have to combine that search with an action to run on each match that deletes the line.
|