help with hard regex please

[cybmole]

this is a regex question - not really a sigil one, but since this is where the helpful regex experts are, I'm hoping one will step in and advise

I want to clean up some .xml files & am thinking regex in notepad++ could do the job if I can figure out the expression.

I want to zap all chunks that look like the example below, where the key search term is X-Fi.

for anything with X-Fi in that 2nd line I want to delete the whole block- ambitious I know but can that be done ?

<RemoteButton>
<Name>X-Fi 24-bit Wheel Button</Name>
<MidiSignal>0A 41 44 09 76</MidiSignal>
<USBSignal>02 C1 44 89 76</USBSignal>
<ButtonType>btKeyboardEvent</ButtonType>
<KeyCode>173</KeyCode>
</RemoteButton>

logically, I have to find the start phrase <RemoteButton>\s*<Name>X-Fi , look forward to locate the matching </Remote... & delete everything found in between. I also have to cope with intermediate line feeds, white space & / characters..

hmm - I seem to be solving my own question as i type...

will a simple (.*) suffice i.e. find
<RemoteButton>\s*<Name>X-Fi(.*)</RemoteButton>


well i can load the xml into sigil & the above almost works, ( it leaves some mysterious de> entries ) but I can't see how to then save as xml ( sigil v4)

i try the same expression in notepad++ & it does NOT work - either N+= cannot do multi-line or it uses different regex syntax ?

[Jellby]

Replace <RemoteButton> and </RemoteButton> with two unused characters, like ¬ and |

Search for ¬\s*<Name>X-Fi[^|]*| (with multi-line)

Replace back ¬ and | with <RemoteButton> and </RemoteButton>

If Notepad does not have multi-line matching, find another editor that does, it's an essential feature. (In Sigil, you can copy-paste the result, instead of saving.)

[cybmole]

in sigil, it worked 100% when I replaced
<RemoteButton>\s*<Name>X-Fi(.*)</RemoteButton>
with
<RemoteButton>\s*<Name>X-Fi(.*)/RemoteButton>

previously it worked on replace once, but not on replace all, it left in trailing each instance of de> from KeyCode> for some obscure reason.

I googled notepad++ it does not have multi-line regex, not sure that any free editor does, except for sigil!

open, edit, copy - paste from code view into notepad++ and save may be the way to go

[Jellby]

Be careful if you have greedy matching or something like:

<RemoteButton>
<Name>X-Fi 24-bit Wheel Button</Name>
<MidiSignal>0A 41 44 09 76</MidiSignal>
<USBSignal>02 C1 44 89 76</USBSignal>
<ButtonType>btKeyboardEvent</ButtonType>
<KeyCode>173</KeyCode>
</RemoteButton>

<RemoteButton>
<Name>foo 24-bit Wheel Button</Name>
<MidiSignal>0A 41 44 09 76</MidiSignal>
<USBSignal>02 C1 44 89 76</USBSignal>
<ButtonType>btKeyboardEvent</ButtonType>
<KeyCode>173</KeyCode>
</RemoteButton>

(no X-Fi in the second button)

[cybmole]

Originally Posted by Jellby

Be careful if you have greedy matching or something like:

<RemoteButton>
<Name>X-Fi 24-bit Wheel Button</Name>
<MidiSignal>0A 41 44 09 76</MidiSignal>
<USBSignal>02 C1 44 89 76</USBSignal>
<ButtonType>btKeyboardEvent</ButtonType>
<KeyCode>173</KeyCode>
</RemoteButton>

<RemoteButton>
<Name>foo 24-bit Wheel Button</Name>
<MidiSignal>0A 41 44 09 76</MidiSignal>
<USBSignal>02 C1 44 89 76</USBSignal>
<ButtonType>btKeyboardEvent</ButtonType>
<KeyCode>173</KeyCode>
</RemoteButton>

(no X-Fi in the second button)

I will have that construction, there will typically be a bunch of X-Fi button defintions, then some other button definitions.

I have not really mastered greedy management - that is why I was concerned about using (.*)

is there a way to write STARTphrase(.*)ENDphrase type searches which makes them nongreedy

this will be useful info for books also for removing redundant SPAN structures

[Jellby]

I believe you can use (.*?) instead to make it ungreedy. (Or using [^¬]* after replacement, as per my suggestion above, which is essentially the same.)

With <span> it may be more complicated, since they can be nested to no end (I assume <RemoteButton>s would not be nested).

[PeterT]

I just went looking and came across http://www.editpadlite.com/ which claims

EditPad Lite is a compact general-purpose text editor. Use EditPad Lite to easily edit any kind of plain text file. EditPad Lite has all the essential features to make text editing a breeze:

• Large file and long line support.
• Full Unicode support, including complex scripts and right-to-left scripts.
• Direct editing of text files using Windows, UNIX, and mac Mac text encodings (code pages) and line breaks.
• Tabbed interface for working with many files.
• Unlimited undo and redo for all open files, even after saving.
• Automatic backup and working copies prevent data loss.
• Powerful search-and-replace with literal search terms and regular expressions that can span multiple lines.

[Jellby]

Vim supports multiline regex too, and much more.

[DiapDealer]

Originally Posted by PeterT

I just went looking and came across http://www.editpadlite.com/ which claims

EditPad Lite is a great regex-capable editor. I use it all the time.

There are some subtle differences between its JGSoft regex engine and Sigils PCRE engine, but they rarely come up in most situations. JGSoft doesn't support /K, and its commands for changing the case of text is a little different.