Whenever I have to edit Perl Mason files, I always have problems indenting lines due to Perl code starting with %. For example:
<div>
<div>
% if( !$something ) {
<strong><% $title %></strong>
% }
</div>
</div>
Any idea how I can tell Vim to ignore the % at the beginning of the line and indent like it wasn't there?
I'm using https://github.com/aming/vim-mason to support the mixed Perl/HTLM syntax, but I don't think it changes anything.
This is Perl code embedded inside HTML, so the indenting comes from $VIMRUNTIME/indent/html.vim. This defines an 'indentexpr', implemented by HtmlIndent().
You need to modify that implementation to ignore % in the first column; whenever it accesses the buffer (getline(), prevnonblank(), shiftwidth()), you need to intercept, find the previous line that does have such % sigil, and return the value for that instead. (If those special lines can also contain HTML tags, you may have to extract those from the Perl code and return only those.) That gets you the indenting you desire.
Unfortunately, it's not trivial, and you have to fork the original implementation. However, if you manage to implement a clean solution, you could suggest adding integration points to the author of indent/html.vim. If there are other languages apart from Mason that use these prefixes on top of HTML, that would be an additional argument for adding such support (and maybe even your wrapper functions).
Related
Not exactly sure how to phrase this but it is incredibly annoying that when I'm writing js for example and open a comment block (/*) or a template string(`) and everything that I have folded in the following lines gets unfolded because it thinks that my comment or template string block might just extend to the very end of the whole file.
At this point I'm already in the habit of writing the end of a comment block before the beginning and copy-pasting my template strings and then editing them rather than making new ones but it would be so much nicer if I could just write without having to worry about such things, notepad++ handled it well, is there any way to make vs code behave decently too?
I would like to create a filter folder, best inside .vim and be able to run a text filter just with one file name:! filter.pl
I put up a Perl text filter to change all special Characters in a LaTeX Math Formula, which is running fine so far - only problem it is running on the whole line not the selected formula, but I can live with it ...
#!/usr/bin/perl -np
use strict;
use warnings;
# this filter transforms all special characters in Mathformular for LaTeX
s/\\/\\backslash /g;
s/([\$\#&%_{}])/\\$1/g;
But to call this filter is cumbersome
: '<,'>!"/Users/username/Library/Mobile Documents/com~apple~CloudDocs/my_vim_cheat_sheet/perl_filter.pl"
Apple put in the path to the iCloud a white space, so I have to put "" around! Where I put a collection of text filters?
Thank you for your answers
marek
You can safely create a subfolder with any name different from ones Vim uses itself (see :h 'rtp'). So this is ok:
:*!$HOME/.vim/filters/perl_filter.pl
Also Vim has a predefined interface for a general purpose filter called 'equalprg'. To make use of it simply set a global-local (i.e. both set and setlocal are meaningful) option equalprg to a fully qualified name of your script. Then hit = in visual mode to apply filter (or ={motion} in normal mode). (Read :h 'equalprg' :h =).
If you need several filters at once, and switching equalprg is not convenient, you can still try different options to reduce typing.
For example, mappings, such as
vnoremap <Leader>f :!/path/to/my/filter<CR>
Then hitting \f (or whatever is your "leader" key set) in the visual mode will result in the executing :'<,'>!/path/to/my/filter (note that the visual selection will be applied automatically).
Another attempt is to set a dedicated environment variable (which will be inherited by all child processes including shell(s). For example,
:let $filters = '~/.vim/filters'
:*!$filters/myfilter.pl
Of course, you can put those set equalprg=... vnoremap ... let $filters=... etc.etc. in your vimrc.
I would like to create a filter folder, best inside .vim and be able to run a text filter just with one file name :! filter.pl
Simply add the script to somewhere within your $PATH. Or, if you really only intend to use that from within Vim, then add that directory to your $PATH in your .vimrc, so you have it available there.
For example, if you'd like to use ~/.vim/scripts for your external Perl or shell scripts, you can use this in your ~/.vimrc:
call setenv('PATH', expand('~/.vim/scripts').':'.$PATH)
After that, you can simply use :'<,'> !filter.pl to run it. And Tab completion will work with the name of the script, type :!fil<Tab> and Vim will complete it to filter.pl, assuming it's a unique prefix.
The snippet above for your .vimrc has one minor issue, that if you :source your .vimrc during Vim runtime, it will keep adding the entry to $PATH multiple times. That doesn't typically break anything, only the entry will become longer, you might run into variable length issues.
You can fix it by checking whether that's present in path or not before updating it, perhaps with something like:
let scripts_dir = expand('~/.vim/scripts')
if index(split($PATH, ':'), scripts_dir) < 0
call setenv('PATH', scripts_dir.':'.$PATH)
endif
But also, about this:
I put up a Perl text filter to change all special Characters in a LaTeX Math Formula
s/\\/\\backslash /g;
s/([\$\#&%_{}])/\\$1/g;
Consider writing that in Vim instead.
In fact, almost the same syntax will work as a Vim function:
function! EscapeLatexMathFormula()
s/\\/\\backslash /eg
s/\([$#&%_{}]\)/\\\1/eg
endfunction
You can call it on a range, with:
:'<,'>call EscapeLatexMathFormula()
Calling it without a range will affect the current line only.
You can also make it into a command, with:
command! -range EscapeLatexMathFormula <line1>,<line2> call EscapeLatexMathFormula()
In which case you can simply use:
:'<,'>EscapeLatexMathFormula
You can use tab-completion for the function and command names (though, of course, you can pick shorter names if you'd like, as well.)
Note that user-defined command names need to start with an uppercase letter. Function names can start with an uppercase letter too (there are more options for function names, but making this global with an uppercase is probably the easiest here.)
It seems that Github does not support the sphinx.ext.doctest syntax in their reStructuredText renderer, which is causing some problems when trying to include a doctest-style code block (.. doctest) in a README.rst which is transcluded into the documentation index (which is rendered by sphinx). If I replace the .. doctest directive with something else, it doesn't render properly as a doctest on Sphinx, but if I don't remove the directive, the code block doesn't render at all (see this gist).
Ideally I'd like to find a solution which just does the right thing in both environments, but failing that, is there a way to fall back to a .. code block or some other supported format (e.g. the rST equivalent of a <NoScript> tag)?
So I have some formatting rule to follow, such as :
Space on each side of operator (*, =, +, %, etc)
No space at the end of a line
No more than 80 chars per line
Is there a way to highlight in red line containing formating error?
The eclipse auto-formating tool is no good because either :
It will change to many line (old code not written by me)
or it won't (only my code)
Because I must follow some "colorfull" guideline :
You must change formating error relative to operators in old code but nothing else
Your code must be correctly formated.
Any ideas?
Thanks
You can select which lines of code you want to format. The Eclipse formatting tool doesn't have to run across the entire file. To do this: select the lines you want to format, then press Ctrl-Shift-F.
You could try using the Eclipse Checkstyle Plugin.
You'll need to configure it with just the rules that you need (the default configuration is very strict, so create a new one with just the rules you care about).
This will highlight all lines with formatting issues. I don't think it's possible to ignore old code using the plugin.
Talk to whoever created that coding guideline. It does not make sense in the long run, because editing code in Eclipse will always apply all current formatting rules (which violates that guideline) or none, if you disable the formatter (which leads to you writing bad code).
If there is really no way around that guideline, then you should split your workflow into 2 phases: Reformat all existing code one time to fulfill that operator guideline. You may use any tool you like, even just a regular expression search and replace might be fine.
After that has been done, configure Eclipse to auto-format only changed lines, but always apply all formattings to each changed line. There is no good reason to not re-format the other 75 characters in an existing line of code, if you already touched 5 characters of it.
I'm writing a program which takes filenames and options on the command line in the usual way, and also can be directed to read arguments from a file. I'm implementing the semi-standard -- to turn off special treatment of subsequent arguments, and # as comment marker.
I also want to implement a marker for 'disregard all arguments from here on', i.e. an end marker. Is there a common/semi-standard way to indicate this? Or, what way would people find least surprising?
I would say implement multiline comments. C style /* */ would only work if you are not using wildcards in paths. XML style <!-- --> or asp <%-- --%> or matlab %{ %} might work. I think XML is the most recognized besides C.
This is also more useful than a simple terminator.
I haven't seen anything like that before. Do you envision the end marker being used during debugging/testing of the options in the file? If so, commenting out the unneeded lines seems like the most reasonable thing to do without adding extra complications.