Does ctags have multiline regex support?

[andreicristianpetcu]

Hello,
I have some Java code

@Subscribe
public void catchEvent(SomeEvent

And I want to match it with something like this in my ~/.ctags

--langdef=javaspring
--langmap=javaspring:.java
--regex-javaspring=/@Subscribe(\s)*([a-z ]+)\s([a-zA-Z]*)\(([a-zA-Z]*)/\3-\4/

This should generate something like "catchEvent-SomeEvent" but it does not. If I remove the newline between @subscribe and public void it gets it.

@Subscribe public void catchEvent(SomeEvent

rm -rf tags && ctags -R . && cat tags|grep Subs
catchEvent-SomeEvent    PathToMyFile.java    /^@Subscribe public void catchEvent(SomeEvent event) {$/;"      r

Do you have any idea if I can do this with the current ctag implementation. Am I doing something wrong or ctags does not support multi-line regex now.

Thank you!

[andreicristianpetcu]

ctags --version
Exuberant Ctags 5.9~svn20110310, Copyright (C) 1996-2009 Darren Hiebert
  Compiled: Oct  7 2014, 13:52:37
  Addresses: <dhiebert@users.sourceforge.net>, http://ctags.sourceforge.net
  Optional compiled features: +wildcards, +regex

[masatake]

As far as I know ctags utilizes the regex engine in line oriented way.
We don't have any concrete plan to extend it to handle multiple line.

[andreicristianpetcu]

Thank you @masatake do you think I can make a workaround like using other regex engine or something?

[masatake]

There are some approaches.

If you don't have enough time, and you know well aobut a script language, you can use xcmd.
See https://github.com/fishman/ctags/blob/master/docs/xcmd.rst.

If you have time and know C language, look at parseJavaAnnotation in c.c and extend it as you want.

If you have enough time, tell us your idea how ctags should deal java's annotation(not only spring.)
#64 and #80 may be related to this issue.

[andreicristianpetcu]

The beauty of ctags is the fact that anybody can extend it with project specific settings by writing just some small regex. Please take a look at what I did for AngularJS (http://bit.ly/16gcKd7). My ctags settings (and vim) figure out AngularJS controllers, directives, modules etc. This is something that IntelliJ cannot. AngularJS is hip now, tomorrow it will not be. I think that ctags should have basic support for languages but not for frameworks. The JavaScript and Java support is great. I just wanted to have something with vim+ctags that no IDE offers. Multiline regex would give me the possibility to lookaround for stuff before and after and build a tag based on that. This is not language specific, it should not be in ctags source code, it should be in people's ~/.ctags on github :P

I have no C skills but I plan to learn. I see no need to extend parseJavaAnnotation now, it is great the way it is. I will check out xcmd and the two issues you pointed out.

Thank you again @masatake

[masatake]

(I think it is nice if we have multiline regex support, however I cannot have an image how to implement it well in the current ctags code base. See lregex.c. )

[andreicristianpetcu]

@masatake I understand implementing it is complicated with the current codebase. I took a look at ---xcmd=COMMAND syntax and I will try to use it. I only read the documentation and it seems it is too much for me. From what I get from the documentation it will be invoked twice, once some sort of init that returns an exit code 0, something else or 77 and after that once for each file. It seems that the COMMAND needs to keep some sort of internal state.

I don't think current implementation is very good for my case since oneliners using grep do not keep state. I will try to use it and post on this issue an example if it works, if not, I will post what issues I had. Thank you for your input.

[masatake]

After learning etags.c of emacs, I know how can we implment multiline regex parser in ctags.
Search "multi_line" in https://searchcode.com/codesearch/view/56189132/.

You can accumulate all input at iFileGetLine of read.c.
You can run multipeline regex matcher at fileClose.

[masatake]

Now Universal-ctags has the most of all facilities for implementing multline regex parser.

[masatake]

It works. Please, try #1224 if you have still interest.