Which error format should be used for Vim and Cppcheck?

问题

I use the following script to integrate Cppcheck with gVim:

" vimcppcheck.vim
"  ===================================================================
"  Code Checking with cppcheck (1)
"  ===================================================================

function! Cppcheck_1()
  set makeprg=cppcheck\ --enable=all\ %
  setlocal errorformat=[%f:%l]:%m
  let curr_dir = expand('%:h')
  if curr_dir == ''
    let curr_dir = '.'
  endif
  echo curr_dir
  execute 'lcd ' . curr_dir
  execute 'make'
  execute 'lcd -'
  exe    ":botright cwindow"
  :copen
endfunction


:menu Build.Code\ Checking.cppcheck :cclose<CR>:update<CR>:call Cppcheck_1() <cr>

Normally this is very good, but this script sometimes creates trouble when checking wrong pointers with Cppcheck.

For example, I have the following C code:

/* test_cppcheck.c */
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>

int main(void) {
  int *ptr01;

  *ptr01 = (int *)malloc((size_t)10 * sizeof(int)); /* FIXME: I intensionally written *ptr01 instead of ptr01 */
  if(ptr01==NULL) {
    fprintf(stderr, "\ndynamic memory allocation failed\n");
    exit(EXIT_FAILURE);
  }
  free(ptr01);
  ptr01 = NULL;
}

The quickfix list shows:

|| Checking test_cppcheck.c...
H:\codes\test_cppcheck.c:11] -> [test_cppcheck.c|12| (warning) Possible null pointer dereference: ptr01 - otherwise   it is redundant to check it against null.
H:\codes\test_cppcheck.c|11| (error) Uninitialized variable: ptr01
H:\codes\test_cppcheck.c|16| (error) Uninitialized variable: ptr01
H:\codes\test_cppcheck.c|12| (error) Uninitialized variable: ptr01
|| Checking usage of global functions..
|| (information) Cppcheck cannot find all the include files (use --check-config for details)

After a lot of Vim errors, a new file '11] -> [test_cppcheck.c' is created in a new buffer. When I double-click the first error, nothing can be done from the quickfix window. It is because of the errorformat as much as I know.

The -> instead of : is creating all the trouble, although I know minor tweaking of this script will fix this issue, but I am tired of doing so.

Please try this first. How can I deal with this?

回答1:

Without the original format of the error this is guesswork, but I think you need to add an alternative to the 'errorformat' definition (these are comma-separated):

setlocal errorformat=[%f:%l]\ ->\ %m,[%f:%l]:%m

PS: You should also use :setlocal for the 'makeprg' option to restrict it to the current buffer, too.

回答2:

Now I'm using the script below, and it's working perfectly as I expected.

This can be a general solution to everyone who is interested to integrate Cppcheck with Vim.

Of course, this script can be improved a lot. But it's a starting point for them.

" vimcppcheck.vim
"  ===================================================================
"  Code Checking with cppcheck (1)
"  Thanks to Mr. Ingo Karkat
"  http://stackoverflow.com/questions/19157270/vim-cppcheck-which-errorformat-to-use
"  ===================================================================

function! Cppcheck_1()
  setlocal makeprg=cppcheck\ --enable=all\ %
  " earlier it was: " setlocal errorformat=[%f:%l]:%m
  " fixed by an advise by Mr. Ingo Karkat
  setlocal errorformat+=[%f:%l]\ ->\ %m,[%f:%l]:%m
  let curr_dir = expand('%:h')
  if curr_dir == ''
    let curr_dir = '.'
  endif
  echo curr_dir
  execute 'lcd ' . curr_dir
  execute 'make'
  execute 'lcd -'
  exe    ":botright cwindow"
  :copen
endfunction


:menu Build.Code\ Checking.cppcheck :cclose<CR>:update<CR>:call Cppcheck_1() <cr>

来源：https://stackoverflow.com/questions/19157270/which-error-format-should-be-used-for-vim-and-cppcheck

标签

vim

cppcheck