complete-tag at the end of a buffer

complete-tag at the end of a buffer

[Sarah Weissman]

I thought I could implement general tab completion using tags in a
custom mode by following the advice found here:
http://emacsblog.org/2007/03/12/tab-completion-everywhere/
and simply replacing the call to dabbrev-expand with a call to
complete-tag. This worked fine, except when trying to do completion at
the end of the buffer, which generates an end-of-buffer error. This
seems to be because tags-completion-at-point-function (called by
complete-tags) is using forward-char without checking that it might go
past the end. To get around this I wrote my own version of
tags-completion-at-point-function, just to add in the extra check
around each call to forward char, and my own version of complete-tag
to call this function, but this seems kind of silly. Am I using the
wrong entry point for completing tags in a normal buffer? Or is this a
bug? Below I've included the original version and my version for
tags-completion-at-point-function, for reference.

(defun tags-completion-at-point-function ()
  "Using tags, return a completion table for the text around point.
If no tags table is loaded, do nothing and return nil."
  (when (or tags-table-list tags-file-name)
    (let ((completion-ignore-case (if (memq tags-case-fold-search '(t nil))
				      tags-case-fold-search
				    case-fold-search))
	  (pattern (funcall (or find-tag-default-function
				(get major-mode 'find-tag-default-function)
				'find-tag-default)))
	  beg)
      (when pattern
	(save-excursion
          (forward-char (1- (length pattern)))
          (search-backward pattern)
          (setq beg (point))
          (forward-char (length pattern))
          (list beg (point) (tags-lazy-completion-table) :exclusive 'no))))))

(defun my-tags-completion-at-point-function ()
  "Using tags, return a completion table for the text around point.
If no tags table is loaded, do nothing and return nil."
  (when (or tags-table-list tags-file-name)
    (let ((completion-ignore-case (if (memq tags-case-fold-search '(t nil))
				      tags-case-fold-search
				    case-fold-search))
	  (pattern (funcall (or find-tag-default-function
				(get major-mode 'find-tag-default-function)
				'find-tag-default)))
	  beg)
      (when pattern
	(save-excursion
	  (save-excursion
	    (end-of-buffer)
	    (setq e (point)))
	  (if (> (+ (point) (1- (length pattern))) e)
	      (end-of-buffer)
	    (forward-char (1- (length pattern))))
          (search-backward pattern)
          (setq beg (point))
	  (if (> (+ (point) (1- (length pattern))) e)
	      (end-of-buffer)
          (forward-char (length pattern)))
          (list beg (point) (tags-lazy-completion-table) :exclusive 'no))))))

Re: complete-tag at the end of a buffer

[Dmitry Gutov]

Sarah Weissman <seweissman@gmail.com> writes:

> I thought I could implement general tab completion using tags in a
> custom mode by following the advice found here:
> http://emacsblog.org/2007/03/12/tab-completion-everywhere/
> and simply replacing the call to dabbrev-expand with a call to
> complete-tag. This worked fine, except when trying to do completion at
> the end of the buffer, which generates an end-of-buffer error. This
> seems to be because tags-completion-at-point-function (called by
> complete-tags) is using forward-char without checking that it might go
> past the end. To get around this I wrote my own version of
> tags-completion-at-point-function, just to add in the extra check
> around each call to forward char, and my own version of complete-tag
> to call this function, but this seems kind of silly. Am I using the
> wrong entry point for completing tags in a normal buffer? Or is this a
> bug? Below I've included the original version and my version for
> tags-completion-at-point-function, for reference.

I'm pretty sure it's a bug.
It should be able to complete anywhere in the buffer.