How to wrap long lines of tooltip

[crotoc]

Thanks for the great plugin!

I use company to auto complete file path. However, it's usually very long and the lines appearing in the tooltip doesn't wrap so very hard to select the right choice. Is it possbile to wrap the lines? Please advice. Thanks!

[dgutov]

Hi!

Wrap how?

The best we could do for this particular case, I guess, is to only treat the last path segment as "prefix".

[crotoc]

How to setup up ignore the prefix you mentioned to show the next part of path? Current setup shows each file/directory in a line, if it's too long I don't know which line is the correct one to complete. Wrapping means show the each line in two lines to show the whole path. Thanks!

[crotoc]

@dgutov For example, if the file path I need is:

Can I setup to just show "dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disTRUE_max20_ws1e6" when I try to auto complete at "~/fs0/subgwas/scz/145loci/gibbs_result/" or can it is shown as:

~/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic
_disTRUE_max20_ws1e6/a_file
~/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic
_disTRUE_max20_ws1e6/b_file
~/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic
_disTRUE_max20_ws1e6/c_file

so that I can know which selection is right. I have searched for so many times and don't identify a way to workaround. Please give me some advice. Thanks!

[dgutov]

There's no existing option, it will be a change in the implementation.

For instance, like this:

diff --git a/company-files.el b/company-files.el
index 8859266..a6339d0 100644
--- a/company-files.el
+++ b/company-files.el
@@ -98,15 +98,15 @@ The values should use the same format as `completion-ignored-extensions'."
 (defvar company-files--completion-cache nil)
 
 (defun company-files--complete (prefix)
-  (let* ((dir (file-name-directory prefix))
-         (file (file-name-nondirectory prefix))
+  (let* ((dir (file-name-directory
+               (company-files--grab-existing-name)))
+         (file prefix)
          (key (list file
                     (expand-file-name dir)
                     (nth 5 (file-attributes dir))))
          (completion-ignore-case read-file-name-completion-ignore-case))
     (unless (company-file--keys-match-p key (car company-files--completion-cache))
-      (let* ((candidates (mapcar (lambda (f) (concat dir f))
-                                 (company-files--directory-files dir file)))
+      (let* ((candidates (company-files--directory-files dir file))
              (directories (unless (file-remote-p dir)
                             (cl-remove-if-not (lambda (f)
                                                 (and (company-files--trailing-slash-p f)
@@ -116,7 +116,9 @@ The values should use the same format as `completion-ignored-extensions'."
              (children (and directories
                             (cl-mapcan (lambda (d)
                                          (mapcar (lambda (c) (concat d c))
-                                                 (company-files--directory-files d "")))
+                                                 (company-files--directory-files
+                                                  (concat dir d)
+                                                  "")))
                                        directories))))
         (setq company-files--completion-cache
               (cons key (append candidates children)))))
@@ -139,7 +141,7 @@ File paths with spaces are only supported inside strings."
   (interactive (list 'interactive))
   (cl-case command
     (interactive (company-begin-backend 'company-files))
-    (prefix (company-files--grab-existing-name))
+    (prefix (file-name-nondirectory (company-files--grab-existing-name)))
     (candidates (company-files--complete arg))
     (location (cons (dired-noselect
                      (file-name-directory (directory-file-name arg))) 1))

I'm tempted to push this, but it does call company-files--grab-existing-name one extra time, which could lead to slower performance over Tramp. Or not, if file-exists-p hits some kind of cache. Should test it out.

[crotoc]

Happy holidays! Just have several days off! Thanks for your reply!

I just merge the changes to the company-files.el in my installed dir but no wrapping or no ignoring of prefix. Am I doing something wrong? Thanks!

[dgutov]

Let's see.

What does M-x company-diag say, at that position in the buffer?

(And happy holidays!)

[crotoc]

The command shows:

Emacs 26.3 (x86_64-pc-linux-gnu) of 2020-03-25 on c418c5199a0f
Company 0.9.13

company-backends: (company-emacs-eclim company-bbdb company-semantic company-clang company-xcode company-cmake company-capf company-files
(company-dabbrev-code company-gtags company-etags company-keywords)
company-oddmuse company-dabbrev)

Used backend: company-files
Major mode: perl-mode
Prefix: "/fs0/subgwas/scz/145loci/gibbs_result/"
Completions:
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws5e5/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disTRUE_max20_ws1e6/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disTRUE_max20_ws5e5/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.5e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.5e-6.5e5.bed.fmt_generic_disFALSE_max20_ws5e5/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.5e-6.5e5.bed.fmt_generic_disTRUE_max20_ws1e6/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.5e-6.5e5.bed.fmt_generic_disTRUE_max20_ws5e5/"
"/fs0/subgwas/scz/145loci/gibbs_result/evaluateTRUE/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/20191127-1315-UIHVO.param"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/all.pp"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/all.pp_Q5/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/ldsc_10000.pdf"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/ldsc_100000.pdf"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/rlt_nearest/"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt.evi.processed"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt.evi.pvalue"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt.evi.weight"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt.generic.processed"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt.generic.weight"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt.genesInRegion"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt_gibbs_sampling_go_rp_0.3_posterior_probability"
"/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt_gibbs_sampling_go_rp_0.3_posterior_probability.RData"
"~/fs0/subgwas/scz/145loci/gibbs_result/dir_scz.145.sub.1e-6.5e5.bed.fmt_generic_disFALSE_max20_ws1e6/scz.145.sub.1e-6.5e5.bed.fmt_gibbs_sampling_go_rp_0.3_posterior_probability.training.
RData"
......

The buffer of completion shows:

Let me know if you need more info! Thanks!

[dgutov]

Okay, that looks legit.

Did you apply the patch? And re-evaluated the definitions? Or restarted Emacs?

[crotoc]

I installed company through EPLA and it's located in ~/.emacs.d/elpa/company-20200725.2348/. So I opened the company-fiiles.el in this dir and shows:

And then, I closed emas and reopened a file and tried the command you mentioned. Is it the correct way to apply the patch?

LISP is hard to read because of so manny brackets:)

[dgutov]

And then, I closed emas and reopened a file and tried the command you mentioned. Is it the correct way to apply the patch?

There are roughly three options:

1. Apply the patch, don't restart Emacs, but visit every affected definition and press C-M-x to re-evaluate the definition.
2. M-x byte-compile-file after applying the patch and saving the file.
3. Set load-prefer-newer to t early in the init file, then you don't have to byte-compile the new version to get it loaded.

[crotoc]

Many thanks!! I use the secend option and it shows:

This is exactly what I want! Appreciate very much!

At last, happy new year, buddy!

[dgutov]

After some testing, there are a couple of problems with this patch (in particular, seeing the Matching input is required error from time to time). You're welcome to use it (though be aware that the direct changes to the files will be overwritten when the package updates), but I'll avoid pushing this change for now.

A better solution is in the works.

[crotoc]

Thanks for the remind! The current patch is enought for me! Looking forward to the perfect solution. 👍

[dgutov]

Let's keep this open in the meantime. 👍

[crotoc]

Sure! I hope I can help but I am still learning lisp.

[crotoc]

Hi, is there any update about this issue? Has it been pushed to the main so that this works instantly after installation without patching? Thanks!

[dgutov]

Not yet, but thanks for the bump.

[dgutov]

Pushed after some more tweaking.

Let me know if you see any problems.

[crotoc]

Thanks very much! Will try :) Can't believe you are still monitoring!