Add ripgrep support

[timfjord]

This PR adds support for ripgrep search tool

I haven't covered the ag edge case that is mentioned in the README but I am happy to do that if I have more details

[AndrewRadev]

Thanks, this looks good! I would certainly appreciate help with ag, if I could only remember what the problem was :D. It might even be fixed already. If you'd like to help, maybe you could just enable it and try it out -- it might work fine now. I would keep an eye out on matches at the ends of files, because that's what past me has written down at least :). Really wish I'd left more info for my future self.

[AndrewRadev]

Trying to get ag to work, I'm running into trouble that's apparently quite recent. I'm not sure if the old issue is fixed, but this is a new one -- mixed delimiters.

Here's one result coming from ack:

bundle/csv/csv.vmb-2926-
bundle/csv/csv.vmb-2927-To have Vim automatically detect csv files, you need to do the following.
bundle/csv/csv.vmb-2928-
bundle/csv/csv.vmb:2929:   1) Create your user runtime directory if you do not have one yet. This
bundle/csv/csv.vmb:2930:      directory needs to be in your 'runtime' path. In Unix this would
bundle/csv/csv.vmb-2931-      typically the ~/.vim directory, while in Windows this is usually your
bundle/csv/csv.vmb-2932-      ~/vimfiles directory. Use :echo expand("~") to find out, what Vim thinks
bundle/csv/csv.vmb-2933-      your user directory is.
--

Now here's one coming from ag:

bundle/csv/csv.vmb:2926-
bundle/csv/csv.vmb:2927-To have Vim automatically detect csv files, you need to do the following.
bundle/csv/csv.vmb:2928-
bundle/csv/csv.vmb:2929:   1) Create your user runtime directory if you do not have one yet. This
bundle/csv/csv.vmb:2930:      directory needs to be in your 'runtime' path. In Unix this would
bundle/csv/csv.vmb:2931-      typically the ~/.vim directory, while in Windows this is usually your
bundle/csv/csv.vmb:2932-      ~/vimfiles directory. Use :echo expand("~") to find out, what Vim thinks
bundle/csv/csv.vmb:2933-      your user directory is.
--

They're very similar, but there's a subtle difference -- the lines that match (as opposed to being "context" lines from the -C flag) have mixed delimiters -- both : and -, like this: bundle/csv/csv.vmb:2927-To have....

The reason this is a problem is because the plugin tries to figure out what the delimiter is so it could validate whether the "filename" and "line number" that are parsed are actually valid filename + line number. This is to work around the issue of the filename having numbers and dashes in it (#15). It's an annoying problem and one that makes me think I need a separate backend setup for rg --json, but it's going to be a relatively large undertaking, maybe.

Anyway, I guess what I'd have to do is replace the split( with a series of matchstr calls and construct a data structure that holds both the pieces and the delimiters between them. The code is around here:

writable_search.vim/autoload/writable_search/file_parser.vim

Lines 17 to 70 in ae34560

	for delimiter in ['-', ':']
	let parts = split(line, delimiter)
	if line[len(line) - 1] == delimiter
	" the line ended in a delimiter, so the split will end up one character
	" short
	call add(parts, '')
	endif
	if len(parts) < 3
	" we expect [filename, line_number, text], so less than 3 parts means
	" this is not a line we can parse, go on to the next delimiter
	continue
	endif
	let delimiter_index = 0
	while delimiter_index < len(parts) - 2
	let filename = join(parts[0:delimiter_index], delimiter)
	let line_number = parts[delimiter_index + 1]
	if line_number !~ '^\d\+$'
	" not a real number, so this isn't the right breakdown
	let delimiter_index += 1
	continue
	endif
	let text = join(parts[delimiter_index + 2:], delimiter)
	if !filereadable(filename)
	" not an existing file, not the right breakdown
	let delimiter_index += 1
	continue
	endif
	if !has_key(self.cache, filename)
	let file_lines = readfile(filename)
	let self.cache[filename] = {'lines': file_lines}
	endif
	let file_lines = self.cache[filename].lines
	if len(file_lines) < line_number
	" the file exists, but the lines don't match up
	let delimiter_index += 1
	continue
	endif
	if file_lines[line_number - 1] != text
	" the file exists, the line exists, but the text doesn't match
	let delimiter_index += 1
	continue
	endif
	return [filename, line_number]
	endwhile
	endfor

If you'd like to try your hand with it, I'd certainly appreciate it! I think it will be kind of annoying to do, though. I'll write it down as a todo for myself for later, but let me know what you think.

[timfjord]

Hm, I was more worried if the ag case is also affecting rg 🙂

It is a bit tricky to dive into this as I am not 100% in it
But what I am definitely going to do is to keep an eye on how the rg integration performs(as I use this plugin as the main tool for serach)
Just a note here, I've added --vimgrep flag to the rg_command to make it works, but it is also available in ag so it might be a good idea to add it to the ag_command too (as far as I understand for ag it is a shortcut for --nogroup --nocolor --column) Most likely it won't fix the above issue but just something to keep in mind

So If I find something strange with the plugin I will let you know