How to search in specified folder or among files of specified type?

[ravisumit33]

Currently to start search I press <leader>/ and it opens Grep. What if I want to search in specific folder or among files of specific extension? I tried typing file:js$ as a pattern in the search box to filter only javascript files but it didn't work.

[drowning-cat]

#1119

[drowning-cat]

#957 (comment)

[ravisumit33]

I understood we can use fzf search syntax in non-live mode. But where does file:.lua$ come from? I don't think its supported by fzf. Or is it?

[drowning-cat]

#957 (comment)

Be aware though that if you use !.txt$, then that will match the last part of the item's search string, which is not always the filename. So to filter out filenames this way, it's better to do file:!.txt$.

https://github.com/folke/snacks.nvim/blob/main/docs/picker.md#-features

✨ Features

• 🔎 over 40 built-in sources
• 🚀 Fast and powerful fuzzy matching engine that supports the fzf search syntax
  • additionally supports field searches like file:lua$ 'function
  • files and grep additionally support adding optiont like foo -- -e=lua

https://github.com/folke/snacks.nvim/blob/main/docs/picker.md#%EF%B8%8F-config

matcher = {
  fuzzy = true, -- use fuzzy matching
  smartcase = true, -- use smartcase
  ignorecase = true, -- use ignorecase
  sort_empty = false, -- sort results when the search string is empty
  filename_bonus = true, -- give bonus for matching file names (last part of the path)
  file_pos = true, -- support patterns like `file:line:col` and `file:line`
  -- the bonusses below, possibly require string concatenation and path normalization,
  -- so this can have a performance impact for large lists and increase memory usage
  cwd_bonus = false, -- give bonus for matching files in the cwd
  frecency = false, -- frecency bonus
  history_bonus = false, -- give more weight to chronological order
},

[drowning-cat]

Picker module supports additional syntax on top of fzf for searching <field>:

group > file:lua$ text:nvim_create_augroup

group > file:lua$ 'nvim_create_augroup

In the examples above, there are two fields: file, text (text: is the same as ').

---

However, it's unclear where these fields are defined—you need to search them in the source code.
I suppose the search fields correspond to the properties assigned to the items.

For example:

picker/source/diagnostics.lua

items[#items + 1] = {
  text = table.concat({ severity or "", tostring(diag.code or ""), file, diag.source or "", diag.message }, " "),
  file = file,
  buf = diag.bufnr,
  is_current = buf == current_buf and 0 or 1,
  is_cwd = file:sub(1, #cwd) == cwd and 0 or 1,
  lnum = diag.lnum,
  severity = diag.severity,
  pos = { diag.lnum + 1, diag.col },
  end_pos = diag.end_lnum and { diag.end_lnum + 1, diag.end_col } or nil,
  item = diag,
  comment = diag.message,
}

They are also listed, at least partially, in picker.md inside sort.fields for each source:

sort = {
  fields = {
    "is_current",
    "is_cwd",
    "severity",
    "file",
    "lnum",
  },
}

[dpetka2001]

@drowning-cat You can press <a-d> to see the debug for the item. It also mentions the fields that it has.

I'm searching for how the item is constructed for the snacks.grep picker, so if you happen to have any clue about that I'd appreciate it. Specifically, I'm trying to understand how exactly the file is transformed to only show the relative path to the current working directory and not the full path of the item.

[dpetka2001]

Never mind, just found out that this is the default output of rg.