telescope-frecency.nvim

A telescope.nvim extension that offers intelligent prioritization when selecting files from your editing history.

Using an implementation of Mozilla's Frecency algorithm (used in Firefox's address bar), files edited frecently are given higher precedence in the list index.

As the extension learns your editing habits over time, the sorting of the list is dynamically altered to prioritize the files you're likely to need.

• Scores shown in finder for demonstration purposes - disabled by default

Frecency: Sorting by 'frequency' and 'recency'

'Frecency' is a score given to each unique file indexed in a file history database.

A timestamp is recorded once per session when a file is first loaded into a buffer.

The score is calculated using the age of the 10 (customizable) most recent timestamps and the total amount of times that the file has been loaded:

Recency values (per timestamp)

These values are customizable in setup().

Timestamp age	Value
4 hours	100
1 day	80
3 days	60
1 week	40
1 month	20
90 days	10

Score calculation

score = frequency * recency_score / max_number_of_timestamps

What about files that are neither 'frequent' or 'recent' ?

Frecency naturally works best for indexed files that have been given a reasonably high score.

New projects or rarely used files with generic names either don't get listed at all or can be buried under results with a higher score.

Frecency tackles this with Workspace Filters:

The workspace filter feature enables you to select from user defined filter tags that map to a directory or collection of directories. Filters are applied by entering :workspace_tag: anywhere in the query. Filter name completion is available by pressing <Tab> after the first : character.

When a filter is applied, results are reduced to entries whose path is a descendant of the workspace directory. The indexed results are optionally augmented with a listing of all files found in a recursive search of the workspace directory. Non-indexed files are given a score of zero and appear below the 'frecent' entries. When a non-indexed file is opened, it gains a score value and is available in future 'frecent' search results.

If the active buffer (prior to the finder being launched) is attached to an LSP server, an automatic LSP tag is available, which maps to the workspace directories provided by the language server.

Requirements

• telescope.nvim (required)
• nvim-web-devicons (optional)
• ripgrep or fd (optional)

NOTE: ripgrep or fd will be used to list up workspace files. They are extremely faster than the native Lua logic. If you don't have them, it fallbacks to Lua code automatically. See the detail for workspace_scan_cmd option.

Installation

Packer.nvim

use {
  "nvim-telescope/telescope-frecency.nvim",
  config = function()
    require("telescope").load_extension "frecency"
  end,
}

Lazy.nvim

{
  "nvim-telescope/telescope-frecency.nvim",
  config = function()
    require("telescope").load_extension "frecency"
  end,
}

If no database is found when running Neovim with the plugin installed, a new one is created and entries from shada v:oldfiles are automatically imported.

Usage

:Telescope frecency

or to map to a key:

-- Bind command
vim.keymap.set("n", "<leader><leader>", "<Cmd>Telescope frecency<CR>")
-- Bind Lua function directly
vim.keymap.set("n", "<leader><leader>", function()
  require("telescope").extensions.frecency.frecency {}
end)

Use a specific workspace tag:

:Telescope frecency workspace=CWD

or

-- Bind command
vim.keymap.set("n", "<leader><leader>", "<Cmd>Telescope frecency workspace=CWD<CR>")
-- Bind Lua function directly
vim.keymap.set("n", "<leader><leader>", function()
  require("telescope").extensions.frecency.frecency {
    workspace = "CWD",
  }
end)

Filter tags are applied by typing the :tag: name (adding surrounding colons) in the finder query. Entering :<Tab> will trigger omnicompletion for available tags.

Dealing with upper case letters

In default, the sorter always ignores upper case letters in your input string. But when 'smartcase' is ON and input string includes one upper case letter at least, it matches against exact the same as you input.

input string	'smartcase' is ON	'smartcase' is OFF
abc	matches abc, ABC, aBc	matches abc, ABC, aBc
aBc	matches aBc	no match
ABC	matches ABC	no match

Configuration

See default configuration for full details on configuring Telescope.

• auto_validate (default: true)

  If true, it removes stale entries count over than db_validate_threshold. See the note for DB maintenance

• db_root (default: vim.fn.stdpath "data")

  Path to parent directory of custom database location. Defaults to $XDG_DATA_HOME/nvim if unset.

• db_safe_mode (default: true)

  If true, it shows confirmation dialog by vim.ui.select() before validating DB. See the note for DB maintenance.

• db_validate_threshold (default: 10)

  It will removes over than this count in validating DB.

• default_workspace (default: nil)

  Default workspace tag to filter by e.g. 'CWD' to filter by default to the current directory. Can be overridden at query time by specifying another filter like ':*:'.

• disable_devicons (default: false)

  Disable devicons (if available)

• hide_current_buffer (default: false)

  If true, it does not show the current buffer in candidates.

• filter_delimiter (default: ":")

  Delimiters to indicate the filter like :CWD:.

• ignore_patterns (default: for non-Windows → { "*.git/*", "*/tmp/*", "term://*" }, for Windows → { [[*.git\*]], [[*\tmp\*]], "term://*" })

  Patterns in this table control which files are indexed (and subsequently which you'll see in the finder results).

• matcher (default: "default")

  CAUTION
  This option is highly experimental.

  In default, it matches against candidates by the so-called “substr matcher”, that is, you should input characters ordered properly. If you set here with "fuzzy", it uses fzy matcher implemented in telescope itself, and combines the result with recency scores. With this, you can select candidates fully fuzzily, besides that, can select easily ones that has higher recency scores.

  See the discussion in #165.

• max_timestamps (default: 10)

  Set the max count of timestamps DB keeps when you open files. It ignores the value and use 10 if you set less than or equal to 0.

  CAUTION When you reduce the value of this option, it removes old timestamps when you open the file. It is reasonable to set this value more than or equal to the default value: 10.

• path_display (default: nil)

  Overwrite the path_display setting in telescope.nvim itself. See :h telescope.defaults.path_display for acceptable values. This setting will be used by these priorities below.

  1. Option specified with the command or Lua code.
  • :Telescope frecency path_display={"absolute"}.
  • or :lua require("telescope").extensions.frecency.frecency { path_display = { "absolute" } }.

  2. opts.extensions.frecency.path_display in setup.

  3. opts.defaults.path_display in setup.

     require("telescope").setup {
       defaults = {
         -- This has the 3rd precedence.
         path_display = { "absolute" },
       },
       extensions = {
         frecency = {
           -- This has the 2nd precedence.
           path_display = { "shorten" },
         },
       },
     }

• recency_values (default: see below)

  Set weighting factors for calculating “frecency”. This option does not affect values already recorded in DB. So you can change these values without spoiling data.

  -- default values
  recency_values = {
    { age = 240, value = 100 }, -- past 4 hours
    { age = 1440, value = 80 }, -- past day
    { age = 4320, value = 60 }, -- past 3 days
    { age = 10080, value = 40 }, -- past week
    { age = 43200, value = 20 }, -- past month
    { age = 129600, value = 10 }, -- past 90 days
  }

• scoring_function (default: see below)

  CAUTION
  This option is highly experimental.

  This will be used only when matcher option is "fuzzy". You can customize the logic to adjust scores between fzy matcher scores and recency ones.

  -- the default value
  ---@param recency integer
  ---@param fzy_score number
  ---@return number
  scoring_function = function(recency, fzy_score)
    local score = (10 / (recency == 0 and 1 or recency)) - 1 / fzy_score
    -- HACK: -1 means FILTERED, so return a bit smaller one.
    return score == -1 and -1.000001 or score
  end,

  NOTE: telescope orders candidates in the ascending order. It also accepts negative numbers, but -1 means the candidates should not be shown.

• show_filter_column (default: true)

  Show the path of the active filter before file paths. In default, it uses the tail of paths for 'LSP' and 'CWD' tags. You can configure this by setting a table for this option.

  -- show the tail for "LSP", "CWD" and "FOO"
  show_filter_column = { "LSP", "CWD", "FOO" }

• show_scores (default : false)

  To see the scores generated by the algorithm in the results, set this to true.

• show_unindexed (default: true)

  Determines if non-indexed files are included in workspace filter results.

• workspace_scan_cmd (default: nil)

  This option can be set values as "LUA"|string[]|nil. With the default value: nil, it uses these way below to make entries for workspace files. It tries in order until it works successfully.

  1. rg -.g '!.git' --files
  2. fdfind -Htf
  3. fd -Htf
  4. Native Lua code (old way)

  If you like another commands, set them to this option, like workspace_scan_cmd = { "find", ".", "-type", "f" }.

  If you prefer Native Lua code, set workspace_scan_cmd = "LUA".

• workspaces (default: {})

  This table contains mappings of workspace_tag -> workspace_directory. The key corresponds to the :tag_name used to select the filter in queries. The value corresponds to the top level directory by which results will be filtered.

Example Configuration:

telescope.setup {
  extensions = {
    frecency = {
      db_root = "/home/my_username/path/to/db_root",
      show_scores = false,
      show_unindexed = true,
      ignore_patterns = { "*.git/*", "*/tmp/*" },
      disable_devicons = false,
      workspaces = {
        ["conf"]    = "/home/my_username/.config",
        ["data"]    = "/home/my_username/.local/share",
        ["project"] = "/home/my_username/projects",
        ["wiki"]    = "/home/my_username/wiki"
      }
    }
  },
}

Note for Database

Location

The default location for the database is $XDG_DATA_HOME/nvim (eg ~/.local/share/nvim/ on linux). This can be configured with the db_root config option. The filename for the database is file_frecency.bin.

Maintenance

By default, frecency will prune files that no longer exist from the database. In certain workflows, switching branches in a repository, that behaviour might not be desired. The following configuration control this behaviour:

db_safe_mode

When this is enabled, the user will be prompted before any entries are removed from the database.

auto_validate

When this to false, stale entries will never be automatically removed.

The command FrecencyValidate can be used to clean the database when auto_validate is disabled.

" clean DB
:FrecencyValidate
" clean DB without prompts to confirm
:FrecencyValidate!

Delete entries

You can delete entries from DB by FrecencyDelete command. This command does not remove the file itself, only from DB.

" delete the current opened file
:FrecencyDelete

" delete the supplied path
:FrecencyDelete /full/path/to/the/file

Note about the compatibility for the former version.

The former version of this plugin has used SQLite3 library to store data. #172 has removed the whole code for that. If you prefer the old SQLite database, you should lock the version to a3e818d with your favorite plugin manager.

-- example for lazy.nvim
{
  "nvim-telescope/telescope-frecency.nvim",
  commit = "a3e818d001baad9ee2f6800d3bbc71c4275364ae",
}

Highlight Groups

TelescopeBufferLoaded
TelescopePathSeparator
TelescopeFrecencyScores
TelescopeQueryFilter

TODO: describe highlight groups

References

• Mozilla: Frecency algorithm