blink-cmp-wiki-links

A blink.cmp completion source that provides fuzzy matching for markdown file names and existing wiki links, automatically formatting them as wiki links ([[filename]]).

✨ Features

• Always-on completion: No need to type [[ - completions appear as you type
• Fuzzy matching: Powered by blink.cmp's performance-optimized fuzzy matcher
• Auto wiki-link formatting: Accepted completions become [[filename]]
• Dual search backends: Combines file discovery (fd) with existing wiki link search (rg)
• Smart deduplication: File results take priority over existing wiki links when duplicates exist
• Smart workspace detection: Automatically finds your notes workspace
• Performance optimized: Uses fd for fast file discovery and rg for existing link detection
• Configurable: Customize filetypes, exclusions, and workspace detection
• Contextual file preview: Shows file content preview starting from the relevant line
• Minimum prefix length: Configurable minimum characters before search starts

📦 Installation

📝 Requirements

• Neovim 0.9+
• blink.cmp
• fd: For fast file discovery
• rg: For searching existing wiki links in files

Using lazy.nvim

{
  "saghen/blink.cmp",
  dependencies = { "l-lin/blink-cmp-wiki-links" },
  opts = {
    sources = {
      default = { "lsp", "path", "snippets", "buffer", "wiki_links" },
      providers = {
        wiki_links = {
          name = "WikiLinks",
          module = "blink-cmp-wiki-links",
          score_offset = 85, -- High priority for wiki links
        },
      },
    },
  },
}

Using other plugin managers

1. Install the plugin using your preferred method
2. Add the wiki_links provider to your blink.cmp configuration

⚙️ Configuration

The configuration of blink-cmp-wiki-links needs to be embedded into the configuration for blink:

{
  "saghen/blink.cmp",
  dependencies = {
    "l-lin/blink-cmp-wiki-links",
  },
  opts = {
    sources = {
      default = { "lsp", "path", "snippets", "buffer", "wiki_links" },
      providers = {
        wiki_links = {
          name = "WikiLinks",
          module = "blink-cmp-wiki-links",
          opts = {
            -- Enable for additional file types
            filetypes = { "markdown", "md", "txt" },
            -- More exclusions
            exclude_paths = { ".git", ".obsidian", "assets", "images", ".trash" },
            -- Specifies how to find the root of the project where the fd
            -- search will start from. Accepts the same options as the marker
            -- given to `:h vim.fs.root()` which offers many possibilities for
            -- configuration. If none can be found, defaults to Neovim's cwd.
            --
            -- Examples:
            -- - ".git" (default)
            -- - { ".git", "package.json", ".root" }
            project_root_marker = ".git",
            -- Require more characters before searching
            prefix_min_len = 3,
            -- Show more lines in preview
            preview_line_length = 20,
            -- Custom icon for the completion item kind
            kind_icon = "",
            fd_opts = {
              -- Additional options to pass to the fd command
              additional_fd_options = {},
            },
            rg_opts = {
              -- The maximum file size of a file that ripgrep should include in
              -- its search. Useful when your project contains large files that
              -- might cause performance issues.
              -- Examples:
              -- "1024" (bytes by default), "200K", "1M", "1G", which will
              -- exclude files larger than that size.
              max_filesize = "1M",
              -- The casing to use for the search in a format that ripgrep accepts.
              search_casing = "--ignore-case",
              -- Additional options to pass to the rg command
              additional_rg_options = {},
            },
          },
          score_offset = 85,
        },
      },
    },
  },
}

🚀 Usage

1. Open any file with a supported filetype (markdown, md by default)
2. Start typing a filename (minimum 3 characters by default)
3. See completions from two sources:
   • Files: Matching filenames from your workspace (higher priority)
   • Existing links: Wiki links already used in your project files
4. Select from the fuzzy-matched completions
5. The selected item becomes [[filename]] automatically
6. Hover over completions to see contextual file content preview:
   • Files: Preview starts from line 1
   • Existing links: Preview starts from the line where the link was found

⚡ Performance

• Dual-backend architecture: Combines fd for file discovery and rg for existing link search
• Concurrent execution: Both backends run simultaneously for optimal performance
• Smart deduplication: Files take priority over existing links to avoid duplicates
• Prefix filtering: Only searches when minimum prefix length is met
• Lazy preview loading: File content loaded only when needed for documentation
• Efficient exclusions: Directory exclusions applied during both fd and rg scanning
• Contextual previews: Uses sed to efficiently extract specific line ranges for preview

📄 License

MIT License - see LICENSE file for details.

👏 Acknowledgements

• obsidian-various-complements-plugin for its ingenious feature of automatically adding wiki links, allowing writers to easily link files without having to remember if a file with that name already exists, and enabling them to maintain their writing flow.
• blink-ripgrep served as inspiration for this codebase.