█▀▀▀█▀▀▀█▀▀▀█▀▀▀█▀▀▀▀▀▀▀█▀▀▀█▀▀▀█▀▀▀▀▀▀▀█▀▀▀▀█▀▀█    █▀▀▀▀█▀▀█▀▀▀█▀▀▀█▀▀▀▀▀█▀▀▀▀▀▀▀█
█   █   █   █   █   ▄▄  █   █   █   ▄▄▄▄█    █  █    █    █  █   █   █▄▄ ▄▄█  ▄ ▄  █
█       █       █       █       █       █  █ █  █    █  █ █  █   █   ███ ███  █ █  █
█   █   █▀▀▀▀   █   █████   █   █   ▀▀▀▀█  █    █    █  █    ██▄   ▄██▀▀ ▀▀█  █ █  █
█▄▄▄█▄▄▄█▄▄▄▄▄▄▄█▄▄▄█████▄▄▄█▄▄▄█▄▄▄▄▄▄▄█▄▄█▄▄▄▄█ ██ █▄▄█▄▄▄▄████▄████▄▄▄▄▄█▄▄█▄█▄▄█

A Neovim plugin for showing TeX-style hyphenation breakpoints in text.

Table of Contents

• Features
• Motivation
• Requirements
• Installation
• Usage
• Configuration
• Pattern Data
• Commands
• API
• Limitations
• License

Features

• Shows hyphenation suggestions for the word under the cursor.
• Uses Knuth-Liang hyphenation patterns, the same family of patterns used by TeX.
• Ships with built-in en-us and en-gb support.
• Supports user-supplied TeX Live-style pattern files.
• Uses inline virtual text to show hints without modifying the buffer.
• Lazily loads core Lua modules and language pattern tables on first use.
• Works in any text buffer; LaTeX editing is the primary motivation.

Motivation

hyphen.nvim can be used in any text buffer, but its primary motivation is LaTeX editing. When a rendered PDF contains an overfull line or an awkward line break, manual discretionary hyphenation can be useful.

For example, when the cursor is on hyphenation, the plugin can show breakpoints equivalent to:

hy-phen-ation

hyphen.nvim currently only shows breakpoints; it does not insert text automatically.

Requirements

• Neovim 0.10 or newer. Required for virt_text_pos = "inline".
• No TeX Live installation is required for built-in en-us and en-gb support.
• TeX Live-style .pat.txt and .hyp.txt files are only needed for custom languages or custom pattern sources.

Installation

Install hyphen.nvim with your preferred Neovim plugin manager.

hyphen.nvim uses plugin-manager-agnostic command-level lazy loading: startup only registers lightweight command callbacks, while the core Lua modules and language pattern tables are loaded on first use.

Usage

Show hyphenation hints for the word under the cursor:

:HyphenShow

Show hints using a specific language:

:HyphenShow en-gb

Clear active hints:

:HyphenClear

By default, hints are cleared on CursorMoved, InsertEnter, and TextChanged. You can change this through clear_event.

Configuration

Default configuration:

require("hyphen").setup({
  clear_event = {"CursorMoved", "InsertEnter", "TextChanged"},
  pattern_dir = {},
  lang = "en-us",
  leftmin = 2,
  rightmin = 3,
  separator = "-",
})

pattern_dir accepts either a single directory:

require("hyphen").setup({
  pattern_dir = "/path/to/dir",
})

or a list of directories:

require("hyphen").setup({
  pattern_dir = {
    "/path/to/dir1",
    "/path/to/dir2",
  },
})

Despite the singular name, pattern_dir may be a string or a list.

Pattern Data

Built-in languages:

• en-us
• en-gb

Custom pattern directories should contain files named like this:

hyph-<lang>.pat.txt
hyph-<lang>.hyp.txt

The .hyp.txt file is optional.

Pattern lookup order:

1. Built-in Lua module: hyphen.lang.<lang>
2. Directories listed in pattern_dir

pattern_dir is only consulted when no built-in Lua module exists for the requested language.

You can convert TeX Live-style text pattern files into Lua modules with:

:HyphenConvert <lang> <dir>

Example:

:HyphenConvert en-us /path/to/patterns

Converted pattern files may have licenses independent of hyphen.nvim. If you redistribute generated lang/<lang>.lua files, preserve the original pattern notices.

Commands

Command	Description
:HyphenShow [lang]	Show hyphenation hints for the word under the cursor.
:HyphenClear	Manually clear active hyphenation hints.
:HyphenConvert <lang> <dir>	Convert TeX Live-style pattern files into a Lua language module.

API

local hyphen = require("hyphen")

Function	Description
hyphen.setup(opts)	Configure the plugin.
hyphen.position(word, opts)	Return 1-indexed hyphenation positions.
hyphen.hyphenate(word, opts)	Return the word with separators inserted.
hyphen.show(opts)	Show virtual text hints for the word under the cursor.
hyphen.clear()	Clear active hints in the current buffer.

Examples:

require("hyphen").position("hyphenation")
-- returns { 2, 6 }

require("hyphen").hyphenate("hyphenation")
-- returns "hy-phen-ation"

Limitations

• Word detection is currently ASCII/English-centric.
• The plugin does not automatically insert text yet, including LaTeX discretionary hyphens (\-).
• The plugin does not automatically detect LaTeX language settings from babel or polyglossia yet.
• Built-in pattern data is currently limited to en-us and en-gb.

License

Distributed under the MulanPSL-2.0 license. See LICENSE for details.

Bundled hyphenation pattern data has independent copyright and license notices in lua/hyphen/lang/.

Why MulanPSL-2.0?

The Mulan Permissive Software License v2 (MulanPSL-2.0) may be less familiar than more widely used licenses. To provide clarity and context, the following table (cited from Choose a License) compares key aspects of MulanPSL v2 with popular licenses including Apache-2.0, BSD-3-Clause, and MIT.

License	Commercial Use	Distribution	Modification	Patent Use	Private Use	Disclose Source	License and Copyright Notice	Network Use is Distribution	Same License	State Changes	Liability	Trademark Use	Warranty
Apache-2.0	🟢	🟢	🟢	🟢	🟢		🔵			🔵	🔴	🔴	🔴
BSD-3-Clause	🟢	🟢	🟢		🟢		🔵				🔴		🔴
MIT	🟢	🟢	🟢		🟢		🔵				🔴		🔴
MulanPSL-2.0	🟢	🟢	🟢	🟢	🟢		🔵				🔴	🔴	🔴

The drafter of the MulanPSL-2.0 license addressed similar concerns in this comment:

Thank you for raising this issue. Please allow me to explain. (I'm the one responsible for drafting MulanPSL-2.0 and getting it approved by OSI.)

Actually at the beginning we just say in the license, english and chinese version have the same legal effect (because we carefully translated the two versions word by word, sentence by sentence). However, the OSI community suggested that IN CASE, in case there is a conflict between the two languages, we should indicate which language prevails.

However, I must say, there is a tiny chance (close to zero) that this circumstance will happen. On the one hand, many people (including technical experts and lawyers) did careful proofreading between english version and chinese version; on the other hand, MulanPSL-2.0 is such a loose license that really doesn't have constrains, what conflict will you expect? We worry about conflict because we worry about legal risk that may bring, but since the legal terms are so loose we hardly see a risk.