ispell-autobuildhash

ispell-autobuildhash [-d] [-i enc] [-o enc] [-v] [-3] [-S] [-s] [-n] [-H name] [-I name] dictname wordlist [affixfile]

-d
    Debug mode, prints detailed processing info.

-i enc
    Input encoding (default: ISO8859-1).

-o enc
    Output encoding (default: ISO8859-1).

-v
    Verbose output during build.

-3
    Build only 3-letter preliminary hash file.

-S
    Skip .sfx (suffix) file creation.

-s
    Skip squeeze file for run-together words.

-n
    Disable word sorting.

-H name
    Custom hash filename (default: dictname.hash).

-I name
    Custom info filename (default: dictname.i.*).

dictname
    Base name for output dictionary files.

wordlist
    Path to input plain-text word list.

affixfile (optional)
    Path to affix rules file.

ispell-autobuildhash is a Perl script that simplifies building hash files for the ispell spell-checker from plain text word lists. It automates the traditionally manual process using buildhash, handling word sorting, contractions, affix rules, n-grams, keep lists, and encoding conversions.

Primarily used by package maintainers to create language-specific dictionaries (e.g., for English, French), it generates essential files like dictname.hash (primary hash), dictname.aff (affixes), dictname.i.* (info), optional .sfx (suffixes), and squeeze files for run-together words.

The tool detects input format, supports ISO8859-1 and UTF-8, and optimizes for ispell's indexing. It's invoked with a dictionary name, wordlist path, and optional affix file, producing files in the current directory. Ideal for Unix-like systems, it ensures compatibility with ispell and older spellers, though large dictionaries require ample RAM and time.

Requires Perl 5; large wordlists need significant CPU/RAM/disk. Output specific to ispell (not hunspell). Run as non-root; test with ispell -vv -d dictname.

dictname.hash (main), dictname.aff (affixes if provided), dictname.i.[lang] (info), optional .sfx/.squeeze/.k.* (keep/contraction).

ispell-autobuildhash -v en_US /usr/share/dict/words en_US.aff
Builds en_US.hash etc. from words file with affixes.

Introduced in ispell 3.1.x (late 1990s) to streamline dictionary packaging. Maintained in GNU ispell and distros like Debian/Ubuntu; evolved for better Unicode support in later versions.

ispell(1), buildhash(1), makedict(1), hunspell(1)