.TH UTFPATGEN 1 "30 May 2026" "utfpatgen 1.0"
.\"=====================================================================
.if t .ds TX \fRT\\h'-0.10m'\\v'0.17v'E\\v'-0.17v'\\h'-0.06m'X\fP
.if n .ds TX TeX
.ie t .ds OX \fIT\h'-0.17m'\v'+0.21m'E\v'-0.21m'\h'-0.04m'X\fP
.el .ds OX TeX
.\"=====================================================================
.SH NAME
utfpatgen \- generate patterns for TeX hyphenation
.SH SYNOPSIS
.B utfpatgen
.I dictionary_file pattern_file patout_file translate_file
.\"=====================================================================
.SH DESCRIPTION
.I UTFpatgen
is an extension to
.BR patgen (1)
for generating patterns from large input alphabets, with an extended
hyphenation level range and native dynamic memory management.
.PP
The program reads a
.I dictionary_file
containing a list of hyphenated words and a
.I pattern_file
containing previously-generated patterns (if any) for a particular
language (not a complete \*(TX source file; see below), and produces the
.I patout_file
with (previously- plus newly-generated) hyphenation patterns for that
language.
.PP
The
.I translate_file
defines language specific values for the parameters
.IR left_hyphen_min " and " right_hyphen_min
used by \*(TX's hyphenation algorithm and the external representation
of the lower and upper case version(s) of all `letters' of that
language.
.PP
Further details of the pattern generation process such as
hyphenation levels and pattern lengths are requested interactively from
the user's terminal. Optionally,
.I UTFpatgen
creates a new dictionary file
.BI pattmp. n
showing the good and bad hyphens found by the generated patterns, where
.I n
is the highest hyphenation level.
.PP
All filenames must be complete; no adding of default
extensions or path searching is done.
.\"=====================================================================
.SH INPUT FORMATS
.TP \w'@@'u+2n
.B Letters
.I UTFpatgen
is able to process any UTF-8 encoded character, or more generally, any
encoding that is prefix-free (no letter is a prefix of another) and does
not use the `0xFF' byte, which has a special meaning in 
.IR UTFpatgen ),
described next:
.TP \w'@@'u+2n
.B Levels and weights
Non-character parts of the text, such as hyphenation levels or weights,
should be represented as a 2-byte sequence `0xFF <value>'. If a file
uses the
.BR patgen (1)
encoding (ASCII numerals), we recommend using
.BR sed (1)
for conversion.
.TP \w'@@'u+2n
.B File formats
The formats and conventions required in the 4 input files (
.I dictionary_file, pattern_file, patout_file, translate_file
) are identical to those in
.BR patgen (1)
with the only exception of level and weight encoding described earlier.
.\"=====================================================================
.SH "SEE ALSO"
Frank Liang,
.IR "Word hy-phen-a-tion by com-puter" ,
STAN-CS-83-977,
Stanford University Ph.D. thesis, 1983,
http://tug.org/docs/liang.
.PP
Donald E. Knuth,
.IR "The \*(OXbook" ,
Addison-Wesley, Appendix H.
.TP
https://ctan.org/pkg/patgen
The original patgen program, by Frank Liang, with system updates by
Peter Breitenlohner.
.TP
https://ctan.org/pkg/hyph-utf8
Collected hyphenation patterns for many languages in many formats.
.TP
https://ctan.org/tex-archive/language/
General CTAN directory for patterns and support for many other languages.
.TP
https://tug.org/TUGboat/Contents/listkeyword.html#CatTAGMultilingualDocumentProcessing
\fITUGboat\fP articles on hyphenation and other aspects of
language-specific document processing.
.\"=====================================================================
.SH AUTHORS
Ondřej Metelka
.br
Released under the MIT license.
.br
https://ctan.org/pkg/utfpatgen