---
name: "swath"
section: "1"
platform: "linux"
package: "swath"
version: "0.6.1-4"
release: "Debian 13 (Trixie)"
platforms: ["linux"]
url: "https://new.carta.tech/swath"
---

# swath(1)

General-purpose Thai word segmentation utility

## NAME

swath - General-purpose Thai word segmentation utility

## SYNOPSIS

**swath** \[options\] \< *infile* \> *outfile*\

## DESCRIPTION

Thai script has no word delimitor.  Applications need to recognize word boundaries before they can do useful things with Thai text, such as line wrapping.

Swath provides word analysis filter to insert word delimitors into a given text stream.  It reads text from standard input, analyzes it for word boundaries by consulting a Thai word list, and outputs to standard output the same text with the predefined word delimitors inserted.

Currently, it can read plain text, HTML, RTF, LaTeX and Lambda (Unicode version of LaTeX with Omega typesetter kernel) documents and insert common word delimitors for each format (pipe \`\|' for plain text). But user can always override this with a preferred delimitor.

## OPTIONS

- **-b \[*delimitor*\]**

  Define the string to be inserted as word delimitor in the output text.

- **-d \[*dict-path*\]**

  Specify alternative dictionary location.  *dict-path* must be either a directory containing the swath dictionary file \`swathdic.tri', or a path to the dictionary file itself.  The dictionary file must be a trie file prepared using [**trietool**(1)](/trietool/1.md) utility from the libdatrie package.

  If this option is given, **swath** will override normal dictionary search and will exit if the given dictionary cannot be found.  Otherwise, if **SWATHDICT** environment is set, it will try to open dictionary from the location specified by its value.  Otherwise, it will try the current working directory, and finally the usual installed location.

- **-f \[*format*\]**

  Specify format of the input.  Possible formats are: html, rtf, latex, lambda.

- **-m \[*scheme*\]**

  Choose word matching scheme when analyzing word boundaries.  Possible schemes are \`long' (for longest or greedy matching) and \`max' (for maximal matching, with least words preferred).  Maximal matching is the default value.

- **-u *input-enc*,*output-enc***

  Specify encodings of the input and the output.  *input-enc* and *output-enc* can be one of 'u' (for UTF-8 encoding) and 't' (for TIS-620 encoding).  Swath will convert the character encoding as necessary. If this option is omitted, TIS-620 encodings on both input and output are assumed.

- **-v, --verbose**

  Turn on verbose mode.

- **-help, --help**

  Show help.

## ENVIRONMENT VARIABLES

- SWATHDICT

  If specified, **swath** will search for dictionary from this location before the usual places (current working directory and usual installed directory, respectively).  This value is overridden by **-d** option.

## EXAMPLES

For LaTeX (to be used with babel-thai package):

$ **swath** -f latex \< thaifile.tex \> thaifile.ttex\
 $ **latex** thaifile.ttex

For HTML (to provide web pages to web browsers that cannot wrap Thai lines properly, but support the \<wbr\> tag):

$ **swath** -f html \< myweb.html \> myweb-wbr.html

To preprocess a Thai UTF-8 encoded LaTeX file for babel-thai with tis620 inputenc:

$ **swath** -f latex -u u,t \< thaifile.tex \> thaifile.ttex\
 $ **latex** thaifile.ttex

This is equivalent to filtering with **[iconv(1)](/iconv/1.md)**:

$ **iconv** -f UTF-8 -t TIS-620 thaifile.tex \| **swath** -f latex \>  thaifile.ttex\
 $ **latex** thaifile.ttex

To use longest matching scheme with LaTeX document:

$ **swath** -f latex -m long \< thaifile.tex \> thaifile.ttex\
 $ **latex** thaifile.ttex

To use an alternative dictionary from libthai:

$ **swath** -f latex -d /usr/share/libthai/thbrk.tri \< thaifile.tex \> thaifile.ttex

## AUTHOR

This manual page was written by Theppitak Karoonboonyanan \<theppitak@gmail.com\>.