Voikkospell
Contents
Installation
m5w/corevoikko, a fork of corevoikko, supports apertium stream format.
To clone it, execute the following command:
git clone https://github.com/m5w/corevoikko.git corevoikko
First, install libvoikko's dependencies. Next, execute the following commands:
cd corevoikko/libvoikko ./configure make sudo make install
If you do not have root privileges or would like to specify where to install libvoikko, execute the following instead: (Otherwise, you are finished with installation.)
cd corevoikko/libvoikko PREFIX="$HOME/install/corevoikko" # e.g. ./configure --prefix="$PREFIX" make make install
Finally, add your "$PREFIX" to your "$PATH" by appending the following lines to your .profile:
PREFIX="$HOME/install/corevoikko" # e.g.
if [ -d "$PREFIX" ]; then
export PATH="$PREFIX/bin:$PATH"
fi
Using voikkospell with apertium Stream Format
Invoke voikkospell with --apertium-stream. voikkospell then expects apertium-stream-formatted input instead of a list of words.
Words in apertium Stream Format
apertium stream format encodes words as lexical units. Each begins with a ^
^ . . .
and ends with a $.
^ . . . $
The word immediately follows the ^,
^word . . .$
and a / immediately follows the word. If the word is unknown, *word follows;
^word/*word$
otherwise, all the word's analyses follow, delimited by /'s.
^word/word<n><sg>/word<vblex><inf>/word<vblex><pres>$
Escaping
To use ^, $, /, <, and > as characters, one must escape them. Each escape sequence begins with a \,
\ . . .
and a character follows. voikkospell then interprets the character literally. Note that the character can be any wide character, including newlines.
To use \'s as characters, one must escape them.
Superblanks
One can also escape multiple characters not encoded in lexical units by encoding them as a superblank. Each superblank begins with a [
[ . . .
and ends with a ].
[ . . . ]
Each ^, $, /, <, and > between the [ and the ] is interpreted literally.
To use [ and ] as characters, one must escape them.
An HTML Example
Let's spellcheck the following webpage.
<!DOCTYPE html>
<html>
<head>
<title>An HTML Example</title>
</head>
<body>
<p>
This is an HTML example.
</p>
</body>
</html>
Running apertium-deshtml on it yields the following.
.[][<!DOCTYPE html>
<html>
<head>
<title>]An HTML Example.[][<\/title>
<\/head>
<body>
<p>
]This is an HTML example..[][
<\/p>
<\/body>
<\/html>
]
Note that all the <'s, >'s, and /'s are encoded as superblanks. In fact, everything except the title and body paragraph is escaped. However, those words are not yet encoded as lexical units. Running lt-proc on the output yields the following, suitable for voikkospell --apertium-stream.
^./.<sent>$[][<!DOCTYPE html>
<html>
<head>
<title>]^An/A<det><ind><sg>$ ^HTML/HTML<n><acr><sp>$ ^Example/Example<n><sg>$^./.<sent>$[][<\/title>
<\/head>
<body>
<p>
]^This/This<det><dem><sg>/This<prn><tn><mf><sg>$ ^is/be<vbser><pri><p3><sg>$ ^an/a<det><ind><sg>$ ^HTML/HTML<n><acr><sp>$ ^example/example<n><sg>$^./.<sent>$^./.<sent>$[][
<\/p>
<\/body>
<\/html>
