Difference between revisions of "Hfst"

From Apertium
Jump to navigation Jump to search
Line 103: Line 103:
 
make CXXFLAGS=-fpermissive
 
make CXXFLAGS=-fpermissive
 
</pre>
 
</pre>
  +
  +
  +
If, when compiling a dictionary, you end up in a "foma" prompt where you can type stuff, you should remove anything related to foma or "hfst-xfst" from your system, and build HFST anew as described above.
   
   
 
For more advices on installation problems, have a look at [https://kitwiki.csc.fi/twiki/bin/view/KitWiki/HfstReadme the Hfst Readme page].
 
For more advices on installation problems, have a look at [https://kitwiki.csc.fi/twiki/bin/view/KitWiki/HfstReadme the Hfst Readme page].
   
See also [[Foma]], [[OpenFST]] and [[SFST]] for compilation problems regarding the back-ends.
+
See also [[Foma]], [[OpenFST]] and [[SFST]] for problems regarding the back-ends.
   
 
==Using==
 
==Using==

Revision as of 14:06, 21 June 2014

hfst is the Helsinki finite-state toolkit. This is formalism-compatible with both lexc and twolc, so, kind of like foma is to xfst. It is currently being used in apertium-sme-nob, apertium-fin-sme, apertium-kaz-tat and in few other pairs which involve Turkic languages.

The IRC channel is #hfst at irc.freenode.net (you may try irc://irc.freenode.net/#hfst if your browser supports it, or enter #hfst into http://webchat.freenode.net/ if you want a web client). The HFST Wiki has some very good documentation (see especially the page HfstReadme when you run into compilation problems).

HFST is actually created as a set of wrappers over several possible back-ends, Foma, OpenFST, SFST, …. The latest versions of HFST include the back-ends you need, so there's no reason to install any of these backends separately.

Building and installing HFST

Install prerequisites

You will need the regular build dependencies:

  • automake, autoconf, libtool, flex, bison, g++, libreadline-dev

If you've already installed apertium/lttoolbox these should be installed already; if not, they should be easily installable with your package manager, e.g.

  • Ubuntu: sudo apt-get install automake autoconf libtool flex bison g++ libreadline-dev
  • Arch Linux: sudo pacman -S base-devel

MacOS X users might need to install XCode (registration required).

Download HFST

Either use the latest release (recommended for users), or go with the bleeding-edge SVN version (recommended for developers).

From SVN

$ svn co svn://svn.code.sf.net/p/hfst/code/trunk/hfst3 hfst3
$ cd hfst3/
$ ./autogen.sh

(The autogen step is only needed when using SVN, not with the tarball.)

Released tarball

Download the latest release, named something like hfst-X.Y.Z.tar.gz, from http://sourceforge.net/projects/hfst/files/, then

$ tar -xzf hfst-X.Y.Z.tgz
$ cd hfst-X.Y.Z/

(replacing X.Y.Z for the version you downloaded)

Configure

In the configure step, you can turn on/off features and backends and such. The OpenFST backend is included in the HFST distribution, while foma and SFST are not and are not recommended since they typically lead to more trouble than it's worth.

For most users, this should work:

$ ./configure --enable-proc --without-foma --enable-lexc --enable-all-tools

The above command will configure it to be installed to /usr/local in the make install step (below).

If you want hfst and back-ends installed somewhere else, you can do

$ ./configure --enable-proc --without-foma --enable-lexc --enable-all-tools  --prefix=/home/USERNAME/local/

Note: When we say USERNAME we mean your username, you need to replace it with your username, if you don't know what it is, you can find out by typing whoami


You can also add --with-unicode-handler=glib (or --with-unicode-handler=ICU) to the ./configure step if you have glib (or ICU) installed and want better Unicode Case_folding.

Compile and install

If your autotools version is older than 1.14 (check with automake --version), first do:

$ scripts/scripts/generate-cc-files.sh

Build by running

$ make


Then you need to install (Note: you need to use sudo make install if you installed it in /usr/local (or did not give a --prefix in the configure step); otherwise, no sudo!)

$ make install

And finally, unless you have a Mac, you may need to do:

$ sudo ldconfig

Troubleshooting

When doing "make" with old autotools (pre 1.14?)

make[5]: *** No rule to make target `xre_parse.hh', needed by `xre_lex.ll'.  Stop.

Run scripts/generate-cc-files.sh and then make again.


If, during the ./configure step, you see

checking for GNU libc compatible malloc... no
[…]
checking for GNU libc compatible realloc... no

and then during make a bunch of errors like:

/usr/local/include/sfst/mem.h:37:57: error: 'malloc' was not declared in this scope

, try the following:

sudo ldconfig
export LD_LIBRARY_PATH=/usr/local/lib
export PKG_CONFIG_PATH=/usr/local/lib/pkgconfig

and then ./configure and make.


If, during make, you see errors like

xre_parse.cc:2293:24: error: invalid conversion from 'const char*' to 'char*' [-fpermissive]

try instead

make CXXFLAGS=-fpermissive


If, when compiling a dictionary, you end up in a "foma" prompt where you can type stuff, you should remove anything related to foma or "hfst-xfst" from your system, and build HFST anew as described above.


For more advices on installation problems, have a look at the Hfst Readme page.

See also Foma, OpenFST and SFST for problems regarding the back-ends.

Using

$ svn co https://victorio.uit.no/langtech/trunk/st/fao
$ cd fao/src
$ make -f Makefile.hfst

$ echo "orð" | hfst-lookup ../bin/fao-morph.hfst
lookup> 
orð	orð+N+Neu+Sg+Nom+Indef
orð	orð+N+Neu+Sg+Acc+Indef
orð	orð+N+Neu+Pl+Nom+Indef
orð	orð+N+Neu+Pl+Acc+Indef

lookup>
$

To compile lexc code, first concatenate all the lexc files:

$ cat fao-lex.txt noun-fao-lex.txt noun-fao-morph.txt adj-fao-lex.txt \
adj-fao-morph.txt verb-fao-lex.txt verb-fao-morph.txt adv-fao-lex.txt \
abbr-fao-lex.txt acro-fao-lex.txt pron-fao-lex.txt punct-fao-lex.txt \
numeral-fao-lex.txt pp-fao-lex.txt cc-fao-lex.txt cs-fao-lex.txt \
interj-fao-lex.txt det-fao-lex.txt > ../tmp/lexc-all.txt

To compile this, just use the hfst-lexc program,

hfst-lexc < ../tmp/lexc-all.txt > ../bin/lexc-fao.bin

To compile the twol rules, just use the hfst-twolc program,

$ hfst-twolc twol-fao.txt > twol-fao.bin

And then to compose the lexicon and rule file, use hfst-compose-intersect:

$ hfst-compose-intersect -l lexc-fao.bin twol-fao.bin -o fao-gen.hfst

This will create a generator, if you want an analyser, you just need to invert the generator with hfst-invert:

$ hfst-invert fao-gen.hfst -o fao-morph.hfst

HFST2 vs HFST3

There have been some changes. Notably:

  • In twol files, a / in alphabetic symbols has to be escaped, e.g. %+Der%/st instead of %+Der/st.
  • In twol files, you can no longer have Sets on the left-hand side of a rule, so write Vx:Vy /<= _ ; where Vx in Set1 Vy in Set2 ; where you before would have Set1:Set2 /<= _ ;
  • The old -r option to hfst-twolc is now uppercase: -R
  • hfst-lookup-optimize is gone, use instead hfst-fst2fst -O -i infile.hfst -o outfile.hfst.ol
  • hfst-lexc needs the outfile option to be before the lexc (input), e.g. hfst-lexc -o outfile.hfst mylexicon.lexc
  • hfst-compose-intersect uses -1 (number one) instead of -l (letter L), and -2 for the rule-file. E.g. hfst-compose-intersect -1 lexicon.hfst -2 rules.twol.hfst -o generator.hfst

See also

External links