Layouts

There are several possible layouts that apertium linguistic data packages may have.

Apertium 3.0

Example using English—Afrikaans:

Filename	Comment
apertium-en-af.af.dix.xml	Afrikaans monolingual dictionary
apertium-en-af.en-af.dix.xml	English—Afrikaans bilingual dictionary
apertium-en-af.en.dix.xml	English monolingual dictionary
apertium-en-af.symbols.xml	List of grammatical symbols
apertium-en-af.af-en.t1x	Afrikaans—English first stage transfer file (transfer)
apertium-en-af.af-en.t2x	Afrikaans—English second stage transfer file (interchunk)
apertium-en-af.af-en.t3x	Afrikaans—English third stage transfer file (postchunk)
apertium-en-af.en-af.t1x	English—Afrikaans first stage transfer file (transfer)
apertium-en-af.en-af.t2x	English—Afrikaans first stage transfer file (interchunk)
apertium-en-af.en-af.t3x	English—Afrikaans first stage transfer file (postchunk)
modes.xml	Modes specification file (see modes)

Using separate include files

One of the main things that separates this method is using separate include files for various bits, most just use it to define symbols, but it could be used for example for a large list of proper nouns, different domains or almost anything.

In all of the dictionary files:

<?xml version="1.0" encoding="UTF-8"?>
<dictionary>
<alphabet>ÄÊËÖÜäêëöüßABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz</alphabet>

  <!-- Symbol definitions -->
  <xi:include xmlns:xi="http://www.w3.org/2001/XInclude" href="apertium-en-af.symbols.xml"/>

  <pardefs>
    
    ...
  
  </pardefs>
  <section id="main" type="inconditional">

    ...

  </section>
</dictionary>

In the apertium-en-af.symbols.xml file:

<?xml version="1.0" encoding="UTF-8"?> <!-- -*- nxml -*- -->

  <sdefs>
    <sdef n="n"       c="Noun"/>
    <sdef n="m"       c="Masculine"/>
    <sdef n="f"       c="Feminine"/>

    ...
  </sdefs>

Then in the Makefile.am, append the following to the TARGETS_COMMON:

TARGETS_COMMON = $(BASENAME).$(LANG1).dix $(BASENAME).$(LANG2).dix $(BASENAME).$(LANG1)-$(LANG2).dix \

And add the following targets underneth:

$(BASENAME).$(LANG1).dix:
        xmllint --xinclude $(BASENAME).$(LANG1).dix.xml > $(BASENAME).$(LANG1).dix
$(BASENAME).$(LANG2).dix:
        xmllint --xinclude $(BASENAME).$(LANG2).dix.xml > $(BASENAME).$(LANG2).dix
$(BASENAME).$(LANG1)-$(LANG2).dix:
        xmllint --xinclude $(BASENAME).$(LANG1)-$(LANG2).dix.xml > $(BASENAME).$(LANG1)-$(LANG2).dix

You may also want to specify a clean-dicts target:

clean-dicts:
        rm $(BASENAME).$(LANG1).dix
        rm $(BASENAME).$(LANG2).dix
        rm $(BASENAME).$(PREFIX1).dix

Dependent on monolingual packages (Apertium 3+)

apertium-kaz-tat depends on the packages apertium-kaz and apertium-tat.

These files are stored in the apertium-kaz repository:

Filename	Comment
apertium-kaz.kaz.lexc	Kazakh monolingual dictionary lexicon (in HFST format)
apertium-kaz.kaz.twol	Kazakh monolingual dictionary two-level rules (in HFST format)
apertium-kaz.kaz.rlx	Kazakh Constraint Grammar disambiguation rules

The Kazakh monolingual files are used by apertium-kaz-tat during compilation, instead of being stored in both repositories. Similarly for apertium-tat.

These files are in the apertium-kaz-tat repository:

Filename	Comment
apertium-kaz-tat.kaz-tat.dix	Kazakh-Tatar bilingual dictionary
apertium-kaz-tat.kaz-tat.lrx	Kazakh→Tatar Lexical selection rules
apertium-kaz-tat.tat-kaz.lrx	Tatar→Kazakh Lexical selection rules
apertium-kaz-tat.kaz-tat.t1x	Kazakh→Tatar first stage transfer file
apertium-kaz-tat.kaz-tat.t2x	Kazakh→Tatar second stage transfer file
apertium-kaz-tat.tat-kaz.t1x	Tatar→Kazakh first stage transfer file
apertium-kaz-tat.tat-kaz.t2x	Tatar→Kazakh second stage transfer file
modes.xml	Modes specification file (see modes)

See Languages#Requiring_a_monolingual_package_as_a_dependency_of_a_pair for the makefile details (or see http://sourceforge.net/p/apertium/svn/HEAD/tree/trunk/apertium-kaz-tat/ for the actual example).