apertium-oc-ca/ca-oc-supervised.make

   1 TAGGER_SUPERVISED_ITERATIONS=0
   2 BASENAME=apertium-oc-ca
   3 LANG1=ca
   4 LANG2=oc
   5 TAGGER=$(LANG1)-tagger-data
   6 PREFIX=$(LANG1)-$(LANG2)
   7
   8 all: $(PREFIX).prob
   9
  10 $(PREFIX).prob: $(BASENAME).$(LANG1).tsx $(TAGGER)/$(LANG1).dic $(TAGGER)/$(LANG1).untagged $(TAGGER)/$(LANG1).tagged $(TAGGER)/$(LANG1).crp
  11         apertium-validate-tagger $(BASENAME).$(LANG1).tsx
  12         apertium-tagger -s $(TAGGER_SUPERVISED_ITERATIONS) \
  13                            $(TAGGER)/$(LANG1).dic \
  14                            $(TAGGER)/$(LANG1).crp \
  15                            $(BASENAME).$(LANG1).tsx \
  16                            $(PREFIX).prob \
  17                            $(TAGGER)/$(LANG1).tagged \
  18                            $(TAGGER)/$(LANG1).untagged;
  19
  20 $(TAGGER)/$(LANG1).dic: $(BASENAME).$(LANG1).dix $(PREFIX).automorf.bin
  21         @echo "Generating $@";
  22         @echo "This may take some time. Please, take a cup of coffee and come back later.";
  23         apertium-validate-dictionary $(BASENAME).$(LANG1).dix
  24         apertium-validate-tagger $(BASENAME).$(LANG1).tsx
  25         lt-expand $(BASENAME).$(LANG1).dix | grep -v "__REGEXP__" | grep -v ":<:" |\
  26         awk 'BEGIN{FS=":>:|:"}{print $$1 ".";}' | apertium-destxt >$(LANG1).dic.expanded
  27         @echo "." >>$(LANG1).dic.expanded
  28         @echo "?" >>$(LANG1).dic.expanded
  29         @echo ";" >>$(LANG1).dic.expanded
  30         @echo ":" >>$(LANG1).dic.expanded
  31         @echo "!" >>$(LANG1).dic.expanded
  32         @echo "42" >>$(LANG1).dic.expanded
  33         @echo "," >>$(LANG1).dic.expanded
  34         @echo "(" >>$(LANG1).dic.expanded
  35         @echo "\\[" >>$(LANG1).dic.expanded
  36         @echo ")" >>$(LANG1).dic.expanded
  37         @echo "\\]" >>$(LANG1).dic.expanded
  38         @echo "¿" >>$(LANG1).dic.expanded
  39         @echo "¡" >>$(LANG1).dic.expanded
  40         lt-proc -a $(PREFIX).automorf.bin <$(LANG1).dic.expanded | \
  41         apertium-filter-ambiguity $(BASENAME).$(LANG1).tsx > $@
  42         rm $(LANG1).dic.expanded;
  43
  44 $(TAGGER)/$(LANG1).crp: $(PREFIX).automorf.bin $(TAGGER)/$(LANG1).crp.txt
  45         apertium-destxt < $(TAGGER)/$(LANG1).crp.txt | lt-proc $(PREFIX).automorf.bin > $(TAGGER)/$(LANG1).crp
  46
  47 $(TAGGER)/$(LANG1).crp.txt:
  48         touch $(TAGGER)/$(LANG1).crp.txt
  49
  50
  51 $(TAGGER)/$(LANG1).tagged:
  52         @echo "Error: File '"$@"' is needed to perform a supervised tagger training" 1>&2
  53         @echo "This file should exist. It is the result of solving the ambiguity from the '"$(TAGGER1)/$(LANG1).tagged.txt"' file" 1>&2
  54         exit 1
  55
  56 $(TAGGER)/$(LANG1).untagged: $(TAGGER)/$(LANG1).tagged.txt $(PREFIX).automorf.bin
  57         cat $(TAGGER)/$(LANG1).tagged.txt | apertium-destxt | lt-proc $(PREFIX).automorf.bin  > $@;
  58
  59 clean:
  60         rm -f $(PREFIX).prob