summaryrefslogtreecommitdiffstats
path: root/JLanguageTool/src/resource/en/make-dict-en.sh
diff options
context:
space:
mode:
authorArno Teigseth <arno@teigseth.no>2011-02-05 08:48:27 +0000
committerArno Teigseth <arno@teigseth.no>2011-02-05 08:48:27 +0000
commit4f3d565a5e5ede6eb6fd1f276d4e8ad37b67b5ce (patch)
tree7af736540eca93034428a975bd850e709fbbe2e5 /JLanguageTool/src/resource/en/make-dict-en.sh
parentecaee85ab5984ebadd56721c295dc26b3335f7ce (diff)
downloadgrammar-norwegian-master.tar.gz
grammar-norwegian-master.tar.bz2
grammar-norwegian-master.tar.xz
added more files, to complete languagetool uploadHEADmaster
Diffstat (limited to 'JLanguageTool/src/resource/en/make-dict-en.sh')
-rw-r--r--JLanguageTool/src/resource/en/make-dict-en.sh25
1 files changed, 25 insertions, 0 deletions
diff --git a/JLanguageTool/src/resource/en/make-dict-en.sh b/JLanguageTool/src/resource/en/make-dict-en.sh
new file mode 100644
index 0000000..c230153
--- /dev/null
+++ b/JLanguageTool/src/resource/en/make-dict-en.sh
@@ -0,0 +1,25 @@
+#!/bin/sh
+
+LANG=POSIX
+TMP_OUTPUT=eng1.txt
+TMP_GET_UNC=english.txt
+TMP_FINAL=eng2.txt
+OUTPUT=english.dict
+OUTPUT_SYNTH=english_synth.dict
+rm $TMP_OUTPUT
+rm $TMP_GET_UNC
+rm $TMP_FINAL
+gawk -f filter_out.awk infl.txt part-of-speech.txt| gawk -f remap.awk >penn.txt
+cat penn.txt manually_added.txt | sort -u >$TMP_GET_UNC
+cp $TMP_GET_UNC $TMP_FINAL
+gawk -f get_unc.awk $TMP_FINAL |sort -u > $TMP_OUTPUT
+#test
+gawk -f test_dict.awk $TMP_OUTPUT
+#create normal dictionary
+cp $TMP_OUTPUT
+gawk -f morph_data.awk $TMP_OUTPUT | fsa_ubuild -O -o $OUTPUT
+echo "Output written to $OUTPUT"
+#create synthesis dictionary
+gawk -f synteza.awk $TMP_OUTPUT | gawk -f morph_data.awk | sort -u | fsa_ubuild -O -o $OUTPUT_SYNTH
+gawk -f tags.awk $TMP_OUTPUT |sort -u >english_tags.txt
+echo "Output written to $OUTPUT_SYNTH" \ No newline at end of file