From a9fb7e0a7b3f2178ef1ff70ffddd243a83d806bc Mon Sep 17 00:00:00 2001 From: =?utf8?q?Reimar=20D=C3=B6ffinger?= Date: Mon, 11 Jan 2016 20:42:39 +0100 Subject: [PATCH] Add support to generate pure translation-to-translation dictionaries. --- EN-trans-dictlist.txt | 1 + generate_dictionaries.sh | 16 ++++++++++++++++ 2 files changed, 17 insertions(+) create mode 100644 EN-trans-dictlist.txt diff --git a/EN-trans-dictlist.txt b/EN-trans-dictlist.txt new file mode 100644 index 0000000..449d5b9 --- /dev/null +++ b/EN-trans-dictlist.txt @@ -0,0 +1 @@ +es Spanish pt Portuguese diff --git a/generate_dictionaries.sh b/generate_dictionaries.sh index 04ea782..4153654 100755 --- a/generate_dictionaries.sh +++ b/generate_dictionaries.sh @@ -7,6 +7,8 @@ FR_DICTS=true #FR_DICTS=false IT_DICTS=true #IT_DICTS=false +EN_TRANS_DICTS=true +#EN_TRANS_DICTS=false # Spanish is unfortunately not yet working SINGLE_DICTS="en de fr it" #SINGLE_DICTS="" @@ -107,3 +109,17 @@ rm -f data/outputs/IT-${lang}.quickdic.${VERSION}.zip done < IT-foreign-dictlist.txt fi + +if $EN_TRANS_DICTS; then +while read langcode1 langname1 langcode2 langname2 ; do +lang1=$(echo $langcode1 | tr '[a-z]' '[A-Z]') +lang2=$(echo $langcode2 | tr '[a-z]' '[A-Z]') +stoplist1="" +stoplist2="" +test -e data/inputs/stoplists/${langcode1}.txt && stoplist1="--lang1Stoplist=data/inputs/stoplists/${langcode1}.txt" +test -e data/inputs/stoplists/${langcode2}.txt && stoplist1="--lang2Stoplist=data/inputs/stoplists/${langcode2}.txt" +./run.sh --lang1=$lang1 --lang2=$lang2 $stoplist1 $stoplist2 --dictOut=data/outputs/${lang1}-${lang2}.quickdic --dictInfo="(EN)Wiktionary-based ${lang1}-${lang2} dictionary." --input1=data/inputs/wikiSplit/en/EN.data --input1Name=enwikitionary --input1Format=EnTranslationToTranslation --input1LangPattern1=${langcode1} --input1LangPattern2=${langcode2} +rm -f data/outputs/${lang1}-${lang2}.quickdic.${VERSION}.zip +7z a -mx=9 data/outputs/${lang1}-${lang2}.quickdic.${VERSION}.zip ./data/outputs/${lang1}-${lang2}.quickdic +done < EN-trans-dictlist.txt +fi -- 2.43.0