10 # Spanish is unfortunately not yet working
11 SINGLE_DICTS="en de fr it"
16 #./run.sh --lang1=EN --dictOut=test --dictInfo=test --input0=data/inputs/wikiSplit/en/EN.data --input0Name=enwikitionary --input0Format=enwiktionary --input0LangPattern=English --input0LangCodePattern=en --input0EnIndex=1 --input0WiktionaryType=EnEnglish
19 # Note: using input1 seems to hang for ZH currently!
20 while read langcode langname ; do
21 lang=$(echo $langcode | tr '[a-z]' '[A-Z]')
22 test "$lang" = "CY" && lang=CI
23 test "$lang" = "CMN" && lang=cmn
24 test "$lang" = "GRC" && lang=grc
25 test "$lang" = "HAW" && lang=haw
26 test "$lang" = "YUE" && lang=yue
29 if test "$lang" = "DE" -o "$lang" = "FR" -o "$lang" = "IT" ; then
30 reverse_dicts="--input3=data/inputs/wikiSplit/$langcode/EN.data --input3Format=WholeSectionToHtmlParser --input3Name=${langcode}wikitionary --input3WiktionaryLang=$lang --input3TitleIndex=1 --input3WebUrlTemplate=http://${langcode}.wiktionary.org/wiki/%s"
34 test -e data/inputs/stoplists/${langcode}.txt && stoplist="--lang2Stoplist=data/inputs/stoplists/${langcode}.txt"
35 ./run.sh --lang1=EN --lang2=$lang --lang1Stoplist=data/inputs/stoplists/en.txt $stoplist --dictOut=data/outputs/EN-${lang}.quickdic --dictInfo="(EN)Wiktionary-based EN-$lang dictionary." --input0=data/inputs/wikiSplit/en/${lang}.data --input0Name=enwikitionary --input0Format=enwiktionary --input0LangPattern=${langname} --input0LangCodePattern=${langcode} --input0EnIndex=1 --input0WiktionaryType=EnForeign --input1=data/inputs/wikiSplit/en/EN.data --input1Name=enwikitionary --input1Format=enwiktionary --input1LangPattern=${langname} --input1LangCodePattern=${langcode} --input1EnIndex=1 --input1WiktionaryType=EnToTranslation --input2=data/inputs/wikiSplit/en/${lang}.data --input2Format=WholeSectionToHtmlParser --input2Name=enwikitionary --input2WiktionaryLang=EN --input2TitleIndex=2 --input2WebUrlTemplate=http://en.wiktionary.org/wiki/%s $reverse_dicts
36 rm -f data/outputs/EN-${lang}.quickdic.${VERSION}.zip
37 7z a -mx=9 data/outputs/EN-${lang}.quickdic.${VERSION}.zip ./data/outputs/EN-${lang}.quickdic
39 done < EN-foreign-dictlist.txt
42 # EnEnglish only makes the dictionary cluttered
43 #./run.sh --lang1=EN --lang1Stoplist=data/inputs/stoplists/en.txt --dictOut=data/outputs/EN.quickdic --dictInfo="Wiktionary-based EN dictionary." --input0=data/inputs/wikiSplit/en/EN.data --input0Name=enwikitionary --input0Format=enwiktionary --input0LangPattern=English --input0LangCodePattern=en --input0EnIndex=1 --input0WiktionaryType=EnEnglish --input2=data/inputs/wikiSplit/en/EN.data --input2Format=WholeSectionToHtmlParser --input2Name=enwikitionary --input2WiktionaryLang=EN --input2TitleIndex=1 --input2WebUrlTemplate=http://en.wiktionary.org/wiki/%s
44 #rm -f data/outputs/EN.quickdic.${VERSION}.zip
45 #7z a -mx=9 data/outputs/EN.quickdic.${VERSION}.zip ./data/outputs/EN.quickdic
47 for langcode in $SINGLE_DICTS ; do
48 lang=$(echo $langcode | tr '[a-z]' '[A-Z]')
49 ./run.sh --lang1=$lang --lang1Stoplist=data/inputs/stoplists/${langcode}.txt --dictOut=data/outputs/${lang}.quickdic --dictInfo="Wiktionary-based ${lang} dictionary." --input1=data/inputs/wikiSplit/${langcode}/${lang}.data --input1Format=WholeSectionToHtmlParser --input1Name=${langcode}wikitionary --input1WiktionaryLang=$lang --input1TitleIndex=1 --input1WebUrlTemplate=http://${langcode}.wiktionary.org/wiki/%s
50 rm -f data/outputs/${lang}.quickdic.${VERSION}.zip
51 7z a -mx=9 data/outputs/${lang}.quickdic.${VERSION}.zip ./data/outputs/${lang}.quickdic
56 while read langcode langname ; do
57 lang=$(echo $langcode | tr '[a-z]' '[A-Z]')
60 if test "$lang" = "FR" -o "$lang" = "IT" ; then
61 reverse_dicts="--input3=data/inputs/wikiSplit/$langcode/DE.data --input3Format=WholeSectionToHtmlParser --input3Name=${langcode}wikitionary --input3WiktionaryLang=$lang --input3TitleIndex=1 --input3WebUrlTemplate=http://${langcode}.wiktionary.org/wiki/%s"
65 test -e data/inputs/stoplists/${langcode}.txt && stoplist="--lang2Stoplist=data/inputs/stoplists/${langcode}.txt"
66 test -e data/inputs/wikiSplit/de/${lang}.data && input0="--input0=data/inputs/wikiSplit/de/${lang}.data --input0Name=dewikitionary --input0Format=enwiktionary --input0LangPattern=${langname} --input0LangCodePattern=${langcode} --input0EnIndex=1 --input0WiktionaryType=EnForeign"
67 test -e data/inputs/wikiSplit/de/${lang}.data && input2="--input2=data/inputs/wikiSplit/de/${lang}.data --input2Format=WholeSectionToHtmlParser --input2Name=dewikitionary --input2WiktionaryLang=DE --input2TitleIndex=2 --input2WebUrlTemplate=http://de.wiktionary.org/wiki/%s"
68 ./run.sh --lang1=DE --lang2=$lang --lang1Stoplist=data/inputs/stoplists/de.txt $stoplist --dictOut=data/outputs/DE-${lang}.quickdic --dictInfo="(DE)Wiktionary-based DE-$lang dictionary." $input0 --input1=data/inputs/wikiSplit/en/EN.data --input1Name=enwikitionary --input1Format=EnTranslationToTranslation --input1LangPattern1=de --input1LangPattern2=${langcode} $input2 $reverse_dicts
69 rm -f data/outputs/DE-${lang}.quickdic.${VERSION}.zip
70 7z a -mx=9 data/outputs/DE-${lang}.quickdic.${VERSION}.zip ./data/outputs/DE-${lang}.quickdic
72 done < DE-foreign-dictlist.txt
76 while read langcode langname ; do
77 lang=$(echo $langcode | tr '[a-z]' '[A-Z]')
80 if test "$lang" = "DE" -o "$lang" = "IT" ; then
81 reverse_dicts="--input3=data/inputs/wikiSplit/$langcode/FR.data --input3Format=WholeSectionToHtmlParser --input3Name=${langcode}wikitionary --input3WiktionaryLang=$lang --input3TitleIndex=1 --input3WebUrlTemplate=http://${langcode}.wiktionary.org/wiki/%s"
85 test -e data/inputs/stoplists/${langcode}.txt && stoplist="--lang2Stoplist=data/inputs/stoplists/${langcode}.txt"
86 ./run.sh --lang1=FR --lang2=$lang --lang1Stoplist=data/inputs/stoplists/fr.txt $stoplist --dictOut=data/outputs/FR-${lang}.quickdic --dictInfo="(FR)Wiktionary-based FR-$lang dictionary." --input0=data/inputs/wikiSplit/fr/${lang}.data --input0Name=frwikitionary --input0Format=enwiktionary --input0LangPattern=${langname} --input0LangCodePattern=${langcode} --input0EnIndex=1 --input0WiktionaryType=EnForeign --input1=data/inputs/wikiSplit/en/EN.data --input1Name=enwikitionary --input1Format=EnTranslationToTranslation --input1LangPattern1=fr --input1LangPattern2=${langcode} --input2=data/inputs/wikiSplit/fr/${lang}.data --input2Format=WholeSectionToHtmlParser --input2Name=frwikitionary --input2WiktionaryLang=FR --input2TitleIndex=2 --input2WebUrlTemplate=http://fr.wiktionary.org/wiki/%s $reverse_dicts
87 rm -f data/outputs/FR-${lang}.quickdic.${VERSION}.zip
88 7z a -mx=9 data/outputs/FR-${lang}.quickdic.${VERSION}.zip ./data/outputs/FR-${lang}.quickdic
90 done < FR-foreign-dictlist.txt
94 while read langcode langname ; do
95 lang=$(echo $langcode | tr '[a-z]' '[A-Z]')
98 if test "$lang" = "FR" -o "$lang" = "DE" ; then
99 reverse_dicts="--input3=data/inputs/wikiSplit/$langcode/IT.data --input3Format=WholeSectionToHtmlParser --input3Name=${langcode}wikitionary --input3WiktionaryLang=$lang --input3TitleIndex=1 --input3WebUrlTemplate=http://${langcode}.wiktionary.org/wiki/%s"
103 test -e data/inputs/stoplists/${langcode}.txt && stoplist="--lang2Stoplist=data/inputs/stoplists/${langcode}.txt"
104 ./run.sh --lang1=IT --lang2=$lang --lang1Stoplist=data/inputs/stoplists/it.txt $stoplist --dictOut=data/outputs/IT-${lang}.quickdic --dictInfo="(IT)Wiktionary-based IT-$lang dictionary." --input0=data/inputs/wikiSplit/it/${lang}.data --input0Name=itwikitionary --input0Format=enwiktionary --input0LangPattern=${langname} --input0LangCodePattern=${langcode} --input0EnIndex=1 --input0WiktionaryType=EnForeign --input1=data/inputs/wikiSplit/en/EN.data --input1Name=enwikitionary --input1Format=EnTranslationToTranslation --input1LangPattern1=it --input1LangPattern2=${langcode} --input2=data/inputs/wikiSplit/it/${lang}.data --input2Format=WholeSectionToHtmlParser --input2Name=itwikitionary --input2WiktionaryLang=IT --input2TitleIndex=2 --input2WebUrlTemplate=http://it.wiktionary.org/wiki/%s $reverse_dicts
105 rm -f data/outputs/IT-${lang}.quickdic.${VERSION}.zip
106 7z a -mx=9 data/outputs/IT-${lang}.quickdic.${VERSION}.zip ./data/outputs/IT-${lang}.quickdic
108 done < IT-foreign-dictlist.txt