remove some city name

diff --git a/dict-dists-extractor/Makefile b/dict-dists-extractor/Makefile
index ede51bf3728cd971147620507355c915313db4b1..6f2bb584c4a3ea98e5d8d7ddf2a96d545ef500e7 100644 (file)
--- a/dict-dists-extractor/Makefile
+++ b/dict-dists-extractor/Makefile
@@ -2,11 +2,17 @@
 
 ROOT_DICT = ../public/dict
 
+LANGS = bg ca da de en eo fr it nl no
+
 all:
+       make $(addprefix check-,$(LANGS))
 
 $(ROOT_DICT):
        mkdir $(ROOT_DICT)
 
+count:
+       wc -l $(ROOT_DICT)/*
+
 #sed -e 's/Ѝ/И/g' 
 $(ROOT_DICT)/bg.json: /usr/share/dict/bulgarian Makefile
        iconv -f Windows-1251 -t utf-8 $< | \
@@ -23,6 +29,7 @@ $(ROOT_DICT)/ca.json: /usr/share/dict/catalan Makefile
        cat $< | \
         grep -v '^.$$' | \
         grep -v "['-]" | \
+        grep -v "^[A-Z]" | \
         sed -e 's/[áàã]/a/g' \
             -e 's/[éè]/e/g' \
             -e 's/[ç]/c/g' \
@@ -41,16 +48,30 @@ $(ROOT_DICT)/da.json: /usr/share/dict/danish Makefile
        cat $< | \
         grep -v '^.$$' | \
         grep -v "['-]" | \
-        sed -e 's/æ/Æ/g' \
-                   -e 's/ø/Ø/g' \
-                       -e 's/å/Å/g' \
-            -e 's/[áä]/a/g' \
-                       -e 's/ð/d/g' \
+        grep -v "^[A-Z]" | \
+        sed -e 's/[áä]/a/g' \
+            -e 's/[ð]/d/g' \
             -e 's/[éëè]/e/g' \
             -e 's/[íì]/i/g' \
             -e 's/[óöô]/o/g' \
             -e 's/[ü]/u/g' | \
-        tr '[a-z]' '[A-Z]' | \
+        tr '[a-zæøå]' '[A-ZÆØÅ]' | \
+        sort | \
+        uniq | \
+        awk 'BEGIN { print "[" } \
+            { print "    \"" $$0 "\"," } \
+            END { print "    \"\"\n]" }' > $@
+
+$(ROOT_DICT)/de.json: /usr/share/dict/ngerman Makefile
+       cat $< | \
+        grep -v '^.$$' | \
+        grep -v "[.'-]" | \
+        sed -e 's/[âà]/a/g' \
+            -e 's/[éê]/e/g' \
+            -e 's/[ñ]/n/g' \
+            -e 's/[œ]/oe/g' \
+            -e 's/[ß]/ss/g' | \
+        tr '[a-zäöü]' '[A-ZÄÖÜ]' | \
         sort | \
         uniq | \
         awk 'BEGIN { print "[" } \
@@ -61,6 +82,7 @@ $(ROOT_DICT)/en.json: /usr/share/dict/british-english-huge Makefile
        cat $< | \
         grep -v '^.$$' | \
         grep -v "[.'-]" | \
+        grep -v "^[A-Z]" | \
         sed -e 's/[áâàäå]/a/g' \
             -e 's/[ç]/c/g' \
             -e 's/[éêèë]/e/g' \
@@ -75,15 +97,27 @@ $(ROOT_DICT)/en.json: /usr/share/dict/british-english-huge Makefile
             { print "    \"" $$0 "\"," } \
             END { print "    \"\"\n]" }' > $@
 
-$(ROOT_DICT)/fr.json: /usr/share/dict/french Makefile
+$(ROOT_DICT)/eo.json: /usr/share/dict/esperanto Makefile
        cat $< | \
         grep -v '^.$$' | \
         grep -v "[.'-]" | \
+        tr '[a-zĉĝĥĵŝŭ]' '[A-ZĈĜĤĴŜŬ]' | \
+        sort | \
+        uniq | \
+        awk 'BEGIN { print "[" } \
+            { print "    \"" $$0 "\"," } \
+            END { print "    \"\"\n]" }' > $@
+
+$(ROOT_DICT)/fr.json: /usr/share/dict/french Makefile
+       cat $< | \
+        grep -v '^.$$' | \
+        grep -v "[.'ú-]" | \
+        grep -v "^[A-Z]" | \
         sed -e 's/[àâ]/a/g' \
             -e 's/[ç]/c/g' \
             -e 's/[éèêë]/e/g' \
             -e 's/[ïî]/i/g' \
-            -e 's/[ô]/o/g' \
+            -e 's/[ôö]/o/g' \
             -e 's/[ùûü]/u/g' | \
         tr '[a-z]' '[A-Z]' | \
         sort | \
@@ -92,6 +126,23 @@ $(ROOT_DICT)/fr.json: /usr/share/dict/french Makefile
             { print "    \"" $$0 "\"," } \
             END { print "    \"\"\n]" }' > $@
 
+$(ROOT_DICT)/it.json: /usr/share/dict/italian Makefile
+       cat $< | \
+        grep -v '^.$$' | \
+        grep -v "[.'-]" | \
+        grep -v "^[A-Z]" | \
+        sed -e 's/[à]/a/g' \
+            -e 's/[éè]/e/g' \
+            -e 's/[ì]/i/g' \
+            -e 's/[ò]/o/g' \
+            -e 's/[ù]/u/g' | \
+        tr '[a-z]' '[A-Z]' | \
+        sort | \
+        uniq | \
+        awk 'BEGIN { print "[" } \
+            { print "    \"" $$0 "\"," } \
+            END { print "    \"\"\n]" }' > $@
+
 $(ROOT_DICT)/nl.json: /usr/share/dict/dutch Makefile
        cat $< | \
         grep -v '^.$$' | \
@@ -110,16 +161,31 @@ $(ROOT_DICT)/nl.json: /usr/share/dict/dutch Makefile
             { print "    \"" $$0 "\"," } \
             END { print "    \"\"\n]" }' > $@
 
+$(ROOT_DICT)/no.json: /usr/share/dict/bokmaal Makefile
+       cat $< | \
+        grep -v '^.$$' | \
+        grep -v "['-]" | \
+        grep -v "^[A-Z]" | \
+        sed -e 's/[áä]/a/g' \
+            -e 's/[ð]/d/g' \
+            -e 's/[éëè]/e/g' \
+            -e 's/[íì]/i/g' \
+            -e 's/[óöô]/o/g' \
+            -e 's/[ü]/u/g' | \
+        tr '[a-zæøå]' '[A-ZÆØÅ]' | \
+        sort | \
+        uniq | \
+        awk 'BEGIN { print "[" } \
+            { print "    \"" $$0 "\"," } \
+            END { print "    \"\"\n]" }' > $@
+
 CHECKER = grep -v '^    "[A-Z]*",$$' | grep -v '\[' | grep -v '\]' | grep -v '""'
 
 check-bg:$(ROOT_DICT)/bg.json
         cat $< | grep -v '"[АОЕИТНПРСВМБДКЛГЪЖЗУЧЯЙХЦШЮФЩЬ]*"' | grep -v '\[' | grep -v '\]' | grep -v '""' || echo "$< is correct"
 
 check-ca:$(ROOT_DICT)/ca.json
-        cat $< | $(CHECKER) | grep -v '"[A-Z]*L·L[A-Z]*"' || echo "$< is correct"
-
-check-da:$(ROOT_DICT)/da.json
-        cat $< | $(CHECKER) | grep -v '"[A-ZÆØÅ]*"' || echo "$< is correct"
+       cat $< | $(CHECKER) | grep -v '"[A-Z]*L·L[A-Z]*"' || echo "$< is correct"
 
 check-%:$(ROOT_DICT)/%.json
        cat $< | $(CHECKER) || echo "$< is correct"