update stijgers2json.py

collect.sh

@@ -143,6 +143,14 @@ do
             | sed -e 's/\.[0-9][0-9]*$//' | sort | uniq \
             | items2count > $part-overige-namen-$ds-$i
 
+        say $part-allewoorden-$ds-$i
+        alto \
+            'fp://node[(@pt and not(@pt="let" or @rel="mwp" or @neclass)) or (@cat="mwu" and not(.//node[@neclass]))]' \
+            'tt:%l\ti%d\t%I' $files \
+            | sed -e 's/pubdate: "[-0-9]*"//' \
+            | sed -e 's/\.[0-9][0-9]*$//' | sort | uniq \
+            | items2count > $part-allewoorden-$ds-$i
+
         # tellingen met tags en postags
 
         say $part-nieuwe-woorden-extra-$ds-$i
@@ -166,18 +174,6 @@ do
             | sed -e 's/\([0-9]\)  */\1\t/' | sort -f -k 2 | sort -n -r -k 1,1 -s \
             > $part-nieuwe-adjww-extra-$ds-$i
 
-        # kale tellingen
-
-        say $part-allewoorden-$ds-$i
-        alto \
-            'fp://node[(@pt and not(@pt="let" or @rel="mwp" or @neclass)) or (@cat="mwu" and not(.//node[@neclass]))]' \
-            'tt:%l\t%I' $files \
-            | sed -e 's/\.[0-9][0-9]*$//' | sort | uniq \
-            | sed -e 's/\t.*//' | uniq -c \
-            | grep -v '^ *1 ' \
-            | sed -e 's/\([0-9]\)  */\1\t/' | sort -f -k 2 | sort -n -r -k 1,1 -s \
-            > $part-allewoorden-$ds-$i
-
     done
 
     # score