namen.sh: opties

namen.sh

@@ -1,9 +1,96 @@
 #!/bin/bash
 
-cd /net/corpora/nlnieuws
+PROGNAME=$0
+usage() {
+    echo "
+gebruik:
 
-select i in `find . -name '*data.dz' | sort`
+    $PROGNAME [opties] -i
+    $PROGNAME [opties] file.data.dz ...
+
+    -i     : kies interactief
+
+overige opties:
+
+    -n int : max aantal resultaten
+    -p     : gebruik pager
+    -s     : tel hits één keer per bericht
+    "
+    exit
+}
+
+SINGLE=0
+SELECT=0
+USEPAGER=0
+LIMIT=0
+while getopts 'sin:p' opt
 do
-    alto "$i" fp:'//node[@cat="mwu" and node[@pt="spec"] and not(@his="normal") and not(@his_1="decap")]' tt:%w | \
+    case "$opt" in
-        sort | uniq -c | sort -nr | head -n 40
+        i)
+            SELECT=1
+            ;;
+        n)
+            LIMIT="$OPTARG"
+            ;;
+        p)
+            USEPAGER=1
+            ;;
+        s)
+            SINGLE=1
+            ;;
+        *)
+            usage
+            ;;
+    esac
 done
+shift "$(($OPTIND -1))"
+
+TAIL=''
+if [ $LIMIT -gt 0 ]
+then
+    TAIL=" | head -n $LIMIT"
+fi
+if [ $USEPAGER = 1 ]
+then
+    TAIL="$TAIL | ${PAGER:-less}"
+fi
+
+EXPR='fp://node[@cat="mwu" and node[@pt="spec"] and not(@his="normal") and not(@his_1="decap")]'
+if [ $SINGLE = 1 ]
+then
+    SORT="sed -e 's/\.[0-9][0-9]*$//' | sort | uniq | sed -e 's/\t.*//'"
+    TEMPLATE='tt:%w\t%I'
+else
+    SORT=sort
+    TEMPLATE='tt:%w'
+fi
+
+search () {
+    eval "alto $@ '$EXPR' '$TEMPLATE' | $SORT | uniq -c | sort -nr 2> /dev/null $TAIL"
+}
+
+for i in "$@"
+do
+    case "$i" in
+        *.data.dz)
+            ;;
+        *)
+            usage
+            ;;
+    esac
+done
+
+if [ $# == 0 -a $SELECT = 1 ]
+then
+    cd /net/corpora/nlnieuws
+    select i in `find . -name '*data.dz' | sort`
+    do
+        search $i
+    done
+elif [ $# -gt 0 -a $SELECT = 0 ]
+then
+    search "$@"
+else
+    usage
+fi
+