update namen.sh: meerdere zoekfuncties

namen.sh

@@ -5,8 +5,19 @@ usage() {
     echo "
 gebruik:
 
-    $PROGNAME [opties] -i
+    $PROGNAME [opties] -x 1|2|3|4|5 -i
-    $PROGNAME [opties] corpusfile(s) en/of corpusdirectory(s)
+    $PROGNAME [opties] -x 1|2|3|4|5 corpusfile(s) en/of corpusdirectory(s)
+
+    -x     : query
+
+       1   : nieuwe namen
+       2   : nieuwe woorden
+       3   : nieuwe woorden met postag en lemma
+       4   : bestaaande locaties
+       5   : bestaande personen
+       6   : bestaande organisaties
+       7   : bestaande andere namen
+       8   : nieuwe adjectieven, deelwoorden en werkwoorden
 
     -i     : kies interactief
 
@@ -25,7 +36,9 @@ SELECT=0
 USEPAGER=0
 LIMIT=0
 VERBOSE=0
-while getopts 'sin:pv' opt
+XN=''
+XVALID=0
+while getopts 'sin:pvx:' opt
 do
     case "$opt" in
         i)
@@ -40,16 +53,73 @@ do
         s)
             SINGLE=1
             ;;
-	v)
+        v)
-	    VERBOSE=1
+            VERBOSE=1
-	    ;;
+            ;;
-        *)
+        x)
-            usage
+            XN="$OPTARG"
             ;;
     esac
 done
 shift "$(($OPTIND -1))"
 
+case $XN in
+    1)
+        # nieuwe namen
+        EXPR='fp://node[((@cat="mwu" and node[@pt="spec"]) or (@pt and @*="eigen" and not(@rel="mwp"))) and not(@his="normal") and not(@his_1="decap" or @his_1="0")]'
+        TEMPLATE='tt:%w'
+        XVALID=1
+        ;;
+    2)
+        # nieuwe woorden
+        EXPR='fp://node[@his and not(@rel="mwp" or @cat="mwu") and not(@his="normal" or @his="name" or @his="prefix_name" or @his_1="decap" or @his_1="0" or @his="skip" or @his="robust_skip" or @his="w_dia" or @his="wo_dia" or @his="within_word_conjunct")]'
+        TEMPLATE='tt:%w'
+        XVALID=1
+        ;;
+    3)
+        # nieuwe woorden met postag en lemma
+        EXPR='fp://node[@his and not(@rel="mwp" or @cat="mwu") and not(@his="normal" or @his="name" or @his="prefix_name" or @his_1="decap" or @his_1="0" or @his="skip" or @his="robust_skip" or @his="w_dia" or @his="wo_dia" or @his="within_word_conjunct")]'
+        TEMPLATE='tt:%w\t%l\t%P'
+        XVALID=1
+        ;;
+    4)
+        # bestaande locaties
+        EXPR='fp://node[(@neclass="LOC" and @his="normal" and not(@rel="mwp")) or (@cat="mwu" and node[@pt="spec" and @neclass="LOC"] and @his="normal")]'
+        TEMPLATE='tt:%l'
+        XVALID=1
+        ;;
+    5)
+        # bestaande personen
+        EXPR='fp://node[(@neclass="PER" and @his="normal" and not(@rel="mwp")) or (@cat="mwu" and node[@pt="spec" and @neclass="PER"] and @his="normal")]'
+        TEMPLATE='tt:%l'
+        XVALID=1
+        ;;
+    6)
+        # bestaande organisaties
+        EXPR='fp://node[(@neclass="ORG" and @his="normal" and not(@rel="mwp")) or (@cat="mwu" and node[@pt="spec" and @neclass="ORG"] and @his="normal")]'
+        TEMPLATE='tt:%l'
+        XVALID=1
+        ;;
+    7)
+        # bestaande andere namen (boeken, films, events, .. )
+        EXPR='fp://node[(@neclass="MISC" and @his="normal" and not(@rel="mwp")) or (@cat="mwu" and node[@pt="spec" and @neclass="MISC"] and @his="normal")]'
+        TEMPLATE='tt:%l'
+        XVALID=1
+        ;;
+    8)
+        # nieuwe adjectieven, deelwoorden en werkwoorden
+        EXPR='fp://node[@pt and @his and not(../@his="normal" or @rel="mwp" or ../@his="name" or ../@his_1="decap") and not(@his="normal" or @his="name" or @his="skip" or @his="robust_skip" or @his="w_dia" or @his="wo_dia" or @his="decap" or @his="within_word_conjunct") and not(@pt="n") ]'
+        TEMPLATE='tt:%w\t%P'
+        XVALID=1
+        ;;
+    '')
+        ;;
+    *)
+        echo Invalid value for option -x
+        exit
+        ;;
+esac
+
 TAIL=''
 if [ $LIMIT -gt 0 ]
 then
@@ -60,22 +130,21 @@ then
     TAIL="$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.*//'"
+    SORT="sed -e 's/\.[0-9][0-9]*$//' | sort | uniq | sed -e 's/\(.*\)\t.*/\1\t/'"
-    TEMPLATE='tt:%w\t%I'
+    TEMPLATE="$TEMPLATE"'\t%I'
 else
     SORT=sort
-    TEMPLATE='tt:%w'
 fi
 
 search () {
+    CMD="alto $@ '$EXPR' '$TEMPLATE' | $SORT | uniq -c | sed -e 's/\([0-9]\)  */\1\t/' | sort -nr 2> /dev/null | tabulate -s '\t' -f plain 2> /dev/null$TAIL"
     if [ $VERBOSE = 1 ]
     then
-        echo "alto $@ '$EXPR' '$TEMPLATE' | $SORT | uniq -c | sort -nr 2> /dev/null$TAIL"
+        echo $CMD
     fi
-    eval "alto $@ '$EXPR' '$TEMPLATE' | $SORT | uniq -c | sort -nr 2> /dev/null$TAIL"
+    eval $CMD
 }
 
 for i in "$@"
@@ -92,14 +161,14 @@ do
     esac
 done
 
-if [ $# == 0 -a $SELECT = 1 ]
+if [ $# == 0 -a $SELECT = 1 -a $XVALID = 1 ]
 then
     cd /net/corpora/nlnieuws
     select i in `find . -name '*data.dz' | sort`
     do
         search $i
     done
-elif [ $# -gt 0 -a $SELECT = 0 ]
+elif [ $# -gt 0 -a $SELECT = 0 -a $XVALID = 1 ]
 then
     search "$@"
 else