grote reorganisatie:

- HLN, NOS, NU, VRT: per week -> per dag
- yyyy-ww -> yyyy.ww
- yyyy*  -> yyyy/yyyy*
etc

NOS/cmd/json2txt/json2txt.go

@@ -28,7 +28,7 @@ type Item struct {
 var (
 	x = e.ExitErr
 
-	reYearWeek = regexp.MustCompile(`^2[0-9][0-9][0-9]-[0-5][0-9]$`)
+	reYearWeek = regexp.MustCompile(`^2[0-9][0-9][0-9]-[01][0-9]-[0-3][0-9]$`)
 )
 
 func main() {
@@ -36,17 +36,17 @@ func main() {
 	var ds string
 	switch len(os.Args) {
 	case 1:
-		year, week := time.Now().AddDate(0, 0, -7).ISOWeek()
-		ds = fmt.Sprintf("%d-%02d", year, week)
+		t := time.Now().AddDate(0, 0, -2)
+		ds = fmt.Sprintf("%d-%02d-%02d", t.Year(), int(t.Month()), t.Day())
 	case 2:
 		if !reYearWeek.MatchString(os.Args[1]) {
-			x(fmt.Errorf("arg must be yyyy-ww"))
+			x(fmt.Errorf("arg must be yyyy-mm-dd"))
 		}
 		ds = os.Args[1]
 	default:
 		x(fmt.Errorf("too many arguments"))
 	}
-	dp := ds[:4] + "/" + ds[5:]
+	dp := strings.ReplaceAll(ds, "-", "/")
 
 	x(os.Chdir("/net/corpora/nlnieuws/NOS/" + dp))
 	x(os.MkdirAll("out", 0777))

NOS/cmd/nos/nos.go

@@ -94,8 +94,7 @@ func main() {
 			}
 		}
 		p(err)
-		year, week := t.ISOWeek()
-		dirname := fmt.Sprintf("/net/corpora/nlnieuws/NOS/%d/%02d", year, week)
+		dirname := fmt.Sprintf("/net/corpora/nlnieuws/NOS/%d/%02d/%02d", t.Year(), int(t.Month()), t.Day())
 		if exists(dirname + "/lock") {
 			continue
 		}

NOS/txt2corpus.sh

@@ -2,17 +2,20 @@
 
 set -e
 
+BASE=/net/corpora/nlnieuws
+PART=$BASE/NOS
+
 unset CDPATH
-PATH=/net/corpora/nlnieuws/bin:/net/aps/bin:$PATH
+PATH=$PART:$BASE/bin:$BASE:/net/aps/bin:$PATH
 export TZ=Europe/Amsterdam
 . /net/aps/etc/alpino-activate.sh > /dev/null
 
 if [ "$1" = "" ]
 then
-    ds=`date -d -7days +%G-%V`
+    ds=`date -d -2days +%Y-%m-%d`
 else
     case "$1" in
-        2[0-9][0-9][0-9]-[0-5][0-9])
+        2[0-9][0-9][0-9]-[01][0-9]-[0-3][0-9])
             ds=$1
             ;;
         *)
@@ -23,10 +26,11 @@ else
 fi
 
 dp=${ds//-//}
+year=${ds%%-*}
+corpus=$PART/corpus/$year/$ds
+mkdir -p $PART/corpus/$year
 
-corpus=/net/corpora/nlnieuws/NOS/corpus/$ds
-
-cd /net/corpora/nlnieuws/NOS/$dp
+cd $PART/$dp
 
 ln -s lock.$$ lock
 if [ "`readlink lock`" != lock.$$ ]
@@ -38,7 +42,7 @@ fi
 rm -fr out
 mkdir out
 
-../../json2txt $ds
+json2txt $ds
 
 rm -f $corpus.lines
 for i in out/*.txt
@@ -53,15 +57,15 @@ cd out
 mkdir xml
 Alpino -flag treebank xml debug=1 end_hook=xml user_max=900000 -parse < $corpus.lines 2> $corpus.log
 
-../../../metadata
+metadata
 
 cd xml
 rm -f $corpus.data.dz $corpus.index
 alto -q -o $corpus.data.dz *.xml
 
 # telling per bericht, niet per zin
-/net/corpora/nlnieuws/namen.sh -x C -s $corpus.data.dz > $corpus.cat.txt
-/net/corpora/nlnieuws/namen.sh -x T -s $corpus.data.dz > $corpus.tag.txt
+query.sh -x C -s $corpus.data.dz > $corpus.cat.txt
+query.sh -x T -s $corpus.data.dz > $corpus.tag.txt
 
 cd ../..
 rm -fr out