diverse aanpassingen

cmd/data2json/data2json.go

@@ -27,18 +27,12 @@ type Data struct {
 }
 
 type Parts struct {
-	NieuweNamen   [][3]any `json:"nieuwe namen"`
-	NieuweWoorden [][3]any `json:"nieuwe woorden"`
-	Personen      [][3]any `json:"personen"`
-	AndereNamen   [][3]any `json:"andere namen"`
-	Locaties      [][3]any `json:"locaties"`
-	Organisaties  [][3]any `json:"organisaties"`
-}
-
-type Value struct {
-	Word  string `json:"word"`
-	Tags  string `json:"tags"`
-	Count int    `json:"count"`
+	NieuweNamen   [][5]any `json:"nieuwe namen"`
+	NieuweWoorden [][5]any `json:"nieuwe woorden"`
+	Personen      [][5]any `json:"personen"`
+	AndereNamen   [][5]any `json:"andere namen"`
+	Locaties      [][5]any `json:"locaties"`
+	Organisaties  [][5]any `json:"organisaties"`
 }
 
 var (
@@ -55,7 +49,7 @@ var (
 		suffix string
 	}{
 		"nieuwe namen":   {"nieuwe-namen", ".t20"},
-		"nieuwe woorden": {"nieuwe-woorden", ".t20"},
+		"nieuwe woorden": {"nieuwe-woorden-extra", ".t20"},
 		"personen":       {"personen", ""},
 		"andere namen":   {"overige-namen", ""},
 		"locaties":       {"locaties", ""},
@@ -127,8 +121,8 @@ func makeParts(source string) *Parts {
 	}
 }
 
-func makeValues(source, part string) [][3]any {
-	v := make([][3]any, 0)
+func makeValues(source, part string) [][5]any {
+	v := make([][5]any, 0)
 
 	filename := fmt.Sprintf("/net/corpora/nlnieuws/data/%s-%s-%d-%02d-%d%s",
 		sources[source],
@@ -149,11 +143,17 @@ func makeValues(source, part string) [][3]any {
 		count, err := strconv.Atoi(strings.TrimSpace(aa[0]))
 		x(err)
 		word := aa[1]
-		var tags string
+		var tags, lemma, postag string
 		if len(aa) > 2 {
 			tags = aa[2]
 		}
-		v = append(v, [3]any{count, word, tags})
+		if len(aa) > 3 {
+			lemma = aa[3]
+		}
+		if len(aa) > 4 {
+			postag = aa[4]
+		}
+		v = append(v, [5]any{count, word, tags, lemma, postag})
 		if lineno == 20 {
 			break
 		}
@@ -173,7 +173,7 @@ func dates() (start, first, last string) {
 	// zoek juiste week
 	var y, w int
 	for {
-		y, w = t.ISOWeek()
+		y, _ = t.ISOWeek()
 		if y < year {
 			t = t.AddDate(0, 12, 0)
 			continue

cmd/items2count/items2count.go

@@ -48,6 +48,12 @@ func main() {
 		word := aa[0]
 		tags := aa[1]
 		lbl := aa[2]
+		if n := len(aa); n > 3 {
+			lbl = aa[n-1]
+			for i := 2; i < n-1; i++ {
+				word += "\t" + aa[i]
+			}
+		}
 		w, ok := words[word]
 		if !ok {
 			w = &Word{
@@ -87,7 +93,13 @@ func main() {
 	})
 
 	for _, w := range wordlist {
-		fmt.Printf("%6d\t%s\t%s\n", w.count, w.word, getTag(w.tags))
+		var tail string
+		i := strings.Index(w.word, "\t")
+		if i > 0 {
+			tail = w.word[i:]
+			w.word = w.word[:i]
+		}
+		fmt.Printf("%6d\t%s\t%s%s\n", w.count, w.word, getTag(w.tags), tail)
 	}
 
 }