mag: tweak script

1 files changed, 22 insertions, 11 deletions

diff --git a/extra/mag/magrefs-mem.go b/extra/mag/magrefs-mem.go
index 403e021..ea7f207 100644
--- a/extra/mag/magrefs-mem.go
+++ b/extra/mag/magrefs-mem.go
@@ -13,6 +13,7 @@ import (
 	"os"
 	"strconv"
 	"strings"
+	"sync/atomic"
 
 	_ "github.com/mattn/go-sqlite3"
 	"github.com/miku/parallel"
@@ -21,6 +22,8 @@ import (
 var (
 	mappingFile = flag.String("f", "", "two column TSV with key and value")
 	cache       = make(map[int]string)
+
+	edgeMissingDOI int64 // number of edges that have one or zero DOI, but not both
 )
 
 func populateCache(r io.Reader, cache map[int]string) error {
@@ -69,27 +72,34 @@ func main() {
 		log.Fatal(err)
 	}
 	pp := parallel.NewProcessor(os.Stdin, os.Stdout, func(p []byte) ([]byte, error) {
-		fields := strings.Split(string(p), "\t")
+		var (
+			s, t, v, w     string
+			source, target int
+			err            error
+			ok             bool
+			complete       = true
+			fields         = strings.Split(string(p), "\t")
+		)
 		if len(fields) < 2 {
 			return nil, nil
 		}
-		s, t := strings.TrimSpace(fields[0]), strings.TrimSpace(fields[1])
-		source, err := strconv.Atoi(s)
-		if err != nil {
+		s, t = strings.TrimSpace(fields[0]), strings.TrimSpace(fields[1])
+		if source, err = strconv.Atoi(s); err != nil {
 			log.Printf("skipping invalid source id: %s (%v)", s, err)
 			return nil, nil
 		}
-		target, err := strconv.Atoi(t)
-		if err != nil {
+		if target, err = strconv.Atoi(t); err != nil {
 			log.Printf("skipping invalid target id: %s (%v)", t, err)
 			return nil, nil
 		}
-		v, ok := cache[source]
-		if !ok {
-			return nil, nil
+		if v, ok = cache[source]; !ok {
+			complete = false
+		}
+		if w, ok = cache[target]; !ok {
+			complete = false
 		}
-		w, ok := cache[target]
-		if !ok {
+		if !complete {
+			atomic.AddInt64(&edgeMissingDOI, 1)
 			return nil, nil
 		}
 		line := fmt.Sprintf("%s\t%s\n", v, w)
@@ -98,4 +108,5 @@ func main() {
 	if err := pp.Run(); err != nil {
 		log.Fatal(err)
 	}
+	log.Printf("edges without DOI: %d", edgeMissingDOI)
 }