rename: skate-to-doi to skate-cleanup

1 files changed, 93 insertions, 0 deletions

diff --git a/skate/cmd/skate-cleanup/main.go b/skate/cmd/skate-cleanup/main.go
new file mode 100644
index 0000000..e1c7f3c
--- /dev/null
+++ b/skate/cmd/skate-cleanup/main.go
@@ -0,0 +1,93 @@
+// Filter to parse out a correctly looking DOI, URL, etc from a field.
+//
+// $ echo "1,xxx 10.123/12312 xxx,3" | skate-to-doi -c doi -d , -f 2
+// 1,10.123/12312,3k
+//
+// We can use this to sanitize fields in the reference dataset.
+
+package main
+
+import (
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"regexp"
+	"runtime"
+	"strings"
+
+	"git.archive.org/martin/cgraph/skate/parallel"
+	"mvdan.cc/xurls/v2"
+)
+
+var (
+	numWorkers     = flag.Int("w", runtime.NumCPU(), "number of workers")
+	batchSize      = flag.Int("b", 100000, "batch size")
+	delimiter      = flag.String("d", "\t", "delimiter")
+	index          = flag.Int("f", 1, "one field to cleanup up a doi, 1-indexed")
+	bestEffort     = flag.Bool("B", false, "only log errors, but do not stop")
+	skipNonMatches = flag.Bool("S", false, "do not emit a line for non-matches")
+	what           = flag.String("c", "doi", "what to clean: doi, url")
+
+	PatDOI    = regexp.MustCompile(`10[.][0-9]{1,8}/[^ ]*[\w]`)
+	rxRelaxed = xurls.Relaxed()
+)
+
+func main() {
+	flag.Parse()
+	var f func([]byte) ([]byte, error)
+	switch *what {
+	case "doi":
+		f = doiFilter
+	case "url":
+		f = urlFilter
+	default:
+		f = doiFilter
+	}
+	pp := parallel.NewProcessor(os.Stdin, os.Stdout, f)
+	pp.NumWorkers = *numWorkers
+	pp.BatchSize = *batchSize
+	if err := pp.Run(); err != nil {
+		log.Fatal(err)
+	}
+}
+
+// urlFilter parses finds the first URL.
+func urlFilter(p []byte) ([]byte, error) {
+	parts := strings.Split(string(p), *delimiter)
+	if len(parts) < *index {
+		msg := fmt.Sprintf("warn: line has too few fields (%d): %s", len(parts), string(p))
+		if *bestEffort {
+			log.Println(msg)
+			return nil, nil
+		} else {
+			return nil, fmt.Errorf(msg)
+		}
+	}
+	url := rxRelaxed.FindString(parts[*index-1])
+	if url == "" && *skipNonMatches {
+		return nil, nil
+	}
+	parts[*index-1] = url
+	return []byte(strings.Join(parts, *delimiter)), nil
+}
+
+// doiFilter finds a DOI
+func doiFilter(p []byte) ([]byte, error) {
+	parts := strings.Split(string(p), *delimiter)
+	if len(parts) < *index {
+		msg := fmt.Sprintf("warn: line has too few fields (%d): %s", len(parts), string(p))
+		if *bestEffort {
+			log.Println(msg)
+			return nil, nil
+		} else {
+			return nil, fmt.Errorf(msg)
+		}
+	}
+	result := PatDOI.FindString(parts[*index-1])
+	if result == "" && *skipNonMatches {
+		return nil, nil
+	}
+	parts[*index-1] = result
+	return []byte(strings.Join(parts, *delimiter)), nil
+}