aboutsummaryrefslogtreecommitdiffstats
path: root/skate/cmd/skate-biblioref-from-wikipedia/main.go
diff options
context:
space:
mode:
Diffstat (limited to 'skate/cmd/skate-biblioref-from-wikipedia/main.go')
-rw-r--r--skate/cmd/skate-biblioref-from-wikipedia/main.go35
1 files changed, 35 insertions, 0 deletions
diff --git a/skate/cmd/skate-biblioref-from-wikipedia/main.go b/skate/cmd/skate-biblioref-from-wikipedia/main.go
new file mode 100644
index 0000000..552b625
--- /dev/null
+++ b/skate/cmd/skate-biblioref-from-wikipedia/main.go
@@ -0,0 +1,35 @@
+package main
+
+import (
+ "flag"
+ "log"
+ "os"
+ "runtime"
+
+ "git.archive.org/martin/cgraph/skate"
+ "git.archive.org/martin/cgraph/skate/parallel"
+ jsoniter "github.com/json-iterator/go"
+)
+
+var (
+ numWorkers = flag.Int("w", runtime.NumCPU(), "number of workers")
+ batchSize = flag.Int("b", 100000, "batch size")
+ json = jsoniter.ConfigCompatibleWithStandardLibrary
+ bytesNewline = []byte("\n")
+)
+
+func main() {
+ flag.Parse()
+ pp := parallel.NewProcessor(os.Stdin, os.Stdout, func(p []byte) ([]byte, error) {
+ var w skate.MinimalCitations
+ if err := json.Unmarshal(p, &w); err != nil {
+ return nil, err
+ }
+ return nil, nil
+ })
+ pp.NumWorkers = *numWorkers
+ pp.BatchSize = *batchSize
+ if err := pp.Run(); err != nil {
+ log.Fatal(err)
+ }
+}