aboutsummaryrefslogtreecommitdiffstats
path: root/skate/cmd/skate-wikipedia-doi
diff options
context:
space:
mode:
authorMartin Czygan <martin.czygan@gmail.com>2021-03-31 23:54:28 +0200
committerMartin Czygan <martin.czygan@gmail.com>2021-03-31 23:54:28 +0200
commitd60dff7db926cc40a288584ac3f9970bb85c30c0 (patch)
treee79829df3f2229354b4dc079ee434274eb09aae0 /skate/cmd/skate-wikipedia-doi
parenta7e0e8fe3a78fa6a9872cefcb7a9eb261128f6f2 (diff)
downloadrefcat-d60dff7db926cc40a288584ac3f9970bb85c30c0.tar.gz
refcat-d60dff7db926cc40a288584ac3f9970bb85c30c0.zip
rename to: skate-wikipedia-doi
Diffstat (limited to 'skate/cmd/skate-wikipedia-doi')
-rw-r--r--skate/cmd/skate-wikipedia-doi/main.go43
1 files changed, 43 insertions, 0 deletions
diff --git a/skate/cmd/skate-wikipedia-doi/main.go b/skate/cmd/skate-wikipedia-doi/main.go
new file mode 100644
index 0000000..15b0d0a
--- /dev/null
+++ b/skate/cmd/skate-wikipedia-doi/main.go
@@ -0,0 +1,43 @@
+package main
+
+import (
+ "flag"
+ "fmt"
+ "log"
+ "os"
+ "runtime"
+ "strings"
+
+ "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
+ }
+ idl := w.ParseIDList()
+ if idl.DOI == "" {
+ return nil, nil
+ }
+ s := fmt.Sprintf("%s\t%s\t%s", strings.TrimSpace(w.PageTitle), idl.DOI, string(p))
+ return []byte(s), nil
+
+ })
+ pp.NumWorkers = *numWorkers
+ pp.BatchSize = *batchSize
+ if err := pp.Run(); err != nil {
+ log.Fatal(err)
+ }
+}