diff options
Diffstat (limited to 'skate/cmd')
-rw-r--r-- | skate/cmd/skate-biblioref-from-wikipedia/main.go | 2 | ||||
-rw-r--r-- | skate/cmd/skate-verify/main.go | 27 |
2 files changed, 25 insertions, 4 deletions
diff --git a/skate/cmd/skate-biblioref-from-wikipedia/main.go b/skate/cmd/skate-biblioref-from-wikipedia/main.go index b51c953..e598491 100644 --- a/skate/cmd/skate-biblioref-from-wikipedia/main.go +++ b/skate/cmd/skate-biblioref-from-wikipedia/main.go @@ -30,7 +30,7 @@ func main() { if idl.DOI == "" { return nil, nil } - s := fmt.Sprintf("%s\t%s", idl.DOI, string(p)) + s := fmt.Sprintf("%s\t%s\t%s", w.PageTitle, idl.DOI, string(p)) return []byte(s), nil }) diff --git a/skate/cmd/skate-verify/main.go b/skate/cmd/skate-verify/main.go index e6fc417..e59d263 100644 --- a/skate/cmd/skate-verify/main.go +++ b/skate/cmd/skate-verify/main.go @@ -16,17 +16,18 @@ import ( "runtime/pprof" "strings" - jsoniter "github.com/json-iterator/go" "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", 10000, "batch size") - mode = flag.String("m", "ref", "mode: exact, ref, bref, zip, bzip") + mode = flag.String("m", "ref", "mode: exact, ref, bref, zip, bzip, wiki") exactReason = flag.String("r", "", "doi, pmid, pmcid, arxiv") provenance = flag.String("p", "join", "provenance info") + wikiFile = flag.String("W", "", "wiki citation file") releasesFile = flag.String("R", "", "releases, tsv, sorted by key (zip mode only)") refsFile = flag.String("F", "", "refs, tsv, sorted by key (zip mode only)") cpuProfile = flag.String("cpuprofile", "", "write cpu profile to file") @@ -90,7 +91,7 @@ func main() { // Take two "sorted key files" (one refs, one releases) and run // verification across groups, generate biblioref file. if *refsFile == "" || *releasesFile == "" { - log.Fatal("zip mode requires -R and -F to be set") + log.Fatal("zip mode requires -F and -R to be set") } f, err := os.Open(*releasesFile) if err != nil { @@ -123,6 +124,26 @@ func main() { if err := pp.Run(); err != nil { log.Fatal(err) } + case "wiki": + // Fixed zip mode for DOI from wikipedia. + if *wikiFile == "" || *releasesFile == "" { + log.Fatal("mode requires -W and -F to be set") + } + f, err := os.Open(*releasesFile) + if err != nil { + log.Fatal(err) + } + defer f.Close() + g, err := os.Open(*wikiFile) + if err != nil { + log.Fatal(err) + } + defer g.Close() + bw := bufio.NewWriter(os.Stdout) + defer bw.Flush() + if err := skate.ZipUnverified(f, g, skate.MatchResult{skate.StatusExact, skate.ReasonDOI}, "wiki", bw); err != nil { + log.Fatal(err) + } default: log.Fatal("not implemented, only: zip, ref, bref") } |