From 09a7e8c9d013f13a1aa1ef4e9b7f397647b79967 Mon Sep 17 00:00:00 2001
From: Martin Czygan <martin.czygan@gmail.com>
Date: Sun, 21 Mar 2021 01:17:38 +0100
Subject: initial import of skate

---
 skate/cmd/skate-derive-key/main.go | 92 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 92 insertions(+)
 create mode 100644 skate/cmd/skate-derive-key/main.go

(limited to 'skate/cmd/skate-derive-key')

diff --git a/skate/cmd/skate-derive-key/main.go b/skate/cmd/skate-derive-key/main.go
new file mode 100644
index 0000000..2375a73
--- /dev/null
+++ b/skate/cmd/skate-derive-key/main.go
@@ -0,0 +1,92 @@
+// skate-derive-key derives a key from release entity JSON documents.
+//
+// $ skate-derive-key < release_entities.jsonlines > docs.tsv
+//
+// Result will be a three column TSV (ident, key, doc), LC_ALL=C sorted by key.
+//
+// ---- ident ---------------    ---- key ------------------------------     ---- doc ----------
+//
+// 4lzgf5wzljcptlebhyobccj7ru    2568diamagneticsusceptibilityofh8n2o10sr    {"abstracts":[],...
+//
+// After this step, a fast "itertools.groupby" or "skate-cluster" on key can yields clusters.
+//
+// Notes
+//
+// Using https://github.com/DataDog/zstd#stream-api, 3700 docs/s for key
+// extraction only; using zstd -T0, we get 21K docs/s; about 13K docs/s; about
+// 32h for 1.5B records.
+//
+// Default sort(1) buffer is 1K, but we'll need G's, e.g. -S35% of 48GB.
+package main
+
+import (
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"runtime"
+	"strings"
+	"time"
+
+	"git.archive.org/martin/cgraph/skate"
+	"git.archive.org/martin/cgraph/skate/parallel"
+)
+
+var (
+	keyFuncName   = flag.String("f", "tsand", "key function name, other: title, tnorm, tnysi, tsand")
+	numWorkers    = flag.Int("w", runtime.NumCPU(), "number of workers")
+	batchSize     = flag.Int("b", 50000, "batch size")
+	verbose       = flag.Bool("verbose", false, "show progress")
+	bestEffort    = flag.Bool("B", false, "best effort")
+	logFile       = flag.String("log", "", "log filename")
+	skipEmptyKeys = flag.Bool("skip-empty-keys", false, "omit docs without keys")
+
+	wsReplacer = strings.NewReplacer("\t", "", "\n", "")
+	keyOpts    = map[string]skate.IdentifierKeyFunc{
+		"title": skate.KeyTitle,
+		"tnorm": skate.KeyTitleNormalized,
+		"tnysi": skate.KeyTitleNysiis,
+		"tsand": skate.KeyTitleSandcrawler,
+	}
+	keyFunc skate.IdentifierKeyFunc
+	ok      bool
+)
+
+func main() {
+	flag.Parse()
+	if keyFunc, ok = keyOpts[*keyFuncName]; !ok {
+		log.Fatal("invalid key func")
+	}
+	if *logFile != "" {
+		f, err := os.OpenFile(*logFile, os.O_CREATE|os.O_APPEND, 0644)
+		if err != nil {
+			log.Fatal(err)
+		}
+		defer f.Close()
+		log.SetOutput(f)
+	}
+	started := time.Now()
+	pp := parallel.NewProcessor(os.Stdin, os.Stdout, func(p []byte) ([]byte, error) {
+		ident, key, err := keyFunc(p)
+		if err != nil {
+			if *bestEffort {
+				log.Printf("keyFunc failed with %v: %v", err, string(p))
+				return nil, nil
+			}
+			return nil, err
+		}
+		ident, key = strings.TrimSpace(ident), strings.TrimSpace(key)
+		if *skipEmptyKeys && key == "" {
+			return nil, nil
+		}
+		v := fmt.Sprintf("%s\t%s\t%s\n", ident, key, wsReplacer.Replace(string(p)))
+		return []byte(v), nil
+	})
+	pp.NumWorkers = *numWorkers
+	pp.BatchSize = *batchSize
+	pp.Verbose = *verbose
+	if err := pp.Run(); err != nil {
+		log.Fatal(err)
+	}
+	log.Printf("took: %s", time.Since(started))
+}
-- 
cgit v1.2.3