update Golang example

Author: hipertracker

README.md

@@ -19,7 +19,7 @@ The following results are for 123 unique utf-8 Bible text files in 23 languages
 5. Elixir 1.13.2  = 7.82s
 6. Ruby 3.1.0     = 8.31s
 
-Golang 1.17    = UNDER REFACTORING, stay tuned
+Golang 1.17.6    = UNDER REFACTORING, stay tuned
 </pre>
 
 ### Conclusion

example-golang/.gitignore

@@ -0,0 +1 @@
+/coverage.out

example-golang/Makefile

@@ -8,14 +8,22 @@ build:
 run: build
 	./${BINARY_NAME}
 
+run-sort: build
+	./${BINARY_NAME} -n 10 -s
+
 test:
+	@go test ./... -v
+
+coverage:
 	@go test ./... -v -coverprofile=coverage.out  
-	
+
+
 cover: test
 	@go tool cover -html=coverage.out
 
 clean:
 	@go clean
 	rm -f coverage.out 
 	rm -f ./${BINARY_NAME}
-	rm -rf ./words
+	rm -rf ./words
+

example-golang/README.md

@@ -4,5 +4,12 @@
 
 ```
 make build
-./main [-n=NUMBER_OF_WORKERS, integer] [-s] 
+./main -n 8
 ```
+
+<pre>
+Usage of ./main:
+  -n int
+     Number of workers to run (zero to match the number of available CPUs) (default 10)
+  -s Sort results
+</pre>

example-golang/app/app.go

@@ -41,7 +41,7 @@ func Run(srcDir, outDir string, numWorkers int, sortResults bool) error {
 		}
 
 		src := file[:len(file)-3] + "txt"
-		dst := filepath.Join(outDir, "extracted-words-for-"+spec.Code+".txt")
+		dst := filepath.Join(outDir, spec.Lang+"-"+spec.Code+".txt")
 
 		wg.Add(1)
 		go extract(src, dst, "POLISH_CI", sortResults, sem, &wg)

example-golang/app/extract.go

@@ -196,3 +196,17 @@ func writeResults(w io.Writer, words []string) error {
 
 	return nil
 }
+
+func ExtractUniqueWords(content string, lang string) ([]string, error) {
+	r := strings.NewReader(content)
+	words, err := collectWords(r)
+	if err != nil {
+		_, _ = fmt.Fprintf(os.Stderr, `collectWords error: %s`, err)
+		return nil, err
+	}
+	less := collate.IndexString(lang)
+	sort.Slice(words, func(i, j int) bool {
+		return less(words[i], words[j])
+	})
+	return words, nil
+}

example-golang/app_test.go

@@ -0,0 +1,21 @@
+package main
+
+import (
+	"fmt"
+	"os"
+	"testing"
+	"wordextractor/app"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func Test_ExtractUniqueWords(t *testing.T) {
+	text := "ćma cześć ser. śmiech!żółw zebra-łuk len Ćma Żółw ser"
+	expected := []string{"cześć", "ćma", "len", "łuk", "ser", "śmiech", "zebra", "żółw"}
+	given, err := app.ExtractUniqueWords(text, "pl")
+	if err != nil {
+		_, _ = fmt.Fprintf(os.Stderr, `ExtractUniqueWords error: %s`, err)
+		return
+	}
+	assert.Equal(t, expected, given, "text should be tokenized into unique words")
+}

example-golang/go.mod

@@ -1,4 +1,4 @@
-module github.com/hipertracker/word_extractor
+module wordextractor
 
 go 1.17
 

example-golang/main.go

@@ -8,7 +8,7 @@ import (
 	"runtime"
 	"time"
 
-	"github.com/hipertracker/word_extractor/app"
+	"wordextractor/app"
 )
 
 const (