Compatibility with https://github.com/caneroj1/stemmer libraryHEADmaster

3 files changed, 76 insertions, 7 deletions

diff --git a/README.md b/README.md
index edd8693..6ff53d4 100644
--- a/README.md
+++ b/README.md
@@ -9,12 +9,29 @@
 
 Этот пакет - реализация [стеммера Портера](https://ru.wikipedia.org/wiki/Стемминг#Стеммер_Портера) для русского языка на Go.
 
+Интерфейс совместим со стеммером https://github.com/caneroj1/stemmer
+
 ## Использование
 
-`основа := StemmerRu.StemWord("слово")`
+`основа := StemmerRu.Stem("слово")`
 
 Преобразует слово на входе в его основу на выходе
 
+Так же, из библиотеки https://github.com/caneroj1/stemmer взяты следющие методы:
+
+```
+ // stem a list of words
+  stems := StemmerRu.StemMultiple(strings)
+
+  // stem a list of words in place, modifying the original slice
+  StemmerRu.StemMultipleMutate(strings)
+  
+  // stem a list of words concurrently. this also stems in place, modifying
+  // the original slice.
+  // NOTE: the order of the strings is not guaranteed to be the same.
+  StemmerRu.StemConcurrent(strings)
+```
+
 ## Пример
 
 ```
diff --git a/stemmer.go b/stemmer.go
index f15281b..80dc592 100644
--- a/stemmer.go
+++ b/stemmer.go
@@ -1,6 +1,7 @@
 package StemmerRu
 
 import (
+	"runtime"
 	"strings"
 )
 
@@ -17,7 +18,7 @@ var (
 	vowels = `аеиоуыэюя`
 )
 
-func StemWord(word string) string {
+func Stem(word string) string {
 
 	word = strings.Replace(word, `ё`, `е`, -1)
 
@@ -27,10 +28,9 @@ func StemWord(word string) string {
 		return word
 	}
 
-
 	R1pos := getRNPart(word, 0)
 	R2pos := getRNPart(word, R1pos)
-	if (R2pos < RVpos) {
+	if R2pos < RVpos {
 		R2pos = 0
 	} else {
 		R2pos -= RVpos
@@ -140,3 +140,55 @@ func getRNPart(word string, startPos int) int {
 
 	return startPos
 }
+
+// Code from https://github.com/caneroj1/stemmer
+
+// StemMultiple accepts a slice of strings and stems each of them.
+func StemMultiple(words []string) (output []string) {
+	output = make([]string, len(words))
+	for idx, word := range words {
+		output[idx] = Stem(word)
+	}
+
+	return
+}
+
+// StemMultipleMutate accepts a pointer to a slice of strings and stems them in place.
+// It modifies the original slice.
+func StemMultipleMutate(words *[]string) {
+	for idx, word := range *words {
+		(*words)[idx] = Stem(word)
+	}
+}
+
+// StemConcurrent accepts a pointer to a slice of strings and stems them in place.
+// It tries to offload the work into multiple threads. It makes no guarantees about
+// the order of the stems in the modified slice.
+func StemConcurrent(words *[]string) {
+	CPUs := runtime.NumCPU()
+	length := len(*words)
+	output := make(chan string)
+	partition := length / CPUs
+
+	var CPU int
+	for CPU = 0; CPU < CPUs; CPU++ {
+		go func(strs []string) {
+			for _, word := range strs {
+				output <- Stem(word)
+			}
+		}((*words)[CPU*partition : (CPU+1)*partition])
+	}
+
+	// if there are leftover words, stem them now
+	if length-(CPU)*partition > 0 {
+		go func(strs []string) {
+			for _, word := range strs {
+				output <- Stem(word)
+			}
+		}((*words)[(CPU)*partition : length])
+	}
+
+	for idx := 0; idx < length; idx++ {
+		(*words)[idx] = <-output
+	}
+}
diff --git a/stemmer_test.go b/stemmer_test.go
index 8b27774..334a329 100644
--- a/stemmer_test.go
+++ b/stemmer_test.go
@@ -1,9 +1,9 @@
 package StemmerRu
 
 import (
-	"testing"
-	"io/ioutil"
 	"encoding/json"
+	"io/ioutil"
+	"testing"
 )
 
 var testFile = `tests.json`
@@ -19,7 +19,7 @@ func TestStemWord(t *testing.T) {
 		t.Error("Can't parse json", err)
 	}
 	for source, expected := range *tests {
-		result := StemWord(source);
+		result := Stem(source)
 		if expected != result {
 			t.Errorf(`Expected "%s" (source: %s) but got "%s"`, result, source, result)
 		}