Initial commit.

README.md

@@ -11,7 +11,7 @@ An n-gram of size 1 is referred to as a "unigram"; size 2 is a "bigram" (or, les
 
 ## Requirements
 
-* PHP >= 7.0,
+* PHP >= 7.0
 
 ## Installation
 
@@ -26,9 +26,9 @@ Let's say you want to find all the N-Gram of size 3 of the string **hello world*
 ```php
 $word = 'hello world';
 $ngram = new \drupol\phpngrams\NGrams();
-$ngrams = $ngram->ngramString($word, 3);
+$ngrams = $ngram->ngramsString($word, 3);
 
-print_r($ngrams);
+print_r(iterator_to_array($ngrams));
 /*
     [0] => hel
     [1] => ell
@@ -49,7 +49,7 @@ Instead of using a string, you may also use an array as input.
 ```php
 $word = ['h', 'e', 'l', 'l', 'o'];
 $ngram = new \drupol\phpngrams\NGrams();
-$ngrams = $ngram->ngramArray($word, 3);
+$ngrams = $ngram->ngramsArray($word, 3);
 
 print_r($ngrams);
 /*
@@ -94,6 +94,11 @@ Array
 */
 ```
 
+To reduce to the maximum the memory footprint, the library returns Generators, if you want to get the complete resulting array, use [iterator_to_array()](https://secure.php.net/manual/en/function.iterator-to-array.php).
+
+The library provides and object Ngrams and a trait NgramsTrait.
+It's up to you to decide how you want to use the library.
+
 ## Tests
 
 Every time the sources are modified, [Travis](https://travis-ci.org/drupol/phpngrams), the continuous integration

spec/drupol/phpngrams/NGramsSpec.php

@@ -18,7 +18,7 @@ public function it_can_get_ngram_from_a_string()
             'he', 'el', 'll', 'lo', 'oh'
         ];
 
-        $this->ngramString('hello', 2)->shouldBe($result);
+        $this->ngramsString('hello', 2)->shouldIterateAs(new \ArrayIterator($result));
     }
 
     public function it_can_get_ngram_from_a_string_with_big_n()
@@ -27,7 +27,7 @@ public function it_can_get_ngram_from_a_string_with_big_n()
             'hello', 'elloh', 'llohe', 'lohel', 'ohell'
         ];
 
-        $this->ngramString('hello', 10)->shouldBe($result);
+        $this->ngramsString('hello', 10)->shouldIterateAs(new \ArrayIterator($result));
     }
 
     public function it_can_get_ngram_from_a_string_without_cycling()
@@ -36,7 +36,7 @@ public function it_can_get_ngram_from_a_string_without_cycling()
             'hel', 'ell', 'llo', 'lo ', 'o w', ' wo', 'wor', 'orl', 'rld'
         ];
 
-        $this->ngramString('hello world', 3, false)->shouldBe($result);
+        $this->ngramsString('hello world', 3, false)->shouldIterateAs(new \ArrayIterator($result));
     }
 
     public function it_can_get_ngram_from_an_array()
@@ -45,7 +45,7 @@ public function it_can_get_ngram_from_an_array()
             ['h', 'e'], ['e','l'], ['l','l'], ['l','o'], ['o','h']
         ];
 
-        $this->ngramArray(['h', 'e', 'l', 'l', 'o'], 2)->shouldBe($result);
+        $this->ngramsArray(['h', 'e', 'l', 'l', 'o'], 2)->shouldIterateAs(new \ArrayIterator($result));
     }
 
     public function it_can_get_ngram_from_an_array_with_big_n()
@@ -58,7 +58,7 @@ public function it_can_get_ngram_from_an_array_with_big_n()
             ['o','h','e','l','l'],
         ];
 
-        $this->ngramArray(['h', 'e', 'l', 'l', 'o'], 10)->shouldBe($result);
+        $this->ngramsArray(['h', 'e', 'l', 'l', 'o'], 10)->shouldIterateAs(new \ArrayIterator($result));
     }
 
     public function it_can_get_ngram_from_an_array_without_cycling()
@@ -75,6 +75,6 @@ public function it_can_get_ngram_from_an_array_without_cycling()
             ['r','l','d'],
         ];
 
-        $this->ngramArray(['h', 'e', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd'], 3, false)->shouldBe($result);
+        $this->ngramsArray(['h', 'e', 'l', 'l', 'o', ' ', 'w', 'o', 'r', 'l', 'd'], 3, false)->shouldIterateAs(new \ArrayIterator($result));
     }
 }

src/NGramsTrait.php

@@ -9,48 +9,48 @@ trait NGramsTrait
      * @param int $n
      * @param bool $cyclic
      *
-     * @return array
+     * @return \Generator
      */
-    public function ngramArray(array $data, $n = 1, $cyclic = true)
+    public function ngramsArray(array $data, $n = 1, $cyclic = true)
     {
-        return $this->doNgram($data, $n, $cyclic);
+        return $this->doNgrams($data, $n, $cyclic);
     }
 
     /**
      * @param $data
      * @param int $n
      * @param bool $cyclic
      *
-     * @return array
+     * @return \Generator
      */
-    public function ngramString($data, $n = 1, $cyclic = true)
+    public function ngramsString($data, $n = 1, $cyclic = true)
     {
-        return array_map(function ($data) {
-            return implode('', $data);
-        }, $this->doNgram(str_split($data), $n, $cyclic));
+        foreach ($this->doNgrams(str_split($data), $n, $cyclic) as $data) {
+            yield implode('', $data);
+        }
     }
 
     /**
      * @param $data
      * @param $n
      * @param $cyclic
      *
-     * @return array
+     * @return \Generator
      */
-    private function doNgram($data, $n, $cyclic)
+    private function doNgrams($data, $n = 1, $cyclic = true)
     {
         $dataLength = count($data);
 
         $n = $n > $dataLength ? $dataLength : $n;
 
-        foreach ($data as $key => $token) {
-            for ($i = $key; $i < $n + $key; $i++) {
+        $length = (false === $cyclic ? $dataLength - $n + 1 : $dataLength);
+
+        for ($j = 0; $j < $length; $j++) {
+            $ngrams = [];
+            for ($i = $j; $i < $n + $j; $i++) {
                 $ngrams[] = $data[$i%$dataLength];
             }
+            yield $ngrams;
         }
-
-        $ngrams = array_chunk($ngrams, $n);
-
-        return (false === $cyclic ? array_slice($ngrams, 0, $dataLength - $n + 1) : $ngrams);
     }
 }