Fix algorithm implementation (#2)

* Added failing test * Simplified code a little * Fix getReachableStates() method (#3) * Defensive programming * Remove redundant call * Cleanup tests --------- Co-authored-by: Martin Auswöger <[email protected]>
Toflar · Oct 6, 2023 · 27fe48f · 27fe48f
1 parent 4abf06e
commit 27fe48f
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 10 deletions.
diff --git a/src/StateSetIndex.php b/src/StateSetIndex.php
@@ -82,7 +82,7 @@ public function findMatchingStates(string $string, int $editDistance): array
         // Initial states
         $states = $this->getReachableStates(0, $editDistance);
 
-        $this->loopOverEveryCharacter($string, function (int $mappedChar, $char) use (&$states, $editDistance) {
+        $this->loopOverEveryCharacter($string, function (int $mappedChar) use (&$states, $editDistance) {
             $statesStar = new CostAnnotatedStateSet(); // This is S∗ in the paper
 
             foreach ($states->all() as $state => $cost) {
@@ -98,7 +98,7 @@ public function findMatchingStates(string $string, int $editDistance): array
                     $newState = (int) ($state * $this->config->getAlphabetSize() + $i);
 
                     if ($this->stateSet->has($newState)) {
-                        if ($i === $this->getAlphabet()->map($char, $this->config->getAlphabetSize())) {
+                        if ($i === $mappedChar) {
                             // Match
                             $statesStarC->add($newState, $cost);
                         } elseif ($cost + 1 <= $editDistance) {
@@ -181,12 +181,14 @@ private function getReachableStates(int $startState, int $editDistance, int $cur
         // A state is always able to reach itself
         $reachable->add($startState, $currentDistance);
 
-        for ($i = 0; $i <= $editDistance; $i++) {
-            for ($c = 0; $c < $this->config->getAlphabetSize(); $c++) {
-                $state = $startState + $c * $i;
-                if ($this->stateSet->has($state)) {
-                    $reachable->add($startState, $currentDistance);
-                }
+        if ($currentDistance >= $editDistance) {
+            return $reachable;
+        }
+
+        for ($c = 1; $c <= $this->config->getAlphabetSize(); $c++) {
+            $state = $startState * $this->config->getAlphabetSize() + $c;
+            if ($this->stateSet->has($state)) {
+                $reachable = $reachable->mergeWith($this->getReachableStates($state, $editDistance, $currentDistance + 1));
             }
         }
 
@@ -203,7 +205,7 @@ private function loopOverEveryCharacter(string $string, \Closure $closure): void
 
         foreach (mb_str_split($indexedSubstring) as $char) {
             $mappedChar = $this->alphabet->map($char, $this->config->getAlphabetSize());
-            $closure($mappedChar, $char);
+            $closure($mappedChar);
         }
     }
 }
diff --git a/tests/StateSetIndexTest.php b/tests/StateSetIndexTest.php
@@ -49,4 +49,17 @@ public function testWithUtf8Alphabet(): void
         $this->assertSame([2710 => ['Mueller'], 2843 => ['Muster', 'Mustermann']], $stateSetIndex->findAcceptedStrings('Mustre', 2));
         $this->assertSame(['Muster'], $stateSetIndex->find('Mustre', 2));
     }
-}
+
+    /**
+     * This use case occurred while testing 2.0.0, which is why this is added as additional test case.
+     */
+    public function testAssassinCanBeFound(): void
+    {
+        $stateSetIndex = new StateSetIndex(new Config(14, 4), new Utf8Alphabet(), new InMemoryStateSet(), new InMemoryDataStore());
+        $stateSetIndex->index(['assassin']);
+
+        $this->assertSame([844, 3380, 13522, 54091], $stateSetIndex->findMatchingStates('assasin', 2));
+        $this->assertSame([54091 => ['assassin']], $stateSetIndex->findAcceptedStrings('assasin', 2));
+        $this->assertSame(['assassin'], $stateSetIndex->find('assasin', 2));
+    }
+}