Support removing strings from the index (#7)

Author: ausi

Diff for: src/DataStore/DataStoreInterface.php

@@ -6,6 +6,8 @@ interface DataStoreInterface
 {
     public function add(int $state, string $string): void;
 
+    public function remove(int $state, string $string): void;
+
     /**
      * Returns the matching strings per state. Key is the state and the value is an array of matching strings
      * for that state.

Diff for: src/DataStore/InMemoryDataStore.php

@@ -14,6 +14,17 @@ public function add(int $state, string $string): void
         $this->data[$state][] = $string;
     }
 
+    public function remove(int $state, string $string): void
+    {
+        $updated = array_values(array_diff($this->data[$state] ?? [], [$string]));
+
+        if ($updated) {
+            $this->data[$state] = $updated;
+        } else {
+            unset($this->data[$state]);
+        }
+    }
+
     public function all(): array
     {
         return $this->data;

Diff for: src/DataStore/NullDataStore.php

@@ -9,6 +9,11 @@ public function add(int $state, string $string): void
         // noop
     }
 
+    public function remove(int $state, string $string): void
+    {
+        // noop
+    }
+
     public function getForStates(array $states = []): array
     {
         return [];

Diff for: src/StateSet/InMemoryStateSet.php

@@ -17,6 +17,11 @@ public function add(int $state): void
         $this->states[$state] = true;
     }
 
+    public function remove(int $state): void
+    {
+        unset($this->states[$state]);
+    }
+
     public function all(): array
     {
         return array_keys($this->states);

Diff for: src/StateSet/StateSetInterface.php

@@ -6,6 +6,8 @@ interface StateSetInterface
 {
     public function add(int $state): void;
 
+    public function remove(int $state): void;
+
     /**
      * @return array<int>
      */

Diff for: src/StateSetIndex.php

@@ -170,6 +170,48 @@ public function index(array $strings): array
         return $assigned;
     }
 
+    /**
+     * Removes an array of strings from the index.
+     */
+    public function removeFromIndex(array $strings): void
+    {
+        foreach ($strings as $string) {
+            unset($this->indexCache[$string]);
+
+            $states = [];
+            $state = 0;
+            $this->loopOverEveryCharacter($string, function (int $mappedChar) use (&$state, &$states) {
+                $states[] = $state = (int) ($state * $this->config->getAlphabetSize() + $mappedChar);
+            });
+
+            $this->dataStore->remove($state, $string);
+
+            foreach (array_reverse($states) as $state) {
+                // If a state is shared with another string or a state exists that follows the current one we must stop
+                // the removal process as all previous states and the current one must be kept.
+                if (isset($this->dataStore->getForStates([$state])[$state]) || $this->hasNextState($state)) {
+                    continue 2;
+                }
+
+                $this->stateSet->remove($state);
+            }
+        }
+    }
+
+    /**
+     * Returns true if a state exists that follows the given state
+     */
+    private function hasNextState(int $state): bool
+    {
+        for ($c = 1; $c <= $this->config->getAlphabetSize(); ++$c) {
+            if ($this->stateSet->has($state * $this->config->getAlphabetSize() + $c)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
     private function getReachableStates(int $startState, int $editDistance, int $currentDistance = 0): CostAnnotatedStateSet
     {
         $reachable = new CostAnnotatedStateSet();

Diff for: tests/StateSetIndexTest.php

@@ -62,4 +62,59 @@ public function testAssassinCanBeFound(): void
         $this->assertSame([54091 => ['assassin']], $stateSetIndex->findAcceptedStrings('assasin', 2));
         $this->assertSame(['assassin'], $stateSetIndex->find('assasin', 2));
     }
+
+    public function testRemoveFromIndex(): void
+    {
+        $stateSetIndex = new StateSetIndex(new Config(6, 4), new Utf8Alphabet(), new InMemoryStateSet(), new InMemoryDataStore());
+        $stateSetIndex->index(['Mueller']);
+
+        $onlyMuellerStates = $stateSetIndex->getStateSet()->all();
+
+        $stateSetIndex->removeFromIndex(['Mueller']);
+
+        $this->assertSame([], $stateSetIndex->getStateSet()->all());
+
+        $stateSetIndex->index(['Müller', 'Muentner', 'Muster', 'Mustermann', 'Mueller']);
+        $stateSetIndex->removeFromIndex(['Müller', 'Muentner', 'Muster', 'Mustermann']);
+
+        $this->assertEquals($onlyMuellerStates, $stateSetIndex->getStateSet()->all());
+        $this->assertSame(['Mueller'], $stateSetIndex->find('Mueler', 1));
+    }
+
+    public function testRemoveFromFullIndex(): void
+    {
+        $stateSetIndex = new StateSetIndex(new Config(5, 4), new Utf8Alphabet(), new InMemoryStateSet(), new InMemoryDataStore());
+        $stateSetIndex->index(['Mueller']);
+
+        $onlyMuellerStates = $stateSetIndex->getStateSet()->all();
+
+        $stateSetIndex->removeFromIndex(['Mueller']);
+
+        $this->assertSame([], $stateSetIndex->getStateSet()->all());
+
+        for ($i = 0; $i < $stateSetIndex->getConfig()->getAlphabetSize(); ++$i) {
+            $strings[] = \IntlChar::chr(97 + $i);
+        }
+
+        for ($length = 1; $length <= $stateSetIndex->getConfig()->getIndexLength(); ++$length) {
+            foreach ($strings as $string) {
+                for ($i = 0; $i < $stateSetIndex->getConfig()->getAlphabetSize(); ++$i) {
+                    $strings[] = $string . \IntlChar::chr(97 + $i);
+                }
+            }
+        }
+
+        // Fill every possible state for the configured length and size
+        $stateSetIndex->index($strings);
+        $stateSetIndex->index(['Mueller']);
+
+        $states = $stateSetIndex->getStateSet()->all();
+        sort($states);
+
+        $this->assertSame(range(1, (((4 * 4 + 4) * 4 + 4) * 4 + 4) * 4 + 4), $states, 'No state should be missing');
+
+        $stateSetIndex->removeFromIndex($strings);
+
+        $this->assertEquals($onlyMuellerStates, $stateSetIndex->getStateSet()->all());
+    }
 }