Merge pull request #8 from vyos/task/e5-buildindexmappings

E5: Add buildIndexMappings() helper

2 files changed, 201 insertions, 0 deletions

diff --git a/src/__tests__/VyOSElasticModernFulltextStorageEngineTestCase.php b/src/__tests__/VyOSElasticModernFulltextStorageEngineTestCase.php
index ff4ade8..e3daab4 100644
--- a/src/__tests__/VyOSElasticModernFulltextStorageEngineTestCase.php
+++ b/src/__tests__/VyOSElasticModernFulltextStorageEngineTestCase.php
@@ -112,4 +112,128 @@ final class VyOSElasticModernFulltextStorageEngineTestCase
       $filter);
   }
 
+  public function testBuildIndexMappingsShape() {
+    $engine = $this->newEngine()->setVersion(7);
+
+    $doc_types = array('TASK', 'DREV');
+    $fields = array('title', 'body', 'comment');
+    $relationships = array('authorPHID', 'projectPHID');
+    $mappings = $engine->buildIndexMappings(
+      $doc_types, $fields, $relationships, 'text');
+
+    // Single typeless mapping with one 'properties' block.
+    $this->assertTrue(isset($mappings['properties']));
+    $this->assertFalse(isset($mappings['TASK']));
+    $this->assertFalse(isset($mappings['DREV']));
+
+    // All three text fields have the multi-analyzer shape.
+    foreach (array('title', 'body', 'comment') as $field) {
+      $this->assertTrue(
+        isset($mappings['properties'][$field]),
+        pht('Field "%s" missing from mappings.', $field));
+      $this->assertEqual(
+        'text',
+        $mappings['properties'][$field]['type'],
+        pht('Field "%s" should be type text.', $field));
+      $this->assertTrue(
+        isset($mappings['properties'][$field]['fields']['raw']),
+        pht('Field "%s" missing raw sub-field.', $field));
+      $this->assertTrue(
+        isset($mappings['properties'][$field]['fields']['keywords']),
+        pht('Field "%s" missing keywords sub-field.', $field));
+      $this->assertTrue(
+        isset($mappings['properties'][$field]['fields']['stems']),
+        pht('Field "%s" missing stems sub-field.', $field));
+    }
+
+    // Both relationships emit as keyword fields with doc_values:false.
+    foreach (array('authorPHID', 'projectPHID') as $rel) {
+      $this->assertEqual(
+        'keyword',
+        $mappings['properties'][$rel]['type'],
+        pht('Relationship "%s" should be keyword type.', $rel));
+      $this->assertEqual(
+        false,
+        $mappings['properties'][$rel]['doc_values'],
+        pht('Relationship "%s" should have doc_values:false.', $rel));
+      $this->assertEqual(
+        'date',
+        $mappings['properties'][$rel.'_ts']['type'],
+        pht('Relationship "%s" missing timestamp field.', $rel));
+      // No include_in_all anywhere.
+      $this->assertFalse(
+        isset($mappings['properties'][$rel]['include_in_all']),
+        pht('Relationship "%s" should not have include_in_all.', $rel));
+    }
+
+    // documentType is a keyword field inside properties.
+    $this->assertEqual(
+      'keyword',
+      $mappings['properties']['documentType']['type']);
+
+    // Standard date fields present.
+    $this->assertEqual(
+      'date',
+      $mappings['properties']['dateCreated']['type']);
+    $this->assertEqual(
+      'date',
+      $mappings['properties']['lastModified']['type']);
+  }
+
+  public function testBuildIndexMappingsRejectsReservedFieldName() {
+    $engine = $this->newEngine()->setVersion(7);
+    $caught = null;
+    try {
+      $engine->buildIndexMappings(
+        array(), array('documentType'), array(), 'text');
+    } catch (Exception $e) {
+      $caught = $e;
+    }
+    $this->assertTrue(
+      $caught !== null,
+      pht('Expected exception when field name collides with reserved key.'));
+  }
+
+  public function testBuildIndexMappingsRejectsReservedRelationshipName() {
+    $engine = $this->newEngine()->setVersion(7);
+    $caught = null;
+    try {
+      $engine->buildIndexMappings(
+        array(), array(), array('lastModified'), 'text');
+    } catch (Exception $e) {
+      $caught = $e;
+    }
+    $this->assertTrue(
+      $caught !== null,
+      pht('Expected exception when relationship name collides with reserved key.'));
+  }
+
+  public function testBuildIndexMappingsRejectsFooTsClash() {
+    // A field named "foo_ts" would clash with the timestamp slot auto-generated
+    // for a relationship named "foo".
+    $engine = $this->newEngine()->setVersion(7);
+    $caught = null;
+    try {
+      $engine->buildIndexMappings(
+        array(), array('foo_ts'), array('foo'), 'text');
+    } catch (Exception $e) {
+      $caught = $e;
+    }
+    $this->assertTrue(
+      $caught !== null,
+      pht('Expected exception for field/relationship timestamp-slot collision.'));
+  }
+
+  public function testBuildIndexMappingsEmptyInputsYieldStandardFields() {
+    $engine = $this->newEngine()->setVersion(7);
+    $mappings = $engine->buildIndexMappings(array(), array(), array(), 'text');
+    $this->assertTrue(isset($mappings['properties']['documentType']));
+    $this->assertTrue(isset($mappings['properties']['dateCreated']));
+    $this->assertTrue(isset($mappings['properties']['lastModified']));
+    $this->assertEqual(
+      'keyword', $mappings['properties']['documentType']['type']);
+    $this->assertEqual('date', $mappings['properties']['dateCreated']['type']);
+    $this->assertEqual('date', $mappings['properties']['lastModified']['type']);
+  }
+
 }
diff --git a/src/engine/VyOSElasticModernFulltextStorageEngine.php b/src/engine/VyOSElasticModernFulltextStorageEngine.php
index 316a02b..fe88adc 100644
--- a/src/engine/VyOSElasticModernFulltextStorageEngine.php
+++ b/src/engine/VyOSElasticModernFulltextStorageEngine.php
@@ -48,6 +48,83 @@ abstract class VyOSElasticModernFulltextStorageEngine
     );
   }
 
+  public function buildIndexMappings(
+    array $doc_types, array $fields, array $relationships, $text_type) {
+
+    // These are emitted as fixed standard fields at the end of the mapping.
+    // Caller-supplied $fields or $relationships must not shadow them.
+    static $reserved = array('documentType', 'dateCreated', 'lastModified');
+
+    $rel_ts_keys = array_map(function($r) { return $r.'_ts'; }, $relationships);
+    $all_caller_keys = array_merge($fields, $relationships, $rel_ts_keys);
+
+    // Check for caller-supplied names that shadow reserved fields.
+    $collisions = array_intersect($all_caller_keys, $reserved);
+    if ($collisions) {
+      throw new Exception(
+        pht(
+          'buildIndexMappings(): caller-supplied field(s) "%s" collide with '.
+          'reserved mapping keys.',
+          implode('", "', array_values($collisions))));
+    }
+
+    // Check for duplicates within caller-supplied keys themselves
+    // (e.g. a field named "foo_ts" that would clash with relationship "foo"'s
+    // implicit timestamp slot).
+    $counts = array_count_values($all_caller_keys);
+    $duplicates = array_keys(array_filter($counts, function($c) { return $c > 1; }));
+    if ($duplicates) {
+      throw new Exception(
+        pht(
+          'buildIndexMappings(): caller-supplied key(s) "%s" appear more '.
+          'than once (check for field/relationship/timestamp-slot collisions).',
+          implode('", "', $duplicates)));
+    }
+
+    $properties = array();
+
+    foreach ($fields as $field) {
+      $properties[$field] = array(
+        'type'   => $text_type,
+        'fields' => array(
+          'raw' => array(
+            'type'                  => $text_type,
+            'analyzer'              => 'english_exact',
+            'search_analyzer'       => 'english',
+            'search_quote_analyzer' => 'english_exact',
+          ),
+          'keywords' => array(
+            'type'     => $text_type,
+            'analyzer' => 'letter_stop',
+          ),
+          'stems' => array(
+            'type'     => $text_type,
+            'analyzer' => 'english_stem',
+          ),
+        ),
+      );
+    }
+
+    foreach ($relationships as $rel) {
+      $properties[$rel] = array(
+        'type'       => 'keyword',
+        'doc_values' => false,
+      );
+      $properties[$rel.'_ts'] = array(
+        'type' => 'date',
+      );
+    }
+
+    $properties['documentType'] = array('type' => 'keyword');
+    $properties['dateCreated']  = array('type' => 'date');
+    $properties['lastModified'] = array('type' => 'date');
+
+    // The $doc_types parameter is part of the signature for symmetry
+    // with the bundled engine's per-type loop, but the typeless API
+    // emits one mapping shared across all doc types.
+    return array('properties' => $properties);
+  }
+
   public function getEngineIdentifier() {
     return 'elasticsearch-modern';
   }