first import

2013-03-12 15:26:34 +01:00
commit 6a402551d7
109 changed files with 32515 additions and 0 deletions
@@ -0,0 +1,185 @@
+<?php
+
+/**
+ * Logic around Solr field schema information.
+ */
+class SearchApiSolrField {
+
+  /**
+   * @var array
+   *   Human-readable labels for Solr schema properties.
+   */
+  public static $schemaLabels = array(
+    'I' => 'Indexed',
+    'T' => 'Tokenized',
+    'S' => 'Stored',
+    'M' => 'Multivalued',
+    'V' => 'TermVector Stored',
+    'o' => 'Store Offset With TermVector',
+    'p' => 'Store Position With TermVector',
+    'O' => 'Omit Norms',
+    'L' => 'Lazy',
+    'B' => 'Binary',
+    'C' => 'Compressed',
+    'f' => 'Sort Missing First',
+    'l' => 'Sort Missing Last',
+  );
+
+  /**
+   * @var stdclass
+   *   The original field object.
+   */
+  protected $field;
+
+  /**
+   * @var array
+   *   An array of schema properties for this field. This will be a subset of
+   *   the SearchApiSolrField::schemaLabels array.
+   */
+  protected $schema;
+
+  /**
+   * Constructor.
+   *
+   * @param stdClass $field
+   *   A field object from Solr's "Luke" servlet.
+   */
+  public function __construct($field) {
+    $this->field = $field;
+  }
+
+  /**
+   * Get the type of the Solr field, according to the Solr schema.
+   *
+   * Note that field types like "text", "boolean", and "date" are conventions,
+   * but their presence and behavior are entirely determined by the particular
+   * schema.xml file used by a Solr core.
+   *
+   * @return string
+   *   The type of the Solr field.
+   */
+  public function getType() {
+    return $this->field->type;
+  }
+
+  /**
+   * Get an array of field properties.
+   *
+   * @return array
+   *   An array of properties describing the solr schema. The array keys are
+   *   single-character codes, and the values are human-readable labels. This
+   *   will be a subset of the SearchApiSolrField::schemaLabels array.
+   */
+  public function getSchema() {
+    if (!isset($this->schema)) {
+      foreach (str_split(str_replace('-', '', $this->field->schema)) as $key) {
+        $this->schema[$key] = isset(self::$schemaLabels[$key]) ? self::$schemaLabels[$key] : $key;
+      }
+    }
+    return $this->schema;
+  }
+
+  /**
+   * Get the "dynamic base" of this field.
+   *
+   * This typically looks like 'ss_*, and is used to aggregate fields based on
+   * "hungarian" naming conventions.
+   *
+   * @return string
+   *   The mask describing the solr aggregate field, if there is one.
+   */
+  public function getDynamicBase() {
+    return isset($this->field->dynamicBase) ? $this->field->dynamicBase : NULL;
+  }
+
+  /**
+   * Determine whether this field may be suitable for use as a key field.
+   *
+   * Unfortunately, it seems like the best way to find an actual uniqueKey field
+   * according to Solr is to examine the Solr core's schema.xml.
+   *
+   * @return boolean
+   *   Whether the field is suitable for use as a key.
+   */
+  public function isPossibleKey() {
+    return !$this->getDynamicBase()
+      && !in_array($this->getType(), array('boolean', 'date', 'text'))
+      && $this->isStored()
+      && !$this->isMultivalued();
+  }
+
+  /**
+   * Determine whether a field is suitable for sorting.
+   *
+   * In order for a field to yield useful sorted results in Solr, it must be
+   * indexed, not multivalued, and not tokenized. It's ok if a field is
+   * tokenized and yields only one token, but there's no general way to check
+   * for that.
+   *
+   * @return boolean
+   *   Whether the field is suitable for sorting.
+   */
+  public function isSortable() {
+    return $this->isIndexed()
+      && !$this->isMultivalued()
+      && !$this->isTokenized();
+  }
+
+  /**
+   * The following functions return information about specific properties of this field.
+   *
+   * @return boolean
+   */
+  public function isIndexed() {
+    $this->getSchema();
+    return isset($this->schema['I']);
+  }
+  public function isTokenized() {
+    $this->getSchema();
+    return isset($this->schema['T']);
+  }
+  public function isStored() {
+    $this->getSchema();
+    return isset($this->schema['S']);
+  }
+  public function isMultivalued() {
+    $this->getSchema();
+    return isset($this->schema['M']);
+  }
+  public function isTermVectorStored() {
+    $this->getSchema();
+    return isset($this->schema['V']);
+  }
+  public function isStoreOffsetWithTermVector() {
+    $this->getSchema();
+    return isset($this->schema['o']);
+  }
+  public function isStorePositionWithTermVector() {
+    $this->getSchema();
+    return isset($this->schema['p']);
+  }
+  public function isOmitNorms() {
+    $this->getSchema();
+    return isset($this->schema['O']);
+  }
+  public function isLazy() {
+    $this->getSchema();
+    return isset($this->schema['L']);
+  }
+  public function isBinary() {
+    $this->getSchema();
+    return isset($this->schema['B']);
+  }
+  public function isCompressed() {
+    $this->getSchema();
+    return isset($this->schema['C']);
+  }
+  public function isSortMissingFirst() {
+    $this->getSchema();
+    return isset($this->schema['f']);
+  }
+  public function isSortMissingLast() {
+    $this->getSchema();
+    return isset($this->schema['l']);
+  }
+}