public function Migration::analyze in Migrate 6.2

Same name and namespace in other branches

1. 7.2 includes/migration.inc \Migration::analyze()

Perform an analysis operation - report on field values in the source.

Return value

array Array of analysis details - each element is keyed by field name and contains an array describing the field values.

File

includes/migration.inc, line 637

Defines the base class for import/rollback processes.

Class

Migration

The base class for all import objects. This is where most of the smarts of the migrate module resides. Migrations are created by deriving from this class, and in the constructor (after calling parent::__construct()) initializing at a minimum the name,…

Code

public function analyze() {

  // The source needs this to find the map table.
  self::$currentMigration = $this;
  try {
    $this->source
      ->rewind();
  } catch (Exception $e) {
    self::displayMessage(t('Migration analysis failed with source plugin exception: !e', array(
      '!e' => $e
        ->getMessage(),
    )));
    self::$currentMigration = NULL;
    return array();
  }

  // Get the documented fields first
  $source_fields = $this->source
    ->fields();
  $analysis = array();
  $field_init = array(
    'is_numeric' => TRUE,
    'min_numeric' => NULL,
    'max_numeric' => NULL,
    'min_strlen' => 0,
    'max_strlen' => 0,
    'distinct_values' => array(),
  );
  foreach ($source_fields as $field_name => $description) {

    // Ignore fields from the map table
    if (substr($field_name, 0, strlen('migrate_map_')) == 'migrate_map_') {
      continue;
    }
    $analysis[$field_name] = $field_init + array(
      'description' => $description,
    );
  }

  // For each data row...
  while ($this->source
    ->valid()) {
    $row = $this->source
      ->current();

    // Cheat for XML migrations, which don't pick up the source values
    // until applyMappings() applies the xpath()
    if (is_a($this, 'XMLMigration')) {
      $this->currentSourceKey = $this->source
        ->getCurrentKey();
      $this->sourceValues = $row;
      $this
        ->applyMappings();
      $row = $this->sourceValues;
      unset($row->xml);
    }

    // For each field in this row...
    foreach ($row as $field_name => $raw_value) {

      // Ignore fields from the map table
      if (substr($field_name, 0, strlen('migrate_map_')) == 'migrate_map_') {
        continue;
      }

      // It might be an array of values, check each value
      if (!is_array($raw_value)) {
        $raw_value = array(
          $raw_value,
        );
      }
      foreach ($raw_value as $field_value) {

        // If this is an undocumented field, initialize it
        if (!isset($analysis[$field_name])) {
          $analysis[$field_name] = $field_init + array(
            'description' => '',
          );
        }

        // Ignore leading/trailing spaces in determing numerics
        $trimmed_value = trim($field_value);
        if (is_numeric($trimmed_value)) {
          $trimmed_value = floatval($trimmed_value);

          // First numeric value, initialize the min/max
          if (is_null($analysis[$field_name]['min_numeric'])) {
            $analysis[$field_name]['min_numeric'] = $trimmed_value;
            $analysis[$field_name]['max_numeric'] = $trimmed_value;
          }
          else {
            $analysis[$field_name]['min_numeric'] = min($trimmed_value, $analysis[$field_name]['min_numeric']);
            $analysis[$field_name]['max_numeric'] = max($trimmed_value, $analysis[$field_name]['max_numeric']);
          }
        }
        elseif ($trimmed_value !== '') {

          // Empty strings are !is_numeric(), but treat as empty rather than
          // assuming we don't have a numeric field
          $analysis[$field_name]['is_numeric'] = FALSE;
        }
        $strlen = strlen($field_value);

        // First string value, initialize both min and max
        if ($analysis[$field_name]['max_strlen'] == 0) {
          $analysis[$field_name]['min_strlen'] = $strlen;
          $analysis[$field_name]['max_strlen'] = $strlen;
        }
        else {
          $analysis[$field_name]['min_strlen'] = min($strlen, $analysis[$field_name]['min_strlen']);
          $analysis[$field_name]['max_strlen'] = max($strlen, $analysis[$field_name]['max_strlen']);
        }

        // Track up to 10 distinct values
        if (count($analysis[$field_name]['distinct_values']) <= 10) {
          $analysis[$field_name]['distinct_values'][$trimmed_value]++;
        }
      }
    }
    try {
      $this->source
        ->next();
    } catch (Exception $e) {
      self::displayMessage(t('Migration analysis failed with source plugin exception: !e. Partial results follow:', array(
        '!e' => $e
          ->getMessage(),
      )));
      self::$currentMigration = NULL;
      return $analysis;
    }
  }
  self::$currentMigration = NULL;
  return $analysis;
}