Changeset 5101

Timestamp:

03/07/2017 01:28:40 AM (9 years ago)

Author:

iandunn

Message:

Events API: Extract country IDs from multi-word inputs to improve matching

This enables successfully matching locations for country codes in inputs like "GB" and "London GB"; and also matching country names of varying length, regardless of the city length, in inputs like "Vancouver Canada", "Santiago De Los Caballeros, Dominican Republic", and "Kaga-Bandoro, Central African Republic".

Location:

sites/trunk/api.wordpress.org/public_html/events/1.0

Files:

• index.php (modified) (2 diffs)
• tests/test-index.php (modified) (2 diffs)

sites/trunk/api.wordpress.org/public_html/events/1.0/index.php

@@ -271 +271 @@
  * Guess the location based on a country identifier inside the given input
  *
+ * This isn't perfect because some of the country names in the database are in a format that regular
+ * people wouldn't type -- e.g., "Venezuela, Bolvarian Republic Of" -- but this will still match a
+ * majority of them.
+ *
+ * Currently, this only works with English names because that's the only data we have.
+ *
  * @param string $location_name
  *
@@ -276 +282 @@
  */
 function guess_location_from_country( $location_name ) {
+        // Check if they entered only the country name, e.g. "Germany" or "New Zealand"
+        $country_code = get_country_code_from_name( $location_name );
+        $location_word_count = str_word_count( $location_name );
+        $location_name_parts = explode( ' ', $location_name );
+
+        // Check if they entered only the country code, e.g., "GB"
+        if ( ! $country_code ) {
+                $valid_country_codes = get_valid_country_codes();
+
+                if ( in_array( $location_name, $valid_country_codes, true ) ) {
+                        $country_code = $location_name;
+                }
+        }
+
         /*
-         * Check if they entered only the country name, e.g. "Germany" or "New Zealand"
-         *
-         * This isn't perfect because some of the country names in the database are in a format that regular
-         * people wouldn't type -- e.g., "Venezuela, Bolvarian Republic Of" -- but this will still match a
-         * majority of them.
-         *
-         * Currently, this only works with English names because that's the only data we have.
+         * Multi-word queries may contain cities, regions, and countries, so try to extract just the country
          */
-        $location_country_code = get_country_code_from_name( $location_name );
-
-        return $location_country_code;
+        if ( ! $country_code && $location_word_count >= 2 ) {
+                // Catch input like "Vancouver Canada"
+                $country_id   = $location_name_parts[ $location_word_count - 1 ];
+                $country_code = get_country_code_from_name( $country_id );
+
+                // Catch input like "London GB"
+                if ( ! $country_code ) {
+                        if ( in_array( $country_id, $valid_country_codes, true ) ) {
+                                $country_code = $country_id;
+                        }
+                }
+        }
+
+        if ( ! $country_code && $location_word_count >= 3 ) {
+                // Catch input like "Santiago De Los Caballeros, Dominican Republic"
+                $country_name = sprintf(
+                        '%s %s',
+                        $location_name_parts[ $location_word_count - 2 ],
+                        $location_name_parts[ $location_word_count - 1 ]
+                );
+                $country_code = get_country_code_from_name( $country_name );
+        }
+
+        if ( ! $country_code && $location_word_count >= 4 ) {
+                // Catch input like "Kaga-Bandoro, Central African Republic"
+                $country_name = sprintf(
+                        '%s %s %s',
+                        $location_name_parts[ $location_word_count - 3 ],
+                        $location_name_parts[ $location_word_count - 2 ],
+                        $location_name_parts[ $location_word_count - 1 ]
+                );
+                $country_code = get_country_code_from_name( $country_name );
+        }
+
+        return $country_code;
+}
+
+/**
+ * Get a list of valid country codes
+ *
+ * @return array
+ */
+function get_valid_country_codes() {
+        global $wpdb;
+
+        return $wpdb->get_col( "SELECT DISTINCT country FROM geoname" );
 }
 

sites/trunk/api.wordpress.org/public_html/events/1.0/tests/test-index.php

@@ -427 +427 @@
                 /*
                  * A combination of city, region, and country are given, along with the locale and timezone
+                 *
+                 * InvalidCity is used in tests that want to bypass the guess_location_from_city() tests and only test the country
                  */
                 '1-word-city-region' => array(
@@ -456 +458 @@
                 ),
 
+                'city-1-word-country' => array(
+                        'input' => array(
+                                'location_name' => 'InvalidCity Canada',
+                                'locale'        => 'en_CA',
+                                'timezone'      => 'America/Vancouver',
+                        ),
+                        'expected' => array(
+                                'country' => 'CA',
+                        ),
+                ),
+
+                'city-2-word-country' => array(
+                        'input' => array(
+                                'location_name' => 'InvalidCity Dominican Republic',
+                                'locale'        => 'es_ES',
+                                'timezone'      => 'America/Santo_Domingo',
+                        ),
+                        'expected' => array(
+                                'country' => 'DO',
+                        ),
+                ),
+
+                'city-3-word-country' => array(
+                        'input' => array(
+                                'location_name' => 'InvalidCity Central African Republic',
+                                'locale'        => 'fr_FR',
+                                'timezone'      => 'Africa/Bangui',
+                        ),
+                        'expected' => array(
+                                'country' => 'CF',
+                        ),
+                ),
+
+                'country-code' => array(
+                        'input' => array(
+                                'location_name' => 'GB',
+                                'locale'        => 'en_GB',
+                                'timezone'      => 'Europe/London',
+                        ),
+                        'expected' => array(
+                                'country' => 'GB',
+                        ),
+                ),
+
+                'city-country-code' => array(
+                        'input' => array(
+                                'location_name' => 'InvalidCity BI',
+                                'locale'        => 'fr_FR',
+                                'timezone'      => 'Africa/Bujumbura',
+                        ),
+                        'expected' => array(
+                                'country' => 'BI',
+                        ),
+                ),
+
 
                 /*