From 91f1ab6272d3ba44dc1181eb0adf5d32f1fed6f0 Mon Sep 17 00:00:00 2001
From: eileen <emcnaughton@wikimedia.org>
Date: Wed, 4 Nov 2020 11:07:37 +1300
Subject: [PATCH] dev/core#2165 Test & fix for Handle emojis less fatally where
 not supported

Fix emoji handling by returning always-false when unwelcomely present
---
 CRM/Core/BAO/SchemaHandler.php                | 19 +++++++++++-
 CRM/Core/DAO.php                              | 16 ++++++++++
 Civi/Test/Api3TestTrait.php                   |  2 +-
 .../phpunit/api/v4/Action/ContactGetTest.php  | 30 +++++++++++++++++++
 4 files changed, 65 insertions(+), 2 deletions(-)

diff --git a/CRM/Core/BAO/SchemaHandler.php b/CRM/Core/BAO/SchemaHandler.php
index cb56200c4b..f4e7892607 100644
--- a/CRM/Core/BAO/SchemaHandler.php
+++ b/CRM/Core/BAO/SchemaHandler.php
@@ -877,7 +877,7 @@ MODIFY      {$columnName} varchar( $length )
    *
    * @return string
    */
-  public static function getInUseCollation() {
+  public static function getInUseCollation(): string {
     if (!isset(\Civi::$statics[__CLASS__][__FUNCTION__])) {
       $dao = CRM_Core_DAO::executeQuery('SHOW TABLE STATUS LIKE \'civicrm_contact\'');
       $dao->fetch();
@@ -886,6 +886,23 @@ MODIFY      {$columnName} varchar( $length )
     return \Civi::$statics[__CLASS__][__FUNCTION__];
   }
 
+  /**
+   * Does the database support utf8mb4.
+   *
+   * Utf8mb4 is required to support emojis but older databases may not have it enabled.
+   *
+   * This is aggressively cached despite just being a string function
+   * as it is expected it might be called many times.
+   *
+   * @return bool
+   */
+  public static function databaseSupportsUTF8MB4(): bool {
+    if (!isset(\Civi::$statics[__CLASS__][__FUNCTION__])) {
+      \Civi::$statics[__CLASS__][__FUNCTION__] = stripos(self::getInUseCollation(), 'utf8mb4') === TRUE;
+    }
+    return \Civi::$statics[__CLASS__][__FUNCTION__];
+  }
+
   /**
    * Get the database collation.
    *
diff --git a/CRM/Core/DAO.php b/CRM/Core/DAO.php
index 13c0c8c65a..2d40afd073 100644
--- a/CRM/Core/DAO.php
+++ b/CRM/Core/DAO.php
@@ -2782,6 +2782,22 @@ SELECT contact_id
    */
   public static function createSQLFilter($fieldName, $filter, $type = NULL, $alias = NULL, $returnSanitisedArray = FALSE) {
     foreach ($filter as $operator => $criteria) {
+      if (!CRM_Core_BAO_SchemaHandler::databaseSupportsUTF8MB4()) {
+        foreach ((array) $criteria as $criterion) {
+          if (!empty($criterion) && !is_numeric($criterion)
+            // The first 2 criteria are redundant but are added as they
+            // seem like they would
+            // be quicker than this 3rd check.
+            && max(array_map('ord', str_split($criterion))) >= 240) {
+            // String contains unsupported emojis.
+            // We return a clause that resolves to false as an emoji string by definition cannot be saved.
+            // note that if we return just 0 for false if gets lost in empty checks.
+            // https://stackoverflow.com/questions/16496554/can-php-detect-4-byte-encoded-utf8-chars
+            return '0 = 1';
+          }
+        }
+      }
+
       if (in_array($operator, self::acceptedSQLOperators(), TRUE)) {
         switch ($operator) {
           // unary operators
diff --git a/Civi/Test/Api3TestTrait.php b/Civi/Test/Api3TestTrait.php
index 176564a686..f84c4c2f98 100644
--- a/Civi/Test/Api3TestTrait.php
+++ b/Civi/Test/Api3TestTrait.php
@@ -173,7 +173,7 @@ trait Api3TestTrait {
 
   /**
    * This function exists to wrap api getValue function & check the result
-   * so we can ensure they succeed & throw exceptions without litterering the test with checks
+   * so we can ensure they succeed & throw exceptions without littering the test with checks
    * There is a type check in this
    *
    * @param string $entity
diff --git a/tests/phpunit/api/v4/Action/ContactGetTest.php b/tests/phpunit/api/v4/Action/ContactGetTest.php
index 649928815e..965a8b88ee 100644
--- a/tests/phpunit/api/v4/Action/ContactGetTest.php
+++ b/tests/phpunit/api/v4/Action/ContactGetTest.php
@@ -102,4 +102,34 @@ class ContactGetTest extends \api\v4\UnitTestCase {
     $this->assertTrue(!empty($limit1->single()['sort_name']));
   }
 
+  /**
+   * Test a lack of fatal errors when the where contains an emoji.
+   *
+   * By default our DBs are not 🦉 compliant. This test will age
+   * out when we are.
+   *
+   * @throws \API_Exception
+   */
+  public function testEmoji(): void {
+    $schemaNeedsAlter = \CRM_Core_BAO_SchemaHandler::databaseSupportsUTF8MB4();
+    if ($schemaNeedsAlter) {
+      \CRM_Core_DAO::executeQuery("
+        ALTER TABLE civicrm_contact MODIFY COLUMN
+        `first_name` VARCHAR(64) CHARACTER SET utf8 COLLATE utf8_unicode_ci DEFAULT NULL COMMENT 'First Name.',
+        CHARSET utf8
+      ");
+    }
+    Contact::get()
+      ->setDebug(TRUE)
+      ->addWhere('first_name', '=', '🦉Claire')
+      ->execute();
+    if ($schemaNeedsAlter) {
+      \CRM_Core_DAO::executeQuery("
+        ALTER TABLE civicrm_contact MODIFY COLUMN
+        `first_name` VARCHAR(64) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci DEFAULT NULL COMMENT 'First Name.',
+        CHARSET utf8mb4
+      ");
+    }
+  }
+
 }
-- 
2.25.1