Validate json integrity (#92)

* validate data integrity for json columns * added comment --------- Co-authored-by: Kevin Tabb <ktabb@Kevins-MacBook-Pro.local>
slackhq · Feb 6, 2023 · 246d1f9 · 246d1f9
1 parent ecdb48e
commit 246d1f9
Show file tree

Hide file tree

Showing 3 changed files with 84 additions and 1 deletion.
diff --git a/src/DataIntegrity.php b/src/DataIntegrity.php
@@ -179,7 +179,28 @@ public static function ensureFieldsPresent(dict<string, mixed> $row, table_schem
               }
             } else {
               $field_value = (string)$row[$field_name];
-              if ($field_length > 0 && \mb_strlen($field_value) > $field_length) {
+
+              // handle json column type validation
+              if ($field_mysql_type === DataType::JSON) {
+                // null is okay
+                if ($row[$field_name] is nonnull) {
+                  if (!Str\is_empty((string)$row[$field_name])) {
+                    // validate json string
+                    $json_obj = \json_decode((string)$row[$field_name]);
+                    if ($json_obj is null) {
+                      // invalid json
+                      throw new SQLFakeRuntimeException(
+                        "Invalid value '{$field_value}' for column '{$field_name}' on '{$schema['name']}', expected json",
+                      );
+                    }
+                  } else {
+                    // empty strings are not valid for json columns
+                    throw new SQLFakeRuntimeException(
+                      "Invalid value '{$field_value}' for column '{$field_name}' on '{$schema['name']}', expected json",
+                    );
+                  }
+                }
+              } else if ($field_length > 0 && \mb_strlen($field_value) > $field_length) {
                 $field_str = \var_export($row[$field_name], true);
                 throw new SQLFakeRuntimeException(
                   "Invalid value '{$field_str}' for column '{$field_name}' on '{$schema['name']}', expected string of size {$field_length}",

diff --git a/tests/InsertQueryTest.php b/tests/InsertQueryTest.php
@@ -240,6 +240,42 @@ final class InsertQueryTest extends HackTest {
       );
   }
 
+  public async function testEmptyStringInsertIntoJsonColumn(): Awaitable<void> {
+    $conn = static::$conn as nonnull;
+    QueryContext::$strictSQLMode = true;
+    expect(() ==> $conn->query("INSERT INTO table_with_json (id, data) VALUES (1, '')"))
+      ->toThrow(
+        SQLFakeRuntimeException::class,
+        "Invalid value '' for column 'data' on 'table_with_json', expected json",
+      );
+  }
+
+  public async function testInvalidJsonStringInsertIntoJsonColumn(): Awaitable<void> {
+    $conn = static::$conn as nonnull;
+    QueryContext::$strictSQLMode = true;
+    expect(() ==> $conn->query("INSERT INTO table_with_json (id, data) VALUES (1, 'abc')"))
+      ->toThrow(
+        SQLFakeRuntimeException::class,
+        "Invalid value 'abc' for column 'data' on 'table_with_json', expected json",
+      );
+  }
+
+  public async function testNullInsertIntoJsonColumn(): Awaitable<void> {
+    $conn = static::$conn as nonnull;
+    QueryContext::$strictSQLMode = true;
+    await $conn->query("INSERT INTO table_with_json (id, data) VALUES (1, NULL)");
+    $result = await $conn->query("SELECT * FROM table_with_json");
+    expect($result->rows())->toBeSame(vec[dict['id' => 1, 'data' => null]]);
+  }
+
+  public async function testValidJsonInsertIntoJsonColumn(): Awaitable<void> {
+    $conn = static::$conn as nonnull;
+    QueryContext::$strictSQLMode = true;
+    await $conn->query("INSERT INTO table_with_json (id, data) VALUES (1, '{\"test\":123}')");
+    $result = await $conn->query("SELECT * FROM table_with_json");
+    expect($result->rows())->toBeSame(vec[dict['id' => 1, 'data' => '{"test":123}']]);
+  }
+
   public async function testDupeInsertNoConflicts(): Awaitable<void> {
     $conn = static::$conn as nonnull;
     await $conn->query(

diff --git a/tests/SharedSetup.php b/tests/SharedSetup.php
@@ -162,6 +162,32 @@ final class SharedSetup {
 				),
 			],
 		),
+		'table_with_json' => shape(
+			'name' => 'table_with_json',
+			'fields' => vec[
+				shape(
+					'name' => 'id',
+					'type' => DataType::BIGINT,
+					'length' => 20,
+					'null' => false,
+					'hack_type' => 'int',
+				),
+				shape(
+					'name' => 'data',
+					'type' => DataType::JSON,
+					'length' => 255,
+					'null' => true,
+					'hack_type' => 'string',
+				),
+			],
+			'indexes' => vec[
+				shape(
+					'name' => 'PRIMARY',
+					'type' => 'PRIMARY',
+					'fields' => keyset['id'],
+				)
+			],
+		),
 		'table_with_more_fields' => shape(
 			'name' => 'table_with_more_fields',
 			'fields' => vec[