[PDO] fix: Explicitly convert PDO queries to utf-8 (#296)

* Explicitly convert query to utf-8 * Fix left side is not undefined * Add a polyfill and suggest a ext-mbstring * Add tests that check for utf-8 encoding
open-telemetry · Oct 2, 2024 · 3dd4f06 · 3dd4f06
1 parent 6123ef4
commit 3dd4f06
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 4 deletions.
diff --git a/src/Instrumentation/PDO/composer.json b/src/Instrumentation/PDO/composer.json
@@ -13,7 +13,11 @@
     "ext-pdo": "*",
     "ext-opentelemetry": "*",
     "open-telemetry/api": "^1.0",
-    "open-telemetry/sem-conv": "^1.24"
+    "open-telemetry/sem-conv": "^1.24",
+    "symfony/polyfill-mbstring": "^1.31"
+  },
+  "suggest": {
+    "ext-mbstring": "For better performance than symfony/polyfill-mbstring"
   },
   "require-dev": {
     "friendsofphp/php-cs-fixer": "^3",

diff --git a/src/Instrumentation/PDO/src/PDOInstrumentation.php b/src/Instrumentation/PDO/src/PDOInstrumentation.php
@@ -70,7 +70,7 @@ public static function register(): void
                 $builder = self::makeBuilder($instrumentation, 'PDO::query', $function, $class, $filename, $lineno)
                     ->setSpanKind(SpanKind::KIND_CLIENT);
                 if ($class === PDO::class) {
-                    $builder->setAttribute(TraceAttributes::DB_STATEMENT, $params[0] ?? 'undefined');
+                    $builder->setAttribute(TraceAttributes::DB_STATEMENT, mb_convert_encoding($params[0] ?? 'undefined', 'UTF-8'));
                 }
                 $parent = Context::getCurrent();
                 $span = $builder->startSpan();
@@ -93,7 +93,7 @@ public static function register(): void
                 $builder = self::makeBuilder($instrumentation, 'PDO::exec', $function, $class, $filename, $lineno)
                     ->setSpanKind(SpanKind::KIND_CLIENT);
                 if ($class === PDO::class) {
-                    $builder->setAttribute(TraceAttributes::DB_STATEMENT, $params[0] ?? 'undefined');
+                    $builder->setAttribute(TraceAttributes::DB_STATEMENT, mb_convert_encoding($params[0] ?? 'undefined', 'UTF-8'));
                 }
                 $parent = Context::getCurrent();
                 $span = $builder->startSpan();
@@ -116,7 +116,7 @@ public static function register(): void
                 $builder = self::makeBuilder($instrumentation, 'PDO::prepare', $function, $class, $filename, $lineno)
                     ->setSpanKind(SpanKind::KIND_CLIENT);
                 if ($class === PDO::class) {
-                    $builder->setAttribute(TraceAttributes::DB_STATEMENT, $params[0] ?? 'undefined');
+                    $builder->setAttribute(TraceAttributes::DB_STATEMENT, mb_convert_encoding($params[0] ?? 'undefined', 'UTF-8'));
                 }
                 $parent = Context::getCurrent();
                 $span = $builder->startSpan();

diff --git a/src/Instrumentation/PDO/tests/Integration/PDOInstrumentationTest.php b/src/Instrumentation/PDO/tests/Integration/PDOInstrumentationTest.php
@@ -187,4 +187,31 @@ public function test_execute_and_fetch_all_spans_linked_to_prepared_statement_sp
         $this->assertCount(1, $fetchAllSpan->getLinks());
         $this->assertEquals($prepareSpan->getContext(), $fetchAllSpan->getLinks()[0]->getSpanContext());
     }
+
+    /**
+     * Tests that the db statement is encoded as UTF-8, this is relevant for grpc exporter which expects UTF-8 encoding.
+     */
+    public function test_encode_db_statement_as_utf8(): void
+    {
+        //setup
+        $db = self::createDB();
+        $db->exec($this->fillDB());
+
+        $non_utf8_id = mb_convert_encoding('rückwärts', 'ISO-8859-1', 'UTF-8');
+
+        $db->prepare("SELECT id FROM technology WHERE id = '{$non_utf8_id}'");
+        $span_db_prepare = $this->storage->offsetGet(2);
+        $this->assertTrue(mb_check_encoding($span_db_prepare->getAttributes()->get(TraceAttributes::DB_STATEMENT), 'UTF-8'));
+        $this->assertCount(3, $this->storage);
+
+        $db->query("SELECT id FROM technology WHERE id = '{$non_utf8_id}'");
+        $span_db_query = $this->storage->offsetGet(3);
+        $this->assertTrue(mb_check_encoding($span_db_query->getAttributes()->get(TraceAttributes::DB_STATEMENT), 'UTF-8'));
+        $this->assertCount(4, $this->storage);
+
+        $db->exec("SELECT id FROM technology WHERE id = '{$non_utf8_id}'");
+        $span_db_exec = $this->storage->offsetGet(4);
+        $this->assertTrue(mb_check_encoding($span_db_exec->getAttributes()->get(TraceAttributes::DB_STATEMENT), 'UTF-8'));
+        $this->assertCount(5, $this->storage);
+    }
 }