From: Michael Wallner <mike@php.net>
Date: Tue, 26 Apr 2022 09:34:59 +0000 (+0200)
Subject: fix userland string serialization
X-Git-Tag: v0.2.0~6
X-Git-Url: https://git.m6w6.name/?a=commitdiff_plain;h=a34b502a9e2beb230de5b8bd78d1ea4bbead57cc;p=awesomized%2Fext-ion

fix userland string serialization
---

diff --git a/ion_private.h b/ion_private.h
index 2d1c54a..344d20e 100644
--- a/ion_private.h
+++ b/ion_private.h
@@ -1950,7 +1950,9 @@ static void php_ion_serialize_string(php_ion_serializer *ser, zend_string *str)
 		ION_STRING is;
 		ION_CHECK(ion_writer_write_string(php_ion_obj(writer, ser->wri)->writer, ion_string_from_zend(&is, str)));
 	} else {
-
+		zval zs;
+		ZVAL_STR(&zs, str);
+		zend_call_method_with_1_params(ser->wri, NULL, NULL, "writeString", NULL, &zs);
 	}
 }
 
diff --git a/tests/~integration/detect_utf8.phpt b/tests/~integration/detect_utf8.phpt
new file mode 100644
index 0000000..e9fee64
--- /dev/null
+++ b/tests/~integration/detect_utf8.phpt
@@ -0,0 +1,56 @@
+--TEST--
+integration: custom serializer which naïvly decides to write strings/clobs/blobs
+--EXTENSIONS--
+ion
+--FILE--
+TEST
+<?php
+
+class AutoUtf8 extends ion\Serializer\Serializer {
+	public function serialize(mixed $data, \ion\Writer\Options|\ion\Writer|array|null $writer = null): mixed {
+		return parent::serialize($data, new class extends \ion\Writer\Buffer\Writer {
+			private function iterate(string $s) : Generator {
+				for ($i = 0; $i < grapheme_strlen($s); ++$i) {
+					yield grapheme_substr($s, $i, 1);
+				}
+			}
+			private function classify(string $c) : bool {
+				if (strlen($c) > 1) {
+					return false;
+				} elseif ($c < "\040") return match ($c) {
+					"\n", "\r", "\t" => false,
+					default => true
+				};
+				return $c >= "\177";
+			}
+			public function writeString(string $value) : void {
+				$dist = [0, 0];
+				foreach ($this->iterate($value) as $c) {
+					if (false === $c) {
+						goto binary;
+					}
+					++$dist[$this->classify($c)];
+				}
+				if ($dist[0] <= $dist[1]) {
+				binary:
+					parent::writeBLob($value);
+				} elseif ($dist[1]) {
+					parent::writeCLob($value);
+				} else {
+					parent::writeString($value);
+				}
+			}
+		});
+	}
+}
+
+$data = ["abc", "äöü", "\nabc\n", "foo\0bar", hex2bin("1f8b08003d96676200034bcacc4b2caae40200f5127b4207000000")];
+echo ion\serialize($data), "\n";
+echo ion\serialize($data, new AutoUtf8), "\n";
+?>
+DONE
+--EXPECTF--
+TEST
+["abc","äöü","\nabc\n","foo\0bar","\x1F%s\x12{B\a\0\0\0"]
+["abc","äöü","\nabc\n",{{"foo\0bar"}},{{H4sIAD2WZ2IAA0vKzEssquQCAPUSe0IHAAAA}}]
+DONE
diff --git a/tests/~integration/no_decimals.phpt b/tests/~integration/no_decimals.phpt
index d938230..c81c3ab 100644
--- a/tests/~integration/no_decimals.phpt
+++ b/tests/~integration/no_decimals.phpt
@@ -1,12 +1,12 @@
 --TEST--
-integration: do not write decimals but native floats only
+integration: custom serializer which does not write decimals but native floats only
 --EXTENSIONS--
 ion
 --FILE--
 TEST
 <?php
 
-class NoDecimals  extends ion\Serializer\Serializer {
+class NoDecimals extends ion\Serializer\Serializer {
 	public function serialize(mixed $data, \ion\Writer\Options|\ion\Writer|array|null $writer = null): mixed {
 		return parent::serialize($data, new class extends \ion\Writer\Buffer\Writer {
 			public function writeDecimal(\ion\Decimal|string $value): void {