From aaafc4e5f3cd6722cc5cfaf20016541c6c61f9d4 Mon Sep 17 00:00:00 2001
From: mark van der Feijst <mark@feyst.nl>
Date: Fri, 18 Oct 2024 16:48:16 +0200
Subject: [PATCH] Improve UNB Interchange Header recognition for syntax
 identifier

---
 src/EDI/Parser.php                | 18 +++++++++---------
 tests/EDITest/InterpreterTest.php |  5 ++++-
 tests/EDITest/ParserTest.php      |  9 ++++++---
 tests/files/example_utf8.edi      |  1 +
 4 files changed, 20 insertions(+), 13 deletions(-)

diff --git a/src/EDI/Parser.php b/src/EDI/Parser.php
index 48042a0..582a6c6 100644
--- a/src/EDI/Parser.php
+++ b/src/EDI/Parser.php
@@ -505,18 +505,18 @@ private function unwrap(string &$string): array
             );
         }
 
+        $unbRegex = sprintf(
+            '/^(UNA[^%1$s]+%1$s\W?\W?)?UNB%2$s(?<syntax_identifier>\w{4})%3$s/m',
+            $this->symbEnd,
+            $this->sepData,
+            $this->sepComp,
+        );
         if (
             ! $this->unbChecked
-            &&
-            \strpos($string, 'UNB') === 0
+            && false !== preg_match($unbRegex, $string, $unbMatches)
+            && isset($unbMatches['syntax_identifier'])
         ) {
-            $this->analyseUNB(
-                (string) \preg_replace(
-                    "#^UNB\+#",
-                    '',
-                    \substr($string, 0, 8)
-                )
-            );
+            $this->analyseUNB($unbMatches['syntax_identifier']);
         }
         if (preg_match_all("/[A-Z0-9]+(?:\?'|$)[\r\n]+/i", $string, $matches, PREG_OFFSET_CAPTURE) > 0) {
             $this->errors[] = 'This file contains some segments without terminators';
diff --git a/tests/EDITest/InterpreterTest.php b/tests/EDITest/InterpreterTest.php
index fe2232c..755bfe4 100644
--- a/tests/EDITest/InterpreterTest.php
+++ b/tests/EDITest/InterpreterTest.php
@@ -53,7 +53,9 @@ public function testServiceSegments()
     public function testBAPLIE()
     {
         $parser = new Parser();
-        $parser->load(__DIR__ . '/../files/D95BBAPLIE.edi')->parse();
+        $parser->load(__DIR__ . '/../files/D95BBAPLIE.edi');
+        $syntaxId = $parser->getSyntaxIdentifier();
+        $parser->parse();
 
         $mapping = new \EDI\Mapping\MappingProvider('D95B');
         $analyser = new Analyser();
@@ -67,6 +69,7 @@ public function testBAPLIE()
         static::assertCount(2, $interpreter->getMessages());
         static::assertCount(0, $interpreter->getErrors());
         static::assertCount(2, $interpreter->getServiceSegments());
+        static::assertSame('UNOA', $syntaxId);
 
         static::assertSame([], $interpreter->getErrors());
     }
diff --git a/tests/EDITest/ParserTest.php b/tests/EDITest/ParserTest.php
index 5799849..1f544d8 100644
--- a/tests/EDITest/ParserTest.php
+++ b/tests/EDITest/ParserTest.php
@@ -208,10 +208,13 @@ public function testLoadFile()
 
     public function testLoadWrappedFile()
     {
-        $p = new Parser();
-        $p->load(__DIR__ . '/../files/example_wrapped.edi')->parse();
-        $result = $p->errors();
+        $parser = new Parser();
+        $parser->load(__DIR__ . '/../files/example_wrapped.edi');
+        $syntaxId = $parser->getSyntaxIdentifier();
+        $parser->parse();
+        $result = $parser->errors();
         static::assertEmpty($result);
+        static::assertSame('IATB', $syntaxId);
     }
 
     public function testUNHWithoutMessageType()
diff --git a/tests/files/example_utf8.edi b/tests/files/example_utf8.edi
index 6585bfb..39a8cd0 100644
--- a/tests/files/example_utf8.edi
+++ b/tests/files/example_utf8.edi
@@ -1,3 +1,4 @@
+UNA:+.? '
 UNB+UNOC:1+1556150:31B+8888888:ZZ+160727:0953+1'
 UNH+142+DESADV:0:96A:UN'
 BGM+351+Y02197250+700101'