Dealing with broken charsets

byrokrat · Oct 13, 2018 · 9583885 · 9583885
1 parent f4dd366
commit 9583885
Show file tree

Hide file tree

Showing 5 changed files with 92 additions and 13 deletions.
diff --git a/spec/Visitor/AmountVisitorSpec.php b/spec/Visitor/AmountVisitorSpec.php
@@ -60,4 +60,17 @@ function it_creates_valid_amounts(Node $node)
 
         $this->beforeAmount($node);
     }
+
+    function it_creates_amounts_from_strings_with_broken_charset(Node $node)
+    {
+        $node->getLineNr()->willReturn(1);
+        $node->hasChild('Object')->willReturn(false);
+        $node->getValueFrom('Text')->willReturn('1230¤');
+
+        $node->addChild(Argument::that(function (Obj $obj) {
+            return (new SEK('-123.00'))->equals($obj->getValue());
+        }))->shouldBeCalled();
+
+        $this->beforeAmount($node);
+    }
 }
diff --git a/src/Parser/Grammar.peg b/src/Parser/Grammar.peg
@@ -404,11 +404,11 @@ grammar Grammar extends MultibyteHack
         return new AutogiroFile('AutogiroAmendmentResponseFile', $open, ...$recs);
     };
 
-    AMENDMENT_OPENING = '01AUTOGIRO    ' S10 date:DATE8 S10 '  MAKULERING/ÄNDRING  ' ^ bgcNr:BGC6 bg:BG10 EOR {
+    AMENDMENT_OPENING = '01AUTOGIRO    ' S10 date:DATE8 S10 '  MAKULERING/'.'NDRING  ' ^ bgcNr:BGC6 bg:BG10 EOR {
         return new Record('Opening', $date, $bgcNr, $bg);
     };
 
-    OLD_AMENDMENT_OPENING = '01' date:DATE8 'AUTOGIRO9900' 'MAK/ÄNDRINGSLISTA   ' ^ S20 bgcNr:BGC6 bg:BG10 EOR {
+    OLD_AMENDMENT_OPENING = '01' date:DATE8 'AUTOGIRO9900' 'MAK/'.'NDRINGSLISTA   ' ^ S20 bgcNr:BGC6 bg:BG10 EOR {
         return new Record('Opening', $date, $bgcNr, $bg);
     };
 

diff --git a/src/Parser/Grammar.php b/src/Parser/Grammar.php
@@ -5836,14 +5836,40 @@ protected function parseAMENDMENT_OPENING()
         if ($_success) {
             $_value148[] = $this->value;
 
-            if (substr($this->string, $this->position, strlen('  MAKULERING/ÄNDRING  ')) === '  MAKULERING/ÄNDRING  ') {
+            if (substr($this->string, $this->position, strlen('  MAKULERING/')) === '  MAKULERING/') {
                 $_success = true;
-                $this->value = substr($this->string, $this->position, strlen('  MAKULERING/ÄNDRING  '));
-                $this->position += strlen('  MAKULERING/ÄNDRING  ');
+                $this->value = substr($this->string, $this->position, strlen('  MAKULERING/'));
+                $this->position += strlen('  MAKULERING/');
             } else {
                 $_success = false;
 
-                $this->report($this->position, '\'  MAKULERING/ÄNDRING  \'');
+                $this->report($this->position, '\'  MAKULERING/\'');
+            }
+        }
+
+        if ($_success) {
+            $_value148[] = $this->value;
+
+            if ($this->position < strlen($this->string)) {
+                $_success = true;
+                $this->value = substr($this->string, $this->position, 1);
+                $this->position += 1;
+            } else {
+                $_success = false;
+            }
+        }
+
+        if ($_success) {
+            $_value148[] = $this->value;
+
+            if (substr($this->string, $this->position, strlen('NDRING  ')) === 'NDRING  ') {
+                $_success = true;
+                $this->value = substr($this->string, $this->position, strlen('NDRING  '));
+                $this->position += strlen('NDRING  ');
+            } else {
+                $_success = false;
+
+                $this->report($this->position, '\'NDRING  \'');
             }
         }
 
@@ -5958,14 +5984,40 @@ protected function parseOLD_AMENDMENT_OPENING()
         if ($_success) {
             $_value149[] = $this->value;
 
-            if (substr($this->string, $this->position, strlen('MAK/ÄNDRINGSLISTA   ')) === 'MAK/ÄNDRINGSLISTA   ') {
+            if (substr($this->string, $this->position, strlen('MAK/')) === 'MAK/') {
+                $_success = true;
+                $this->value = substr($this->string, $this->position, strlen('MAK/'));
+                $this->position += strlen('MAK/');
+            } else {
+                $_success = false;
+
+                $this->report($this->position, '\'MAK/\'');
+            }
+        }
+
+        if ($_success) {
+            $_value149[] = $this->value;
+
+            if ($this->position < strlen($this->string)) {
+                $_success = true;
+                $this->value = substr($this->string, $this->position, 1);
+                $this->position += 1;
+            } else {
+                $_success = false;
+            }
+        }
+
+        if ($_success) {
+            $_value149[] = $this->value;
+
+            if (substr($this->string, $this->position, strlen('NDRINGSLISTA   ')) === 'NDRINGSLISTA   ') {
                 $_success = true;
-                $this->value = substr($this->string, $this->position, strlen('MAK/ÄNDRINGSLISTA   '));
-                $this->position += strlen('MAK/ÄNDRINGSLISTA   ');
+                $this->value = substr($this->string, $this->position, strlen('NDRINGSLISTA   '));
+                $this->position += strlen('NDRINGSLISTA   ');
             } else {
                 $_success = false;
 
-                $this->report($this->position, '\'MAK/ÄNDRINGSLISTA   \'');
+                $this->report($this->position, '\'NDRINGSLISTA   \'');
             }
         }
 

diff --git a/src/Parser/Parser.php b/src/Parser/Parser.php
@@ -25,7 +25,7 @@
 use byrokrat\autogiro\Visitor\VisitorInterface;
 use byrokrat\autogiro\Tree\Node;
 use byrokrat\autogiro\Exception\ParserException;
-use \ForceUTF8\Encoding;
+use ForceUTF8\Encoding;
 
 final class Parser implements ParserInterface
 {

diff --git a/src/Visitor/AmountVisitor.php b/src/Visitor/AmountVisitor.php
@@ -47,11 +47,25 @@ public function beforeAmount(Node $node): void
         }
 
         try {
-            $node->addChild(new Obj($node->getLineNr(), SEK::createFromSignalString($signalStr)));
+            $invertSign = false;
+
+            // due to charset issues unknown trailing signal chars are treated as 'å'
+            if (!preg_match('/^[0-9åJKLMNOPQR]$/', mb_substr($signalStr, -1))) {
+                $signalStr = mb_substr($signalStr, 0, -1) . '0';
+                $invertSign = true;
+            }
+
+            $object = SEK::createFromSignalString($signalStr);
+
+            if ($invertSign) {
+                $object = $object->getInverted();
+            }
+
+            $node->addChild(new Obj($node->getLineNr(), $object));
         } catch (AmountException $e) {
             $this->getErrorObject()->addError(
                 "Invalid signaled amount %s on line %s",
-                $signalStr,
+                (string)$node->getValueFrom('Text'),
                 (string)$node->getLineNr()
             );
         }