From 35c88120b7a6edebb54037a3bb15c5de3a5c8a24 Mon Sep 17 00:00:00 2001
From: Luciano Hanna El Adji <lucianohanna@id.uff.br>
Date: Thu, 17 Jun 2021 04:28:42 -0300
Subject: [PATCH] Don't throw an exception if Encoding doesn't have
 BaseEncoding (#433)

* Dont crash when a Encoding doesn't have BaseEncoding

* Create a specific Exception for encoding not found

* Fix phpdoc and change exception in catch

* Improve test

* Create a test for Encoding without BaseEncoding

* Update src/Smalot/PdfParser/Font.php

Co-authored-by: Konrad Abicht <hi@inspirito.de>

* Update tests/Integration/FontTest.php

Co-authored-by: Konrad Abicht <hi@inspirito.de>

* Move EncodingNotFoundException to src/Smalot/PdfParser/Exception

Co-authored-by: Konrad Abicht <hi@inspirito.de>
---
 src/Smalot/PdfParser/Encoding.php                   |  5 +++--
 .../Exception/EncodingNotFoundException.php         |  7 +++++++
 src/Smalot/PdfParser/Font.php                       | 11 +++++++++--
 tests/Integration/EncodingTest.php                  |  3 ++-
 tests/Integration/FontTest.php                      | 13 +++++++++++++
 5 files changed, 34 insertions(+), 5 deletions(-)
 create mode 100644 src/Smalot/PdfParser/Exception/EncodingNotFoundException.php

diff --git a/src/Smalot/PdfParser/Encoding.php b/src/Smalot/PdfParser/Encoding.php
index 369d3af6..b6537f34 100644
--- a/src/Smalot/PdfParser/Encoding.php
+++ b/src/Smalot/PdfParser/Encoding.php
@@ -33,6 +33,7 @@
 use Exception;
 use Smalot\PdfParser\Element\ElementNumeric;
 use Smalot\PdfParser\Encoding\PostScriptGlyphs;
+use Smalot\PdfParser\Exception\EncodingNotFoundException;
 
 /**
  * Class Encoding
@@ -148,7 +149,7 @@ public function __toString()
     /**
      * @return string
      *
-     * @throws \Exception
+     * @throws EncodingNotFoundException
      */
     protected function getEncodingClass()
     {
@@ -157,7 +158,7 @@ protected function getEncodingClass()
         $className = '\\Smalot\\PdfParser\\Encoding\\'.$baseEncoding;
 
         if (!class_exists($className)) {
-            throw new Exception('Missing encoding data for: "'.$baseEncoding.'".');
+            throw new EncodingNotFoundException('Missing encoding data for: "'.$baseEncoding.'".');
         }
 
         return $className;
diff --git a/src/Smalot/PdfParser/Exception/EncodingNotFoundException.php b/src/Smalot/PdfParser/Exception/EncodingNotFoundException.php
new file mode 100644
index 00000000..c43a66c1
--- /dev/null
+++ b/src/Smalot/PdfParser/Exception/EncodingNotFoundException.php
@@ -0,0 +1,7 @@
+<?php
+
+namespace Smalot\PdfParser\Exception;
+
+class EncodingNotFoundException extends \Exception
+{
+}
diff --git a/src/Smalot/PdfParser/Font.php b/src/Smalot/PdfParser/Font.php
index 5e227dcc..5496d07d 100644
--- a/src/Smalot/PdfParser/Font.php
+++ b/src/Smalot/PdfParser/Font.php
@@ -31,6 +31,7 @@
 namespace Smalot\PdfParser;
 
 use Smalot\PdfParser\Encoding\WinAnsiEncoding;
+use Smalot\PdfParser\Exception\EncodingNotFoundException;
 
 /**
  * Class Font
@@ -107,9 +108,15 @@ public function translateChar($char, $use_default = true)
             \strlen($char) < 2
             && $this->has('Encoding')
             && $this->get('Encoding') instanceof Encoding
-            && WinAnsiEncoding::class === $this->get('Encoding')->__toString()
         ) {
-            $fallbackDecoded = self::uchr($dec);
+            try {
+                if (WinAnsiEncoding::class === $this->get('Encoding')->__toString()) {
+                    $fallbackDecoded = self::uchr($dec);
+                }
+            } catch (EncodingNotFoundException $e) {
+                // Encoding->getEncodingClass() throws EncodingNotFoundException when BaseEncoding doesn't exists
+                // See table 5.11 on PDF 1.5 specs for more info
+            }
         }
 
         return $use_default ? self::MISSING : $fallbackDecoded;
diff --git a/tests/Integration/EncodingTest.php b/tests/Integration/EncodingTest.php
index f6ef6464..9dc9edc3 100644
--- a/tests/Integration/EncodingTest.php
+++ b/tests/Integration/EncodingTest.php
@@ -37,6 +37,7 @@
 use Smalot\PdfParser\Element;
 use Smalot\PdfParser\Encoding;
 use Smalot\PdfParser\Encoding\StandardEncoding;
+use Smalot\PdfParser\Exception\EncodingNotFoundException;
 use Smalot\PdfParser\Header;
 use Tests\Smalot\PdfParser\TestCase;
 
@@ -62,7 +63,7 @@ public function testGetEncodingClass()
      */
     public function testInitGetEncodingClassMissingClassException()
     {
-        $this->expectException(Exception::class);
+        $this->expectException(EncodingNotFoundException::class);
         $this->expectExceptionMessage('Missing encoding data for: "invalid"');
 
         $header = new Header(['BaseEncoding' => new Element('invalid')]);
diff --git a/tests/Integration/FontTest.php b/tests/Integration/FontTest.php
index c4c1ce94..4175ed04 100644
--- a/tests/Integration/FontTest.php
+++ b/tests/Integration/FontTest.php
@@ -379,4 +379,17 @@ public function testXmlContent()
 
         $this->assertEquals('Example PDF', $text);
     }
+
+    /**
+     * Create an instance of Header containing an instance of Encoding that doesn't have a BaseEncoding.
+     * Test if the Font won't raise a exception because Encoding don't have BaseEncoding.
+     */
+    public function testEncodingWithoutBaseEncoding()
+    {
+        $document = new Document();
+        $header = new Header(['Encoding' => new Encoding($document)]);
+        $font = new Font(new Document(), $header);
+        $font->setTable([]);
+        $this->assertEquals('?', $font->translateChar('a'));
+    }
 }