handle unbalanced parentheses for string tokenization

when the close parenthesis is unbalanced and precedes a line break followed by '/' or '>' we assume the bracket to be unbalanced and finish reading the string.

src/UglyToad.PdfPig.Tests/Tokenization/Scanner/PdfTokenScannerTests.cs

@@ -298,17 +298,36 @@ endobj
 << /S 1245 >>
 
 stream
-%¥×³®í»š}%§X{{tøNåÝž¶ö¢ÖÞgrehtyyy$&%&£$££(*¾–~´¼
 %¥×³®í»š}%§X{{tøNåÝž¶ö¢ÖÞgrehtyyy$&%&£$££(*¾–~´¼
-endobj";
+endstream
 endobj";
 
             var scanner = GetScanner(s);
 
-
             var token = ReadToEnd(scanner)[0];
 
-            var stream = Assert.IsType<StreamToken>(token.Data);
+            Assert.Equal(12655, token.Number.ObjectNumber);
+
+            var stream = Assert.IsType<StreamToken>(token.Data);
+
+            Assert.Equal("1245", stream.StreamDictionary.Data["S"].ToString());
+
+            Assert.Equal("%¥×³®í»š}%§X{{tøNåÝž¶ö¢ÖÞgrehtyyy$&%&£$££(*¾–~´¼", Encoding.UTF8.GetString(stream.Data.ToArray()));
+        }
+
+        [Fact]
+        public void ReadsStreamWithoutBreakBeforeEndstream()
+        {
+            const string s = @"
+1 0 obj
+12
+endobj
+
+7 0 obj
+<< /Length 288
+   /Filter /FlateDecode >>
+stream
+xœ]‘ËjÃ0E÷ÿÃ,ÓEð#NÒ€1¤N^ôA~€-]A-YYøï+Ï4¡t#qfîFWQY*­Dïv5:è”–§ñjB‹½Òa¤ •p7¤K	ƒÈûëyr8Tº!Ïà  úð‚ÉÙVG9¶ø@Å7+Ñ*ÝÃ곬¹T_ùÆ
µƒ8Š$vË̗Ƽ6BDöu%½B¹yí$—Ù ¤\Hx71JœL#Ð6ºÇ0È㸀ü|. µüßõÏ""WÛ‰¯Æ.êÄ«ã8;¤iL°!Ø %É`K°ßì¸ÃöÜá
Ü)	[‚#CFðÄ°#(yƒg^ÿ¶æò
 ÿž“¸Zë#¢?¢h–P”Æû?šÑï÷ø
¯‰Šendstream
 endobj
 

src/UglyToad.PdfPig.Tokenization/StringTokenizer.cs

@@ -23,14 +23,14 @@
                 return false;
             }
 
-            int numberOfBrackets = 1;
-            bool isEscapeActive = false;
-            bool isLineBreaking = false;
+            var numberOfBrackets = 1;
+            var isEscapeActive = false;
+            var isLineBreaking = false;
 
-            bool octalModeActive = false;
+            var octalModeActive = false;
 
             short[] octal = { 0, 0, 0 };
-            int octalsRead = 0;
+            var octalsRead = 0;
 
             while (inputBytes.MoveNext())
             {
@@ -85,8 +85,7 @@
                         }
 
                         // TODO: Check for other ends of string where the string is improperly formatted. See commented method
-                        // numberOfBrackets = CheckForEndOfString(inputBytes, numberOfBrackets);
-
+                        numberOfBrackets = CheckForEndOfString(numberOfBrackets, inputBytes);
 
                         break;
                     case '(':
@@ -185,7 +184,7 @@
 
         private static void LeftShiftOctal(char nextOctalChar, int octalsRead, short[] octals)
         {
-            for (int i = octalsRead; i > 0; i--)
+            for (var i = octalsRead; i > 0; i--)
             {
                 octals[i] = octals[i - 1];
             }
@@ -240,5 +239,40 @@
                     break;
             }
         }
+
+        private static int CheckForEndOfString(int numberOfBrackets, IInputBytes bytes)
+        {
+            const byte lineFeed = 10;
+            const byte carriageReturn = 13;
+
+            var braces = numberOfBrackets;
+            var nextThreeBytes = new byte[3];
+
+            var startAt = bytes.CurrentOffset;
+
+            var amountRead = bytes.Read(nextThreeBytes);
+
+            // Check the next 3 bytes if available
+            // The following cases are valid indicators for the end of the string
+            // 1. Next line contains another COSObject: CR + LF + '/'
+            // 2. COSDictionary ends in the next line: CR + LF + '>'
+            // 3. Next line contains another COSObject: CR + '/'
+            // 4. COSDictionary ends in the next line: CR + '>'
+            if (amountRead == 3 && nextThreeBytes[0] == carriageReturn)
+            {
+                if ((nextThreeBytes[1] == lineFeed && (nextThreeBytes[2] == '/') || nextThreeBytes[2] == '>')
+                    || nextThreeBytes[1] == '/' || nextThreeBytes[1] == '>')
+                {
+                    braces = 0;
+                }
+            }
+
+            if (amountRead > 0)
+            {
+                bytes.Seek(startAt);
+            }
+
+            return braces;
+        }
     }
 }