Merge pull request #182 from Setasign/development

Development

src/PdfParser/PdfParser.php

@@ -25,6 +25,7 @@
 use setasign\Fpdi\PdfParser\Type\PdfString;
 use setasign\Fpdi\PdfParser\Type\PdfToken;
 use setasign\Fpdi\PdfParser\Type\PdfType;
+use setasign\Fpdi\PdfParser\Type\PdfTypeException;
 
 /**
  * A PDF parser class
@@ -258,25 +259,25 @@ public function readValue($token = null, $expectedType = null)
         switch ($token) {
             case '(':
                 $this->ensureExpectedType($token, $expectedType);
-                return PdfString::parse($this->streamReader);
+                return $this->parsePdfString();
 
             case '<':
                 if ($this->streamReader->getByte() === '<') {
                     $this->ensureExpectedType('<<', $expectedType);
                     $this->streamReader->addOffset(1);
-                    return PdfDictionary::parse($this->tokenizer, $this->streamReader, $this);
+                    return $this->parsePdfDictionary();
                 }
 
                 $this->ensureExpectedType($token, $expectedType);
-                return PdfHexString::parse($this->streamReader);
+                return $this->parsePdfHexString();
 
             case '/':
                 $this->ensureExpectedType($token, $expectedType);
-                return PdfName::parse($this->tokenizer, $this->streamReader);
+                return $this->parsePdfName();
 
             case '[':
                 $this->ensureExpectedType($token, $expectedType);
-                return PdfArray::parse($this->tokenizer, $this);
+                return $this->parsePdfArray();
 
             default:
                 if (\is_numeric($token)) {
@@ -291,13 +292,7 @@ public function readValue($token = null, $expectedType = null)
                                         );
                                     }
 
-                                    return PdfIndirectObject::parse(
-                                        (int) $token,
-                                        (int) $token2,
-                                        $this,
-                                        $this->tokenizer,
-                                        $this->streamReader
-                                    );
+                                    return $this->parsePdfIndirectObject((int)$token, (int)$token2);
                                 case 'R':
                                     if (
                                         $expectedType !== null &&
@@ -309,7 +304,7 @@ public function readValue($token = null, $expectedType = null)
                                         );
                                     }
 
-                                    return PdfIndirectObjectReference::create((int) $token, (int) $token2);
+                                    return PdfIndirectObjectReference::create((int)$token, (int)$token2);
                             }
 
                             $this->tokenizer->pushStack($token3);
@@ -351,6 +346,65 @@ public function readValue($token = null, $expectedType = null)
         }
     }
 
+    /**
+     * @return PdfString
+     */
+    protected function parsePdfString()
+    {
+        return PdfString::parse($this->streamReader);
+    }
+
+    /**
+     * @return false|PdfHexString
+     */
+    protected function parsePdfHexString()
+    {
+        return PdfHexString::parse($this->streamReader);
+    }
+
+    /**
+     * @return bool|PdfDictionary
+     * @throws PdfTypeException
+     */
+    protected function parsePdfDictionary()
+    {
+        return PdfDictionary::parse($this->tokenizer, $this->streamReader, $this);
+    }
+
+    /**
+     * @return PdfName
+     */
+    protected function parsePdfName()
+    {
+        return PdfName::parse($this->tokenizer, $this->streamReader);
+    }
+
+    /**
+     * @return false|PdfArray
+     * @throws PdfTypeException
+     */
+    protected function parsePdfArray()
+    {
+        return PdfArray::parse($this->tokenizer, $this);
+    }
+
+    /**
+     * @param int $objectNumber
+     * @param int $generationNumber
+     * @return false|PdfIndirectObject
+     * @throws Type\PdfTypeException
+     */
+    protected function parsePdfIndirectObject($objectNumber, $generationNumber)
+    {
+        return PdfIndirectObject::parse(
+            $objectNumber,
+            $generationNumber,
+            $this,
+            $this->tokenizer,
+            $this->streamReader
+        );
+    }
+
     /**
      * Ensures that the token will evaluate to an expected object type (or not).
      *
@@ -359,7 +413,7 @@ public function readValue($token = null, $expectedType = null)
      * @return bool
      * @throws Type\PdfTypeException
      */
-    private function ensureExpectedType($token, $expectedType)
+    protected function ensureExpectedType($token, $expectedType)
     {
         static $mapping = [
             '(' => PdfString::class,

src/PdfParser/StreamReader.php

@@ -113,6 +113,12 @@ public function __construct($stream, $closeStream = false)
             );
         }
 
+        if (fseek($stream, 0) === -1) {
+            throw new \InvalidArgumentException(
+                'Given stream is not seekable!'
+            );
+        }
+
         $this->stream = $stream;
         $this->closeStream = $closeStream;
         $this->reset();

src/PdfParser/Type/PdfArray.php

@@ -25,7 +25,7 @@ class PdfArray extends PdfType
      *
      * @param Tokenizer $tokenizer
      * @param PdfParser $parser
-     * @return bool|self
+     * @return false|self
      * @throws PdfTypeException
      */
     public static function parse(Tokenizer $tokenizer, PdfParser $parser)

src/PdfParser/Type/PdfHexString.php

@@ -21,7 +21,7 @@ class PdfHexString extends PdfType
      * Parses a hexadecimal string object from the stream reader.
      *
      * @param StreamReader $streamReader
-     * @return bool|self
+     * @return false|self
      */
     public static function parse(StreamReader $streamReader)
     {

src/PdfParser/Type/PdfIndirectObject.php

@@ -22,17 +22,17 @@ class PdfIndirectObject extends PdfType
     /**
      * Parses an indirect object from a tokenizer, parser and stream-reader.
      *
-     * @param int $objectNumberToken
-     * @param int $objectGenerationNumberToken
+     * @param int $objectNumber
+     * @param int $objectGenerationNumber
      * @param PdfParser $parser
      * @param Tokenizer $tokenizer
      * @param StreamReader $reader
-     * @return bool|self
+     * @return self|false
      * @throws PdfTypeException
      */
     public static function parse(
-        $objectNumberToken,
-        $objectGenerationNumberToken,
+        $objectNumber,
+        $objectGenerationNumber,
         PdfParser $parser,
         Tokenizer $tokenizer,
         StreamReader $reader
@@ -50,8 +50,8 @@ public static function parse(
         }
 
         $v = new self();
-        $v->objectNumber = (int) $objectNumberToken;
-        $v->generationNumber = (int) $objectGenerationNumberToken;
+        $v->objectNumber = (int) $objectNumber;
+        $v->generationNumber = (int) $objectGenerationNumber;
         $v->value = $value;
 
         return $v;

src/PdfParser/Type/PdfStream.php

@@ -213,18 +213,16 @@ protected function extractStream()
     }
 
     /**
-     * Get the unfiltered stream data.
+     * Get all filters defined for this stream.
      *
-     * @return string
-     * @throws FilterException
-     * @throws PdfParserException
+     * @return PdfType[]
+     * @throws PdfTypeException
      */
-    public function getUnfilteredStream()
+    public function getFilters()
     {
-        $stream = $this->getStream();
         $filters = PdfDictionary::get($this->value, 'Filter');
         if ($filters instanceof PdfNull) {
-            return $stream;
+            return [];
         }
 
         if ($filters instanceof PdfArray) {
@@ -233,6 +231,24 @@ public function getUnfilteredStream()
             $filters = [$filters];
         }
 
+        return $filters;
+    }
+
+    /**
+     * Get the unfiltered stream data.
+     *
+     * @return string
+     * @throws FilterException
+     * @throws PdfParserException
+     */
+    public function getUnfilteredStream()
+    {
+        $stream = $this->getStream();
+        $filters = $this->getFilters();
+        if ($filters === []) {
+            return $stream;
+        }
+
         $decodeParams = PdfDictionary::get($this->value, 'DecodeParms');
         if ($decodeParams instanceof PdfArray) {
             $decodeParams = $decodeParams->value;
@@ -308,6 +324,21 @@ public function getUnfilteredStream()
                     $stream = $filterObject->decode($stream);
                     break;
 
+                case 'Crypt':
+                    if (!$decodeParam instanceof PdfDictionary) {
+                        break;
+                    }
+                    // Filter is "Identity"
+                    $name = PdfDictionary::get($decodeParam, 'Name');
+                    if (!$name instanceof PdfName || $name->value !== 'Identity') {
+                        break;
+                    }
+
+                    throw new FilterException(
+                        'Support for Crypt filters other than "Identity" is not implemented.',
+                        FilterException::UNSUPPORTED_FILTER
+                    );
+
                 default:
                     throw new FilterException(
                         \sprintf('Unsupported filter "%s".', $filter->value),

src/PdfParser/Type/PdfString.php

@@ -78,6 +78,36 @@ public static function ensure($string)
         return PdfType::ensureType(self::class, $string, 'String value expected.');
     }
 
+    /**
+     * Escapes sequences in a string according to the PDF specification.
+     *
+     * @param string $s
+     * @return string
+     */
+    public static function escape($s)
+    {
+        // Still a bit faster, than direct replacing
+        if (
+            \strpos($s, '\\') !== false ||
+            \strpos($s, ')')  !== false ||
+            \strpos($s, '(')  !== false ||
+            \strpos($s, "\x0D") !== false ||
+            \strpos($s, "\x0A") !== false ||
+            \strpos($s, "\x09") !== false ||
+            \strpos($s, "\x08") !== false ||
+            \strpos($s, "\x0C") !== false
+        ) {
+            // is faster than strtr(...)
+            return \str_replace(
+                ['\\',   ')',   '(',   "\x0D", "\x0A", "\x09", "\x08", "\x0C"],
+                ['\\\\', '\\)', '\\(', '\r',   '\n',   '\t',   '\b',   '\f'],
+                $s
+            );
+        }
+
+        return $s;
+    }
+
     /**
      * Unescapes escaped sequences in a PDF string according to the PDF specification.
      *

src/Tcpdf/Fpdi.php

@@ -251,7 +251,7 @@ protected function writePdfType(PdfType $value)
         if ($value instanceof PdfString) {
             $string = PdfString::unescape($value->value);
             $string = $this->_encrypt_data($this->currentObjectNumber, $string);
-            $value->value = \TCPDF_STATIC::_escape($string);
+            $value->value = PdfString::escape($string);
         } elseif ($value instanceof PdfHexString) {
             $filter = new AsciiHex();
             $string = $filter->decode($value->value);

tests/functional/PdfParser/Type/PdfStreamTest.php

@@ -423,4 +423,25 @@ public function testGetUnfilteredStream($file, $objectNumber, $expectedResult)
 
         $this->assertEquals($expectedResult, $stream->getUnfilteredStream());
     }
+
+    public function testParseWithCryptFilter()
+    {
+        $in = "123 0 obj\n<</Filter /Crypt /Length 5>>\nstream\nHello\nendstream\nendobj";
+
+        $stream = StreamReader::createByString($in);
+
+        // set position and prepare dictionary (equals to result)
+        $stream->setOffset(45);
+        $this->assertSame("\n", $stream->getByte()); // this is the \n after the stream keyword
+
+        $dict = PdfDictionary::create([
+            'Filter' => PdfName::create('Crypt'),
+            'Length' => PdfNumeric::create(5)
+        ]);
+
+        $result = PdfStream::parse($dict, $stream);
+
+        $this->assertSame($dict, $result->value);
+        $this->assertSame('Hello', $result->getUnfilteredStream());
+    }
 }

tests/unit/PdfParser/DummyFaultyStreamWrapper.php

@@ -0,0 +1,52 @@
+<?php
+
+namespace setasign\Fpdi\unit\PdfParser;
+
+/**
+ * This stream wrapper is created to simulate a stream wrapper that is not seekable (while its metadata report that
+ * it is seekable).
+ */
+class DummyFaultyStreamWrapper
+{
+    public $context;
+
+    function stream_open($path, $mode, $options, &$opened_path)
+    {
+        return true;
+    }
+
+    function stream_read($count)
+    {
+        return '';
+    }
+
+    function stream_write($data)
+    {
+        return 0;
+    }
+
+    function stream_tell()
+    {
+        return 0;
+    }
+
+    function stream_eof()
+    {
+        return false;
+    }
+
+    function stream_seek($offset, $whence)
+    {
+        return false;
+    }
+
+    function stream_metadata($path, $option, $var)
+    {
+        return true;
+    }
+
+    function stream_stat()
+    {
+        return [];
+    }
+}