From d8bbf8a21ddcbbc1a6161194139d517e87032f4f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Gr=C3=A9gory=20Soutad=C3=A9?= <soutade@gmail.com>
Date: Sat, 18 Dec 2021 17:27:35 +0100
Subject: [PATCH] Manage stream data reading (without filter interpretation)

---
 include/uPDFTypes.h | 26 +++++++++++++++++++++-----
 src/uPDFTypes.cpp   | 45 ++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 65 insertions(+), 6 deletions(-)

diff --git a/include/uPDFTypes.h b/include/uPDFTypes.h
index b9adf9f..8e94ee3 100644
--- a/include/uPDFTypes.h
+++ b/include/uPDFTypes.h
@@ -266,15 +266,31 @@ namespace uPDFParser
     class Stream : public DataType
     {
     public:
-	Stream(int startOffset, int endOffset):
-	    DataType(DataType::TYPE::STREAM), startOffset(startOffset),
-	    endOffset(endOffset)
+	Stream(Dictionary& dict, int startOffset, int endOffset, unsigned char* data=0, unsigned int dataLength=0,
+	       bool freeData=false, int fd=0):
+	    DataType(DataType::TYPE::STREAM), dict(dict), fd(fd),
+	    startOffset(startOffset), endOffset(endOffset),
+	    _data(data), _dataLength(dataLength), freeData(false)
 	{}
-	virtual DataType* clone() {return new Stream(startOffset, endOffset);}
-	virtual std::string str() { return "stream\nendstream\n";}
+
+	~Stream() {
+	    if (_data && freeData) delete[] _data;
+	}
+	
+	virtual DataType* clone() {return new Stream(dict, startOffset, endOffset,
+						     _data, _dataLength, false, fd);}
+	virtual std::string str();
+	unsigned char* data();
+	unsigned int dataLength() {return _dataLength;}
+	void setData(unsigned char* data, unsigned int dataLength, bool freeData=false);
 
     private:
+	Dictionary& dict;
+	int fd;
 	int startOffset, endOffset;
+	unsigned char* _data;
+	unsigned int _dataLength;
+	bool freeData;
     };
 
     class Null : public DataType
diff --git a/src/uPDFTypes.cpp b/src/uPDFTypes.cpp
index 087e308..a4b6835 100644
--- a/src/uPDFTypes.cpp
+++ b/src/uPDFTypes.cpp
@@ -17,6 +17,8 @@
   along with uPDFParser. If not, see <http://www.gnu.org/licenses/>.
 */
 
+#include <unistd.h>
+
 #include "uPDFTypes.h"
 #include "uPDFParser_common.h"
 
@@ -71,7 +73,7 @@ namespace uPDFParser
 	    res += (*it)->str();
 	}
 	    
-	return res + std::string("]");
+	return res + " ]";
     }
 
     void Dictionary::addData(const std::string& key, DataType* value)
@@ -93,4 +95,45 @@ namespace uPDFParser
 	    
 	return res + std::string(">>\n"); 
    }
+
+    std::string Stream::str()
+    {
+	std::string res = "stream\n";
+	const char* streamData = (const char*)data(); // Force reading if not in memory
+	res += std::string(streamData, _dataLength);
+	res += "endstream\n";
+
+	return res;
+    }
+    
+    unsigned char* Stream::data()
+    {
+	if (!_data)
+	{
+	    if (!fd)
+		EXCEPTION(INVALID_STREAM, "Accessing data, but no file descriptor supplied");
+
+	    _dataLength = endOffset - startOffset;
+	    _data = new unsigned char[_dataLength];
+	    freeData = true;
+
+	    lseek(fd, startOffset, SEEK_SET);
+	    int ret = ::read(fd, _data, _dataLength);
+
+	    if ((unsigned int)ret != _dataLength)
+		EXCEPTION(INVALID_STREAM, "Not enough data to read (" << ret << ")");
+	}
+	
+	return _data;
+    }
+
+    void Stream::setData(unsigned char* data, unsigned int dataLength, bool freeData)
+    {
+	if (_data && freeData)
+	    delete[] _data;
+	
+	this->_data = data;
+	this->_dataLength = dataLength;
+	this->freeData = freeData;
+    }
 }