PdfView - Peeking into the Internals of PDFs

// BRawPdf.cpp : Defines the class behaviors for the application.
//

#include "stdafx.h"
#include "BRawPdf.h"

CBRawPdf::CBRawPdf()
{
	m_abFile = NULL;
}

void CBRawPdf::Close()
{
	if (m_abFile != NULL)
		delete m_abFile;
	m_abFile = NULL;
}

void CBRawPdf::Open(CString strFilePath)
{
	Close();

	CFile file;
	if (!file.Open(strFilePath, CFile::modeRead))
	{
	    AfxMessageBox("Cannot open file.");
		AfxThrowUserException();
	}

	ULONGLONG ullLength = (ULONGLONG)file.GetLength();

	m_dwSize = (DWORD)ullLength;
	m_abFile = new BYTE[m_dwSize];
	file.Read(m_abFile, m_dwSize);
	file.Close();

	m_dwPos = 0;
}

DWORD CBRawPdf::GetPos()
{
	return m_dwPos;
}

void CBRawPdf::GotoEnd()
{
	m_dwPos = m_dwSize - 1;
}

void CBRawPdf::GotoPos(DWORD dwPos)
{
	m_dwPos = dwPos;
}

void CBRawPdf::GotoLine(CString strLine)
{
	for (;;)
	{
		if (m_abFile[m_dwPos] == strLine[0])
		{
			BOOL bOk = TRUE;
			for (DWORD i = 1; i < (DWORD)strLine.GetLength(); i++)
			{
				if (m_abFile[(int)(m_dwPos + i)] != strLine[(int)i])
					bOk = FALSE;
			}
			if (bOk == TRUE)
				break;
		}
		m_dwPos++;
	}
}

BOOL CBRawPdf::CharInSet(char c, CString strSet)
{
	for (DWORD i = 0; i < (DWORD)strSet.GetLength(); i++)
	{
		if (c == strSet[(int)i])
			return TRUE;
	}
	return FALSE;
}

BOOL CBRawPdf::_GetNextToken(CString& strToken, BOOL bCheckLink)
{
	strToken = "";

	// Skip White
	for (;;)
	{
		if (m_dwPos >= m_dwSize)
			return FALSE;

		if (CharInSet(m_abFile[m_dwPos], "\r\n ") == false)
			break;

		m_dwPos++;
	}

	if (bCheckLink)
	{
		DWORD dwPos = m_dwPos;

		CString strObject;
		CString strGeneration;
		CString strType;
		_GetNextToken(strObject, FALSE);
		if ((strObject.GetLength() > 0) && (strObject[0] >= '0') && (strObject[0] <= '9'))
		{
			_GetNextToken(strGeneration, FALSE);
			_GetNextToken(strType, FALSE);
			if (strType == "R")
			{
				strToken.Format("%s %s R", strObject.GetBuffer(0), strGeneration.GetBuffer(0));
				return TRUE;
			}
		}
		m_dwPos = dwPos;
	}

	for(;;)
	{
		if (CharInSet(m_abFile[m_dwPos], "\r\n "))
			return TRUE;

		if (!memcmp((BYTE*)&m_abFile[m_dwPos], "<<", 2))
		{
			if (strToken != "")
				return TRUE;
			strToken = "<<";
			m_dwPos += 2;
			return TRUE;
		}

		if (!memcmp((BYTE*)&m_abFile[m_dwPos], ">>", 2))
		{
			if (strToken != "")
				return TRUE;
			strToken = ">>";
			m_dwPos += 2;
			return TRUE;
		}

		if (CharInSet(m_abFile[m_dwPos], "/[])>"))
		{
			if (strToken != "")
				return TRUE;
		}
		if (m_abFile[m_dwPos] == '(')
		{
			if (strToken != "")
				return TRUE;

			strToken = "";
			DWORD dwDepth = 0;
			BOOL bBinary = FALSE;
			BOOL bUnicode = FALSE;

			if (m_dwPos + 2 <= m_dwSize)
			{
				if ((m_abFile[m_dwPos + 1] == 0xFE) && (m_abFile[m_dwPos + 2] == 0xFF))
				{
					strToken += "(";
					bUnicode = TRUE;
					m_dwPos += 3;
				}
			}

			for (int i = 0;; i++)
			{
				if (m_dwPos >= m_dwSize)
				{
					AfxMessageBox("Unterminated string.");
					AfxThrowUserException();
				}

				if (bUnicode)
				{
					if (m_abFile[m_dwPos] == ')')
					{
						strToken += CString((char)m_abFile[m_dwPos]);
						m_dwPos++;
						break;
					}
					m_dwPos++;
				}

				if (m_abFile[m_dwPos] > 127)
					bBinary = TRUE;

				strToken += CString((char)m_abFile[m_dwPos]);
				if (m_abFile[m_dwPos] == '(')
					dwDepth++;
				if (m_abFile[m_dwPos] == ')')
				{
					dwDepth--;
					if ((dwDepth <= 0) || bBinary)
					{
						m_dwPos++;
						break;
					}
				}
				m_dwPos++;
			}
			return TRUE;
		}

		if (m_abFile[m_dwPos] == '<')
		{
			if (strToken != "")
				return TRUE;

			strToken = "";
			DWORD dwDepth = 0;
			for (;;)
			{
				if (m_dwPos >= m_dwSize)
				{
					AfxMessageBox("Unterminated string.");
					AfxThrowUserException();
				}

				strToken += CString((char)m_abFile[m_dwPos]);
				if (m_abFile[m_dwPos] == '<')
					dwDepth++;
				if (m_abFile[m_dwPos] == '>')
				{
					dwDepth--;
					if (dwDepth <= 0)
					{
						m_dwPos++;
						break;
					}
				}
				m_dwPos++;
			}
			return TRUE;
		}

		strToken += CString((char)m_abFile[m_dwPos]);
		if (CharInSet(m_abFile[m_dwPos], "[]()<>"))
		{
			m_dwPos++;
			return TRUE;
		}
		m_dwPos++;
	}
}

void CBRawPdf::GetNextToken(CString& strToken)
{
	_GetNextToken(strToken, TRUE);
}

void CBRawPdf::AssertNextToken(CString strExpectedToken)
{
	CString strToken;
	GetNextToken(strToken);
	if (strToken == strExpectedToken)
		return;

    AfxMessageBox("Unexpected token.");
	AfxThrowUserException();
}

BOOL CBRawPdf::GetNextLine(CString& strLine)
{
	strLine = "";
	for (;;)
	{
		if (m_dwPos >= m_dwSize)
			return FALSE;

		if ((m_abFile[m_dwPos] == '\r') && (m_abFile[m_dwPos + 1] == '\n'))
		{
			m_dwPos += 2;
			break;
		}
		else if ((m_abFile[m_dwPos] == '\r') || (m_abFile[m_dwPos] == '\n'))
		{
			m_dwPos++;
			break;
		}
		strLine += m_abFile[m_dwPos];
		m_dwPos++;
	}
	return TRUE;
}

void CBRawPdf::AssertNextLine(CString strExpectedLine)
{
	CString strLine;
	GetNextLine(strLine);
	if (strLine == strExpectedLine)
		return;

    AfxMessageBox("Unexpected line.");
	AfxThrowUserException();
}

void CBRawPdf::GetPreviousLine(CString& strLine)
{
	strLine = "";
	if ((m_abFile[m_dwPos - 1] == '\r') && ((m_abFile[m_dwPos] == '\n')))
		m_dwPos -= 2;
	else if ((m_abFile[m_dwPos] == '\r') || (m_abFile[m_dwPos] == '\n'))
		m_dwPos--;

	for (;;)
	{
		if (m_dwPos < 0)
			return;

		if ((m_abFile[m_dwPos - 1] == '\r') && ((m_abFile[m_dwPos] == '\n')))
			break;
		else if ((m_abFile[m_dwPos] == '\r') || (m_abFile[m_dwPos] == '\n'))
			break;

		strLine = CString((char)m_abFile[m_dwPos]) + strLine;
		m_dwPos--;
	}
	return;
}

void CBRawPdf::AssertPreviousLine(CString strExpectedLine)
{
	CString strLine;
	GetPreviousLine(strLine);
	if (strLine == strExpectedLine)
		return;

    AfxMessageBox("Unexpected line.");
	AfxThrowUserException();
}

BOOL CBRawPdf::IsOpen()
{
	if (m_abFile == NULL)
		return FALSE;

	return TRUE;
}

BOOL CBRawPdf::GetMem(BYTE* pabBuffer, DWORD dwPos, DWORD dwSize)
{
	::memcpy(pabBuffer, &m_abFile[dwPos], dwSize);
	return TRUE;
}