public static string GetTJContent(string rawContent, CMapToUnicode cMapToUnicode, EncodingDifferenceToUnicode encodingDifferenceToUnicode)
{
string content;
string rawArray = rawContent.Remove(rawContent.Length - 2).Trim();
if (string.IsNullOrWhiteSpace(rawArray))
{
return(null);
}
PdfArrayDataType pdfArrayDataType = PdfArrayDataType.Parse(rawArray);
content = string.Empty;
foreach (string item in pdfArrayDataType.Elements.Where(_ => _ is string))
{
string escapedContent;
escapedContent = item.Trim();
content +=
PdfHexStringDataType.IsStartChar(escapedContent) ?
PdfFontHelper.ToUnicode(PdfHexStringDataType.GetHexContent(escapedContent), cMapToUnicode, encodingDifferenceToUnicode).ToString() :
PdfFontHelper.ToUnicode(PdfStringDataType.GetContentFromEscapedContent(escapedContent), cMapToUnicode, encodingDifferenceToUnicode);
}
if (content.Contains("Media"))
{
Console.WriteLine();
}
return(content);
}
public static BFRange Parse(string s, ref int startPosition)
{
string sBeginChar;
string sUnicodeChar;
Statement.SkipSpace(s, ref startPosition);
sBeginChar = PdfHexStringDataType.GetRawData(s, ref startPosition);
Statement.SkipSpace(s, ref startPosition);
sUnicodeChar = PdfHexStringDataType.GetRawData(s, ref startPosition);
int beginChar = int.Parse(sBeginChar.Substring(1, sBeginChar.Length - 2), NumberStyles.HexNumber);
int unicodeChar = int.Parse(sUnicodeChar.Substring(1, sUnicodeChar.Length - 2), NumberStyles.HexNumber);
return(new BFRange()
{
BeginChar = beginChar,
EndChar = beginChar,
UnicodeChar = unicodeChar
});
}
public static string GetNextStatement(string content, ref int i)
{
string statement = "";
bool readingStatement = false;
if (i >= content.Length)
{
return(null);
}
while (i < content.Length)
{
if (content[i] == ' ')
{
if (readingStatement)
{
i++;
return(statement);
}
statement += " ";
i++;
}
else if (content[i] == '\n')
{
i++;
return(statement);
}
else if (readingStatement && IsSeparator(content[i]))
{
return(statement);
}
else if (PdfNumericDataType.IsStartChar(content, i))
{
// string parameter
statement += PdfNumericDataType.GetRawData(content, ref i);
}
else if (PdfStringDataType.IsStartChar(content, i))
{
// string parameter
statement += PdfStringDataType.GetRawData(content, ref i);
}
else if (PdfArrayDataType.IsStartChar(content, i))
{
// array parameter
statement += PdfArrayDataType.GetRawData(content, ref i);
}
else if (PdfHexStringDataType.IsStartChar(content, i))
{
// hex string parameter
statement += PdfHexStringDataType.GetRawData(content, ref i);
}
else if (content[i] == 't' && i + 5 < content.Length && content.Substring(i, 5) == "true ")
{
// boolean true parameter
statement += "true ";
i += 5;
}
else if (content[i] == 'f' && i + 6 < content.Length && content.Substring(i, 6) == "false ")
{
// boolean false parameter
statement += "false ";
i += 6;
}
else if (PdfObjectDataType.IsStartChar(content, i))
{
// hex string parameter
statement += PdfObjectDataType.GetRawData(content, ref i);
}
else
{
statement += content[i];
readingStatement = true;
i++;
}
}
return(statement);
}
public override void CloseMultiLineStatement()
{
Lines = new List <TextObjectStatementLine>();
TextObjectStatementLine actualLineSettings = new TextObjectStatementLine();
Matrix textTransformMatrix = Matrix.Identity;
Point position = new Point();
float leadingParameter = 0;
int pageRotation = PdfReader.GetPageRotation(PageNumber);
for (int index = 0; index < RawContent.Count; index++)
{
string rawContent = RawContent[index];
if (rawContent.EndsWith("Tm"))
{
Matrix matrix;
if (Matrix.TryParse(rawContent, out matrix))
{
textTransformMatrix = matrix;
}
}
else if (rawContent.EndsWith("Tf"))
{
string[] fontParameters = rawContent.Split(' ');
if (fontParameters.Length < 3)
{
// Try to retrieve from previous line. This is global a parsing issue
if (index < 1)
{
continue;
}
fontParameters = (RawContent[index - 1].Trim() + " " + rawContent.Trim()).Split(' ');
if (fontParameters.Length < 3)
{
continue;
}
}
float fontSize;
if (float.TryParse(fontParameters[fontParameters.Length - 2], NumberStyles.Any, NumberFormatInfo.InvariantInfo, out fontSize))
{
actualLineSettings.FontHeight = fontSize;
}
actualLineSettings.CMapToUnicode = PdfFontHelper.GetFontCMapToUnicode(PdfReader, PageNumber, fontParameters[fontParameters.Length - 3]);
actualLineSettings.EncodingDifferenceToUnicode = EncodingDifferenceToUnicode.Parse(PdfFontHelper.GetFont(PdfReader, PageNumber, fontParameters[fontParameters.Length - 3]));
}
else if (rawContent.EndsWith("Td"))
{
float tx;
float ty;
string[] parameters = rawContent.Split(' ');
if (
float.TryParse(parameters[0], NumberStyles.Any, NumberFormatInfo.InvariantInfo, out tx) &&
float.TryParse(parameters[1], NumberStyles.Any, NumberFormatInfo.InvariantInfo, out ty))
{
textTransformMatrix = new Matrix(1, 0, 0, 1, tx, ty);
}
}
else if (rawContent.EndsWith("TD"))
{
float tx;
float ty;
string[] parameters = rawContent.Split(' ');
if (
float.TryParse(parameters[0], NumberStyles.Any, NumberFormatInfo.InvariantInfo, out tx) &&
float.TryParse(parameters[1], NumberStyles.Any, NumberFormatInfo.InvariantInfo, out ty))
{
textTransformMatrix = new Matrix(1, 0, 0, 1, tx, ty) * textTransformMatrix;
leadingParameter = -ty;
}
}
else if (rawContent.EndsWith("TL"))
{
float tl;
string[] parameters = rawContent.Split(' ');
if (
float.TryParse(parameters[0], NumberStyles.Any, NumberFormatInfo.InvariantInfo, out tl))
{
leadingParameter = tl;
}
}
else if (rawContent.EndsWith("T*"))
{
textTransformMatrix = new Matrix(1, 0, 0, 1, 0, -leadingParameter) * textTransformMatrix;
}
else if (rawContent.EndsWith("TJ"))
{
string content = GetTJContent(rawContent, actualLineSettings.CMapToUnicode, actualLineSettings.EncodingDifferenceToUnicode);
if (string.IsNullOrEmpty(content))
{
continue;
}
var line = actualLineSettings.Clone();
line.FontHeight =
line.FontHeight * textTransformMatrix.a * (pageRotation == 90 || pageRotation == 270 ? BaseTransformMatrix.b : BaseTransformMatrix.a);
line.Position = BaseTransformMatrix.TransformPoint(new Point(textTransformMatrix.TransformX(position.X, position.Y + line.FontHeight), textTransformMatrix.TransformY(position.X, position.Y + line.FontHeight))).Rotate(pageRotation);
line.Content = content;
Lines.Add(line);
}
else if (rawContent.Trim().EndsWith("Tj"))
{
string escapedContent;
escapedContent = rawContent.Trim();
escapedContent = escapedContent.Remove(escapedContent.Length - 2);
string content = PdfHexStringDataType.IsStartChar(escapedContent) ? PdfHexStringDataType.GetContent(escapedContent) : PdfStringDataType.GetContentFromEscapedContent(escapedContent);
var line = actualLineSettings.Clone();
line.FontHeight =
line.FontHeight * textTransformMatrix.a * (pageRotation == 90 || pageRotation == 270 ? BaseTransformMatrix.b : BaseTransformMatrix.a);
line.Position = BaseTransformMatrix.TransformPoint(new Point(textTransformMatrix.TransformX(position.X, position.Y + line.FontHeight), textTransformMatrix.TransformY(position.X, position.Y + line.FontHeight))).Rotate(pageRotation);
line.Content = PdfFontHelper.ToUnicode(content, line.CMapToUnicode, line.EncodingDifferenceToUnicode);
Lines.Add(line);
}
}
}