/**
* Put all data from given day into given table
*/
public static void collectDataforDay(DateTime dateTime, SQLiteConnection m_dbConnection)
{
string zeroMonth = dateTime.Month < 10 ? "0" : "";
string zeroDay = dateTime.Day < 10 ? "0" : "";
string date = zeroMonth + dateTime.Month + "/" + zeroDay + dateTime.Day + "/" + dateTime.Year;
PdfReader reader;
try
{
reader = new PdfReader("http://www.equibase.com/premium/eqbPDFChartPlus.cfm?RACE=A&BorP=P&TID=SAR&CTRY=USA&DT=" + date + "&DAY=D&STYLE=EQB");
}
catch (Exception e)
{
Console.WriteLine("CAPTCHA TIME");
Console.ReadKey();
Console.ReadKey();
reader = new PdfReader("http://www.equibase.com/premium/eqbPDFChartPlus.cfm?RACE=A&BorP=P&TID=SAR&CTRY=USA&DT=" + date + "&DAY=D&STYLE=EQB");
}
StringBuilder builder = new StringBuilder();
for (int x = 1; x <= reader.NumberOfPages; x++)
{
PdfDictionary page = reader.GetPageN(x);
IRenderListener listener = new SBTextRenderer(builder);
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(listener);
PdfDictionary pageDic = reader.GetPageN(x);
PdfDictionary resourcesDic = pageDic.GetAsDict(PdfName.RESOURCES);
processor.ProcessContent(ContentByteUtils.GetContentBytesForPage(reader, x), resourcesDic);
}
if (pages.Count != 0)
{
DataHandler handler = new DataHandler(dateTime, pages, m_dbConnection);
Thread thread = new Thread(new ThreadStart(handler.extractPdfData));
thread.Start();
thread.Join();
reader.Dispose();
pages.Clear();
}
else
{
// If there were no races on this particular day, simply skip it! :D
Console.WriteLine("Invalid Date: " + date);
}
}
public static void PopulateOperators(PdfContentStreamProcessor contentProcessor,
PdfCleanUpRenderListener pdfCleanUpRenderListener) {
foreach (String oper in contentProcessor.RegisteredOperatorStrings) {
PdfCleanUpContentOperator contentOperator = new PdfCleanUpContentOperator(pdfCleanUpRenderListener);
contentOperator.originalContentOperator = contentProcessor.RegisterContentOperator(oper, contentOperator);
}
}
public void extractSnippets(String src, String dest)
{
TextWriter output = new StreamWriter(new FileStream(dest, FileMode.Create));
PdfReader reader = new PdfReader(src);
IRenderListener listener = new MyTextRenderListener(output);
PdfContentStreamProcessor processor =
new PdfContentStreamProcessor(listener);
PdfDictionary pageDic = reader.GetPageN(1);
PdfDictionary resourcesDic = pageDic.GetAsDict(PdfName.RESOURCES);
processor.ProcessContent(ContentByteUtils.GetContentBytesForPage(reader, 1), resourcesDic);
output.Flush();
output.Close();
reader.Close();
}
virtual public void WeirdHyphensTest() {
PdfReader reader = TestResourceUtils.GetResourceAsPdfReader(TEST_RESOURCES_PATH, "WeirdHyphens.pdf");
List<String> textChunks = new List<String>();
IRenderListener listener = new MyTextRenderListener(textChunks);
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(listener);
PdfDictionary pageDic = reader.GetPageN(1);
PdfDictionary resourcesDic = pageDic.GetAsDict(PdfName.RESOURCES);
processor.ProcessContent(ContentByteUtils.GetContentBytesForPage(reader, 1), resourcesDic);
/**
* This assertion makes sure that encoding has been read properly from FontDescriptor.
* If not the vallue will be "\u0000 14".
*/
Assert.AreEqual("\u0096 14", textChunks[18]);
reader.Close();
}
private void ProcessBytes(
string resourceName,
int pageNumber)
{
PdfReader pdfReader = TestResourceUtils.GetResourceAsPdfReader(TEST_RESOURCES_PATH, resourceName);
PdfDictionary pageDictionary = pdfReader.GetPageN(pageNumber);
PdfDictionary resourceDictionary = pageDictionary.GetAsDict(PdfName.RESOURCES);
PdfObject contentObject = pageDictionary.Get(PdfName.CONTENTS);
byte[] contentBytes = ReadContentBytes(contentObject);
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(_renderListener);
processor.ProcessContent(contentBytes, resourceDictionary);
}
public void TestType3FontWidth() {
String inFile = "type3font_text.pdf";
LineSegment origLineSegment = new LineSegment(new Vector(20.3246f, 769.4974f, 1.0f), new Vector(151.22923f, 769.4974f, 1.0f));
PdfReader reader = TestResourceUtils.GetResourceAsPdfReader(TEST_RESOURCES_PATH, inFile);
TextPositionRenderListener renderListener = new TextPositionRenderListener();
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(renderListener);
PdfDictionary pageDic = reader.GetPageN(FIRST_PAGE);
PdfDictionary resourcesDic = pageDic.GetAsDict(PdfName.RESOURCES);
processor.ProcessContent(ContentByteUtils.GetContentBytesForPage(reader, FIRST_PAGE), resourcesDic);
Assert.AreEqual(renderListener.LineSegments[FIRST_ELEMENT_INDEX].GetStartPoint()[FIRST_ELEMENT_INDEX],
origLineSegment.GetStartPoint()[FIRST_ELEMENT_INDEX], 1/2f);
Assert.AreEqual(renderListener.LineSegments[FIRST_ELEMENT_INDEX].GetEndPoint()[FIRST_ELEMENT_INDEX],
origLineSegment.GetEndPoint()[FIRST_ELEMENT_INDEX], 1/2f);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber aw = (PdfNumber)operands[0];
PdfNumber ac = (PdfNumber)operands[1];
PdfString str = (PdfString)operands[2];
List <PdfObject> twOperands = new List <PdfObject>(1);
twOperands.Insert(0, aw);
setTextWordSpacing.Invoke(processor, null, twOperands);
List <PdfObject> tcOperands = new List <PdfObject>(1);
tcOperands.Insert(0, ac);
setTextCharacterSpacing.Invoke(processor, null, tcOperands);
List <PdfObject> tickOperands = new List <PdfObject>(1);
tickOperands.Insert(0, str);
moveNextLineAndShowText.Invoke(processor, null, tickOperands);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfName fontResourceName = (PdfName)operands[0];
float size = ((PdfNumber)operands[1]).FloatValue;
PdfDictionary fontsDictionary = processor.resources.GetAsDict(PdfName.FONT);
CMapAwareDocumentFont font;
PdfObject fontObject = fontsDictionary.Get(fontResourceName);
if (fontObject is PdfDictionary)
{
font = processor.GetFont((PdfDictionary)fontObject);
}
else
{
font = processor.GetFont((PRIndirectReference)fontObject);
}
processor.Gs().font = font;
processor.Gs().fontSize = size;
}
/**
* Searches for a tag in a page.
*
* @param tag
* the name of the tag
* @param obj
* an identifier to find the marked content
* @param page
* a page dictionary
* @throws IOException
*/
public virtual void ParseTag(String tag, PdfObject obj, PdfDictionary page)
{
// if the identifier is a number, we can extract the content right away
if (obj is PdfNumber)
{
PdfNumber mcid = (PdfNumber)obj;
RenderFilter filter = new MarkedContentRenderFilter(mcid.IntValue);
ITextExtractionStrategy strategy = new SimpleTextExtractionStrategy();
FilteredTextRenderListener listener = new FilteredTextRenderListener(strategy, new RenderFilter[] { filter });
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(
listener);
processor.ProcessContent(PdfReader.GetPageContent(page), page
.GetAsDict(PdfName.RESOURCES));
outp.Write(XMLUtil.EscapeXML(listener.GetResultantText(), true));
}
// if the identifier is an array, we call the parseTag method
// recursively
else if (obj is PdfArray)
{
PdfArray arr = (PdfArray)obj;
int n = arr.Size;
for (int i = 0; i < n; i++)
{
ParseTag(tag, arr[i], page);
if (i < n - 1)
{
outp.WriteLine();
}
}
}
// if the identifier is a dictionary, we get the resources from the
// dictionary
else if (obj is PdfDictionary)
{
PdfDictionary mcr = (PdfDictionary)obj;
ParseTag(tag, mcr.GetDirectObject(PdfName.MCID), mcr
.GetAsDict(PdfName.PG));
}
}
virtual public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
processor.Gs().colorSpaceStroke = (PdfName)operands[0];
}
public void HandleXObject(PdfContentStreamProcessor processor, PdfStream xobjectStream, PdfIndirectReference refi)
{
ImageRenderInfo renderInfo = ImageRenderInfo.CreateForXObject(processor.Gs().ctm, refi);
processor.renderListener.RenderImage(renderInfo);
}
public void HandleXObject(PdfContentStreamProcessor processor, PdfStream xobjectStream, PdfIndirectReference refi) {
PdfDictionary colorSpaceDic = processor.resources.GetAsDict(PdfName.COLORSPACE);
ImageRenderInfo renderInfo = ImageRenderInfo.CreateForXObject(processor.Gs().ctm, refi, colorSpaceDic);
processor.renderListener.RenderImage(renderInfo);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfName xobjectName = (PdfName)operands[0];
processor.DisplayXObject(xobjectName);
}
public void Invoke(PdfContentStreamProcessor processor,
PdfLiteral oper, List<PdfObject> operands)
{
PdfObject properties = operands[1];
processor.BeginMarkedContent((PdfName)operands[0], GetPropertiesDictionary(properties, processor.resources));
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
processor.textMatrix = null;
processor.textLineMatrix = null;
processor.EndText();
}
public void HandleXObject(PdfContentStreamProcessor processor, PdfStream xobjectStream, PdfIndirectReference refi)
{
// ignore XObject subtype
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfNumber render = (PdfNumber)operands[0];
processor.Gs().renderMode = render.IntValue;
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfName fontResourceName = (PdfName)operands[0];
float size = ((PdfNumber)operands[1]).FloatValue;
PdfDictionary fontsDictionary = processor.resources.GetAsDict(PdfName.FONT);
CMapAwareDocumentFont font;
PdfObject fontObject = fontsDictionary.Get(fontResourceName);
if (fontObject is PdfDictionary)
font = processor.GetFont((PdfDictionary)fontObject);
else
font = processor.GetFont((PRIndirectReference)fontObject);
processor.Gs().font = font;
processor.Gs().fontSize = size;
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
float tx = ((PdfNumber)operands[0]).FloatValue;
float ty = ((PdfNumber)operands[1]).FloatValue;
Matrix translationMatrix = new Matrix(tx, ty);
processor.textMatrix = translationMatrix.Multiply(processor.textLineMatrix);
processor.textLineMatrix = processor.textMatrix;
}
virtual public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber rise = (PdfNumber)operands[0];
processor.Gs().rise = rise.FloatValue;
}
virtual public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber charSpace = (PdfNumber)operands[0];
processor.Gs().characterSpacing = charSpace.FloatValue;
}
virtual public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
processor.Gs().strokeColor = GetColor(processor.Gs().colorSpaceStroke, operands);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
processor.textMatrix = null;
processor.textLineMatrix = null;
processor.EndText();
}
public void Invoke(PdfContentStreamProcessor processor,
PdfLiteral oper, List <PdfObject> operands)
{
processor.EndMarkedContent();
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfNumber leading = (PdfNumber)operands[0];
processor.Gs().leading = leading.FloatValue;
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfNumber scale = (PdfNumber)operands[0];
processor.Gs().horizontalScaling = scale.FloatValue/100f;
}
public virtual void SpaceTrimColumnTextTest() {
Document doc = new Document(PageSize.A4, 50, 30, 50, 30);
PdfWriter writer = PdfWriter.GetInstance(doc, new FileStream(OUTSPTRIMCT, FileMode.Create));
doc.Open();
Phrase under = new Phrase();
under.Font = new Font(Font.FontFamily.TIMES_ROMAN, 12, Font.UNDERLINE);
under.Add(new Chunk(" 1 1 9 "));
Paragraph underlineTest = new Paragraph(under);
underlineTest.KeepTogether = true;
doc.Add(underlineTest);
doc.Close();
writer.Close();
PdfReader reader = new PdfReader(OUTSPTRIMCT);
MyTextRenderListener listener = new MyTextRenderListener();
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(listener);
PdfDictionary pageDic = reader.GetPageN(1);
PdfDictionary resourcesDic = pageDic.GetAsDict(PdfName.RESOURCES);
processor.ProcessContent(ContentByteUtils.GetContentBytesForPage(reader, 1), resourcesDic);
Assert.IsTrue(listener.GetText().Length == 60, "Unexpected text length");
}
public void Invoke(PdfContentStreamProcessor processor,
PdfLiteral oper, List<PdfObject> operands)
{
processor.BeginMarkedContent((PdfName)operands[0], new PdfDictionary());
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
textMoveNextLine.Invoke(processor, null, new List <PdfObject>(0));
showText.Invoke(processor, null, operands);
}
public void Invoke(PdfContentStreamProcessor processor,
PdfLiteral oper, List<PdfObject> operands)
{
processor.EndMarkedContent();
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber render = (PdfNumber)operands[0];
processor.Gs().renderMode = render.IntValue;
}
public void HandleXObject(PdfContentStreamProcessor processor, PdfStream stream, PdfIndirectReference refi) {
PdfDictionary resources = stream.GetAsDict(PdfName.RESOURCES);
// we read the content bytes up here so if it fails we don't leave the graphics state stack corrupted
// this is probably not necessary (if we fail on this, probably the entire content stream processing
// operation should be rejected
byte[] contentBytes;
contentBytes = ContentByteUtils.GetContentBytesFromContentObject(stream);
PdfArray matrix = stream.GetAsArray(PdfName.MATRIX);
new PushGraphicsState().Invoke(processor, null, null);
if (matrix != null){
float a = matrix.GetAsNumber(0).FloatValue;
float b = matrix.GetAsNumber(1).FloatValue;
float c = matrix.GetAsNumber(2).FloatValue;
float d = matrix.GetAsNumber(3).FloatValue;
float e = matrix.GetAsNumber(4).FloatValue;
float f = matrix.GetAsNumber(5).FloatValue;
Matrix formMatrix = new Matrix(a, b, c, d, e, f);
processor.Gs().ctm = formMatrix.Multiply(processor.Gs().ctm);
}
processor.ProcessContent(contentBytes, resources);
new PopGraphicsState().Invoke(processor, null, null);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber scale = (PdfNumber)operands[0];
processor.Gs().horizontalScaling = scale.FloatValue / 100f;
}
public void HandleXObject(PdfContentStreamProcessor processor, PdfStream xobjectStream, PdfIndirectReference refi) {
// ignore XObject subtype
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
processor.gsStack.Pop();
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfNumber wordSpace = (PdfNumber)operands[0];
processor.Gs().wordSpacing = wordSpace.FloatValue;
}
virtual public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
processor.Gs().fillColor = GetColor(4, operands);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
// ignore the oper
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
PdfName dictionaryName = (PdfName)operands[0];
PdfDictionary extGState = processor.resources.GetAsDict(PdfName.EXTGSTATE);
if (extGState == null)
throw new ArgumentException(MessageLocalization.GetComposedMessage("resources.do.not.contain.extgstate.entry.unable.to.process.oper.1", oper));
PdfDictionary gsDic = extGState.GetAsDict(dictionaryName);
if (gsDic == null)
throw new ArgumentException(MessageLocalization.GetComposedMessage("1.is.an.unknown.graphics.state.dictionary", dictionaryName));
// at this point, all we care about is the FONT entry in the GS dictionary
PdfArray fontParameter = gsDic.GetAsArray(PdfName.FONT);
if (fontParameter != null){
CMapAwareDocumentFont font = processor.GetFont((PRIndirectReference)fontParameter[0]);
float size = fontParameter.GetAsNumber(1).FloatValue;
processor.Gs().font = font;
processor.Gs().fontSize = size;
}
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfString str = (PdfString)operands[0];
processor.DisplayPdfString(str);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
GraphicsState gs = processor.gsStack.Peek();
GraphicsState copy = new GraphicsState(gs);
processor.gsStack.Push(copy);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber leading = (PdfNumber)operands[0];
processor.Gs().leading = leading.FloatValue;
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
float a = ((PdfNumber)operands[0]).FloatValue;
float b = ((PdfNumber)operands[1]).FloatValue;
float c = ((PdfNumber)operands[2]).FloatValue;
float d = ((PdfNumber)operands[3]).FloatValue;
float e = ((PdfNumber)operands[4]).FloatValue;
float f = ((PdfNumber)operands[5]).FloatValue;
Matrix matrix = new Matrix(a, b, c, d, e, f);
GraphicsState gs = processor.gsStack.Peek();
gs.ctm = matrix.Multiply(gs.ctm);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfNumber wordSpace = (PdfNumber)operands[0];
processor.Gs().wordSpacing = wordSpace.FloatValue;
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
processor.Gs().fillColor = GetColor(4, operands);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
processor.textMatrix = new Matrix();
processor.textLineMatrix = processor.textMatrix;
processor.BeginText();
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
processor.Gs().colorSpaceStroke = (PdfName)operands[0];
}
public void Invoke(PdfContentStreamProcessor processor,
PdfLiteral oper, List <PdfObject> operands)
{
processor.BeginMarkedContent((PdfName)operands[0], new PdfDictionary());
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
processor.Gs().strokeColor = GetColor(processor.Gs().colorSpaceStroke, operands);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List <PdfObject> operands)
{
PdfName xobjectName = (PdfName)operands[0];
processor.DisplayXObject(xobjectName);
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
processor.gsStack.Pop();
}
public override void ParseTag(String tag, PdfObject obj, PdfDictionary page) {
if (obj is PdfNumber) {
PdfNumber mcid = (PdfNumber) obj;
RenderFilter filter = new MyMarkedContentRenderFilter(mcid.IntValue);
ITextExtractionStrategy strategy = new SimpleTextExtractionStrategy();
FilteredTextRenderListener listener = new FilteredTextRenderListener(
strategy, filter);
PdfContentStreamProcessor processor = new PdfContentStreamProcessor(
listener);
processor.ProcessContent(PdfReader.GetPageContent(page), page
.GetAsDict(PdfName.RESOURCES));
outp.Write(XMLUtil.EscapeXML(listener.GetResultantText(), true));
}
else {
base.ParseTag(tag, obj, page);
}
}
public void Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List<PdfObject> operands) {
processor.textMatrix = new Matrix();
processor.textLineMatrix = processor.textMatrix;
processor.BeginText();
}
