public void Process()
{
var pipelineText = _context.GetPipelineText <TextStructure>();
var resultPipeline = pipelineText
.ConvertText <CreateTextSegments, TextSegment>()
.ConvertText <FilterTextWithFontsSegments, TextSegment>()
.ConvertText <AfterFilterTextSegments, TextSegment>();
_context.SetPipelineText <TextSegment>(resultPipeline);
}
public void Process()
{
var pipelineText = _context.GetPipelineText <TextSegment>();
var resultPipeline = pipelineText
.ConvertText <CreateContent, TextSegment>();
_context.SetPipelineText <TextSegment>(resultPipeline);
}
public void Process()
{
Pipeline pipeline = _context.GetPipeline();
var pipelineText = pipeline.Input($"{_context.InputFilePrefix}.pdf")
.AllPages <CreateTextLines>(GetLines)
.ConvertText <CreateTextLineIndex, TextLine>(true);
_context.SetPipelineText <TextLine>(pipelineText);
}
public void Process()
{
var pipelineText = _context.GetPipelineText <TextSegment>();
var resultPipeline = pipelineText
.ConvertText <CreateStructText, TextSegment>(true)
.Log <AnalyzeSegmentTextVersion>($"{_context.OutputFilePrefix}-text-version.txt");
_context.SetPipelineText <TextSegment>(resultPipeline);
}
public void Process()
{
var pipelineText = _context.GetPipelineText <TextSegment>();
var resultPipeline = pipelineText
.ConvertText <CreateTreeSegments, TextSegment>(true)
//.ConvertText<MergeTreeSegments, TextSegment>(true)
.Log <AnalyzeSegmentTitles>($"{_context.OutputFilePrefix}-analyze-segment-titles.txt")
.Log <AnalyzeTreeStructure>($"{_context.OutputFilePrefix}-tree.txt");
_context.SetPipelineText <TextSegment>(resultPipeline);
_context.AddOutput("tree", $"{_context.OutputFilePrefix}-tree.txt");
}
public void Process()
{
var pipelineText = _context.GetPipelineText <TextSegment>();
var resultPipeline = pipelineText
.ConvertText <CreateTreeSegments, TextSegment>(true)
.ShowPdf <ShowTitleSegment>($"{_context.OutputFilePrefix}-titles.pdf")
.Log <AnalyzeSegmentTitles>($"{_context.OutputFilePrefix}-analyze-segment-titles.txt")
.Log <AnalyzeTreeStructureFontSize2>($"{_context.OutputFilePrefix}-tree-fontsize.txt")
.Log <AnalyzeTreeStructure2>($"{_context.OutputFilePrefix}-tree.txt");
_context.SetPipelineText <TextSegment>(resultPipeline);
_context.AddOutput("titles", $"{_context.OutputFilePrefix}-titles.pdf");
_context.AddOutput("tree", $"{_context.OutputFilePrefix}-tree.txt");
}
public void Process()
{
var pipelineText = _context.GetPipelineText <TextLine>();
var resultPipeline = pipelineText
.ConvertText <PreCreateStructures, TextLine2>()
.ConvertText <CreateStructures3, TextStructure>()
.ConvertText <PreCreateTextSegments, TextStructureAgg>()
.ConvertText <AggregateStructures, TextStructure>(true)
.ShowPdf <ShowStructureCentral>($"{_context.OutputFilePrefix}-show-central.pdf")
.Log <AnalyzeStructures>($"{_context.OutputFilePrefix}-analyze-structures.txt")
.Log <AnalyzeStructuresCentral>($"{_context.OutputFilePrefix}-analyze-structures-central.txt")
.ConvertText <CreateTextSegments, TextSegment>();
_context.SetPipelineText <TextSegment>(resultPipeline);
}
public void Process()
{
Pipeline pipeline = _context.GetPipeline();
var result = pipeline.Input($"{_context.InputFilePrefix}.pdf")
.AllPages <CreateTextLines>(GetLines)
.ConvertText <CreateTextLineIndex, TextLine>(true)
.ConvertText <PreCreateStructures, TextLine2>()
.ConvertText <CreateStructures3, TextStructure>()
.ConvertText <PreCreateTextSegments, TextStructureAgg>()
.ConvertText <AggregateStructures, TextStructure>(true)
.ShowPdf <ShowStructureCentral>($"{_context.OutputFilePrefix}-show-central.pdf")
.Log <AnalyzeStructures>($"{_context.OutputFilePrefix}-analyze-structures.txt")
.Log <AnalyzeStructuresCentral>($"{_context.OutputFilePrefix}-analyze-structures-central.txt");
_context.SetPipelineText <TextStructure>(result);
_context.AddOutput("show-central", $"{_context.OutputFilePrefix}-show-central.pdf");
}