public static void Process(string basename, string inputfolder, string outputfolder)
{
PdfReaderException.ContinueOnException();
var conteudos = GetTextLines(basename, inputfolder, outputfolder, out Execution.Pipeline pipeline)
.Log <AnalyzeLines>($"{outputfolder}/{basename}-lines.txt")
.ConvertText <CreateStructures, TextStructure>()
.Log <AnalyzeStructuresCentral>($"{outputfolder}/{basename}-central.txt")
//.PrintAnalytics($"bin/{basename}-print-analytics.txt")
.ConvertText <CreateTextSegments, TextSegment>()
.Log <AnalyzeSegmentTitles>($"{outputfolder}/{basename}-tree.txt")
.Log <AnalyzeSegmentStats>($"{outputfolder}/{basename}-segments-stats.txt")
.Log <AnalyzeSegments2>($"{outputfolder}/{basename}-segments.csv")
.ConvertText <CreateTreeSegments, TextSegment>()
.ConvertText <TransformConteudo, Conteudo>()
.ToList();
//Create XML
var createArticle = new TransformArtigo();
var artigos = createArticle.Create(conteudos);
createArticle.CreateXML(artigos, outputfolder, basename);
pipeline.Done();
}
public static void RunCreateArtigosJson(IVirtualFS virtualFS, string basename, string inputfolder, string tmpfolder, string outputfolder)
{
VirtualFS.ConfigureFileSystem(virtualFS);
PdfReaderException.ContinueOnException();
Pipeline pipeline = new Pipeline();
var conteudo = GetTextLines(pipeline, basename, inputfolder, tmpfolder) // use temp folder
.ConvertText <CreateTextLineIndex, TextLine>()
.ConvertText <PreCreateStructures, TextLine2>()
.ConvertText <CreateStructures2, TextStructure>()
.ConvertText <PreCreateTextSegments, TextStructureAgg>()
.ConvertText <AggregateStructures, TextStructure>()
.ConvertText <CreateTextSegments, TextSegment>()
.ConvertText <CreateTreeSegments, TextSegment>()
.Log <AnalyzeSegmentTitles>($"{tmpfolder}/{basename}/segment-titles-tree.txt")
.ConvertText <TransformConteudo, Conteudo>()
.ToList();
var createArticle = new TransformArtigo();
var artigos = createArticle.Create(conteudo);
createArticle.CreateJson(artigos, $"{outputfolder}/{basename}", basename);
}
public void Process()
{
var pipelineText = _context.GetPipelineText <TextSegment>();
var conteudo = pipelineText
.ConvertText <TransformConteudo2, Conteudo>()
.ToList();
var createArticle = new TransformArtigo();
var artigos = createArticle.Create(conteudo);
createArticle.CreateXML(artigos, $"{_context.OutputFolder}/{_context.Basename}/artigos", _context.Basename);
_context.AddOutput("artigosGN", $"{_context.OutputFolder}/{_context.Basename}/artigos/{_context.Basename}-artigo{{0}}.xml");
}
public static void Process(string basename, string inputfolder, string outputfolder)
{
PdfReaderException.ContinueOnException();
var conteudos = GetTextLines(basename, inputfolder, outputfolder, out Execution.Pipeline pipeline)
.ConvertText <CreateTextLineIndex, TextLine>()
.Log <AnalyzeLines>($"{outputfolder}/{basename}-lines.txt")
.Log <AnalyzeLinesCenterRight>($"{outputfolder}/{basename}-;lines-center-right.txt")
.ConvertText <CreateStructures, TextStructure>()
.Log <AnalyzeStructuresCentral>($"{outputfolder}/{basename}-central.txt")
//.PrintAnalytics($"bin/{basename}-print-analytics.txt")
.ConvertText <CreateTextSegments, TextSegment>()
.Log <AnalyzeSegmentTitles>($"{outputfolder}/{basename}-tree.txt")
.Log <AnalyzeSegmentStats>($"{outputfolder}/{basename}-segments-stats.txt")
.Log <AnalyzeSegments2>($"{outputfolder}/{basename}-segments.csv")
.ConvertText <CreateTreeSegments, TextSegment>()
.Log <AnalyzeTreeStructure>($"{outputfolder}/{basename}-tree-hier.txt")
.ConvertText <TransformConteudo, Conteudo>()
.ToList();
//Create XML
var createArticle = new TransformArtigo();
var artigos = createArticle.Create(conteudos);
createArticle.CreateXML(artigos, outputfolder, basename);
var validation = pipeline.Statistics.Calculate <ValidateFooter, StatsPageFooter>();
var layout = (ValidateLayout)pipeline.Statistics.Calculate <ValidateLayout, StatsPageLayout>();
var overlap = (ValidateOverlap)pipeline.Statistics.Calculate <ValidateOverlap, StatsBlocksOverlapped>();
var pagesLayout = layout.GetPageErrors().ToList();
var pagesOverlap = overlap.GetPageErrors().ToList();
var pages = pagesLayout.Concat(pagesOverlap).Distinct().OrderBy(t => t).ToList();
if (pages.Count > 0)
{
ExtractPages($"{outputfolder}/{basename}-output", $"{outputfolder}/{basename}-parser-errors", pages);
}
pipeline.Done();
}
public static void ProcessStats(string basename)
{
//PdfWriteText.Test();
//return;
Console.WriteLine();
Console.WriteLine("Program4 - Processing with hierachy");
Console.WriteLine();
// Extract(1);
Examples.FollowText(basename);
Examples.ShowHeaderFooter(basename);
ExamplesWork.PrintAllSteps(basename, "bin", "bin");
PdfReaderException.ContinueOnException();
var conteudos = GetTextLinesWithPipelineBlockset(basename, out Execution.Pipeline pipeline)
.Log <AnalyzeLines>($"bin/{basename}-lines.txt")
.ConvertText <CreateStructures, TextStructure>()
.Log <AnalyzeStructuresCentral>($"bin/{basename}-central.txt")
//.PrintAnalytics($"bin/{basename}-print-analytics.txt")
.ConvertText <CreateTextSegments, TextSegment>()
.Log <AnalyzeSegmentTitles>($"bin/{basename}-tree.txt")
.Log <AnalyzeSegmentStats>($"bin/{basename}-segments-stats.txt")
.Log <AnalyzeSegments2>($"bin/{basename}-segments.csv")
.ConvertText <CreateTreeSegments, TextSegment>()
.ConvertText <TransformConteudo, Conteudo>()
.ToList();
var validation = pipeline.Statistics.Calculate <ValidateFooter, StatsPageFooter>();
//Create XML
var createArticle = new TransformArtigo();
var artigos = createArticle.Create(conteudos);
createArticle.CreateXML(artigos, basename);
var validator = new ProgramValidatorXML();
validator.ValidateArticle("bin");
}
public static void Process(string basename, string inputfolder, string outputfolder)
{
logDir = Directory.CreateDirectory($"{outputfolder}/Log").FullName;
xmlDir = Directory.CreateDirectory($"{outputfolder}/XMLs").FullName;
errors = Directory.CreateDirectory($"{outputfolder}/PDF-Errors").FullName;
PdfReaderException.ContinueOnException();
//ExamplesWork.PrintAllSteps(basename, inputfolder, outputfolder);
var conteudos = GetTextLines(basename, inputfolder, outputfolder, out Execution.Pipeline pipeline)
.Log <AnalyzeLines>($"{logDir}/{basename}-lines.txt")
.ConvertText <CreateTextLineIndex, TextLine>()
.ConvertText <CreateStructures, TextStructure>()
.Log <AnalyzeStructuresCentral>($"{logDir}/{basename}-central.txt")
.ShowPdf <ShowStructureCentral>($"{logDir}/{basename}-show-central.pdf")
.PrintAnalytics($"{logDir}/{basename}-print-analytics.txt")
.ConvertText <CreateTextSegments, TextSegment>()
.Log <AnalyzeSegmentTitles>($"{logDir}/{basename}-tree.txt")
.Log <AnalyzeSegmentStats>($"{logDir}/{basename}-segments-stats.txt")
.Log <AnalyzeSegments2>($"{logDir}/{basename}-segments.csv")
.ConvertText <CreateTreeSegments, TextSegment>()
.Log <AnalyzeTreeStructure>($"{logDir}/{basename}-tree-data.txt")
.ConvertText <TransformConteudo, Conteudo>()
.ConvertText <AggregateAnexo, Conteudo>()
.ConvertText <AggregateSingularBody, Conteudo>()
.ToList();
//Create XML
var createArticle = new TransformArtigo();
var artigos = createArticle.Create(conteudos);
createArticle.CreateXML(artigos, xmlDir, basename);
pipeline.ExtractOutput <ShowParserWarnings>($"{errors}/{basename}-parser-errors.pdf");
pipeline.Done();
var validator = new ProgramValidatorXML();
validator.ValidateArticle(outputfolder);
}
