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);
}
public static void Validate(string basename, string inputfolder, string outputfolder)
{
var validator = new ProgramValidatorXML();
validator.ValidateArticle($"{inputfolder}");
}