/// <summary>
/// Parses the contents of a raw, untagged DocFile into a new Document instance.
/// </summary>
/// <param name="doc">The raw, untagged DocFile to parse.</param>
/// <returns>
/// The contents of the DocFile composed into a fully reified
/// <see cref="Document"/> instance.
/// </returns>
public Document DocumentFromDoc(DocFile doc)
{
var docx = new DocToDocXConverter(doc).ConvertFile() as DocXFile;
var txt = new DocxToTextConverter(docx).ConvertFile();
return(new TaggedSourceParser(new TaggedFile(new SharpNLPTagger(this.TaggerMode, txt.FullPath).ProcessFile())).LoadDocument(txt.NameSansExt));
}
/// <summary>
/// Returns a Task<string> which when awaited yields all of the text in the DocXFile.
/// </summary>
/// <returns>A Task<string> which when awaited yields all of the text in the DocXFile.</returns>
public override async Task <string> LoadTextAsync()
{
var converter = new DocxToTextConverter(this);
var txtFile = await converter.ConvertFileAsync().ConfigureAwait(false);
return(await txtFile.LoadTextAsync().ConfigureAwait(false));
}
public void ConvertFileTest()
{
var target = new DocxToTextConverter(Input);
TxtFile actual;
actual = target.ConvertFile();
Check.That(actual.FullPath).Satisfies(File.Exists);
}
public async Task ConvertFileAsyncTest()
{
var target = new DocxToTextConverter(Input);
TxtFile actual;
actual = await target.ConvertFileAsync();
Check.That(FileInfo(actual.FullPath)).Satisfies(x => x.Exists);
}
public void Execute(IndexFieldEvent e)
{
var umbracoFileName = e.UmbracoProperty.Value != null?e.UmbracoProperty.Value.ToString() : string.Empty;
var docPath = ConfigurationManager.AppSettings["Dexter:DocumentPath"];
var filePath = string.IsNullOrWhiteSpace(docPath)
? umbracoFileName
: umbracoFileName.Replace("~/media", ConfigurationManager.AppSettings["Dexter:DocumentPath"])
.Replace("/media", ConfigurationManager.AppSettings["Dexter:DocumentPath"]);
var text = string.Empty;
switch (System.IO.Path.GetExtension(filePath))
{
case ".pdf":
text = new PdfToTextConverter().Convert(filePath);
break;
case ".doc":
text = new DocToTextConverter().Convert(filePath);
break;
case ".xls":
text = new XlsToTextConverter().Convert(filePath);
break;
case ".docx":
text = new DocxToTextConverter().Convert(filePath);
break;
case ".xlsx":
text = new XlsxToTextConverter().Convert(filePath);
break;
case ".pptx":
text = new PptxToTextConverter().Convert(filePath);
break;
case ".ppt":
text = System.IO.Path.GetFileName(filePath);
break;
case ".zip":
text = new ZipToTextConverter().Convert(filePath);
break;
}
e.Value = string.Join(" ", text.Split(new[] { ' ' }).Except(IGNORE));
}
/// <summary>
/// Returns a single string containing all of the text in the DocXFile.
/// </summary>
/// <returns>A string containing all of the text in the DocXFile.</returns>
public override string LoadText()
{
var converter = new DocxToTextConverter(this);
return(converter.ConvertFile().LoadText());
}
public void DocxToTextConverterConstructorTest()
{
var target = new DocxToTextConverter(Input);
Check.That(target.Original.FullPath).IsEqualTo(Input.FullPath);
}
