/// <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));
}
public void DocToDocXConverterConstructorTest()
{
var infile = Input;
var target = new DocToDocXConverter(infile);
Check.That(target.Original).IsEqualTo(infile);
}
public void ConvertFileTest()
{
var infile = Input;
var target = new DocToDocXConverter(infile);
InputFile actual;
actual = target.ConvertFile();
Check.That(File.Exists(actual.FullPath)).IsTrue();
}
public async Task ConvertFileAsyncTest()
{
var infile = Input;
var target = new DocToDocXConverter(infile);
InputFile actual;
actual = await target.ConvertFileAsync();
Check.That(File.Exists(actual.FullPath)).IsTrue();
}
/// <summary>
/// Returns a single string containing all of the text in the DocFile.
/// </summary>
/// <returns>A string containing all of the text in the DocFile.</returns>
public override string LoadText()
{
var todocXConverter = new DocToDocXConverter(this);
try
{
return(todocXConverter.ConvertFile().LoadText());
}
catch (IOException e)
{
throw CreateFileConversionFailureException("DOCX", e);
}
}
/// <summary>
/// Returns a Task<string> which when awaited yields all of the text in the DocFile.
/// </summary>
/// <returns>A Task<string> which when awaited yields all of the text in the DocFile.</returns>
public override async Task <string> LoadTextAsync()
{
var toDocXConverter = new DocToDocXConverter(this);
try
{
var docx = await toDocXConverter.ConvertFileAsync().ConfigureAwait(false);
return(await docx.LoadTextAsync().ConfigureAwait(false));
}
catch (IOException e)
{
throw CreateFileConversionFailureException("DOCX", e);
}
}
