public static POCO.DocumentText ExtractText(IFormFile upload)
{
POCO.DocumentText text = new POCO.DocumentText();
// Open the presentation as read-only.
using (PresentationDocument presentationDocument = PresentationDocument.Open(upload.OpenReadStream(), false))
{
// Check for a null document object.
if (presentationDocument == null)
{
throw new ArgumentNullException("presentationDocument");
}
int numSlides = CountSlides(presentationDocument);
for (int i = 0; i < numSlides; i++)
{
// Get the text for the slide
string slideText = GetSlideIdAndText(presentationDocument, i);
// Create a new OfficePart object
POCO.DocumentPart slide = new POCO.DocumentPart();
slide.partnumber = i;
slide.body = slideText;
slide.header = string.Empty;
slide.footer = string.Empty;
// Add to our parts collection
text.parts.Add(slide);
}
}
return(text);
}
private static DocumentText ProcessFile(MemoryStream memstream)
{
int runCounter = 0;
POCO.DocumentText text = new POCO.DocumentText();
try
{
// Open the document as read-only.
using (WordprocessingDocument wordDocument = WordprocessingDocument.Open(memstream, false))
{
// Check for a null document object.
if (wordDocument == null)
{
throw new ArgumentNullException("wordDocument");
}
int paraCounter = 0;
foreach (var paragraph in wordDocument.MainDocumentPart.RootElement.Descendants <Paragraph>())
{
paraCounter++;
string paraText = string.Empty;
foreach (var run in paragraph.Elements <Run>())
{
runCounter++;
foreach (var texttype in run.Elements <TextType>())
{
paraText += texttype.Text;
}
//string textContent = run.Elements<DocumentFormat.OpenXml.Wordprocessing.TextType>().Aggregate("", (s, t) => s + t.Text);
}
// Check if any text was found
if (paraText != string.Empty)
{
POCO.DocumentPart part = new POCO.DocumentPart();
part.body = paraText;
part.partnumber = paraCounter;
text.parts.Add(part);
}
}
}
}
catch (OpenXmlPackageException packageEx)
{
if (packageEx.ToString().Contains("Invalid Hyperlink"))
{
MemoryStream fixedMemStream = new MemoryStream();
fixedMemStream = Castlepoint.Text.FileHandlers.Utils.FixInvalidUri(memstream);
text = ProcessFile(fixedMemStream);
}
}
return(text);
}
public static POCO.DocumentText ExtractText(IFormFile upload, ILogger logger)
{
POCO.DocumentText text = new POCO.DocumentText();
try
{
// Read the bytes from the stream
System.IO.MemoryStream memstream = new System.IO.MemoryStream();
upload.OpenReadStream().CopyTo(memstream);
text = ProcessFile(memstream);
return(text);
}
catch (Exception exWordExtractText)
{
logger.LogError("Word Extract Text: " + exWordExtractText.Message);
throw;
}
}
private static DocumentText ProcessFile2(MemoryStream memstream, string fileName, ILogger logger)
{
int runCounter = 0;
POCO.DocumentText text = new POCO.DocumentText();
// Auto-detect format, supports:
// - Binary Excel files (2.0-2003 format; *.xls)
// - OpenXml Excel files (2007 format; *.xlsx)
using (var reader = ExcelReaderFactory.CreateReader(memstream))
{
var result = reader.AsDataSet();
logger.LogInformation("ProcessFile: tables=" + result.Tables.Count.ToString() + " filename=" + fileName);
int sheetCounter = 0;
foreach (System.Data.DataTable table in result.Tables)
{
sheetCounter++;
logger.LogInformation("ProcessFile: processing sheet#" + sheetCounter.ToString() + " filename=" + fileName);
StringBuilder sbSheet = new StringBuilder();
foreach (System.Data.DataRow row in table.Rows)
{
// Append each row of data as tab-separated
sbSheet.Append(string.Join("\t", row.ItemArray));
sbSheet.AppendLine();
}
// Add this sheet as a new part
POCO.DocumentPart part = new POCO.DocumentPart();
part.body = sbSheet.ToString();
part.partnumber = sheetCounter;
text.parts.Add(part);
}
}
return(text);
}
public static POCO.DocumentText GetCommentsFromDocument(WordprocessingDocument wordDocument)
{
POCO.DocumentText text = new POCO.DocumentText();
WordprocessingCommentsPart commentsPart =
wordDocument.MainDocumentPart.WordprocessingCommentsPart;
int counterComments = 0;
if (commentsPart != null && commentsPart.Comments != null)
{
foreach (Comment comment in commentsPart.Comments.Elements <Comment>())
{
counterComments++;
POCO.DocumentPart part = new POCO.DocumentPart();
part.body = comment.InnerText;
part.partnumber = counterComments;
text.parts.Add(part);
}
}
return(text);
}
public async Task <IActionResult> Post([FromForm] UploadFile upload)
{
try
{
// Validate the upload
if (upload.file == null ||
upload.filename == null ||
upload.mimetype == null)
{
return(new BadRequestResult());
}
long fileLength = upload.file.Length;
if (fileLength > 50000000)
{
return(new BadRequestResult());
}
POCO.DocumentText text = new POCO.DocumentText();
Stopwatch st = new Stopwatch();
// Check the mimetype
switch (upload.mimetype.ToLower().Trim())
{
case "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
case "application/vnd.openxmlformats-officedocument.wordprocessingml.template":
// DOCX
st.Start();
text = Word.ExtractText(upload.file, _logger);
st.Stop();
_logger.LogInformation("TEXT [" + st.ElapsedMilliseconds + "ms] DOCX: " + upload.filename);
break;
case "application/vnd.openxmlformats-officedocument.presentationml.presentation":
case "application/vnd.openxmlformats-officedocument.presentationml.template":
case "application/vnd.openxmlformats-officedocument.presentationml.slideshow":
// PPTX
st.Start();
text = PowerPoint.ExtractText(upload.file);
st.Stop();
_logger.LogInformation("TEXT [" + st.ElapsedMilliseconds + "ms] PPTX: " + upload.filename);
break;
case "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet":
case "application/vnd.openxmlformats-officedocument.spreadsheetml.template":
// XLSX
st.Start();
text = Excel.ExtractText(upload.file, _logger);
st.Stop();
_logger.LogInformation("TEXT [" + st.ElapsedMilliseconds + "ms] XLSX: " + upload.filename);
break;
default:
return(new BadRequestResult());
}
// Serialize the extract result
string jsonText = JsonConvert.SerializeObject(text);
ObjectResult result = new ObjectResult(jsonText);
return(result);
}
catch (Exception exPost)
{
_logger.LogError("ExtractTextController: " + exPost.Message);
return(StatusCode((int)System.Net.HttpStatusCode.InternalServerError));
}
}
