/// <summary>
/// Based on method `ReplaceTextsWithTranslation` (line 667 onwards) in
/// TranslationAssistant.Business/DocumentTranslationManager.cs in
/// MicrosoftTranslator/DocumentTranslator
/// </summary>
private static async Task ReplaceTextsWithTranslation(List <DocumentFormat.OpenXml.Drawing.Text> texts, ITextTranslator textTranslator, string to, string from)
{
if (texts.Count() > 0)
{
// Extract text for translation
var values = texts.Select(text => text.Text);
// Do translation
var translations = await textTranslator.TranslateTexts(values, to, from);
// Apply translations to document by iterating through both lists and
// replacing the original text with its translation
using (var textsEnumerator = texts.GetEnumerator())
{
using (var translationsEnumerator = translations.GetEnumerator())
{
while (textsEnumerator.MoveNext() && translationsEnumerator.MoveNext())
{
textsEnumerator.Current.Text = translationsEnumerator.Current;
}
}
}
}
}
/// <summary>
/// Translates an Word document
///
/// Based on method `ProcessWordDocument` (line 726 onwards) in
/// TranslationAssistant.Business/DocumentTranslationManager.cs in
/// MicrosoftTranslator/DocumentTranslator
/// </summary>
public async Task TranslateDocument(MemoryStream memoryStream, ITextTranslator textTranslator, string to, string from = null, bool ignoreHidden = false)
{
using (WordprocessingDocument doc = WordprocessingDocument.Open(memoryStream, true))
{
// Simply the Word document mark-up
OpenXmlPowerTools.SimplifyMarkupSettings settings = new OpenXmlPowerTools.SimplifyMarkupSettings
{
AcceptRevisions = true,
NormalizeXml = true, //setting this to false reduces translation quality, but if true some documents have XML format errors when opening
RemoveBookmarks = true,
RemoveComments = true,
RemoveContentControls = true,
RemoveEndAndFootNotes = true,
RemoveFieldCodes = true,
RemoveGoBackBookmark = true,
//RemoveHyperlinks = false,
RemoveLastRenderedPageBreak = true,
RemoveMarkupForDocumentComparison = true,
RemovePermissions = false,
RemoveProof = true,
RemoveRsidInfo = true,
RemoveSmartTags = true,
RemoveSoftHyphens = true,
RemoveWebHidden = true,
ReplaceTabsWithSpaces = false
};
OpenXmlPowerTools.MarkupSimplifier.SimplifyMarkup(doc, settings);
}
var texts = new List <Text>();
using (WordprocessingDocument doc = WordprocessingDocument.Open(memoryStream, true))
{
// Find all text nodes in the document (body, headers & footers)
var body = doc.MainDocumentPart.Document.Body;
texts.AddRange(body.Descendants <Text>()
.Where(text => !String.IsNullOrEmpty(text.Text) && text.Text.Length > 0));
var headers = doc.MainDocumentPart.HeaderParts.Select(p => p.Header);
foreach (var header in headers)
{
texts.AddRange(header.Descendants <Text>().Where(text =>
!String.IsNullOrEmpty(text.Text) && text.Text.Length > 0));
}
var footers = doc.MainDocumentPart.FooterParts.Select(p => p.Footer);
foreach (var footer in footers)
{
texts.AddRange(footer.Descendants <Text>().Where(text =>
!String.IsNullOrEmpty(text.Text) && text.Text.Length > 0));
}
if (ignoreHidden)
{
texts.RemoveAll(t => t.Parent.Descendants <Vanish>().Any());
}
// Extract text strings for translation
var values = texts.Select(text => text.Text);
// Do Translation
var translations = await textTranslator.TranslateTexts(values, to, from);
// Apply translations to document by iterating through both lists and
// replacing the original with its translation
using (var textsEnumerator = texts.GetEnumerator())
{
using (var translationsEnumerator = translations.GetEnumerator())
{
while (textsEnumerator.MoveNext() && translationsEnumerator.MoveNext())
{
textsEnumerator.Current.Text = translationsEnumerator.Current;
}
}
}
}
}
/// <summary>
/// Translates an PowerPoint document
///
/// Based on method `ProcessPowerPointDocument` (line 569 onwards) in
/// TranslationAssistant.Business/DocumentTranslationManager.cs in
/// MicrosoftTranslator/DocumentTranslator
/// </summary>
public async Task TranslateDocument(MemoryStream memoryStream, ITextTranslator textTranslator, string to, string from = null)
{
using (PresentationDocument doc = PresentationDocument.Open(memoryStream, true))
{
var texts = new List <DocumentFormat.OpenXml.Drawing.Text>();
var notes = new List <DocumentFormat.OpenXml.Drawing.Text>();
var comments = new List <Comment>();
var slideParts = doc.PresentationPart.SlideParts;
if (slideParts != null)
{
// Find all text items, notes and comments in all slides
foreach (var slidePart in slideParts)
{
if (slidePart.Slide != null)
{
var slide = slidePart.Slide;
ExtractTextContent(texts, slide);
var commentsPart = slidePart.SlideCommentsPart;
if (commentsPart != null)
{
comments.AddRange(commentsPart.CommentList.Cast <Comment>());
}
var notesPart = slidePart.NotesSlidePart;
if (notesPart != null)
{
ExtractTextContent(notes, notesPart.NotesSlide);
}
}
}
// Translate and replace the text items
await ReplaceTextsWithTranslation(texts, textTranslator, to, from);
// Translate and replace the notes
await ReplaceTextsWithTranslation(notes, textTranslator, to, from);
// Translate and replace the comments
if (comments.Count() > 0)
{
// Extract text from comment for translation
var values = comments.Select(text => text.InnerText);
// Do translation
var translatedComments = await textTranslator.TranslateTexts(values, to, from);
// Apply translations to document by iterating through both lists and
// replacing the original comment text with its translation
using (var commentsEnumerator = comments.GetEnumerator())
{
using (var translationsEnumerator = translatedComments.GetEnumerator())
{
while (commentsEnumerator.MoveNext() && translationsEnumerator.MoveNext())
{
commentsEnumerator.Current.Text = new DocumentFormat.OpenXml.Presentation.Text
{
Text = translationsEnumerator.Current
};
}
}
}
}
}
}
}
/// <summary>
/// Translates an Excel document
///
/// Based on method `ProcessExcelDocument` (line 424 onwards) in
/// TranslationAssistant.Business/DocumentTranslationManager.cs in
/// MicrosoftTranslator/DocumentTranslator
/// </summary>
public async Task TranslateDocument(MemoryStream memoryStream, ITextTranslator textTranslator, string to, string from = null)
{
using (SpreadsheetDocument document = SpreadsheetDocument.Open(memoryStream, true))
{
// Find all string items in the spreadsheet
List <DocumentFormat.OpenXml.Spreadsheet.Text> texts = new List <DocumentFormat.OpenXml.Spreadsheet.Text>();
foreach (SharedStringItem si in document.WorkbookPart.SharedStringTablePart.SharedStringTable.Elements <SharedStringItem>())
{
if (si != null && si.Text != null && !String.IsNullOrEmpty(si.Text.Text))
{
texts.Add(si.Text);
}
else if (si != null)
{
texts
.AddRange(si.Elements <DocumentFormat.OpenXml.Spreadsheet.Run>()
.Where(item => (item != null && item.Text != null && !String.IsNullOrEmpty(item.Text.Text)))
.Select(item => item.Text));
}
}
// Extract text for translation
var textValues = texts.Select(item => item.Text);
// Do the translation
var translations = await textTranslator.TranslateTexts(textValues, to, from);
// Apply translations to document by iterating through both lists and
// replacing the original text with its translation
using (var textsEnumerator = texts.GetEnumerator())
{
using (var translationEnumerator = translations.GetEnumerator())
{
while (textsEnumerator.MoveNext() && translationEnumerator.MoveNext())
{
textsEnumerator.Current.Text = translationEnumerator.Current;
}
}
}
// Refresh all the shared string references.
var tables = document.WorkbookPart.GetPartsOfType <WorksheetPart>()
.Select(part => part.TableDefinitionParts)
.SelectMany(_tables => _tables);
foreach (var table in tables)
{
foreach (TableColumn col in table.Table.TableColumns)
{
col.Name = translations.Take(int.Parse(col.Id)).Last();
}
table.Table.Save();
}
// Find all comments
WorkbookPart workBookPart = document.WorkbookPart;
List <DocumentFormat.OpenXml.Spreadsheet.Comment> comments = new List <DocumentFormat.OpenXml.Spreadsheet.Comment>();
foreach (var commentsPart in workBookPart.WorksheetParts.SelectMany(sheet => sheet.GetPartsOfType <WorksheetCommentsPart>()))
{
comments.AddRange(commentsPart.Comments.CommentList.Cast <Comment>());
}
// Extract text for translation
var commentValues = comments.Select(item => item.InnerText).ToArray();
// Do the translation
var translatedComments = await textTranslator.TranslateTexts(commentValues, to, from);
// Apply translations to document by iterating through both lists and
// replacing the original comment text with its translation
using (var commentsEnumerator = comments.GetEnumerator())
{
using (var translationEnumerator = translations.GetEnumerator())
{
while (commentsEnumerator.MoveNext() && translationEnumerator.MoveNext())
{
var text = translationEnumerator.Current;
commentsEnumerator.Current.CommentText = new CommentText
{
Text = new DocumentFormat.OpenXml.Spreadsheet.Text {
Text = text
}
};
}
}
}
}
}