public static string ConvertDocx2Html(string fileInput, string fileOutput, int margin)
{
var fileInfo = new FileInfo(fileInput);
string fullFilePath = fileInfo.FullName;
string htmlText = string.Empty;
try
{
htmlText = ParseDOCX(fileInfo, margin);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (FileStream fs = new FileStream(fullFilePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
htmlText = ParseDOCX(fileInfo, margin);
}
}
var writer = File.CreateText(fileOutput + ".html");
writer.WriteLine(htmlText.ToString());
writer.Dispose();
return(fileOutput);
}
private async Task <IActionResult> ConvertDocxToHtmlWithImageAsync(string fileName, byte[] fileBytes)
{
try
{
var docFilePath = Path.Combine(_tempPath, fileName);
System.IO.File.WriteAllBytes(docFilePath, fileBytes);
string htmlText = string.Empty;
try
{
htmlText = ParseDOCX(docFilePath);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (FileStream fs = new FileStream(docFilePath, FileMode.Open, FileAccess.Read))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
htmlText = ParseDOCX(docFilePath);
}
}
var outfile = Path.Combine(_tempPath, ("htmldata" + ".html"));
var writer = System.IO.File.CreateText(outfile);
writer.WriteLine(htmlText.ToString());
writer.Dispose();
System.IO.File.Delete(docFilePath);
return(File(await System.IO.File.ReadAllBytesAsync(outfile), "application/octet-stream", (Path.GetFileNameWithoutExtension(fileName) + ".html")));
}
catch (Exception ex)
{
return(StatusCode(StatusCodes.Status500InternalServerError));
}
}
/// <summary>
/// Code obtained from CodeProject on GitHub
/// https://github.com/zaagan/Docx-Html-Docx/blob/master/DocxToHTML.Converter/HTMLConverter.cs
/// </summary>
/// <param name="fullFilePath"></param>
/// <returns>html of document</returns>
public string ConvertToHtml(string fullFilePath)
{
if (string.IsNullOrEmpty(fullFilePath) || Path.GetExtension(fullFilePath) != ".docx")
{
return("Unsupported format");
}
FileInfo fileInfo = new FileInfo(fullFilePath);
string htmlText = string.Empty;
try
{
htmlText = ParseDOCX(fileInfo);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (FileStream fs = new FileStream(fullFilePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
htmlText = ParseDOCX(fileInfo);
}
}
return(htmlText);
}
public static Dictionary <uint, string[]> GetTextOnSlides(string presentationFile)
{
var slides = new Dictionary <uint, string[]>();
try
{
using (PresentationDocument presentationDocument = PresentationDocument.Open(presentationFile, false))
{
PresentationPart presentationPart = presentationDocument.PresentationPart;
// Verify that the presentation part and presentation exist.
if (presentationPart != null && presentationPart.Presentation != null)
{
// Get the Presentation object from the presentation part.
Presentation presentation = presentationPart.Presentation;
// Verify that the slide ID list exists.
if (presentation.SlideIdList != null)
{
// Get the collection of slide IDs from the slide ID list.
var slideIds = presentation.SlideIdList.ChildElements;
foreach (SlideId sld in slideIds)
{
string slidePartRelationshipId = sld.RelationshipId;
// Get the specified slide part from the relationship ID.
SlidePart slidePart = (SlidePart)presentationPart.GetPartById(slidePartRelationshipId);
// Pass the slide part to the next method, and
// then return the array of strings that method
// returns to the previous method.
slides.Add(sld.Id, GetAllTextInSlide(slidePart) ?? new string[] { });
}
}
}
}
return(slides);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (FileStream fs = new FileStream(presentationFile, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
return(GetTextOnSlides(presentationFile));
}
throw e;
}
}
public DataTable techBaseTable = new DataTable(); // Cюда записана техническая база
public bool Optimize()
{
try // код помещён в блок try, потому что возможно исключение типа OpenXmlPackageException - Invalid Hyperlink
{
using (SpreadsheetDocument spreadSheetDocument = SpreadsheetDocument.Open(techBasePath, true))
{
techBaseTable = DataTableFromExcel.GetDataTableFromSpreadSheet(spreadSheetDocument);
}
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink")) // если исключение сожержит "Invalid Hyperlink"
{
using (FileStream fs = new FileStream(techBasePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => UriFixer.FixUri(brokenUri)); // чиним Uri
}
using (SpreadsheetDocument spreadSheetDocument = SpreadsheetDocument.Open(techBasePath, true))
{
techBaseTable = DataTableFromExcel.GetDataTableFromSpreadSheet(spreadSheetDocument);
}
}
}
catch (Exception)
{
return(false);
}
foreach (DataRow row in techBaseTable.Rows) // Оптимизация тех. базы - убираем цифры
{
string product = System.Convert.ToString(row[1]); // Название продукта
string productWithoutDigits = "";
for (int i = 0; i < product.Length; i++) // Убираем числа
{
if (!char.IsDigit(product[i]))
{
productWithoutDigits += product[i];
}
}
row[1] = productWithoutDigits;
}
return(true);
}
static string[] stringSeparators = new string[] { ", " }; // разделитель строк, содержащих интересуюие продукты
public static string Fill(ref List <ProductsOfInterest> products, List <string> sheets_names, List <ExcelFile> files, Report report)
{
products = new List <ProductsOfInterest>();
foreach (string sheetFromListBox in sheets_names)
{
foreach (ExcelFile file in files)
{
foreach (string sheetFromFile in file.Sheets)
{
if (sheetFromListBox.Equals(sheetFromFile))
{
try // код помещён в блок try, потому что возможно исключение типа OpenXmlPackageException - Invalid Hyperlink
{
using (SpreadsheetDocument spreadSheetDocument = SpreadsheetDocument.Open(file.FilePath, true))
{
HelpFill(spreadSheetDocument, sheetFromFile, file, ref products, report);
}
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink")) // если исключение сожержит "Invalid Hyperlink"
{
using (FileStream fs = new FileStream(file.FilePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => UriFixer.FixUri(brokenUri)); // чиним Uri
}
using (SpreadsheetDocument spreadSheetDocument = SpreadsheetDocument.Open(file.FilePath, true))
{
HelpFill(spreadSheetDocument, sheetFromFile, file, ref products, report);
}
}
}
catch (Exception e)
{
return(file.FilePath);
}
}
}
}
}
return(null);
}
public string ConvertToHTML(byte[] wordBytes, string fileName)
{
string htmlText = string.Empty;
try
{
htmlText = ParseDOCX(wordBytes, fileName);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (MemoryStream ms = new MemoryStream(wordBytes))
{
UriFixer.FixInvalidUri(ms, brokenUri => FixUri(brokenUri));
}
htmlText = ParseDOCX(wordBytes, fileName);
}
}
return(htmlText);
}
public static string GetHtmlfromWord(string fileInput, int margin)
{
var fileInfo = new FileInfo(fileInput);
string fullFilePath = fileInfo.FullName;
string htmlText = string.Empty;
try
{
htmlText = ParseDOCX(fileInfo, margin);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (FileStream fs = new FileStream(fullFilePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
htmlText = ParseDOCX(fileInfo, margin);
}
}
return(htmlText);
}
public static string ConvertDocx2Pdf(string fileInput, string fileOutput, int margin)
{
//Convert docx to html
var fileInfo = new FileInfo(fileInput);
string fullFilePath = fileInfo.FullName;
string htmlText = string.Empty;
try
{
htmlText = ParseDOCX(fileInfo, margin);
}
catch (OpenXmlPackageException e)
{
if (e.ToString().Contains("Invalid Hyperlink"))
{
using (FileStream fs = new FileStream(fullFilePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
htmlText = ParseDOCX(fileInfo, margin);
}
}
var writer = File.CreateText(fileOutput + ".html");
writer.WriteLine(htmlText.ToString());
writer.Dispose();
// Convert html to pdf
var doc = new HtmlToPdfDocument()
{
GlobalSettings =
{
DocumentTitle = "Business Document",
ColorMode = DinkToPdf.ColorMode.Color,
Orientation = DinkToPdf.Orientation.Portrait,
PaperSize = PaperKind.A4,
Out = fileOutput + ".pdf",
Margins =
{
Top = 2,
Bottom = 2,
Right = 1.5,
Left = 1,
Unit = Unit.Centimeters
}
},
Objects =
{
new ObjectSettings()
{
PagesCount = true,
HtmlContent = File.ReadAllText(fileOutput + ".html"),
WebSettings = { DefaultEncoding = "utf-8" },
FooterSettings ={ FontSize =9, Center = "Trang [page] / [toPage]", Line = true }
}
}
};
var converter = new BasicConverter(new PdfTools());
converter.Convert(doc);
return(fileOutput);
}
/// <summary>
/// プロパティを読み込みます
/// </summary>
/// <param name="filePath"></param>
/// <param name="propertiesTable"></param>
/// <returns>取得したプロパティ一覧</returns>
public Dictionary <string, string> ReadProperties(ClsFilePropertyList list, string fileType, string str_exception) // 20171011 修正（ファイル読込時のエラー理由を保存）
{
Dictionary <string, string> retTable = new Dictionary <string, string>();
System.IO.FileInfo fi = new System.IO.FileInfo(list.filePath);
if (fi.Length == 0)
{
// ファイルサイズがゼロの場合プロパティが存在しないのでエラー
// 20171011 追加 （読取専用、ファイ存在しない以外のエラー）
//error_reason = ListForm.LIST_VIEW_NA;
return(retTable);
}
SpreadsheetDocument excel = null;
WordprocessingDocument word = null;
PresentationDocument ppt = null;
CoreFilePropertiesPart coreFileProperties;
// 20171011 修正（xlsxとdocxのみファイルを開いている状態でファイルアクセスするとエラーになる為の回避対応）
try
{
// ファイルのプロパティ領域を開く
switch (fileType)
{
case "Excel": // エクセルの場合
excel = SpreadsheetDocument.Open(list.filePath, false);
coreFileProperties = excel.CoreFilePropertiesPart;
break;
case "Word": // ワードの場合
word = WordprocessingDocument.Open(list.filePath, false);
coreFileProperties = word.CoreFilePropertiesPart;
break;
case "PowerPoint": // パワポの場合
ppt = PresentationDocument.Open(list.filePath, false);
coreFileProperties = ppt.CoreFilePropertiesPart;
break;
default:
// 異常なファイル
// 20171228 追加（エラー理由保存）
//error_reason = ListForm.LIST_VIEW_NA;
return(retTable);
}
NameTable nt = new NameTable();
XmlNamespaceManager nsManager = new XmlNamespaceManager(nt);
nsManager.AddNamespace(PropertiesKeyList.STR_TAG_CP, PropertiesSchemaList.CorePropertiesSchema);
nsManager.AddNamespace(PropertiesKeyList.STR_TAG_DC, PropertiesSchemaList.DcPropertiesSchema);
nsManager.AddNamespace(PropertiesKeyList.STR_TAG_DCTRMS, PropertiesSchemaList.DctermsPropertiesSchema);
nsManager.AddNamespace(PropertiesKeyList.STR_TAG_DCMITYPE, PropertiesSchemaList.DcmitypePropertiesSchema);
nsManager.AddNamespace(PropertiesKeyList.STR_TAG_XSI, PropertiesSchemaList.XsiPropertiesSchema);
XmlDocument xdoc = new XmlDocument();
xdoc.Load(coreFileProperties.GetStream());
// プロパティのキーリストを作成
List <string> propertieslist = PropertiesKeyList.getPropertiesKeyList();
// 全キーリストを見て存在するデータを取得
foreach (string key in propertieslist)
{
// 書き込み先のキーワードを指定
string searchString = string.Format(PropertiesKeyList.STR_CORE_PROPERTIES + "{0}", key);
// 書き込み先を検索
XmlNode xNode = xdoc.SelectSingleNode(searchString, nsManager);
if (xNode != null)
{
// 読み込む
retTable.Add(key, xNode.InnerText);
}
}
// ファイルのプロパティ領域を閉じる
if (excel != null)
{
excel.Close();
}
if (word != null)
{
word.Close();
}
if (ppt != null)
{
ppt.Close();
}
// プロパティ取得
string[] propertyData = null;
if (retTable.ContainsKey(PropertiesKeyList.STR_KEYWORDS) != false)
{
propertyData = retTable[PropertiesKeyList.STR_KEYWORDS].Split(';');
}
if (propertyData != null)
{
list.fileSecrecy = propertyData[0].TrimEnd();
// 機密区分が登録されているか判定
if (!string.IsNullOrEmpty(list.fileSecrecy))
{
// 登録あり
if (propertyData.Count() >= 1)
{
list.fileClassification = propertyData[1].TrimEnd();
}
if (propertyData.Count() >= 2)
{
list.fileOfficeCode = propertyData[2].TrimEnd();
}
// 他事務所ファイル判定
if (Globals.ThisAddIn.clsCommonSettings.strOfficeCode != list.fileOfficeCode.Trim())
{
// 他事務所ファイルの場合、登録なし扱にする
list.fileSecrecy = "";
}
}
}
else
{
list.fileSecrecy = "Notting";
}
}
#if false
#region HyperLink修復
// ■ ADD TSE Kitada
// HyperLinkが破損している場合に、そのリンクを書き直して正常にOPEN出来るようにする。
// 但し、ドキュメントの中身を直接書き換える処理のため、見送る。
catch (OpenXmlPackageException ope)
{
if (ope.ToString().Contains("Invalid Hyperlink"))
{
// HyperLinkの破損が原因なので、内部のリンクを修正する
using (FileStream fs = new FileStream(filePath, FileMode.OpenOrCreate, FileAccess.ReadWrite))
{
UriFixer.FixInvalidUri(fs, brokenUri => FixUri(brokenUri));
}
if (count >= 1)
{
// 2回実行してダメだったので終了
error_reason = ListForm.LIST_VIEW_NA;
}
else
{
// もう一度トライ
retTable = ReadProperties(filePath, filetype, ref error_reason, ref str_exception, 1);
}
}
if (excel != null)
{
excel.Close();
}
if (word != null)
{
word.Close();
}
if (ppt != null)
{
ppt.Close();
}
}
#endregion
#endif
catch (Exception e)
{
str_exception += "file : " + list.filePath + "\r\n\r\n";
str_exception += "error : " + e.ToString();
// xlsxとdocxのみファイルを開いている状態でファイルアクセスした場合
// ファイルが開かれている場合のエラー
//error_reason = ListForm.LIST_VIEW_NA;
if (excel != null)
{
excel.Close();
}
if (word != null)
{
word.Close();
}
if (ppt != null)
{
ppt.Close();
}
return(retTable);
}
return(retTable);
}
