/// <summary>
/// Raised when the Chrome process exits
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void _chromeProcess_Exited(object sender, EventArgs e)
{
try
{
// ReSharper disable once AccessToModifiedClosure
if (_chromeProcess == null)
{
return;
}
WriteToLog("Chrome exited unexpectedly, arguments used: " + string.Join(" ", DefaultArguments));
WriteToLog("Process id: " + _chromeProcess.Id);
WriteToLog("Process exit time: " + _chromeProcess.ExitTime.ToString("yyyy-MM-ddTHH:mm:ss.fff"));
var exception =
ExceptionHelpers.GetInnerException(Marshal.GetExceptionForHR(_chromeProcess.ExitCode));
WriteToLog("Exception: " + exception);
throw new ChromeException("Chrome exited unexpectedly, " + exception);
}
catch (Exception exception)
{
_chromeEventException = exception;
if (_chromeProcess != null)
{
_chromeProcess.Exited -= _chromeProcess_Exited;
}
_chromeWaitEvent.Set();
}
}
/// <summary>
/// Opens the <paramref name="inputFile" /> and returns it as an <see cref="WordInterop.Document" /> object
/// </summary>
/// <param name="inputFile">The file to open</param>
/// <param name="repairMode">When true the <paramref name="inputFile" /> is opened in repair mode</param>
/// <returns></returns>
private WordInterop.Document OpenDocument(string inputFile, bool repairMode)
{
Logger.WriteToLog($"Opening document '{inputFile}'{(repairMode ? " with repair mode" : string.Empty)}");
try
{
WordInterop.Document document;
var extension = Path.GetExtension(inputFile);
if (extension != null && extension.ToUpperInvariant() == ".TXT")
{
document = _word.Documents.OpenNoRepairDialog(inputFile, false, true, false, "dummy password",
Format: WordInterop.WdOpenFormat.wdOpenFormatUnicodeText,
OpenAndRepair: repairMode,
NoEncodingDialog: true);
}
else
{
document = _word.Documents.OpenNoRepairDialog(inputFile, false, true, false, "dummy password",
OpenAndRepair: repairMode,
NoEncodingDialog: true);
}
// This will lock or unlock all form fields in a Word document so that auto fill
// and date/time field do or don't get updated automatic when converting
if (document.Fields.Count > 0)
{
Logger.WriteToLog("Locking all form fields against modifications");
foreach (WordInterop.Field field in document.Fields)
{
field.Locked = true;
}
}
Logger.WriteToLog("Document opened");
return(document);
}
catch (Exception exception)
{
Logger.WriteToLog(
$"ERROR: Failed to open document, exception: '{ExceptionHelpers.GetInnerException(exception)}'");
if (repairMode)
{
throw new OCFileIsCorrupt("The file '" + Path.GetFileName(inputFile) +
"' seems to be corrupt, error: " +
ExceptionHelpers.GetInnerException(exception));
}
return(OpenDocument(inputFile, true));
}
}
/// <summary>
/// Converts the <paramref name="inputFile"/> to PDF and saves it as the <paramref name="outputFile"/>
/// </summary>
/// <param name="inputFile">The Microsoft Office file</param>
/// <param name="outputFile">The output file with full path</param>
/// <returns>Returns true when the conversion is succesfull, false is retournerd when an exception occurred.
/// The exception can be retrieved with the <see cref="GetErrorMessage"/> method</returns>
public bool ConvertFromCom(string inputFile, string outputFile)
{
try
{
_errorMessage = string.Empty;
Convert(inputFile, outputFile);
return(true);
}
catch (Exception exception)
{
_errorMessage = ExceptionHelpers.GetInnerException(exception);
return(false);
}
}
/// <summary>
/// Opens the <paramref name="inputFile"/> and returns it as an <see cref="WordInterop.Document"/> object
/// </summary>
/// <param name="word">The <see cref="WordInterop.Application"/></param>
/// <param name="inputFile">The file to open</param>
/// <param name="repairMode">When true the <paramref name="inputFile"/> is opened in repair mode</param>
/// <returns></returns>
private static WordInterop.Document Open(WordInterop._Application word,
string inputFile,
bool repairMode)
{
try
{
WordInterop.Document document;
var extension = Path.GetExtension(inputFile);
if (extension != null && extension.ToUpperInvariant() == ".TXT")
{
document = word.Documents.OpenNoRepairDialog(inputFile, false, true, false, "dummypassword",
Format: WordInterop.WdOpenFormat.wdOpenFormatUnicodeText,
OpenAndRepair: repairMode,
NoEncodingDialog: true);
}
else
{
document = word.Documents.OpenNoRepairDialog(inputFile, false, true, false, "dummypassword",
OpenAndRepair: repairMode,
NoEncodingDialog: true);
}
// This will lock or unlock all form fields in a Word document so that auto fill
// and date/time field do or don't get updated automaticly when converting
if (document.Fields.Count > 0)
{
foreach (WordInterop.Field field in document.Fields)
{
field.Locked = true;
}
}
return(document);
}
catch (Exception exception)
{
if (repairMode)
{
throw new OCFileIsCorrupt("The file '" + Path.GetFileName(inputFile) +
"' seems to be corrupt, error: " +
ExceptionHelpers.GetInnerException(exception));
}
return(Open(word, inputFile, true));
}
}
/// <summary>
/// This method will delete the automatic created Resiliency key. Word uses this registry key
/// to make entries to corrupted documents. If there are to many entries under this key Word will
/// get slower and slower to start. To prevent this we just delete this key when it exists
/// </summary>
private static void DeleteAutoRecoveryFiles()
{
try
{
// HKEY_CURRENT_USER\Software\Microsoft\Office\14.0\Word\Resiliency\DocumentRecovery
var version = string.Empty;
switch (VersionNumber)
{
// Word 2003
case 11:
version = "11.0";
break;
// Word 2017
case 12:
version = "12.0";
break;
// Word 2010
case 14:
version = "14.0";
break;
// Word 2013
case 15:
version = "15.0";
break;
// Word 2016
case 16:
version = "16.0";
break;
}
var key = @"Software\Microsoft\Office\" + version + @"\Word\Resiliency";
if (Registry.CurrentUser.OpenSubKey(key, false) != null)
{
Registry.CurrentUser.DeleteSubKeyTree(key);
}
}
catch (Exception exception)
{
EventLog.WriteEntry("OfficeConverter", ExceptionHelpers.GetInnerException(exception), EventLogEntryType.Error);
}
}
/// <summary>
/// Opens the <paramref name="inputFile" /> and returns it as an <see cref="PowerPointInterop.Presentation" /> object
/// </summary>
/// <param name="inputFile">The file to open</param>
/// <param name="repairMode">When true the <paramref name="inputFile" /> is opened in repair mode</param>
/// <returns></returns>
/// <exception cref="OCFileIsCorrupt">
/// Raised when the <paramref name="inputFile" /> is corrupt and can't be opened in
/// repair mode
/// </exception>
private PowerPointInterop.Presentation OpenPresentation(string inputFile, bool repairMode)
{
try
{
return(_powerPoint.Presentations.Open(inputFile, MsoTriState.msoTrue, MsoTriState.msoTrue,
MsoTriState.msoFalse));
}
catch (Exception exception)
{
if (repairMode)
{
throw new OCFileIsCorrupt("The file '" + Path.GetFileName(inputFile) +
"' seems to be corrupt, error: " +
ExceptionHelpers.GetInnerException(exception));
}
return(OpenPresentation(inputFile, true));
}
}
/// <summary>
/// This method reads the <paramref name="inputFile" /> and when the file is supported it will do the following: <br />
/// - Extract the HTML, RTF (will be converted to html) or TEXT body (in these order) <br />
/// - Puts a header (with the sender, to, cc, etc... (depends on the message type) on top of the body so it looks
/// like if the object is printed from Outlook <br />
/// - Reads all the attachents <br />
/// And in the end writes everything to the given <paramref name="outputFolder" />
/// </summary>
/// <param name="inputFile">The vcf file</param>
/// <param name="outputFolder">The folder where to save the extracted vcf file</param>
/// <param name="hyperlinks">When true hyperlinks are clickable, otherwhise they are written as plain text</param>
/// <param name="culture"></param>
public string[] ExtractToFolderFromCom(string inputFile,
string outputFolder,
bool hyperlinks = false,
string culture = "")
{
try
{
if (!string.IsNullOrEmpty(culture))
{
SetCulture(culture);
}
return(ExtractToFolder(inputFile, outputFolder));
}
catch (Exception exception)
{
_errorMessage = ExceptionHelpers.GetInnerException(exception);
return(new string[0]);
}
}
/// <summary>
/// If you want to run this code on a server then the following folders must exist, if they don't
/// then you can't use Excel to convert files to PDF
/// </summary>
/// <exception cref="OCConfiguration">Raised when the needed directory could not be created</exception>
private static void CheckIfSystemProfileDesktopDirectoryExists()
{
return;
if (Environment.Is64BitOperatingSystem)
{
var x64DesktopPath = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Windows),
@"SysWOW64\config\systemprofile\desktop");
if (!Directory.Exists(x64DesktopPath))
{
try
{
Directory.CreateDirectory(x64DesktopPath);
}
catch (Exception exception)
{
throw new OCConfiguration("Can't create folder '" + x64DesktopPath +
"' Excel needs this folder to work on a server, error: " +
ExceptionHelpers.GetInnerException(exception));
}
}
}
var x86DesktopPath = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Windows),
@"System32\config\systemprofile\desktop");
if (!Directory.Exists(x86DesktopPath))
{
try
{
Directory.CreateDirectory(x86DesktopPath);
}
catch (Exception exception)
{
throw new OCConfiguration("Can't create folder '" + x86DesktopPath +
"' Excel needs this folder to work on a server, error: " +
ExceptionHelpers.GetInnerException(exception));
}
}
}
/// <summary>
/// This method will delete the automatic created Resiliency key. PowerPoint uses this registry key
/// to make entries to corrupted presentations. If there are to many entries under this key PowerPoint will
/// get slower and slower to start. To prevent this we just delete this key when it exists
/// </summary>
private void DeleteResiliencyKeys()
{
Logger.WriteToLog("Deleting PowerPoint resiliency keys from the registry");
try
{
// HKEY_CURRENT_USER\Software\Microsoft\Office\14.0\PowerPoint\Resiliency\DocumentRecovery
var key = $@"Software\Microsoft\Office\{_versionNumber}.0\PowerPoint\Resiliency";
if (Registry.CurrentUser.OpenSubKey(key, false) != null)
{
Registry.CurrentUser.DeleteSubKeyTree(key);
Logger.WriteToLog("Resiliency keys deleted");
}
else
{
Logger.WriteToLog("There are no keys to delete");
}
}
catch (Exception exception)
{
Logger.WriteToLog($"Failed to delete resiliency keys, error: {ExceptionHelpers.GetInnerException(exception)}");
}
}
/// <summary>
/// Extract the given <paramref name="inputFile"/> to the given <paramref name="outputFolder"/>
/// </summary>
/// <param name="inputFile">The input file</param>
/// <param name="outputFolder">The folder where to save the extracted web archive</param>
/// <param name="options"><see cref="ExtractorOptions"/></param>
/// <param name="logStream">When set then logging is written to this stream</param>
/// <returns></returns>
/// <exception cref="WAEInvalidFile">Raised when a required resource is not found in the web archive</exception>
/// <exception cref="WAEResourceMissing">Raised when a required resource is not found in the web archive</exception>
/// <exception cref="FileNotFoundException">Raised when the <paramref name="inputFile"/> is not found</exception>
/// <exception cref="DirectoryNotFoundException">Raised when the <paramref name="outputFolder"/> does not exist</exception>
public string Extract(string inputFile, string outputFolder, ExtractorOptions options = ExtractorOptions.None, Stream logStream = null)
{
if (logStream != null)
{
Logger.LogStream = logStream;
}
try
{
if (!Directory.Exists(outputFolder))
{
throw new DirectoryNotFoundException($"The output folder '{outputFolder}' does not exist");
}
var reader = new PList.BinaryPlistReader();
IDictionary archive;
try
{
archive = reader.ReadObject(inputFile);
}
catch (Exception exception)
{
throw new WAEInvalidFile($"The file '{inputFile}' is not a valid Safari web archive", exception);
}
if (!archive.Contains(WebMainResource))
{
var message = $"Can't find the resource '{WebMainResource}' in the webarchive";
Logger.WriteToLog(message);
throw new WAEResourceMissing(message);
}
var mainResource = (IDictionary)archive[WebMainResource];
var webPageFileName = Path.Combine(outputFolder, "webpage.html");
Logger.WriteToLog($"Getting main web page from '{WebMainResource}'");
var webPage = ProcessMainResource(mainResource, out var mainUri);
#if (DEBUG)
File.WriteAllText(webPageFileName, webPage);
#endif
if (!archive.Contains(WebSubresources))
{
Logger.WriteToLog("Web archive does not contain any sub resources");
}
else
{
var subResources = (object[])archive[WebSubresources];
var count = subResources.Length;
Logger.WriteToLog($"Web archive has {count} sub resource{(count > 1 ? "s" : string.Empty)}");
foreach (IDictionary subResource in subResources)
{
ProcessSubResources(subResource, outputFolder, mainUri, options, ref webPage);
}
}
if (!archive.Contains(WebSubframeArchives))
{
Logger.WriteToLog("Web archive does not contain any sub frame archives");
}
else
{
var subFrameResources = (object[])archive[WebSubframeArchives];
var subFrameResourcesCount = subFrameResources.Length;
Logger.WriteToLog($"Web archive has {subFrameResourcesCount} sub frame resource{(subFrameResourcesCount > 1 ? "s" : string.Empty)}");
var i = 1;
foreach (IDictionary subFrameResource in subFrameResources)
{
var subFrameMainResource = (IDictionary)subFrameResource[WebMainResource];
Logger.WriteToLog($"Getting web page from sub frame resource '{WebMainResource}'");
var subFrameResourceWebPage = ProcessSubFrameMainResource(subFrameMainResource, out var frameName, out var subFrameMainUri);
var subFrameOutputFolder = Path.Combine(outputFolder, $"subframe_{i}");
Logger.WriteToLog($"Creating folder '{subFrameOutputFolder}' for iframe '{frameName}' content");
Directory.CreateDirectory(subFrameOutputFolder);
i += 1;
var subFrameSubResources = (object[])subFrameResource[WebSubresources];
if (subFrameSubResources == null)
{
Logger.WriteToLog("Web archive sub frame does not contain any sub resources");
}
else
{
var subFrameSubResourcesCount = subFrameSubResources.Length;
Logger.WriteToLog($"Web archive sub frame has {subFrameSubResourcesCount} sub resource{(subFrameSubResourcesCount > 1 ? "s" : string.Empty)}");
foreach (IDictionary subFrameSubResource in subFrameSubResources)
{
ProcessSubResources(subFrameSubResource, subFrameOutputFolder, subFrameMainUri, options,
ref subFrameResourceWebPage);
}
}
var subFrameWebPageFileName = Path.Combine(subFrameOutputFolder, "webpage.html");
var subFrameWebPageRelativeUri = $"subframe_{i}/webpage.html";
var subFrameUri = subFrameMainUri.ToString();
var subFrameUriWithoutScheme = subFrameUri.Replace($"{subFrameMainUri.Scheme}:", string.Empty);
var subFrameUriWithoutMainUri = subFrameUri.Replace($"{mainUri.Scheme}://{mainUri.Host}{mainUri.AbsolutePath}", string.Empty);
if (webPage.Contains(subFrameUri))
{
Logger.WriteToLog($"Replacing '{subFrameUri}' with '{subFrameWebPageRelativeUri}'");
webPage = webPage.Replace(subFrameUri, subFrameWebPageRelativeUri);
}
else if (webPage.Contains(subFrameUriWithoutScheme))
{
Logger.WriteToLog($"Replacing '{subFrameUriWithoutScheme}' with '{subFrameWebPageRelativeUri}'");
webPage = webPage.Replace(subFrameUriWithoutScheme, $"{subFrameWebPageRelativeUri}");
}
else if (webPage.Contains(subFrameUriWithoutMainUri))
{
Logger.WriteToLog($"Replacing '{subFrameUriWithoutMainUri}' with '{subFrameWebPageRelativeUri}'");
webPage = webPage.Replace(subFrameUriWithoutMainUri, $"{subFrameWebPageRelativeUri}");
}
else
{
Logger.WriteToLog($"Could not find any resources with url '{subFrameUri}' in the web page");
}
File.WriteAllText(subFrameWebPageFileName, subFrameResourceWebPage);
}
}
File.WriteAllText(webPageFileName, webPage);
return(webPageFileName);
}
catch (Exception exception)
{
Logger.WriteToLog(ExceptionHelpers.GetInnerException(exception));
throw;
}
}
/// <summary>
/// Opens the <paramref name="inputFile"/> and returns it as an <see cref="ExcelInterop.Workbook"/> object
/// </summary>
/// <param name="excel">The <see cref="ExcelInterop.Application"/></param>
/// <param name="inputFile">The file to open</param>
/// <param name="extension">The file extension</param>
/// <param name="repairMode">When true the <paramref name="inputFile"/> is opened in repair mode</param>
/// <returns></returns>
/// <exception cref="OCCsvFileLimitExceeded">Raised when a CSV <paramref name="inputFile"/> has to many rows</exception>
private static ExcelInterop.Workbook Open(ExcelInterop._Application excel,
string inputFile,
string extension,
bool repairMode)
{
try
{
switch (extension.ToUpperInvariant())
{
case ".CSV":
var count = File.ReadLines(inputFile).Count();
var excelMaxRows = MaxRows;
if (count > excelMaxRows)
{
throw new OCCsvFileLimitExceeded("The input CSV file has more then " + excelMaxRows +
" rows, the installed Excel version supports only " +
excelMaxRows + " rows");
}
string separator;
ExcelInterop.XlTextQualifier textQualifier;
GetCsvSeperator(inputFile, out separator, out textQualifier);
switch (separator)
{
case ";":
excel.Workbooks.OpenText(inputFile, Type.Missing, Type.Missing,
ExcelInterop.XlTextParsingType.xlDelimited,
textQualifier, true, false, true);
return(excel.ActiveWorkbook);
case ",":
excel.Workbooks.OpenText(inputFile, Type.Missing, Type.Missing,
ExcelInterop.XlTextParsingType.xlDelimited, textQualifier,
Type.Missing, false, false, true);
return(excel.ActiveWorkbook);
case "\t":
excel.Workbooks.OpenText(inputFile, Type.Missing, Type.Missing,
ExcelInterop.XlTextParsingType.xlDelimited, textQualifier,
Type.Missing, true);
return(excel.ActiveWorkbook);
case " ":
excel.Workbooks.OpenText(inputFile, Type.Missing, Type.Missing,
ExcelInterop.XlTextParsingType.xlDelimited, textQualifier,
Type.Missing, false, false, false, true);
return(excel.ActiveWorkbook);
default:
excel.Workbooks.OpenText(inputFile, Type.Missing, Type.Missing,
ExcelInterop.XlTextParsingType.xlDelimited, textQualifier,
Type.Missing, false, true);
return(excel.ActiveWorkbook);
}
default:
if (repairMode)
{
return(excel.Workbooks.Open(inputFile, false, true,
Password: "******",
IgnoreReadOnlyRecommended: true,
AddToMru: false,
CorruptLoad: ExcelInterop.XlCorruptLoad.xlRepairFile));
}
return(excel.Workbooks.Open(inputFile, false, true,
Password: "******",
IgnoreReadOnlyRecommended: true,
AddToMru: false));
}
}
catch (COMException comException)
{
if (comException.ErrorCode == -2146827284)
{
throw new OCFileIsPasswordProtected("The file '" + Path.GetFileName(inputFile) +
"' is password protected");
}
throw new OCFileIsCorrupt("The file '" + Path.GetFileName(inputFile) +
"' could not be opened, error: " + ExceptionHelpers.GetInnerException(comException));
}
catch (Exception exception)
{
if (repairMode)
{
throw new OCFileIsCorrupt("The file '" + Path.GetFileName(inputFile) +
"' could not be opened, error: " +
ExceptionHelpers.GetInnerException(exception));
}
return(Open(excel, inputFile, extension, true));
}
}
/// <summary>
/// Returns a dictionary with all the property mappings
/// </summary>
/// <param name="propertyIdents">List with all the named property idents, e.g 8005, 8006, 801C, etc...</param>
/// <returns></returns>
internal List <MapiTagMapping> GetMapping(IEnumerable <string> propertyIdents)
{
var result = new List <MapiTagMapping>();
var entryStreamBytes = GetStreamBytes(MapiTags.EntryStream);
var stringStreamBytes = GetStreamBytes(MapiTags.StringStream);
if (entryStreamBytes.Length == 0)
{
return(result);
}
foreach (var propertyIdent in propertyIdents)
{
try
{
// To read the correct mapped property we need to calculate the offset in the entry stream
// The offset is calculated bij subtracting 32768 (8000 hex) from the named property and
// multiply the outcome with 8
var identValue = ushort.Parse(propertyIdent, NumberStyles.HexNumber);
var entryOffset = (identValue - 32768) * 8;
if (entryOffset > entryStreamBytes.Length)
{
continue;
}
string entryIdentString;
// We need the first 2 bytes for the mapping, but because the nameStreamBytes is in little
// endian we need to swap the first 2 bytes
if (entryStreamBytes[entryOffset + 1] == 0)
{
var entryIdent = new[] { entryStreamBytes[entryOffset] };
entryIdentString = BitConverter.ToString(entryIdent).Replace("-", string.Empty);
}
else
{
var entryIdent = new[] { entryStreamBytes[entryOffset + 1], entryStreamBytes[entryOffset] };
entryIdentString = BitConverter.ToString(entryIdent).Replace("-", string.Empty);
}
var stringOffset = ushort.Parse(entryIdentString, NumberStyles.HexNumber);
if (stringOffset >= stringStreamBytes.Length)
{
//Debug.Print($"{propertyIdent} - {entryIdentString}");
result.Add(new MapiTagMapping(propertyIdent, entryIdentString));
continue;
}
// Read the first 4 bytes to determine the length of the string to read
var stringLength = 0;
var len = stringStreamBytes.Length - stringOffset;
if (len == 1)
{
var bytes = new byte[1];
Buffer.BlockCopy(stringStreamBytes, stringOffset, bytes, 0, len);
stringLength = bytes[0];
}
if (len == 2)
{
stringLength = BitConverter.ToInt16(stringStreamBytes, stringOffset);
}
if (len == 3)
{
var bytes = new byte[3];
Buffer.BlockCopy(stringStreamBytes, stringOffset, bytes, 0, len);
stringLength = Bytes2Int(bytes[2], bytes[1], bytes[0]);
}
else if (len >= 4)
{
stringLength = BitConverter.ToInt32(stringStreamBytes, stringOffset);
}
if (stringOffset + stringLength >= stringStreamBytes.Length)
{
result.Add(new MapiTagMapping(propertyIdent, entryIdentString));
continue;
}
var str = string.Empty;
// Skip 4 bytes and start reading the string
stringOffset += 4;
for (var i = stringOffset; i < stringOffset + stringLength; i += 2)
{
var chr = BitConverter.ToChar(stringStreamBytes, i);
str += chr;
}
// Remove any null character
str = str.Replace("\0", string.Empty);
result.Add(new MapiTagMapping(str, propertyIdent, true));
}
catch (Exception exception)
{
Logger.WriteToLog(ExceptionHelpers.GetInnerException(exception));
throw;
}
}
return(result);
}
/// <summary>
/// Start Chrome headless with the debugger set to the given port
/// </summary>
/// <remarks>
/// If Chrome is already running then this step is skipped
/// </remarks>
/// <exception cref="ChromeException"></exception>
private void StartChromeHeadless()
{
if (IsChromeRunning)
{
return;
}
var starting = true;
var userName = string.Empty;
if (_userName.Contains("\\"))
{
userName = _userName.Split('\\')[1];
}
var domain = _userName.Split('\\')[0];
WriteToLog($"Starting Chrome from location {_chromeExeFileName}");
WriteToLog($"{_chromeExeFileName} {string.Join(" ", DefaultArguments)}");
_chromeProcess = new Process();
var processStartInfo = new ProcessStartInfo
{
FileName = _chromeExeFileName,
Arguments = string.Join(" ", DefaultArguments),
UseShellExecute = false,
CreateNoWindow = true,
ErrorDialog = false,
WindowStyle = ProcessWindowStyle.Hidden,
// ReSharper disable once AssignNullToNotNullAttribute
WorkingDirectory = Path.GetDirectoryName(_chromeExeFileName),
RedirectStandardOutput = true,
RedirectStandardError = true
};
if (!string.IsNullOrWhiteSpace(_userName))
{
WriteToLog($"Starting Chrome with user '{userName}' on domain '{domain}'");
processStartInfo.Domain = domain;
processStartInfo.UserName = userName;
var secureString = new SecureString();
foreach (var t in _password)
{
secureString.AppendChar(t);
}
processStartInfo.Password = secureString;
}
_chromeProcess.StartInfo = processStartInfo;
var waitEvent = new ManualResetEvent(false);
_chromeProcess.OutputDataReceived += (sender, args) =>
{
if (!string.IsNullOrWhiteSpace(args.Data))
{
WriteToLog($"Error: {args.Data}");
}
};
_chromeProcess.ErrorDataReceived += (sender, args) =>
{
if (args.Data == null)
{
return;
}
if (args.Data.StartsWith("DevTools listening on"))
{
// DevTools listening on ws://127.0.0.1:50160/devtools/browser/53add595-f351-4622-ab0a-5a4a100b3eae
_browser = new Browser(new Uri(args.Data.Replace("DevTools listening on ", string.Empty)));
WriteToLog("Connected to dev protocol");
waitEvent.Set();
}
else if (!string.IsNullOrWhiteSpace(args.Data))
{
WriteToLog($"Error: {args.Data}");
}
};
_chromeProcess.Exited += (sender, args) =>
{
// ReSharper disable once AccessToModifiedClosure
if (!starting)
{
return;
}
WriteToLog("Chrome exited unexpectedly, arguments used: " + string.Join(" ", DefaultArguments));
WriteToLog("Process id: " + _chromeProcess.Id);
WriteToLog("Process exit time: " + _chromeProcess.ExitTime.ToString("yyyy-MM-ddTHH:mm:ss.fff"));
var exception =
ExceptionHelpers.GetInnerException(Marshal.GetExceptionForHR(_chromeProcess.ExitCode));
WriteToLog("Exception: " + exception);
throw new ChromeException("Chrome exited unexpectedly, " + exception);
};
_chromeProcess.EnableRaisingEvents = true;
_chromeProcess.Start();
WriteToLog("Chrome process started");
_chromeProcess.BeginErrorReadLine();
_chromeProcess.BeginOutputReadLine();
if (_conversionTimeout.HasValue)
{
waitEvent.WaitOne(_conversionTimeout.Value);
}
else
{
waitEvent.WaitOne();
}
starting = false;
WriteToLog("Chrome started");
}
/// <summary>
/// Start Chrome headless with the debugger set to the given port
/// </summary>
/// <remarks>
/// If Chrome is already running then this step is skipped
/// </remarks>
/// <exception cref="ChromeException"></exception>
private void StartChromeHeadless()
{
if (IsChromeRunning())
{
return;
}
WriteToLog($"Starting Chrome from location {_chromeExeFileName}");
_chromeProcess = new Process();
var processStartInfo = new ProcessStartInfo
{
FileName = _chromeExeFileName,
Arguments = string.Join(" ", _defaultArguments),
UseShellExecute = false,
RedirectStandardOutput = true,
RedirectStandardError = true,
};
if (!string.IsNullOrWhiteSpace(_userName))
{
var userName = string.Empty;
if (_userName.Contains("\\"))
{
userName = _userName.Split('\\')[1];
}
var domain = _userName.Split('\\')[0];
WriteToLog($"Starting Chrome with user '{userName}' on domain '{domain}'");
processStartInfo.Domain = domain;
processStartInfo.UserName = userName;
var secureString = new SecureString();
foreach (var t in _password)
{
secureString.AppendChar(t);
}
processStartInfo.Password = secureString;
}
_chromeProcess.StartInfo = processStartInfo;
var waitEvent = new ManualResetEvent(false);
_chromeProcess.ErrorDataReceived += (sender, args) =>
{
if (args.Data == null)
{
return;
}
if (args.Data.StartsWith("DevTools listening on"))
{
// DevTools listening on ws://127.0.0.1:50160/devtools/browser/53add595-f351-4622-ab0a-5a4a100b3eae
_communicator = new Browser(new Uri(args.Data.Replace("DevTools listening on ", string.Empty)));
WriteToLog("Connected to dev protocol");
waitEvent.Set();
}
else if (!string.IsNullOrWhiteSpace(args.Data))
{
WriteToLog($"Error: {args.Data}");
}
};
_chromeProcess.Exited += (sender, args) =>
{
WriteToLog("Chrome process: " + _chromeExeFileName);
WriteToLog("Arguments used: " + string.Join(" ", _defaultArguments));
var exception = ExceptionHelpers.GetInnerException(Marshal.GetExceptionForHR(_chromeProcess.ExitCode));
WriteToLog("Exception: " + exception);
throw new ChromeException("Could not start Chrome, " + exception);
};
_chromeProcess.Start();
_chromeProcess.BeginErrorReadLine();
waitEvent.WaitOne();
WriteToLog("Chrome started");
}
/// <summary>
/// Start Chrome headless
/// </summary>
/// <remarks>
/// If Chrome is already running then this step is skipped
/// </remarks>
/// <exception cref="ChromeException"></exception>
public void EnsureRunning()
{
lock (mutex)
{
if (IsChromeRunning)
{
WriteToLog($"Chrome is already running on PID {_chromeProcess.Id}... skipped");
return;
}
_chromeEventException = null;
var workingDirectory = Path.GetDirectoryName(_chromeExeFileName);
WriteToLog(
$"Starting Chrome from location '{_chromeExeFileName}' with working directory '{workingDirectory}'");
WriteToLog($"\"{_chromeExeFileName}\" {string.Join(" ", DefaultArguments)}");
_chromeProcess = new Process();
var processStartInfo = new ProcessStartInfo
{
FileName = _chromeExeFileName,
Arguments = string.Join(" ", DefaultArguments),
UseShellExecute = false,
CreateNoWindow = true,
ErrorDialog = false,
WindowStyle = ProcessWindowStyle.Hidden,
// ReSharper disable once AssignNullToNotNullAttribute
WorkingDirectory = workingDirectory,
RedirectStandardOutput = true,
RedirectStandardError = true
};
try
{
processStartInfo.LoadUserProfile = false;
if (!string.IsNullOrWhiteSpace(_userName))
{
var userName = string.Empty;
var domain = string.Empty;
if (_userName.Contains("\\"))
{
userName = _userName.Split('\\')[1];
domain = _userName.Split('\\')[0];
}
WriteToLog($"Starting Chrome with username '{userName}' on domain '{domain}'");
processStartInfo.Domain = domain;
processStartInfo.UserName = userName;
var secureString = new SecureString();
foreach (var t in _password)
{
secureString.AppendChar(t);
}
processStartInfo.Password = secureString;
processStartInfo.LoadUserProfile = true;
}
}
catch (Exception ex)
{
_logger?.LogWarning($"Failed to set user info {ex.Message}");
}
processStartInfo.Environment[UniqueEnviromentKey] = UniqueEnviromentKey;
_chromeProcess.StartInfo = processStartInfo;
_chromeWaitEvent = new ManualResetEvent(false);
_chromeProcess.OutputDataReceived += _chromeProcess_OutputDataReceived;
_chromeProcess.ErrorDataReceived += _chromeProcess_ErrorDataReceived;
_chromeProcess.Exited += _chromeProcess_Exited;
_chromeProcess.EnableRaisingEvents = true;
try
{
_chromeProcess.Start();
}
catch (Exception exception)
{
WriteToLog("Could not start the Chrome process due to the following reason: " +
ExceptionHelpers.GetInnerException(exception));
throw;
}
WriteToLog("Chrome process started");
_chromeProcess.BeginErrorReadLine();
_chromeProcess.BeginOutputReadLine();
if (_conversionTimeout.HasValue)
{
_chromeWaitEvent.WaitOne(_conversionTimeout.Value);
}
else
{
_chromeWaitEvent.WaitOne();
}
if (_chromeEventException != null)
{
WriteToLog("Exception: " + ExceptionHelpers.GetInnerException(_chromeEventException));
throw _chromeEventException;
}
WriteToLog("Chrome started");
}
}
/// <summary>
/// Stops Word
/// </summary>
private void StopWord()
{
if (IsWordRunning)
{
Logger.WriteToLog("Stopping Word");
try
{
_word.Quit(false);
}
catch (Exception exception)
{
Logger.WriteToLog($"Word did not shutdown gracefully, exception: {ExceptionHelpers.GetInnerException(exception)}");
}
var counter = 0;
// Give Word 2 seconds to close
while (counter < 200)
{
if (!IsWordRunning)
{
break;
}
counter++;
Thread.Sleep(10);
}
if (IsWordRunning)
{
Logger.WriteToLog($"Word did not shutdown gracefully in 2 seconds ... killing it on process id {_wordProcess.Id}");
_wordProcess.Kill();
Logger.WriteToLog("Word process killed");
}
else
{
Logger.WriteToLog("Word stopped");
}
}
if (_word != null)
{
Marshal.ReleaseComObject(_word);
_word = null;
}
_wordProcess = null;
GC.Collect();
GC.WaitForPendingFinalizers();
}
/// <summary>
/// Stops PowerPoint
/// </summary>
private void StopPowerPoint()
{
if (IsPowerPointRunning)
{
Logger.WriteToLog("Stopping PowerPoint");
try
{
_powerPoint.Quit();
}
catch (Exception exception)
{
Logger.WriteToLog($"PowerPoint did not shutdown gracefully, exception: {ExceptionHelpers.GetInnerException(exception)}");
}
var counter = 0;
// Give PowerPoint 2 seconds to close
while (counter < 200)
{
if (!IsPowerPointRunning)
{
break;
}
counter++;
Thread.Sleep(10);
}
if (IsPowerPointRunning)
{
Logger.WriteToLog(
$"PowerPoint did not shutdown gracefully in 2 seconds ... killing it on process id {_powerPointProcess.Id}");
_powerPointProcess.Kill();
Logger.WriteToLog("PowerPoint process killed");
}
else
{
Logger.WriteToLog("PowerPoint stopped");
}
}
if (_powerPoint != null)
{
Marshal.ReleaseComObject(_powerPoint);
_powerPoint = null;
}
_powerPointProcess = null;
GC.Collect();
GC.WaitForPendingFinalizers();
}
