Console application: How to update the display without flicker? - c#-4.0

Using C# 4 in a Windows console application that continually reports progress how can I make the "redraw" of the screen more fluid?
I'd like to do one of the following:
- Have it only "redraw" the part of the screen that's changing (the progress portion) and leave the rest as is.
- "Redraw" the whole screen but not have it flicker.
Currently I re-write all the text (application name, etc.). Like this:
Console.Clear();
WriteTitle();
Console.WriteLine();
Console.WriteLine("Deleting:\t{0} of {1} ({2})".FormatString(count.ToString("N0"), total.ToString("N0"), (count / (decimal)total).ToString("P2")));
Which causes a lot of flickering.

Try Console.SetCursorPosition. More details here: How can I update the current line in a C# Windows Console App?

static void Main(string[] args)
{
Console.SetCursorPosition(0, 0);
Console.Write("################################");
for (int row = 1; row < 10; row++)
{
Console.SetCursorPosition(0, row);
Console.Write("# #");
}
Console.SetCursorPosition(0, 10);
Console.Write("################################");
int data = 1;
System.Diagnostics.Stopwatch clock = new System.Diagnostics.Stopwatch();
clock.Start();
while (true)
{
data++;
Console.SetCursorPosition(1, 2);
Console.Write("Current Value: " + data.ToString());
Console.SetCursorPosition(1, 3);
Console.Write("Running Time: " + clock.Elapsed.TotalSeconds.ToString());
Thread.Sleep(1000);
}
Console.ReadKey();
}

I know this question is a bit old but I found if you set Console.CursorVisible = false then the flickering stops as well.

Here's a simple working demo that shows multi-line usage without flickering. It shows the current time and a random string every second.
private static void StatusUpdate()
{
var whiteSpace = new StringBuilder();
whiteSpace.Append(' ', 10);
var random = new Random();
const string chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789";
var randomWord = new string(Enumerable.Repeat(chars, random.Next(10)).Select(s => s[random.Next(s.Length)]).ToArray());
while (true)
{
Console.SetCursorPosition(0, 0);
var sb = new StringBuilder();
sb.AppendLine($"Program Status:{whiteSpace}");
sb.AppendLine("-------------------------------");
sb.AppendLine($"Last Updated: {DateTime.Now}{whiteSpace}");
sb.AppendLine($"Random Word: {randomWord}{whiteSpace}");
sb.AppendLine("-------------------------------");
Console.Write(sb);
Thread.Sleep(1000);
}
}
The above example assumes your console window is blank to start. If not, make sure to use Console.Clear() first.
Technical Note:
SetCursorPosition(0,0) places the cursor back to the top (0,0) so the next call to Console.Write will start from line 0, char 0. Note, it doesn't delete the previous content before writing. As an example, if you write "asdf" over a previous line such as "0123456", you'll end up with something like "asdf456" on that line. For that reason, we use a whiteSpace variable to ensure any lingering characters from the previous line are overwritten with blank spaces. Adjust the length of the whiteSpace variable to meet your needs. You only need the whiteSpace variable for lines that change.
Personal Note:
For my purposes, I wanted to show the applications current status (once a second) along with a bunch of other status information and I wanted to avoid any annoying flickering that can happen when you use Console.Clear(). In my application, I run my status updates behind a separate thread so it constantly provides updates even though I have numerous other threads and long running tasks going at the same time.
Credits:
Thanks to previous posters and dtb for the random string generator used in the demo.
How can I generate random alphanumeric strings in C#?

You could try to hack something together using the core libraries.
Rather than waste your time for sub-standard results, I would check out this C# port of the ncurses library (which is a library used for formatting console output):
Curses Sharp

I think you can use \r in Windows console to return the beginning of a line.
You could also use SetCursorPosition.

I would recommend the following extension methods. They allow you to use a StringBuilder to refresh the console view without any flicker, and also tidies up any residual characters on each line
The Problem: The following demo demonstrates using a standard StringBuilder, where updating lines that are shorter than the previously written line get jumbled up. It does this by writing a short string, then a long string on a loop:
public static void Main(string[] args)
{
var switchTextLength = false;
while(true)
{
var sb = new StringBuilder();
if (switchTextLength)
sb.AppendLine("Short msg");
else
sb.AppendLine("Longer message");
sb.UpdateConsole();
switchTextLength = !switchTextLength;
Thread.Sleep(500);
}
}
Result:
The Solution: By using the extension method provided below, the issue is resolved
public static void Main(string[] args)
{
var switchTextLength = false;
while(true)
{
var sb = new StringBuilder();
if (switchTextLength)
sb.AppendLineEx("Short msg");
else
sb.AppendLineEx("Longer message");
sb.UpdateConsole();
switchTextLength = !switchTextLength;
Thread.Sleep(500);
}
}
Result:
Extension Methods:
public static class StringBuilderExtensions
{
/// <summary>
/// Allows StrinbBuilder callers to append a line and blank out the remaining characters for the length of the console buffer width
/// </summary>
public static void AppendLineEx(this StringBuilder c, string msg)
{
// Append the actual line
c.Append(msg);
// Add blanking chars for the rest of the buffer
c.Append(' ', Console.BufferWidth - msg.Length - 1);
// Finish the line
c.Append(Environment.NewLine);
}
/// <summary>
/// Combines two StringBuilders using AppendLineEx
/// </summary>
public static void AppendEx(this StringBuilder c, StringBuilder toAdd)
{
foreach (var line in toAdd.ReadLines())
{
c.AppendLineEx(line);
}
}
/// <summary>
/// Hides the console cursor, resets its position and writes out the string builder
/// </summary>
public static void UpdateConsole(this StringBuilder c)
{
// Ensure the cursor is hidden
if (Console.CursorVisible) Console.CursorVisible = false;
// Reset the cursor position to the top of the console and write out the string builder
Console.SetCursorPosition(0, 0);
Console.WriteLine(c);
}
}

I actually had this issue so I made a quick simple method to try and eliminate this.
static void Clear(string text, int x, int y)
{
char[] textChars = text.ToCharArray();
string newText = "";
//Converts the string you just wrote into a blank string
foreach(char c in textChars)
{
text = text.Replace(c, ' ');
}
newText = text;
//Sets the cursor position
Console.SetCursorPosition(x, y);
//Writes the blank string over the old string
Console.WriteLine(newText);
//Resets cursor position
Console.SetCursorPosition(0, 0);
}
It actually worked surprisingly well and I hope it may work for you!

Naive approach but for simple applications is working:
protected string clearBuffer = null; // Clear this if window size changes
protected void ClearConsole()
{
if (clearBuffer == null)
{
var line = "".PadLeft(Console.WindowWidth, ' ');
var lines = new StringBuilder();
for (var i = 0; i < Console.WindowHeight; i++)
{
lines.AppendLine(line);
}
clearBuffer = lines.ToString();
}
Console.SetCursorPosition(0, 0);
Console.Write(clearBuffer);
Console.SetCursorPosition(0, 0);
}

Console.SetCursorPosition(0, 0); //Instead of Console.Clear();
WriteTitle();
Console.WriteLine();
Console.WriteLine("Deleting:\t{0} of {1} ({2})".FormatString(count.ToString("N0")

Related

Getting object name and randomly placing it to Text UI - Unity

I am a beginner in Unity and I am currently making a simple game. I have a problem managing the flow of my minigame. The minigame is simply finding an object, when I found and tap on the item name below will be shaded or marked or there can be animation just to indicate that the item is found.
What I want to do is to get the name of the objects that need to be found and set them randomly in the three (3) item names below. like every time this minigame opens the names of the items are randomly placed in the 3 texts. And when the item is found the name below will be marked or shaded or anything that will indicate it is found, but for now I will just set it inactive for easier indication. How can I properly do this whole process?
The objects inside the scene are button for them to have onCLick() events
Correction: the term choices are wrong because they just display the name of the items that you need to find, just in case you get confused with the term choice in my minigame. I will fix it.
Here is the visuals for the minigame:
The script I currently have for when the objects was clicked:
using UnityEngine;
using UnityEngine.UI;
using UnityEngine.EventSystems;
using System.Collections.Generic;
using TMPro;
public class ClickObject : MonoBehaviour
{
[SerializeField] Button pillowBtn, pcBtn, lampBtn;
// [SerializeField] TextMeshProUGUI choice1, choice2, choice3;
// FROM THE SOLUTION
[SerializeField] List<GameObject> gameObjectWanted;
[SerializeField] List<TextMeshProUGUI> textBoxes;
// METHOD NAMES IS TEMPORARY
public void pillowClicked()
{
Debug.Log("you found the " + EventSystem.current.currentSelectedGameObject.name);
}
public void desktopClicked()
{
Debug.Log("you found the " + EventSystem.current.currentSelectedGameObject.name);
}
public void lampClicked()
{
Debug.Log("you found the " + EventSystem.current.currentSelectedGameObject.name);
}
}
You asked for a lot and I hope I understood your intention.
first, if you want to randomly choose an amount of game objects from your game, I think the best way to do it is by adding the refernece of all the game objects you want to choose from radomly inside a list of game objects and then randomly take a game object of the list and make it a child of another game object I call "fatherGoTranform" on my code like that:
[SerializeField] List<GameObject> gameObjectWanted;
[SerializeField] float numOfGO = 4;
[SerializeField] Transform fatherGoTranform;
void Start()
{
for(int i=0;i<numOfGO;i++)
{
int index = Random.Range(0, gameObjectWanted.Count-1);
GameObject currentGO = gameObjectWanted[index ];
currentGO.transform.parent = fatherGoTranform;
gameObjectWanted.RemoveAt(index);
}
}
and then to click on a game object and the do with what you want try this:
void Update()
{
//Check for mouse click
if (Input.GetMouseButtonDown(0))
{
RaycastHit raycastHit;
Ray ray = Camera.main.ScreenPointToRay(Input.mousePosition);
if (Physics.Raycast(ray, out raycastHit, 100f))
{
if (raycastHit.transform != null)
{
//Our custom method.
CurrentClickedGameObject(raycastHit.transform.gameObject);
}
}
}
}
I have not checked the code so if there is an error tell me and I will fix it
This is the solution that worked for my problem. I randomly placed numbers to the list according to childCount and every index indicate the index of the text that I want to put them on which I get as a Transform child.
using UnityEngine;
using UnityEngine.UI;
using UnityEngine.EventSystems;
using System.Collections.Generic;
using TMPro;
public class ClickObject : MonoBehaviour
{
// [SerializeField] Button pillowBtn, pcBtn, lampBtn;
[SerializeField] GameObject choices, v1, finishPanel;
// RANDOM NUMBER HOLDER FOR CHECKING
private int randomNumber, foundCount;
// RANDOMIZED NUMBER LIST HOLDER
public List<int> RndmList = new List<int>();
private void Awake()
{
foundCount = 0;
RndmList = new List<int>(new int[v1.transform.childCount]);
for (int i = 0; i < v1.transform.childCount; i++)
{
randomNumber = UnityEngine.Random.Range(0, (v1.transform.childCount) + 1);
while (RndmList.Contains(randomNumber))
{
randomNumber = UnityEngine.Random.Range(0, (v1.transform.childCount) + 1);
}
RndmList[i] = randomNumber;
// Debug.Log(v1.transform.GetChild(randomNumber-1).name);
choices.transform.GetChild(i).GetComponentInChildren<TextMeshProUGUI>().text = v1.transform.GetChild(randomNumber - 1).name;
}
}
public void objectFound()
{
string objectName = EventSystem.current.currentSelectedGameObject.name;
Debug.Log("you found the " + objectName);
for (int i = 0; i < choices.transform.childCount; i++)
{
if (objectName == choices.transform.GetChild(i).GetComponentInChildren<TextMeshProUGUI>().text)
{
// Debug.Log(i);
// choices.transform.GetChild(i).gameObject.SetActive(false);
// choices.transform.GetChild(i).GetComponentInChildren<TextMeshProUGUI>().color = Color.gray;
if(RndmList.Contains(i+1))
{
// Debug.Log(i);
// Debug.Log(v1.transform.GetChild(RndmList[i]-1).name);
choices.transform.GetChild(i).GetComponentInChildren<TextMeshProUGUI>().color = Color.gray;
v1.transform.GetChild(RndmList[i]-1).GetComponent<Button>().enabled = false;
foundCount++;
}
}
}
if(foundCount == v1.transform.childCount)
{
finishPanel.SetActive(true);
}
}
}

Inserting uris into Gtk.Clipboard with vala

I am currently trying to implement copy and paste for my application, the problem is that i can only plaintext or images to the clipboard according to the documentation of Gtk.Clipboard: https://valadoc.org/gtk+-3.0/Gtk.Clipboard.html set_text / set_image.
But then there is also this method https://valadoc.org/gtk+-3.0/Gtk.Clipboard.set_with_data.html set_with_data, which i think i can use for adding a uri or an array of uris. But i can't figure out how and didn't find any good examples either.
UPDATE
Using the given answer i can fill the clipboard with an array of uris, but i can read them, when i try it just calls the get_func again and refills it.
CTRL C pressed
clipboard get_func called
Received: file:///home/marcel/Downloads/.gitignore
CTRL V pressd
clipboard get_func called
Received: file:///home/marcel/Downloads
Try Pasting: file:///home/marcel/Downloads
This is the code i use for testing CTRL + V:
print ("\nCTRL V pressd\n");
clipboard.request_uris ((clipboard, uris) => {
foreach ( string content in uris ) {
print ("Try Pasting: ");
print (content);
print ("\n");
}
});
and this is the relevant part of the get_func for CTRL + C:
clipboard.set_with_owner (
clipboard_targets,
(clipboard, selection_data, info, user_data_or_owner) => {
print ("clipboard get_func called\n");
var w = user_data_or_owner as Window;
File[] files = { w.get_selected_file () };
switch ( info ) {
case ClipboardProtocol.TEXT_URI_LIST:
print ("Received: ");
string[] uris = {};
foreach ( var file in files ) {
print (file.get_uri ());
print ("\n");
uris += file.get_uri ();
}
selection_data.set_uris (uris);
break;
As you can see in the terminal output above, it just refills the clipboard, throwing away its previous values.
As requested I am providing both an example for writing URIs to clipboard and getting URIs from clipboard. These examples are basically command line programs that get / set the clipboard immediately. In an actual GUI application you would probably react to a button press or, to catch CtrlC / CtrlV events, use Gtk.Widget.add_events() and get / set the clipboard when handling the Gtk.Widget.event signal.
Getting the clipboard
You can request URIs from the X11 clipboard using Gtk.Clipboard.request_uris (). This function takes a callback that will be called once the URIs are available.
Example:
public void main (string[] args) {
Gtk.init (ref args);
Gdk.Display display = Gdk.Display.get_default ();
Gtk.Clipboard clipboard = Gtk.Clipboard.get_for_display (display, Gdk.SELECTION_CLIPBOARD);
clipboard.request_uris (recieved_func);
Gtk.main ();
}
/* Gtk.ClipboardURIRecievedFunc */
private void recieved_func (Gtk.Clipboard clipboard, string[] uris) {
foreach (var uri in uris) {
print (uri + "\n");
}
Gtk.main_quit ();
}
To be compiled with valac clipget.vala --pkg=gtk+-3.0
Setting the clipboard
Theory:
From the Qt4 documentation:
Since there is no standard way to copy and paste files between
applications on X11, various MIME types and conventions are currently
in use. For instance, Nautilus expects files to be supplied with a
x-special/gnome-copied-files MIME type with data beginning with the
cut/copy action, a newline character, and the URL of the file.
Gtk.Clipboard does not pre-implement setting the clipboard for copying / cutting files. As you said, there is no such Gtk.Clipboard.set_uris().
Instead, you should set the clipboard by providing a callback that X11 gets the clipboard contents from once requested.
These are the steps required:
Create a bunch of Gtk.TargetEntrys that specify what clipboard protocols your app can handle. You'll want to handle the protocolstext/uri-list, x-special/gnome-copied-files and UTF8_STRING. Each TargetEntry is identified by its info field, so that number should be unique (see enum ClipboardProtocol in the example below)
Implement a method of the type Gtk.ClipboardGetFunc. This method should fill the Gtk.SelectionData object that is passed with the file paths to copy / cut. Check for the info parameter to set the SelectionData argument according to the protocol specified.
Register the callback and the protocols implemented to X11 using Gtk.Clipboard.set_with_owner or Gtk.Clipboard.set_with_data
Example:
enum ClipboardProtocol {
TEXT_URI_LIST,
GNOME_COPIED_FILES,
UTF8_STRING
}
public void main (string[] args) {
Gtk.init (ref args);
Gdk.Display display = Gdk.Display.get_default ();
Gtk.Clipboard clipboard = Gtk.Clipboard.get_for_display (display, Gdk.SELECTION_CLIPBOARD);
var clipboard_targets = new Gtk.TargetEntry[3];
Gtk.TargetEntry target_entry = { "text/uri-list", 0, ClipboardProtocol.TEXT_URI_LIST };
clipboard_targets[0] = target_entry;
target_entry = { "x-special/gnome-copied-files", 0, ClipboardProtocol.GNOME_COPIED_FILES };
clipboard_targets[1] = target_entry;
target_entry = { "UTF8_STRING", 0, ClipboardProtocol.UTF8_STRING };
clipboard_targets[2] = target_entry;
var owner = new Object ();
var rc = clipboard.set_with_owner (
clipboard_targets,
get_func,
clear_func,
owner
);
assert (rc);
clipboard.store ();
Gtk.main ();
}
/* Gtk.ClipboardGetFunc */
private void get_func (
Gtk.Clipboard clipboard,
Gtk.SelectionData selection_data,
uint info,
void* user_data_or_owner
) {
print ("GET FUNC!\n");
File my_file = File.new_for_path ("/home/lukas/tmp/test.txt");
File my_2nd_file = File.new_for_path ("/home/lukas/tmp/test2.txt");
File[] files = { my_file, my_2nd_file };
switch (info) {
case ClipboardProtocol.TEXT_URI_LIST:
string[] uris = {};
foreach (var file in files) {
uris += file.get_uri ();
}
selection_data.set_uris (uris);
break;
case ClipboardProtocol.GNOME_COPIED_FILES:
var prefix = "copy\n";
//var prefix = "cut\n";
/* use one of the above */
var builder = new StringBuilder (prefix);
for (int i = 0; i < files.length; i++) {
builder.append (files[i].get_uri ());
/* dont put the newline if this is the last file */
if (i != files.length - 1)
builder.append_c ('\n');
}
selection_data.set (
selection_data.get_target (),
8,
builder.data
);
break;
case ClipboardProtocol.UTF8_STRING:
var builder = new StringBuilder ();
foreach (var file in files) {
builder.append (file.get_parse_name ());
}
builder.append_c ('\n');
selection_data.set_text (builder.str, -1);
break;
default:
assert_not_reached ();
}
Gtk.main_quit ();
}
/* Gtk.ClipboardClearFunc */
private void clear_func (Gtk.Clipboard clipboard, void* data) {
;
}
To be compiled with valac clipset.vala --pkg=gtk+-3.0
A couple of notes:
In my example, I could only test x-special/gnome-copied-files since I only have Nautilus installed at the moment. I adapted all of the protocols from the Thunar source code (see sources below) but they might still require troubleshooting*
If you do not want to go through the trouble of implementing this yourself, you could also use the xclip command line tool: https://askubuntu.com/a/210428/345569 However, IMHO implementing this yourself is a little more elegant.
Sources:
Article from the Ubuntu Forums: https://ubuntuforums.org/archive/index.php/t-2135919.html
Thunar source code (especially thunar/thunar/thunar-clipboard-manager.c): https://github.com/xfce-mirror/thunar/blob/3de231d2dec33ca48b73391386d442231baace3e/thunar/thunar-clipboard-manager.c
Qt4 documentation: http://doc.qt.io/archives/qt-4.8/qclipboard.html

Spell checker in android

I want to create an application in which it checks if the word typed by user is correct or not using Google Dictionary ?
i have used the below link . But the problem with the given example is that it suggests the different words. I don't want suggestion, instead i want to only check that word entered is correct or not.
I haven't worked on it yet. But you can probably modify it as:
When you get the suggestions, instead of appending them to StringBuilder, and showing that StringBuilder to MainView, just compare all suggestions with your input string of edittext.
If it matches, then the spell is correct, else the spell is incorrect.
Code snippet:
#Override
public void onGetSuggestions(final SuggestionsInfo[] arg0) {
isSpellCorrect = false;
final StringBuilder sb = new StringBuilder();
for (int i = 0; i < arg0.length; ++i) {
// Returned suggestions are contained in SuggestionsInfo
final int len = arg0[i].getSuggestionsCount();
if(editText1.getText().toString().equalsIgnoreCase(arg0[i].getSuggestionAt(j))
{
isSpellCorrect = true;
break;
}
}
}
Hope this helps.

Removing part of a String^ in MFC C++

So far I've only written console applications. My first application using MFC (in Visual Studio 2010) is basically a form with two multiline boxes (using String[] arrays noted with String^) and a button to activate text processing. It should search the String^ for a [, look for the ] behind it and delete all characters between them (including the []). With 'normal' C++ strings, this isn't difficult. String^ however is more like an object and MSDN tells me to make use of the Remove method. So, I tried to implement it.
public ref class Form1 : public System::Windows::Forms::Form
{
public:
Form1(void)
{
InitializeComponent();
//
//TODO: Add the constructor code here
//
}
String^ DestroyCoords(String^ phrase)
{
int CoordsStart = 0;
int CoordsEnd = 0;
int CharCount = 0;
for each (Char ch in phrase)
{
if (ch == '[')
CoordsStart = CharCount;
if (ch == ']')
{
CoordsEnd = CharCount;
//CoordsEnd = phrase->IndexOf(ch);
phrase->Remove( CoordsStart , CoordsEnd-CoordsStart );
}
CharCount++;
}
return phrase;
}
The button using the method:
private: System::Void button1_Click(System::Object^ sender, System::EventArgs^ e) {
TempString = String::Copy(BoxInput->Text);
DestroyCoords(TempString);
BoxOutput->Text = TempString;
The function seems to hit the correct places at the correct time, but the phrase->Remove() method is doing absolutely nothing..
I'm no OO hero (as said, I normally only build console applications), so it's probably a rookie mistake. What am I doing wrong?
In C++/CLI, System::String is immutable, so Remove creates a new String^. This means you'll need to assign the results:
phrase = phrase->Remove( CoordsStart , CoordsEnd-CoordsStart );
The same is true in your usage:
TempString = DestroyCoords(TempString);
BoxOutput->Text = TempString;
Note that this will still not work, as you'd need to iterate through your string in reverse (as the index will be wrong after the first removal).
No MFC here, that's the C++/CLI that Microsoft uses for writing .NET programs in C++.
The .NET System::String class is immutable, so any operations you expect to modify the string actually return a new string with the adjustment made.
A further problem is that you're trying to modify a container (the string) while iterating through it. Instead of using Remove, have a StringBuilder variable and copy across the parts of the string you want to keep. This means only a single copy and will be far faster than repeated calls to Remove each of which makes a copy. And it won't interfere with iteration.
Here's the right approach:
int BracketDepth = 0;
StringBuilder sb(phrase->Length); // using stack semantics
// preallocated to size of input string
for each (Char ch in phrase)
{
if (ch == '[') { // now we're handling nested brackets
++BracketDepth;
}
else if (ch == ']') { // and complaining if there are too many closing brackets
if (!BracketDepth--) throw gcnew Exception();
}
else if (!BracketDepth) { // keep what's not brackets or inside brackets
sb.Append(ch);
}
}
if (BracketDepth) throw gcnew Exception(); // not enough closing brackets
return sb.ToString();

.NET 4.0 BackgroundWorker class and unusual behavior

I am running into some strange behavior in the backgroundworker class that leads me to believe that I don't fully understand how it works. I assumed that the following code sections were more or less equal except for some extra features that the BackgroundWorker implements (like progress reporting, etc.):
section 1:
void StartSeparateThread(){
BackgroundWorker bw = new BackgroundWorker();
bw.DoWork += new DoWorkEventHandler(bw_DoWork);
bw.RunWorkerAsync();
}
void bw_DoWork(object sender, DoWorkEventArgs e)
{
//Execute some code asynchronous to the thread that owns the function
//StartSeparateThread() but synchronous to itself.
var SendCommand = "SomeCommandToSend";
var toWaitFor = new List<string>(){"Various","Possible","Outputs to wait for"};
var SecondsToWait = 30;
//this calls a function that sends the command over the NetworkStream and waits
//for various responses.
var Result=SendAndWaitFor(SendCommand,toWaitFor,SecondsToWait);
}
Section 2:
void StartSeparateThread(){
Thread pollThread = new Thread(new ThreadStart(DoStuff));
pollThread.Start();
}
void DoStuff(object sender, DoWorkEventArgs e)
{
//Execute some code asynchronous to the thread that owns the function
//StartSeparateThread() but synchronous to itself.
var SendCommand = "SomeCommandToSend";
var toWaitFor = new List<string>(){"Various","Possible","Outputs to wait for"};
var SecondsToWait = 30;
//this calls a function that sends the command over the NetworkStream and waits
//for various responses.
var Result=SendAndWaitFor(SendCommand,toWaitFor,SecondsToWait);
}
I was using Section 1 to run some code that sent a string over a networkstream and waited for a desired response string, capturing all output during that time. I wrote a function to do this that would return the networkstream output, the index of the the sent string, as well as the index of the desired response string. I was seeing some strange behavior with this so I changed the function to only return when both the send string and the output string were found, and that the index of the found string was greater than the index of the sent string. It would otherwise loop forever (just for testing). I would find that the function would indeed return but that the index of both strings were -1 and the output string was null or sometimes filled with the expected output of the previous call. If I were to make a guess about what was happening, it would be that external functions called from within the bw_DoWork() function are run asynchronously to the thread that owns the bw_DoWork() function. As a result, since my SendAndWaitFor() function was called multiple times in succession. the second call would be run before the first call finished, overwriting the results of the first call after they were returned but before they could be evaluated. This seems to make sense because the first call would always run correctly and successive calls would show the strange behavior described above but it seems counter intuitive to how the BackgroundWorker class should behave. Also If I were to break within the SendAndWaitFor function, things would behave properly. This again leads me to believe there is some multi-threading going on within the bwDoWork function itself.
When I change the code in the first section above to the code of the second section, things work entirely as expected. So, can anyone who understands the BackgroundWorker class explain what could be going on? Below are some related functions that may be relevant.
Thanks!
public Dictionary<string, string> SendAndWaitFor(string sendString, List<string> toWaitFor, int seconds)
{
var toReturn = new Dictionary<string, string>();
var data = new List<byte>();
var enc = new ASCIIEncoding();
var output = "";
var FoundString = "";
//wait for current buffer to clear
output = this.SynchronousRead();
while(!string.IsNullOrEmpty(output)){
output = SynchronousRead();
}
//output should be null at this point and the buffer should be clear.
//send the desired data
this.write(enc.GetBytes(sendString));
//look for all desired strings until timeout is reached
int sendIndex=-1;
int foundIndex = -1;
output += SynchronousRead();
for (DateTime start = DateTime.Now; DateTime.Now - start < new TimeSpan(0, 0, seconds); )
{
//wait for a short period to allow the buffer to fill with new data
Thread.Sleep(300);
//read the buffer and add it to the output
output += SynchronousRead();
foreach (var s in toWaitFor)
{
sendIndex = output.IndexOf(sendString);
foundIndex = output.LastIndexOf(s);
if (foundIndex>sendIndex)
{
toReturn["sendIndex"] = sendIndex.ToString();
toReturn["foundIndex"] = sendIndex.ToString();
toReturn["Output"] = output;
toReturn["FoundString"] = s;
return toReturn;
}
}
}
//Set this to loop infinitely while debuging to make sure the function was only
//returning above
while(true){
}
toReturn["sendIndex"]="";
toReturn["foundIndex"]="";
toReturn["Output"] =output;
toReturn["FoundString"] = "";
return toReturn;
}
public void write(byte[] toWrite)
{
var enc = new ASCIIEncoding();
var writeString = enc.GetString(toWrite);
var ns = connection.GetStream();
ns.Write(toWrite, 0, toWrite.Length);
}
public string SynchronousRead()
{
string toReturn = "";
ASCIIEncoding enc = new ASCIIEncoding();
var ns = connection.GetStream();
var sb = new StringBuilder();
while (ns.DataAvailable)
{
var buffer = new byte[4096];
var numberOfBytesRead = ns.Read(buffer, 0, buffer.Length);
sb.AppendFormat("{0}", Encoding.ASCII.GetString(buffer, 0, numberOfBytesRead));
toReturn += sb.ToString();
}
return toReturn;
}
All data to be used by a background worker should be passed in through the DoWorkEventArgs and nothing should be pulled off of the class (or GUI interface).
In looking at your code I could not identify where the property(?) connnection was being created. My guess is that connection is created on a different thread, or may be pulling read information, maybe from a GUI(?) and either one of those could cause problems.
I suggest that you create the connection instance in the dowork event and not pull an existing one off of a different thread. Also verify that the data connection works with does not access any info off of a GUI, but its info is passed in as its made.
I discuss an issue with the Background worker on my blog C# WPF: Linq Fails in BackgroundWorker DoWork Event which might show you where the issue lies in your code.

Resources