Inconsistent sorting in ExtendedStringComparer leads to crash

[Ronin22222]

Every time I click on the Library at the top of Comicrack, the program crashes. I'm given the option to try to resume or restart. Restart just closes the program. Resume gets me back into the program, but nothing is displayed on the right side where the books are supposed to be listed

• Version: 0.9.184
• Commit: 551b3ca

Additional context
I first noticed this crashing issue when I made a smart list that shows books with series that 1) start with a space 2) end with a space and 3) have double spaces in the series name so I can clean them to group series properly for easier scraping. I usually don't go to the main library because my collection is large at the moment while I organize it and clean up junk files, but when I did today, the problem is the same and seems to be the source of the issue. I apologize for trimming the crash dump, but github wouldn't let me post it because of the length

If you have a crash dump paste it between ```

Crash dump

Application: ComicRack Community Edition
Version    : 0.9.184
Assembly   : 1.0.9616.33184
OS         : Microsoft Windows NT 10.0.26200.0 64
.NET       : 4.0.30319.42000
Processors : 24
Workingset : 2.08 GB
THREADEXCEPTION
--------------------
ArgumentException
Void ThrowOrIgnoreBadComparer(System.Object)
Unable to sort because the IComparer.Compare() method returns inconsistent results. Either a value does not compare equal to itself, or one value repeatedly compared to another value yields different results. IComparer: 'cYo.Common.Text.ExtendedStringComparer'.
   at System.Collections.Generic.IntrospectiveSortUtilities.ThrowOrIgnoreBadComparer(Object comparer)
   at System.Collections.Generic.GenericArraySortHelper`1.Sort(T[] keys, Int32 index, Int32 length, IComparer`1 comparer)
   at System.Array.Sort[T](T[] array, Int32 index, Int32 length, IComparer`1 comparer)
   at System.Collections.Generic.List`1.Sort(Int32 index, Int32 count, IComparer`1 comparer)
   at cYo.Projects.ComicRack.Engine.Controls.SearchBrowserControl.BuildList(Int32 start)
   at cYo.Projects.ComicRack.Engine.Controls.SearchBrowserControl.HandleSelectionChange(ListView lv)
   at System.Windows.Forms.ListView.OnItemSelectionChanged(ListViewItemSelectionChangedEventArgs e)
   at System.Windows.Forms.ListView.WmReflectNotify(Message& m)
   at System.Windows.Forms.ListView.WndProc(Message& m)
   at cYo.Common.Windows.Forms.ListViewEx.WndProc(Message& m)
   at System.Windows.Forms.NativeWindow.Callback(IntPtr hWnd, Int32 msg, IntPtr wparam, IntPtr lparam)

--------------------
Report generated at: 5/13/2026 9:26:39 PM

[maforget]

The rest of the crash dump aren't really relevant anyway.

It seems related to the search browser at the top. Keeping it close might prevent the crash.

Some error when sorting values from that smart list when building the list. Never seen that error so it might be a particular case from your list and your particular data.

My guess is that it may be related to the change from #103. Since we aren't trimming the names anymore it might have trouble sorting multiple variant of spaces. Different value (with different spacing) might give the same result when it shouldn't. I think that might be what that errors means. I would need to test that ExtendedStringComparer with your specific data.

Could you please share that smart lists? Do you have an example of the data it's trying to sort?

[Ronin22222]

It's a simple smart list

Name "#Spaces"
Match All
{
Match Any
{
[Series] starts with "#Template",
[Series] ends with "#Template",
[Series] contains "#Template"
},
Not [File Directory] contains "E:\Comics\"
}

I leave it with #Template as placeholders because Comicrack won't save just spaces after a restart. I replace the 1st and 2nd with a single space and the 3rd with a double space and then let Data Manager correct the bad spaces. I've been fixing badly done smart lists like this one with regex. I used ^\s+ and \s+$ for the leading/trailing spaces and got the same crash.

This crash doesn't happen when I make a smart list to point to a specific title like 'Superman' or any other English title, but does crash when I go to the main library. The only thing I can think of is there are bad characters in some of the file names that can't be handled correctly. I've gotten some files from Nyaa, Usenet, and Libgen recently with non standard English file names. I've deleted all of these files that I can find but still have the issue with the main library. Here's an example of what I mean. The listed text is the highlighted file in the pic I attached. These characters don't translate well apparently

[���][����][��].Vol.01

I'm exploring the problem further, but the more I look, it looks like an issue on my end and not necessarily with Comicrack. Sorry to bother you

[maforget]

the more I look, it looks like an issue on my end and not necessarily with Comicrack. Sorry to bother you

It is related to your data somehow, but it does seem to be an actual issue with the Comparer if you got this error. It might be fringe and probably related to these weird strings. But there is still a potential inconsistency that arises. Probably doesn't have anything to do with the smart list itself, but just the data it returned. Could be some logic issue in the code, with these strings of characters.

ExtendedStringComparer

ComicRackCE/cYo.Common/Text/ExtendedStringComparer.cs

Lines 54 to 145 in fbdccd9

	public static int Compare(string s1, string s2, ExtendedStringComparison compareMode)
	{
	if (string.IsNullOrEmpty(s1))
	return !string.IsNullOrEmpty(s2) ? -1 : 0;
	if (string.IsNullOrEmpty(s2))
	return 1;
	if (s1 == s2)
	return 0;
	int i1 = 0;
	int i2 = 0;
	bool zeroesFirst = (compareMode & ExtendedStringComparison.ZeroesFirst) != 0;
	bool ignoreCase = (compareMode & ExtendedStringComparison.IgnoreCase) != 0;
	bool ordinal = (compareMode & ExtendedStringComparison.Ordinal) != 0;
	bool ignoreArticles = (compareMode & ExtendedStringComparison.IgnoreArticles) != 0;
	int result = 0;
	if (ignoreArticles)
	{
	i1 = s1.IndexAfterArticle();
	i2 = s2.IndexAfterArticle();
	result = Math.Sign(i2 - i1);
	}
	int length = s1.Length;
	int length2 = s2.Length;
	bool isLetterOrDigit_i1 = char.IsLetterOrDigit(s1[i1]);
	bool isLetterOrDigit_i2 = char.IsLetterOrDigit(s2[i2]);
	if (isLetterOrDigit_i1 && !isLetterOrDigit_i2)
	return 1;
	if (!isLetterOrDigit_i1 && isLetterOrDigit_i2)
	return -1;
	do
	{
	char c1 = s1[i1];
	char c2 = s2[i2];
	bool isDigit_c1 = char.IsDigit(c1);
	bool isDigit_c2 = char.IsDigit(c2);
	if (!isDigit_c1 && !isDigit_c2)
	{
	if (c1 != c2)
	{
	bool isLetter_c1 = char.IsLetter(c1);
	bool isLetter_c2 = char.IsLetter(c2);
	if (isLetter_c1 && isLetter_c2)
	{
	int num = ordinal
	? (char.ToUpper(c1) - char.ToUpper(c2))
	: string.Compare(s1, i1, s2, i2, 1, ignoreCase);
	if (num != 0)
	return num;
	}
	else
	{
	if (isLetter_c1 || isLetter_c2)
	return isLetter_c1 ? 1 : -1;
	int num2 = ordinal
	? (c1 - c2)
	: string.Compare(s1, i1, s2, i2, 1);
	if (num2 != 0)
	return num2;
	}
	}
	}
	else
	{
	if (!(isDigit_c1 && isDigit_c2))
	return isDigit_c1 ? -1 : 1;
	int num3 = CompareNumbers(s1, length, ref i1, s2, length2, ref i2, zeroesFirst);
	if (num3 != 0)
	return num3;
	}
	i1++;
	i2++;
	if (i1 >= length)
	return i2 >= length2 ? result : -1;
	}
	while (i2 < length2);
	return 1;
	}

That code is called with IgnoreArticles & IgnoreCase. So it might be related to articles. One of the change I made was to add support for articles like L'. Maybe some of these strings has an apostrophe and it messing with it.

But these filenames are weird, seems like an encoding issue, probably some Unicode characters that came out like that. Might be related.

[Ronin22222]

When I was fixing that smart list to use regex instead of what I was doing so it would be saved and work properly I used \s{2,} for the double+ spaces. It pulled up some non standard spaces that show as a square in Comicrack and if you change the file name it shows as invisible spaces. What I mean by that is for Example it will look like Superman Man of Steel which looks correct, but if you move the cursor between words it takes 2 or more presses of the arrow key to navigate past what looks like a single space.

I'm running Data Manager right now to correct this and will see if it fixes the issue, but it will take a while with the size of my collection. I'll look at the L' files that I have when I'm done with this. I do know there's 2 different apostrophes that I've seen in file names. The standard ' that you have on your keyboard and a 'smart apostrophe' that looks like this ’ that leans more to the right. Could that be an issue with the new code?

[maforget]

There is not really that much new code. Unless you've modified the IgnoredArticles list it should only use L'.

From what I gather this errors arises when some order is inconsistent. So if A > B & B > C, then A > C, but if A < C then it freaks out. It might just be because of some weird character the files are using. Specially if you have characters that aren't visible or are interpreted as some Unicode value (even if you don't see it yourself, the program might).

Probably related to these string that are not encoded correctly but are interpreted by the Comparer in an inconsistent manner. Coupled with some logic problem in the comparer with these specific characters in these combination.

[Ronin22222]

I'm working on narrowing down the issue. When I get done with seeing if Data Manager fixes it, I'll look in the individual directories to see where the crash occurs. My unscraped files are broken down into directories based on the first letter, so Ant-Man would be in \Comics Unsorted\A , Superman would be in \Comics Unsorted\S and so-on. If I find the right directory, I can further break it down if needed if it's not obvious until I find it

When I do find the issue if it looks like something that should work but doesn't I'll report it but I think it's just some trash file that made it into my collection

Thanks for taking the time

[Ronin22222]

It turns out it was unicode characters in the file names being read as non-standard spaces in the series name. After running Data Manager to clear these spaces everything's working normally. It's an edge case since unicode characters aren't common in file names and probably not worth looking further into. Thank you for your time

[maforget]

It made me found an issue with the ExtendedStringComparer, there was a tie-breaker for articles length that was backwards.

So instead of:
Chien
Le Chien
The Chien

it would be:
The Chien
Le Chien
Chien