Printable View
Now that ChooseOption is done similar as UpText there will be bugs. This thread is more for reports that it's completely broken as there are some cases that I haven't been able to fix without breaking something else.
Known Bugs
- Misreading of 'kt' or 'rt' and some others it returns 'M' instead. This is because the letters are actually connected in some cases. [Possible future fix, add 'rt' etc together which if found would return 2 letters - major rework for font system?]
I was thinking, what if you join two fonts such as "r" and "t" and when found on screen, it'll recognize it as rt.. that's one way to get around the mistakes no?
I thought about something similar. Actually adding a 'rt' character to fonts and then it returns 2 letters instead of one if its found. However idk if the text system can use/do that without major changes.Quote:
Originally Posted by ggzz
I think we'll need a SplitTPA with a ``leash'' - it can't differentiate itself more than n coordinates from some starty. (Something similar to x for a general case) That would prevent the SplitTPA from ``jumping'' to the next text. (You could even update this leash dynamically by using the mean of all points currently in the part of the atpa)
Sounds like fun. :D
This below fixes the problem just its only if they don't change the Height (for now its the same everytime).
Simba Code:Function GetChooseOptions(TextType: string): Array of TOptions;
var
B, BB, sB, BigBox: TBox;
TPA, TPA1, TPA2, TPA3: TPointArray;
ATPA, tempatpa: T2DPointArray;
I, L, target, bmp,w,h, cts: Integer;
BoxColors: TIntegerArray;
begin
target := GetImageTarget;
GetClientDimensions(B.X2, B.Y2);
B := IntToBox(0, 0, B.X2-1, B.Y2-1);
cts := GetColorToleranceSpeed;
ColorToleranceSpeed(1);
// Black, Top Blue, grey
BoxColors := [131843, 3682593, 3552822];
SetLength(ATPA, Length(BoxColors));
for i := 0 to High(BoxColors) do
FindColorsTolerance(ATPA[i], BoxColors[i], B.X1, B.Y1, B.X2, B.Y2, 0);
TPA := MergeATPA(ATPA);
If Length(TPA) < 100 Then
begin
ColorToleranceSpeed(cts);
Exit;
end;
ATPA := SplitTPA(TPA, 4);
For i := 0 to High(ATPA) do
begin
B := GetTPABounds(ATPA[i]);
if ((B.x2-B.x1) > 90) and ((B.y2-B.y1) > 18) then
Break
else
B := IntToBox(0, 0, 0, 0);
end;
If (B.x2 = 0) then
begin
ColorToleranceSpeed(cts);
Exit;
end;
BigBox := B;
SetColorToleranceSpeed(3);
FindColorsTolerance(TPA2, 2827284, B.X1, B.Y1, B.X2, B.Y2, 16);//Background
FindColorsTolerance(TPA1, 6968358, B.X1, B.Y1, B.X2, B.Y2, 10);//Blue Highlight
TPA2 := CombineTPA(TPA2, TPA1);
sB := GetTPABounds(TPA2);
ReturnPointsNotInTPAWrap(TPA2, B, TPA);
FindColorsTolerance(TPA3, 460291, B.X1, B.Y1, B.X2, B.Y2, 6);//shadow
SetColorToleranceSpeed(cts);
bmp := CreateBitmap(B.X2-B.X1+2, B.Y2-B.Y1+2);
OffsetTPA(TPA, Point(-B.X1, -B.Y1));
OffsetTPA(TPA3, Point(-B.X1, -B.Y1));
FastDrawClear(bmp, 0);
DrawTPABitmap(bmp, TPA, clRed);//Text
DrawTPABitmap(bmp, TPA3, clPurple);//Shadow
ocr_FilterUpTextByCharacteristics(GetMufasaBitmap(bmp));
GetbitmapSize(bmp,w,h);
SetTargetBitmap(bmp); // new target for findcolorstolerance.
//findcolorstolerance(TPA, clRed, 0, 0, w-1,h-1, 0);
//ATPA := SplitTPAEx(TPA, 15, 2);
SetLength(ATPA, H/16);
for i := 0 to High(ATPA) do
FindColorsTolerance(ATPA[i], clRed, 0, 2+i*16, W-1, 18+i*16, 0);//+2 to y to line it up
L := High(ATPA);
SortATPAFromFirstPointY(ATPA, Point(w/2,0));
SetArrayLength(Result, Length(ATPA));
Result[0].BigBox := BigBox;
for i := 0 to high(ATPA) do
begin
TPA := ATPA[i];
tempatpa := SplitTPAEx(TPA, 1, 10); // Split one into characters
SortATPAFromFirstPointX(tempatpa, Point(0, 0));
Result[i].Str := GetTextATPA(tempatpa, 5, 'UpCharsEx'); //Writeln(Result[i].Str);
BB := GetTPABounds(ATPA[i]);
Result[i].Bounds := IntToBox(BB.X1+B.X1, BB.Y1+B.Y1, BB.X2+B.X1, BB.Y2+B.Y1);
setlength(tempatpa,0);
setlength(TPA,0);
end;
ColorToleranceSpeed(cts);
SetImageTarget(target);
FreeBitmap(bmp);
end;
I'd push it now but don't have GIT setup correctly on my laptop.
How about the frequent crashes? Anyone have any ideas on this? It's pretty annoying; if you have a script that calls chooseoption a lot it will crash script relatively quickly.
I asked Wizzup about it too. I don't get why it says, "DrawATPA" and then only now and then it'll be on the DrawTPA line (in the function). I have yet to reproduce the error, so I haven't figured out whats causing it.
Any particular script this happens a lot in?
Any type of script that power drops items. (Power miner/chopper). Yet, I can't get it to reproduce a lot.
Quote:
Originally Posted by Kyle Undefined
From what I've read, the FindNPCChatText out of range is very common. Happening with quite a few people when dropping, like you said. I got the SortATPAFromFirstPointY last night as well, but it was during MSI's walking.
When dropping huh? Maybe when text is highlighted? Its a pain to debug when its so random. I'll mess around with it tonight before I go to sleep.
I actually just noticed that sometimes if the option is already highlighted, ChooseOption will return false (not choose the option). That would explain why people are getting this error on the last items (when the 'drop' option is already highlighted).Quote:
Originally Posted by Narcle
Don't know why it would throw an out of range though.
Yeah it might not be grabbing the text then if the Highlighted color changed. I just don't have the time to get all the highlighted colors to make sure the correct tolerance is set. (cts 3 works really nice for it too)
Should fix the out of range, should it not?
If the length of a[i] was 0, it calls a[i][0], which would be an out of range. Same thing will have to be changed in SortATPAFromFirstPointX.Simba Code:procedure SortATPAFromFirstPointY(var a: T2DPointArray; const From: TPoint);
var
i, l: Integer;
DistArr: TIntegerArray;
begin
l := high(a);
if (l < 0) then
exit;
setLength(DistArr, l + 1);
for i := 0 to l do
begin
if (length(a[i]) <= 0) then // prevent out of range errors
continue;
DistArr[i] := Round(Sqr(From.y - a[i][0].y));
end;
QuickATPASort(DistArr, a, 0, l, True);
end;
E: Also, that function is right below FindNPCChatText. My guess is that Simba's just giving the wrong line number sometimes.
That makes sense then. If the hightlighted chooseoption isn't found (I search in boxes) then that ATPA[x] (of highlighted area) would be empty making it out of range in that function.Quote:
Originally Posted by Coh3n
Both need to be added. I'm guessing in rare cases the FindColors just happens to not see highlighted color - a tolerance increase and a better average color should fix it.
I can run some tests if you'd like.
The biggest help would be picking all the different highlighted colors so we can get a average. Then you'd basically have it fixed. I still want to play with CTS3 more. I've just be playing with it painting on smart to see how high the tolerance needs to be for various functions (switching to it).Quote:
Originally Posted by Kyle Undefined
(if this is indeed the problem and i'm leaning towards it)
I'll do some tweaking and commit a fix shortly.
E: I've fixed the out of ranges, but didn't have time to tweak the colors. But at least the runtime errors should be gone.
Nah error is still there after that fix. I figured it out. Its getting double points passed to ReturnPointsNotInTPAWrap which errors out either in that function or in DrawTPABitmap.
I tried using ClearDoubleTPA - it fixes it but makes it delay by ~1400 ms. Trying to lower tolerances to see if it'll stop overlapping TPoints found.
I'll have to get back to this when I have more free time.
Edit:
Pushed a fix, was working well for me. (only searches for one color now - yay cts3)
Awesome work Narcle! How does it handle the Choose Option box? I noticed the UpText and ChooseOption text, whatever color the UpText is, the ChooseOption text color is the same and the hover is the same aswell, but only lighter.
I grabbed the various colors of the dark background of the ChooseOption got the best color for that then did the same when its Highlighted. Got the best color of the 2 together and CTS3 with 18 tolerance it'll get all the pixels around the text (including if its highlighted). Find the gaps and that's the letters. I also get the shadows of text to check it as well. Right now it works so well that you probably don't even need shadows but its better with it in for now, gives it a fail safe of sorts.Quote:
Originally Posted by Kyle Undefined
Awesome! :)
Does this mean the OP is fixed as well?
For now yes. I'll edit it out. Hopefully I can let this thread die >.>Quote:
Originally Posted by Coh3n
So I've noticed a bug with choose option..
There has to be an infinite loop somewhere in there because sometimes it will right click and instead of choosing the option, it just sits there with the menu open. Happens about once every 48 hrs for me assuming I run my script that long. So it is rare but it keeps happening more than often enough for me to feel the need to report it.
How to fix: Well it sits there, I disable smart and moved the mouse away from the menu, the script recognizes the menu isn't there and right clicks and chooses the option correctly. Seems moving the mouse away from the menu fixes it. So..?
Oh and another thing, the withdraw function definitely needs an update as it lags like hell!! I do not lag.. only that function lags me AND it doesn't always withdraw even if the DTM is found. I've tested it 100 times in a row to see if the dtm was found, it was found 100% of the time, but the item was only withdrawn 70% of the time. The more I run it, the more the function lags.
For ChooseOption
Screenshot when it gets like that?
Which Withdraw function? You mention DTM finding too. Because the current Withdraw function shouldn't have anything in it to "lag." Only thing to make it lag is if you use UpText (which'll wait 500ms) or ChooseOption bug.
SRL's WithdrawEx and Withdraw don't use DTMs at all. If you're searching for a DTM in in a loop, make sure you have a wait(100+random(100)) each time. That's probably what's causing your lag.
function WithdrawItem(Ident: integer; IdentType: string; Amount: integer; UpText: String; Tol: TIntegerArray): boolean;
Can be used like:
WithdrawItem(DTMHere, 'dtm', 10, 'potion', [10]); and that lags bad.
Its rare to get the chooseoption thing so I can't catch it all the time for a screenshot but I'll try.
Oh wow, didn't know there was a function like that, but I only looked in bank.simba. You'd think that's where all the banking functions would be. :pQuote:
Originally Posted by ggzz
So, you're saying using that function it doesn't always find the DTM, but when you use your function, it always works? Can you post the code you're using to test?
http://i.imgur.com/rDUHy.pngQuote:
Originally Posted by Coh3n
Picture of the bug.. See.. It refuses to climb even when the options are there..
I'm using ChooseOptionMulti(['adder', 'up Lad', 'limb-up', 'dder', 'limb', 'mb-up']);
I actually caught it this time.. The thing is, ChooseOption is working 99% of the time.. the other 1% of the time, it gets in an infinite loop where it right clicks, sees the options, n then does nothing.. gets stuck in an infini loops until the menu goes away.. Now for that particular screenie, when it doesn't find the option, it moves the mouse away so that the menu disappears.. then tries again.. and fails for an infinite amount of times until I help it.
This bug is extremely rare and that's why it took me soo long to catch it doing that in a screenie.
For the inventory one with the banking and the lag, if I use:
FindDTM(DTMInteger, X, Y, .............) it'll find the dtm's 100% of the time.. but when I use it in combination with withdrawex, a massive lag happens which might be causing it to fail:
Simba Code:Pouch := DTMFromString('m1gAAAHic42JgYDjIxMBwjAlC70PCIP4RKD4JxJyMDAxcQMwLxSA+GxIGiXEAcWFJPENxeSKDv78tGOcXxzIUlMQxFJUlMGSkhYDFyqqSGbiAdhODGYnECAAAECwT5g==');
NOBC:= ItemAmount('inv', 'dtm', Pouch, []);
while ((NOBC < NOB) or (Tries < 5)) do
begin
if (NOBC < NOB) then
if FindDTM(Pouch, X, Y, MBX1, MBY1, MBX2, MBY2) then
begin
MMouse(X, Y, 0, 0);
if (WaitUptextMulti(['yip pouch', 'unyip', 'Bunyip pouch', 'yip po'], 1200)) then
begin
BP:= MSTPointToBankPoint(Point(X, Y));
WithdrawEx(BP.X, BP.Y, NOB, ['yip pouch', 'unyip', 'Bunyip pouch', 'yip po']);
end;
end;
Inc(Tries);
T:= 0;
MarkTime(T);
while (TimeFromMark(T) < 2000) and (ItemAmount('inv', 'dtm', Pouch, []) < NOB) do
wait(1);
NOBC:= ItemAmount('inv', 'dtm', Pouch, []);
if (NOBC >= NOB) then
break;
end;
Tries:= 0;
It finds it like 80% of the time and lags extremely hard!! Lags so bad that sometimes I have to restart Simba or my computer.. I get this white outline of smart to the upper left of my screen when it does that.
So instead I just use my own custom functions for withdrawing.
EDIT: I finally caught it with the summoning potion..
http://i.imgur.com/CLBVc.png
That right there is one infinite loop in chooseoptionmulti.. it will not move until i get rid of the menu.. then when the menu is gone, it will right click and drink.. Which it did. It drank after I made menu go away (it right clicked and drank perfectly fine because the option was found :S)
Getting the same problem.
Here This is Mine Working:
http://villavu.com/forum/attachment....1&d=1330178455
This is my friends not working:
http://villavu.com/forum/attachment....1&d=1330178455
His Looks Darker and bolder.
I'm picking up on All.
Mat
Why make 2 threads?
Edit: Use Multi?
I deleted the other one I put it here as It more appropriate here, I tried that :/.
Have you tried using WaitOption?
I had it is a repeat
Just ends up logging out.Simba Code:Repeat
Wait(50+random(10));
Until(ChooseOption('All');
That's weird. Can you try WaitOption('All', RandomRange(850, 1250)); please?
Nope :/
Is your friend playing in low graphics? Maybe, if you want to know if it is really ChooseOption that isn't working. Ask your friend to Change his X value for something like 4500 and Change your Chooseoption for the number 4500.
Simba Code:2: ChooseOption('4500');
If it doesn't work after that, I can't think about a solution.
So its moving the mouse away but the menu doesn't disappear? Or does it disappear but clicks again?
What happens is that for the first screenshot, it will move the mouse away and the menu disappears but when it tries again, it will never choose the option to climb.. It'll do this an infinite amount of times.
For the second screenshot, it never moves the mouse away so the menu never disappears.. when I have to disable and re-enable smart.. When that is done, it tries again and chooses the right option. If I don't disable/re-enable smart, it'll stay there for an infinite amount of time with the menu open.
@ggzz: For that banking function you posted, try increasing the wait(1) to something like wait(100).