Hi,

1- How can TSynCompletion's ItemList be filled depending on what is before the dot(.)? I know the Lazarus IDE does it, but I couldn't figure it out.

TEdit.|

If Ctrl+Space is pressed, then the list of TEdit properties should popup. Currentstring is empty. How can we read the word before the dot?

2- Any reason why TSynCompletion is not a palette component, and has to be created through code?

Thanks,
Stephano

There is an example on how to use it, in the example folder.

It should show you which callbacks are triggered, so you can place code of your own there.

2- Any reason why TSynCompletion is not a palette component, and has to be created through code?

Yes and no.

It needs some real refactor, and deprecating a method/property will allow people to change there code easily (if enough time is given before the final remove) => but as soon as stuff get's into the lfm, it gets way harder.
Though I have no idea, when, if ever, I get to that.

Martin

I checked the 3 examples before posting. They show how to use CurrentString (which is after the dot), but not how to read the token before the dot.

You've done a already great job on Synedit!

Thanks,
Stephano

ok, I may be able to look into it later.

but what about using LogicalCaretXY, then getting the line, and reading it?

possible subtracting the length of CurrentString.

I am almost there (by giving it another shot at the ide sources):

   Line:=SynEdit1.LineText;
   p1:=SynEdit1.LogicalCaretXY;
   if (p1.x>length(Line)+1) or (Line='') then exit;
   SynEdit1.GetWordBoundsAtRowCol(p1, x1, x2);
   // There must be a saner way than subtracting 2
   p1.x := x1-2;
   SynEdit1.GetWordBoundsAtRowCol(p1, x1, x2);
   WordToken := copy(Line, x1, x2);

However, what is the correct way to go back to the previous token? Subtracting 2 is not the way.

Thanks,
Stephano

The caret is at the end of currentstring.
So your dot should be 1 before. But if you allow spaces after the dot, then it can be anything before..., and maybe you even allow it on the previous line?

Note that LogicalCaret is 1 based, and 1 is before the first char. Also you must use LogicalCaret, normal Caret will not work if there are multybyte chars, or tabs.
- LogicalCaret: BYTE (not char) pos in text
-Caret (sometime PhysicalCaret): screenpos, a tab is one byte in the text, but may be 4 chars on the screen // an a-umlaut is 2 bytes in the text, but 1 on the screen)

If your line is
ABC.def
And the caret is between the A and B, then LogicalCaret.X = 2 (1 would be before the A)

So if you are behind the "def" LC.x = 8
x := LC.x - len(CurrentString);

makes x = 8 - 3 = 5
LineString[5] = 'd'

so you need to subtract another one to see the char before currentstring
x:= LC.x-length(CurrentString)
while (x> 1) and (linestring[x] in[#32,#9]) do dec(x);

Now you should be on the first none space, before the currentstring....

and so on,
It all depends on what you believe a word/token is, what are the boundaries...

You can look at GetWordBoundsAtRowCol....

maybe it expects to be NOT at the end of the word, so by substracting one more (-2) you get into the middle (or to the start) of the word???

Not sure, haven't looked at

x:= LC.x-length(CurrentString)  -1

somehow the -1 got lost