Probably incorrect horizontal cursor positioning with certain CSI sequences

[cubiclesoft]

It appears to be possible to put the buffer.x position into an odd state with certain CSI sequences.

Sample

\E[4000G\E[B

Or:

\E[4000G\E[e

Analysis

    InputHandler.prototype.cursorCharAbsolute = function (params) {
        var param = params[0];
        if (param < 1) {
            param = 1;
        }
        this._terminal.buffer.x = param - 1;
    };

    InputHandler.prototype.cursorDown = function (params) {
        var param = params[0];
        if (param < 1) {
            param = 1;
        }
        this._terminal.buffer.y += param;
        if (this._terminal.buffer.y >= this._terminal.rows) {
            this._terminal.buffer.y = this._terminal.rows - 1;
        }
        if (this._terminal.buffer.x >= this._terminal.cols) {
            this._terminal.buffer.x--;
        }
    };

    InputHandler.prototype.vPositionRelative = function (params) {
        var param = params[0];
        if (param < 1) {
            param = 1;
        }
        this._terminal.buffer.y += param;
        if (this._terminal.buffer.y >= this._terminal.rows) {
            this._terminal.buffer.y = this._terminal.rows - 1;
        }
        if (this._terminal.buffer.x >= this._terminal.cols) {
            this._terminal.buffer.x--;
        }
    };

Not quite sure what the purpose of the x-- is. If the goal is to always cap it within the max columns, then that won't necessarily work properly here.

On an unrelated note, cursorDown and vPositionRelative contain identical code. cursorForward and hPositionRelative also contain identical code. Also, cursorPosition and hVPosition appear to have logically identical code but hVPosition is shorter and more readable, IMO.

[Tyriar]

I think the x-- is because printing can push the cursor to equal this._terminal.cols and only on the next print will it move to the following line.

[jerch]

Yep, the x-- tries to fix the faulty cursor position handling on sequence basis atm, a more general fix is needed as pointed out by #1434.

[Tyriar]

@jerch is this a dupe of #1434?

[jerch]

Kinda, its a part of it.

[Tyriar]

@jerch just the code duplication part is not?

[jerch]

Well the code duplication part is the try to fix the wrong positioning for certain methods, while others have not this "sanity" check yet. I guess there are more that can show this overflow (not checked), some methods even try to fix it on enter with a precheck (print for example always checks this before inserting chars to the buffer).
It happens to work atm with this mixture of pre- and postcheck/corrections, still the cursor is sometimes off (top shows the cursor wrongly in first column during screen refreshs). #1434 tried to address this but I kinda got distracted from it (and the fix was more an ugly hack).