>Psychtoolbox>PsychBasic

[nx, ny, textbounds, cache, wordbounds] = DrawFormattedText2(tstring, key-value pairs)
or:
[nx, ny, textbounds, cache, wordbounds] = DrawFormattedText2(cache, key-value pairs)

When called with a string, the following key-value pairs are understood:
win [, sx][, sy][, xalign][, yalign][, xlayout][, baseColor][, wrapat][, transform][, vSpacing][, righttoleft][, winRect][, resetStyle][, cacheOnly]
Those enclosed in square braces are optional.

When called with a cache struct, the following optional key-value pair
arguments are accepted:
[, win][, sx][, sy][, xalign][, yalign][, transform][, winRect][, cacheOnly]

example call:
DrawFormattedText2(‘test text’, ‘win’,window_pointer,’baseColor’,[255 0 0])

Draws a string of text ‘tstring’ into Psychtoolbox window ‘win’. Allows
some formatting and precise positioning. Only works with the FTGL-based
text plugin (see help DrawTextPlugin)), activate it with
Screen(‘Preference’,’TextRenderer’, 1);
This function is _not_ made to be fast. Test carefully if you want it to
work within a screen refresh if you throw a lot of text at it.

The text string ‘tstring’ may contain newline characters ‘\n’.
Whenever a newline character ‘\n’ is encountered, a linefeed and
carriage return is performed, breaking the text string into lines.

The text may also contain the formatting tags listed below. Each tag
changes the formatting of the text from that point onward. Tags do not
need to be closed. At the exit of the function, the state of the font
renderer (font, text size, etc) is reset to the state upon function entry
or the state stored in the cache if drawing from cache.

• <i>: toggles italicization
• <b>: toggles bolding
• <u>: toggles underlining
• <color=colorFmt> switches to a new color
• <font=name> switches to a new font
• <size=number> switches to a new font size
  The <i>, <b> and <u> tags toggle whether text is italicized, bolded or
  underlined and remain active (possibly until the end of the input string)
  until the same <i>, <b> or <u> tag is encountered again.
  The <color>, <font> and <size> tags can be provided empty (i.e., without
  argument), in which case they cause to revert back to the color, font or
  size active before the previous switch. Multiple of these in a row go
  back further in history (until start color, font, size is reached).
  To escape a tag, prepend it with a slash, e.g., /<color>. If you want a
  slash right in front of a tag, escape it by making it a double /:
  //<color>. No other slashes should be escaped.

<color>’s argument can have one of two formats, it can either be a
hexadecimal string (HEX), or a comma-separated floating point array
(FPN). If a HEX input is provided, it should encode colors using a 1, 2,
6 or 8 hexadecimal digit string. If 1 or 2 digits are provided, this is
interpreted as a grayscale value. 6 hexadecimal digits are interpreted as
R, G and B values (2 digits each). 8 digits further includes an alpha
value. Using the HEX values, color values range from 0 to 255. Example:
<color=ff0000> changes the text color to red (ff corresponds to 255). The
FPN format consists of 1, 3 or 4 comma separated floating point color
values (luminance, RGB, or RGBA) that can take on any value, but will be
processed according to the current color settings for the window as
indicated by Screen(‘ColorRange’). Typically, the values provided will
range from 0.0–1.0 per component. Floating point values should include
the decimal point to ensure that single element FPN values are parsed
correctly. Examples: <color=1.,0.,0.> and <color=.5>. If a floating point
value is specified but Screen(‘ColorRange’) returns 255, indicating 8bit
color values are used for the window, the floating point color values
provided are automatically scaled and rounded to the 0-255 range. If the
HEX format is used but Screen(‘ColorRange’) returns 1.0, indicating
floating point color values are used for the window, the uint8 color
values are automatically converted to the 0.0-1.0 range.
A size/font command before a newline can change the height of the line on
which it occurs. So if you want to space two words ‘test’ and ‘text’
vertically by white space equivalent to an 80pts line, use:
‘test\n<size=80>\n<size>text’. There is an empty line between the two new
lines, and the size=80 says that this line has height of 80pts in the
selected font.

‘sx’ and ‘sy’ provide a location on the screen with respect to which the
textbox is positioned. ‘sx’ and ‘sy’ can be a pixel location provided as
a number. Alternatively, ‘sx’ can also be ‘left’ (default), ‘center’, or
‘right’ signifying the left side, horizontal middle, or right side of the
window rect. ‘sy’ can also be ‘top’ (default), ‘center’, or ‘bottom’
signifying the top, vertical middle, or bottom of the window rect.

‘xalign’ and ‘yalign’ are text strings indicating how the textbox should
be aligned to the screen location provided in ‘sx’ and ‘sy’. For
‘xalign’, ‘left’ (default), signifies that the left of the text’s
bounding box is aligned to the screen location; ‘center’ that the
bounding box is centered on this location; and ‘right’ that the right of
the text’s bounding box is aligned with this location. For ‘yalign’,
‘top’ (default), signifies that the top of the text’s bounding box is
aligned to the screen location; ‘center’ that the bounding box is
centered on this location; and ‘bottom’ that the bottom of the text’s
bounding box is aligned with this location.

‘xlayout’ indicates how each line is positioned horizontally in the
bounding box. If ‘left’ (default), all lines are aligned to the left of
the text’s bounding box; if ‘center’, all lines are centered in the
bounding box; and if ‘right’, all lines are aligned to the right of the
bounding box. Justification options are currently not supported.

‘baseColor’ is the color in which the text will be drawn (until changed
by a <color> format call). This is also the color that will remain active
after invocation of this function.

‘wrapat’, if provided, will automatically break text strings longer than
‘wrapat’ characters into newline separated strings of roughly ‘wrapat’
characters. This is done by calling the WrapString function (See ‘help
WrapString’). ‘wrapat’ mode may not work reliably with non-ASCII text
strings, e.g., UTF-8 encoded uint8 strings on all systems. It also does
not necessarily lead to lines of roughly equal length, unless using a
monospace font.

‘transform’ allows transforming the text to be drawn as a whole. The
‘transform’ input is a cell array of key-value parameters, indicating
which transform to do in which order. The following transform are
supported:
‘translate’, [dx dy]: translates text by dx horizontally and dy
vertically
‘flip’, axis : mirrors text horizontally if axis is 1, vertically
if axis is 2, and along both axes if number is 3
(which is equal to a 180 deg rotation)
‘scale’, [sx sy] : scale text horizontally by sx and vertically by sy.
Set to 1 if you want no scaling.
‘rotate’, angle : Rotate text by angle (degrees). Note that for the
PTB screen, a positive rotation is clockwise.
example {‘translate’,[100 0],’rotate’,45} first translates text by 100
pixels, then rotates it by 45 degree clockwise. Note that the order of
operations is important. The above is equal to
{‘rotate’,45,’translate’,100*sqrt([2 2])/2}. Advanced note: for OpenGL
transform are applied in the reverse order from how they’re specified.
That is not the case for this interface, transform are applied in the
order specified.

The optional argument ‘vSpacing’ sets the spacing between the lines.
Default value is 1.

The optional argument ‘winRect’ allows to specify a [left top right
bottom] rectangle, in which the text should be placed etc. By default, the
rectangle of the whole ‘win’dow is used.

‘resetStyle’. If true, we reset the base text style to normal before
interpreting formatting commands that are present in the input text
string. If not (false), active text style at function entry is taken into
account when processing style toggle tags

‘cache’. Upon invocation of the function, it provides an optional output
‘cache’. Providing this output back to DrawFormattedText2 instead of the
‘tstring’ input allows direct drawing of the exact same text without all
the preprocessing having to be done again, potentially saving significant
time (and simplifying the call syntax). In this mode, a subset of the
below arguments can be used to, e.g., draw to a different window or
reposition the text. This cache can be generated without any actual
drawing being done by setting the ‘cacheOnly’ argument to true. The cache
is an implementation detail and is subject to change at any time.
When drawing from cache, the text can be repositioned with the ‘sx’,
‘sy’, ‘xalign’, ‘yalign’ and ‘winRect’ inputs described above. If only
‘sx’ and ‘sy’ are provided, these are taken to be offsets to move the
bounding box of the cached text. ‘sx’ and ‘sy’ must be numerical in this
case and cannot be empty. If ‘xalign’ and/or ‘yalign’ are provided,
full-fledged parsing of ‘sx’, ‘sy’, ‘xalign’ and ‘yalign’ is done, same
as during a normal call to DrawFormattedText2. The bounding box is then
repositioned according to these for inputs. The winRect argument is only
used in this case. It is optional (defaulting to the whole windows), and
works as described above, specifying the rect to which ‘xalign’ and
‘yalign’ apply. The input option ‘transform’ can furthermore be set,
which appends additional transformations to what is already in the cache.

Return variables:

The function returns the new (nx, ny) position of the text drawing cursor
and the bounding rectangle ‘textbounds’ of the drawn string. (nx,ny) can
be used as new start position for connecting further text strings to the
bottom of the drawn text string. Calculation of textbounds is
approximative, so it may give wrong results with some text fonts and
styles on some operating systems, depending on the various settings. The
optional ‘cache’ output argument is discussed above.

When rotating by angles that are not a multiple of 90 degrees, the
bounding box may not be accurate (too large). The returned bounding box
is the rect that tightly fits the rotated original bounding box, not the
box that tightly fits the rotated ink of the letters.

The optional return argument ‘wordbounds’, if assigned in the calling
function, returns a n-by-4 matrix of per-word bounding boxes. Each row
defines a [left,top,right,bottom] rectangle with the bounding box of a
word in the text string, ie. row 1 = first word, row 2 = 2nd word, …
white-space characters delimit single words, as do style changes and line-
feeds, and these delimiters are not taken into account for the bounding box,
ie. they don’t get their own bounding boxes. The white-space separating
successive words is as defined by the function isspace(tstring), or by a
change of text style, color, formatting, etc. Use of ‘wordbounds’ may cause
a significant slow-down in text drawing, so only assign this return argument
if you actually need it. A current limitation is that returned bounding boxes
will be likely incorrect if you apply multiple transformations like ‘scale’
‘rotate’, ‘translate’ and ‘flip’ at once. A single transformation will work,
but multiple ones will cause misplaced per word bounding boxes. If you want
to get proper ‘wordbounds’ when drawing text from the ‘cache’ then you must
assign ‘wordbounds’ already in the DrawFormattedText2() invocation which
returns the ‘cache’, otherwise bounding boxes might be wrong.

One difference in the return values from this function and
DrawFormattedText is that the new (nx, ny) position of the text drawing
cursor output is the baseline of the text. So to use (nx,ny) as the new
start position for connecting further text strings, you need to draw
these strings with yPositionIsBaseline==true. Another difference is that
the returned textbounds bounding box includes the height of an empty line
at the end if the input string ended with a carriage return.
DrawFormattedText only moved (nx,ny) but did not include the empty line
in the bounding box. The empty line is also taken into account when
centering text

Further Notes:

Please note that while positioning and bounding boxes are pixel accurate
with the fonts tested during development, i cannot guarantee this is the
case with all fonts you throw at it. Also note that this function is not
made to be fast. It has a make draw from cache mode so that all
preprocessing is done only once and stored in a cache from which the text
can be drawn directly. Nonetheless, especially if you throw a lot of
formatting at it, this function may still not be fast in this mode. Use
with caution in timing critical paths (test and measure to know if its
fast enough). That said, with reasonable inputs it should manage to draw
text to screen well within the inter frame interval of most screens.

The function employs clipping by default. Text lines that are detected as
laying completely outside the ‘win’dow or optional ‘winRect’ will not be
drawn, but clipped away. This allows to draw multi-page text (multiple
screen heights) without too much loss of drawing speed. If you find the
clipping to interfere with text layout of exotic texts/fonts at exotic
sizes and formatting, you can define the global variable…
global ptb_drawformattedtext2_disableClipping;
… and set it like this …
ptb_drawformattedtext2_disableClipping = 1;
… to disable the clipping.

Regardless of the clipping setting, the optional 3rd return parameter
‘textbounds’ always covers the complete text.

See DrawFormattedText2Demo for a usage example.