There used to be a problem with the setfont and setrtl script commands.
Namely, if you used them in between text boxes naïvely, without any
careful thought, then the fading out text box would suddenly gain the
font of the new one. A kludge solution to this was implemented by simply
blocking the script until the existing text box faded out before
switching the font or RTL, and shipped for 2.4.0.
However, a better solution is to simply bake the font flags in to the
text box, so that way, if the level font switches, then the text box
keeps its font.
This is only for custom levels, because in the main game, the font in a
text box needs to be able to change depending on language. But it seems
like custom level translations weren't much on the roadmap, and so even
the existing hack didn't support changing the font based on translation
(even though translation of custom level cutscenes is supported). So
baking the font flags into the text box here doesn't make things any
worse.
It also makes things better, arguably, by allowing multiple text boxes
to exist on screen at once with different fonts.
Maybe in the future we'll need a flag that specifies that the font
should change depending on language if a translation in said language
exists for the text box, or something like that.
For people that want to override the fonts of every existing text box on
screen, you can specify "all" as the second parameter of setfont or
setrtl to do so.
This fixes a segmentation fault caused by an out-of-bounds indexing
caused by an attempt to unwordwrap a string that starts with two
newlines.
The problem here is that in the branch of the function
string_unwordwrap() where `consecutive_newlines == 1`, the function does
not check that the string `result` isn't empty before attempting to
index `result.size()-1`. If `result` is empty, then `result.size()` is
0, and `result.size()-1` becomes -1, and indexing a string at position
-1 is always undefined behavior.
Funnily enough, a similar indexing happens just a few lines down, but
this time, there is a check to make sure that the string isn't empty
first. I'm unsure of how Dav999 forgot that check a few lines earlier.
This situation can happen in practice, with custom level localizations.
I made a level with a filename of testloc.vvvvvv and created a file at
lang/fr/levels/testloc/custom_cutscenes.xml with the following content:
<?xml version="1.0" encoding="UTF-8"?>
<cutscenes>
<cutscene id="test" explanation="">
<dialogue speaker="cyan" english="This is text..." translation="blarg"/>
</cutscene>
</cutscenes>
Then I switched to French, created a script named `test`, and created a
text box that started with two newlines (so in total, the text box must
be at least 3 lines in length). Running the script triggers the segfault
when the text box is created. (Well, technically, on my machine, it
triggers an assertion fail in libstdc++ and aborts, but that's basically
the same thing.)
To fix this while still preserving the exact amount of newlines, if
`result` is empty, we add a newline instead of attempting to index the
string.
These are functions used in other files that are not on the
GraphicsResources class but are implemented inside the
GraphicsResources.cpp file. For some reason they were never put in the
GraphicsResources.h file until now, even though it's a perfectly good
header file to put them in.
If text is set to be centered, but is so long that it starts running
offscreen on both sides, the print function instead makes the text
start no further left than the left border of the screen (x=0).
This is because text running offscreen at the end only is more readable
and looks less sloppy than running offscreen at both sides.
For RTL, the opposite applies, so it now also works oppositely for RTL
prints, where centered strings will only run offscreen on the left side
of the screen.
They're invisible in font::print(), but they were still considered
characters with widths in the width function. This change made the
levels screen look better in RTL too - I was wondering why the level
options were too far left.
This lets you mirror the X axis specifically in RTL languages, so the
left border is 320 and the right border is 0, and invert the meaning of
PR_LEFT (0) and PR_RIGHT. Most of the time this is not necessary,
it's just for stuff where a label is followed by a different print,
like "Font: " followed by the font name, time trial time displays, etc
Again, the RTL property controls whether textboxes will be
right-aligned, and that kind of stuff. It can't be font-bound, since
Space Station supports Hebrew characters and we want to be able to
support, say, a Hebrew translation or Hebrew levels in the future
without having to make a dedicated (or duplicated) font for it.
Therefore it's a property of both the language pack as well as custom
levels - like custom levels already had a <font> tag, they now also
have an <rtl> tag that sets this property.
Right now, we'll have to hardcode it so the menu option for the Arabic
font sets the <rtl> property to 1, and all the other options set it to
0. But it's future-proof in that we can later decide to split the
option for Space Station into an LTR option and an RTL option (so both
"english/..." and "עברית" would select Space Station, but one sets the
RTL property to 0 and the other sets it to 1).
I'm now using SheenBidi to reorder RTL and bidirectional text properly
at text rendering time! For Arabic this is still missing reshaping, but
everything's looking really promising now!
The code changes are really non-invasive. The changes to Font.cpp are
absolutely minimal:
1305+ if (bidi_should_transform(text))
1306+ {
1307+ text = bidi_transform(text);
1308+ }
There's now a FontBidi.cpp, which implements these two functions,
notably bidi_transform(), which takes a UTF-8 encoded string and
returns another UTF-8 encoded string that has bidi reorderings and
reshapings applied.
In that function, SheenBidi gives us information about where in the
input string runs start and end, and on a basic level, all we need to
do there is to concatenate the parts together in the order that we're
given them, and to reverse the RTL runs (recognizable by odd levels).
As this is a proof-of-concept, bidi_should_transform() still always
returns true, applying the bidi algorithm to all languages and all
strings. I'm thinking of enabling bidi only when the language/font
metadata enables RTL (which could be for the interface or for a custom
level), or outside of that, at least when RTL characters are detected
(such as Arabic or Hebrew Unicode blocks).
This code was introduced by Dav999 in
abf12632bb (PR #1077), but it contains a
memory error. I spotted this with Valgrind.
The problem comes from the fact that `max_codepoint` is calculated from
the width and height of the surface (which will have the same width and
height as the source `font.png` from the filesystem). Let's work through
an example using a typical 128 by 128 `font.png` and an 8 by 8 glyph.
`chars_per_line` is calculated by dividing the width of the image
(`temp_surface->w`, or 128) by `f->glyph_w` (8), yielding 16.
`max_codepoint` is calculated by first calculating the height of the
image divided by the height of the glyph - which here just happens to be
the same as `chars_per_line` (16) since we have a square `font.png` -
and then multiplying the result by `chars_per_line`. 16 times 16 is 256.
Now it is important to recognize here that this is the _amount_ of
glyphs in `font.png`. It is _not_ the last codepoint in the image. To
see why, consider the fact that codepoint 0 is contained in the image.
If we have codepoint 0, then we can't have codepoint 256, because that
would imply that we have 257 codepoints, but clearly, we don't. If we
try to read codepoint 256, then after working through the calculations
to read the glyphs, we would be trying to read from pixel columns 0
through 7 and pixel rows 128 through 135... in a 128 by 128 image...
which is clearly incorrect.
Therefore, it's incorrect to write the upper bound of the for-loop
iterating over every codepoint as `codepoint <= max_codepoint` instead
of `codepoint < max_codepoint`.
VVVVVV 2.2 only supported displaying characters 00-7F with its font
system. VVVVVV 2.3 added support for unicode, by supplying a font.txt
with all the characters that are in the font image. But 2.3 made
another change that I didn't immediately realize, even after reading
the code: if font.txt is not present, then the font is not assumed to
have _only_ 00-7F, but _all_ of unicode, as far as the image dimensions
allow.
However, an inconsistency I _did_ notice is how unknown characters
would be rendered in 2.3. If a font had a font.txt, then any unknown
characters would be shown as a '?'. If a font had no font.txt however,
then suddenly any unknown characters would just come out as a space.
I fixed this behavior with the new font system; but what was actually
happening for characters to come out blank is that characters up to
U+00FF, which _were_ technically in the font image but as fully
transparent, would be shown as they were in the image, and characters
beyond U+00FF wouldn't be shown since they were outside of the image.
I don't really want to show blank characters for any character between
80-FF if it is technically inside the image, because pretty much every
single ASCII-only font.png in existence (including the one in data.zip)
contains a blank lower half, just because the font in the game had
always had this specific resolution. (We didn't want to do things that
might crash the game because something was different from what it
expected...)
We have had some confusing occasions before with the old behavior where
the fonts weren't correctly packaged or something (like when the
Catalan translator was sent the first version of the translator pack,
or when people customize their fonts wrong) and special characters were
just blank spaces.
So, instead, for characters beyond 7F, I decided to consider them part
of the font, as long as they are not blank. That means, if a character
beyond the ASCII range has any (non-alpha-0) pixels, then it will be
added, otherwise it won't be. This is just to handle legacy fonts, and
the case where all fonts are missing and the one from data.zip is used;
new fonts should just use .fontmeta or .txt to define their characters.
This commit removes the `NO_EDITOR` and `NO_CUSTOM_LEVELS` defines,
which cleans up the code a lot, and they weren't really needed anyways.
This commit also disables the editor on the Steam Deck, and adds a
program argument to re-enable the editor, `-enable-editor`.
This will actually do several things:
(1) Make the tile size checks apply to the appropriate graphics files
once again.
(2) Make the game print a fallback error message if the error message
hasn't been set on the levelDirError error screen.
(3) Use levelDirError for graphics errors too.
(4) Make the error message for tile size checks failing specify both
width and height, not just a square dimension.
(5) Make the error messages mentioned above translatable.
It turns out that (1) didn't happen after #923 was merged, since #923
removed needing to process a tilesheet into a vector of surfaces for all
graphics files except sprites.png and flipsprites.png. Thus, the game
ended up only checking the correct tile sizes for those files only.
In the process of fixing this, I also got rid of the PROCESS_TILESHEET
macros and turned them into two different functions: One to make the
array, and one to check the tile size of the tilesheet.
I also did (2) just in case FILESYSTEM_levelDirHasError() returns false
even though we know we have an error.
And (3) is needed so things are unified and we have one user-facing
error message system when users load levels. To facilitate this, I
removed the title string, since it's really not needed.
Unfortunately, (1) doesn't apply to font.png again, but that's because
of the new font stuff and I'm not sure what Dav999 has in store for
error checking. But that's also why I did (4), because it looks like
tile sizes in font.png files can be different (i.e. non-square).
This removes the `addnull` argument from `FILESYSTEM_loadFileToMemory`
and `FILESYSTEM_loadAssetToMemory`, and makes it so a null terminator is
always appended no matter what.
This simplifies things and removes the need for callers to make the
decision about null termination and what its implications are. Then you
get cases where null termination might not happen when it should be,
such as the one df577c59ef (#947) fixed.
When FIQ added the `addnull` argument in
5862af4445 (#117), I'm guessing he did it
because he wanted to be cautious about adding the null terminator to
every file, so he only did it for XML files, which was the only case
needed at the time. But really, there's no downsides to always appending
a null terminator. In fact, it's already always done whenever the STDIN
buffer is loaded.
An example is Maximally Misleading Miserable Misadventure, which has a
font.txt which includes all ASCII characters starting with a 0x00 byte.
This would accidentally null-terminate the string too early.
Instead, we now use the total length of the file again, and keep
getting the next UTF-8 codepoint until the file ends. We still need to
null-terminate it - it protects against incomplete sequences getting
the UTF-8 decoder to read out of bounds.
This was an oversight when we migrated to the new UTF-8 system - it
expects a null-terminated string, but the utfcpp implementation worked
with a pointer to the end of the file instead.
I also added an assert in FILESYSTEM_loadFileToMemory() so this is less
likely to happen again - because there should be no valid reason to
have a NULL pointer for the total file size, as well as not wanting a
null terminator to be added at the end of the file.
See the previous two commits, a lot of the time we don't need
std::string objects to be passed to these functions because we already
have C strings.
Commit 1/3: font::print_wrap
Commit 2/3: font::print
-> Commit 3/3: font::len
Turns out I was overplaying my hand a little when changing font::print
from std::string to const char*, so instead, I'll overload the
function: it can take either a const char* (the main function) or a
std::string (a wrapper). This means any C string that's printed
everywhere else (which is common, especially because loc::gettext gives
them) no longer needs to be converted to a std::string object each call.
Commit 1/3: font::print_wrap
-> Commit 2/3: font::print
Commit 3/3: font::len
We no longer need to pass a std::string object to the print and len
functions - in fact, we often only have a C string that we want to
print or get the visual width of (that C string most often comes from
loc::gettext), and it's a bit wasteful to wrap it in a new std::string
object on every print/len call.
This does mean adding a few more .c_str()s, but there's not many places
where a std::string is being passed to these functions, and we already
use .c_str() sometimes.
-> Commit 1/3: font::print_wrap
Commit 2/3: font::print
Commit 3/3: font::len
The hashmap would get populated with the name of each font, as each
font was being added. Unfortunately, adding a font would also realloc
the storage for fonts, in which the names are also stored... Possibly
invalidating the pointers to the names. This is now fixed by populating
the hashmap after all the fonts are added.
I'm also planning to change the argument types of font::len,
font::print and font::print_wrap from const std::string&s to
const char*s, but I'll do that separately.
There is no overlap in side effects between this line and the switch
statement after it, but it did result in adding the width of a final
null terminator or newline to the width of the current line, which is
a waste because those widths both 1) require trying to find
non-existent characters in the font and 2) will not be used.
I found this out because I added a debug print in find_glyphinfo(), and
something was requesting lots of codepoint 0 from the font.
In a button glyph font (like buttons_8x8.fontmeta) you can now specify
<type>buttons</type> to indicate that it's a button glyphs font. In a
normal font, you can specify <fallback>buttons_8x8</fallback>. This
will make it such that if a character is not found in the main font,
it will instead be looked for in buttons_8x8. If not found there
either, the main font's U+FFFD or '?' will be used as before.
This makes find_font_by_name() not O(n). It's not really a big deal,
because there won't be many fonts, but it'd make a function in the next
commit (finding the given fallback font for each font by name) O(n^2).
It's easy enough to add the hashmap.
Even though it would be a bizarre combination, declaring no character
set (neither via <chars> nor via font.txt) meant that <special>
couldn't be used because the ASCII fallback charset would be loaded
after special ranges were processed. Now, all the methods of loading
the charset are attempted sequentially, and only afterwards, the
special ranges are loaded.
There used to be two ways of fading in/out text in VVVVVV:
- Local code that modifies the R, G and B values of the text
- Keeping the RGB values the same and using the alpha channel
The latter approach is only used once, for [Press ENTER to return to
editor]. The former approach causes problems with colored (button)
glyphs: there's no way for the print function to tell from the RGB
values whether a color is "full Viridian-cyan" or "Viridian-cyan faded
out 50%", so I added the flag PR_COLORGLYPH_BRI(value) to tell the
print function that the color brightness is reduced to match the
brightness of colored glyphs to the brightness of the rest of the text.
However, there were already plans to make the single use of alpha
consistent with the rest of the game and the style, so PR_ALPHA(value)
could be removed, as well as the bit signifying whether the brightness
or alpha value is used. For the editor text, I simply copied the "Press
{button} to teleport" behavior of hiding the text completely if it
becomes darker than 100/255.
Another simplification is to make the print function handle not just
the brightness of the color glyphs while local code handled the
brightness of the normal text color, but to make the print function
handle both. That way, the callsite can simply pass in the full colors
and the brightness flag, and the flag name can be made a lot simpler as
well: PR_BRIGHTNESS(value).
If, somehow, a single character is wider than the limit, next_wrap
would get you stuck in an infinite loop by refusing to update the start
index and giving a line length of 0. Now, it just gives you a line with
that single character.
I also made some small readability improvements: I renamed next_wrap_s
to next_wrap_buf, and added comments at the top of both functions
explaining what they do.
By default, when you open the level editor to start a new level, the
level font will now match your VVVVVV language; so if you're, say,
Japanese, then you can make Japanese levels from the get-go. If you
want to make levels for a different target audience, you can change the
font via a new menu (map settings > change description > change font).
The game will remember this choice and it will become the new initial
level font.
If a custom level doesn't specify a font, it should be the 8x8 font.
But the main game can't specify a font, it's just the interface font
because that's for the language that the game is in.
They need to know how wide the text is going to be in a particular
font, so font::string_wordwrap and font::string_wordwrap_balanced now
take a flags argument like all the printing and dimensions-getting
functions. next_wrap and next_wrap_s take a Font* now, they're internal
to Font.cpp so they can take a Font and avoid double flag-parsing. But
if any non-Font.cpp code needs next_wrap/next_wrap_s in the future, I'd
just make a public wrapper that takes a uint32_t flags and passes the
Font* to the internal functions.
None of the structs in the new font system ended up being "publicly"
accessible, they were all treated as implementation details for
Font.cpp to use, so these structs are now fully defined in Font.cpp
only.
Only three deprecated functions remain:
- Graphics::Print
- Graphics::PrintWrap
- Graphics::bigprint
I also fixed multiline transparent textboxes having their outlines
overlap the text itself, and fixed textboxclass::padtowidth assuming
glyph widths of 8 (it made the hints at the start of intermission 1
run offscreen for example)
I especially focused on graphics.len and the print calls around them,
because graphics.len calls appear a bit less often, might be overlooked
when migrating print calls (thus possibly using different fonts by
accident) and are often used for some kind of right-alignment or
centering which can be changed into PR_RIGHT or PR_CEN with a different
X anyway.
Notably, I also added a new function to generate these kinds of
sliders: ....[]............
Different languages means that the slider for analogue stick
sensitivity needs to be longer to fit possibly long words for
Low/Medium/High, and then different font sizes means that the longer
slider won't fit onscreen in a language that needs a 12-wide font. So
slider_get() can take a "target width", which dynamically changes the
number of characters depending on the width of them in the interface
font.
I kinda forgot that I could force the 8x8 font instead of adapting the
characters in the slider to the font, and other ideas (like using
different characters or a more graphical progress bar) have been
brought up on Discord, so this might all change again sooner or later.
meta.xml can now have a <font> tag, which gives the name of the font
that the language needs. This will directly control the interface font
when the language is active, and will soon also control the font used
for each option on the language screen.
Also added some borders to more of the text in room name translator
mode, fixed a positioning issue if the interface font is not 8x8, and
migrated the trophy texts to font::print_wrap (including
PR_COLORGLYPH_BRI that still needed to be done)
Some textboxes need to be in the level font (like room names, cutscene
dialogue, etc - even in the main game), and some need to be in the
interface font (like when you collect a shiny trinket or crewmate). So
most of these textboxes now have graphics.textboxprintflags(font_flag)
as appropriate.
RoomnameTranslator.cpp is now also migrated to the new print system -
in room name translator mode, the room name is now displayed in the 8x8
font if it's untranslated and the level font if it is.
Level text such as room names, text box content, and the contents of
the script editor need to be displayed in the level-specific font, and
tweaked to look right. This involves displaying less lines in the
script editor, making text boxes bigger, displaying some text higher
and some text lower. This is still unfinished, but it's the real start
of a migration to font::print functions!
find_font_by_name() just finds the index of a given font name. This
index is supposed to be stored and reused, because the font (for a
language/level) won't be changed very often. So this function would
only run when getting the language metadata, when loading a level, etc.
All global fonts and all custom fonts in a level are now loaded, and
added to their respective "vectors". The selected font is still always
as the global font.png, and the custom level font also isn't selected
yet, but it's now easier to implement that.
Also, I added FILESYSTEM_enumerateAssets, which #902 already has but I
needed it now. I also rewrote it to not use std::vector<std::string>.
That was my idea, it's also how FILESYSTEM_getLanguageCodes worked,
so for symmetry, that function is getting changed as well.
The font::len function now handles the printing scale, so it can
immediately return the scaled length instead of having the caller
calculate it. The print function now handles CJK low/high flags and
vertically centers CJK text by default (instead of letting it stick
out on the bottom).
The following functions were moved directly:
- next_wrap
- next_wrap_s
- string_wordwrap
- string_wordwrap_balanced
- string_unwordwrap
These ones will probably still need get a flags argument, except for
string_unwordwrap (since they need to know what font we're talking
about.
The implementation of graphics.len has also been moved to Font.cpp,
but graphics.len still exists for now and is deprecated.
graphics.PrintWrap is now also deprecated. An advantage of the new
version (with flags) is that it'll be possible to do things like put
a border around wrapped text, wrap text at larger scales, etc, but
these things don't work perfectly yet.
This commit also has some other fixes, like the default advance of
6 pixels for characters 0x00-0x1F in 8x8 fonts.
