The issue is that the \button fragments in SatCom are replaced with
invalid UTF-8 bytes, hence those bytes need to be handled as a special
case when checking for UTF-8 character length.
When the text is processed e.g. in `Gfx::CText::Justify`, there are
generally two arrays: `text` and `format`. The second one is interpreted
as a font for a given character, e.g. `text[i]` has font `format[i]`.
Now, in the loops the index for `text` is increased by the length of
the character, e.g. 1 to 4 bytes, whereas index for `format` is
only incremented. It seems there's an assumption that `format` treats
multibyte characters as one byte, which doesn't seem to be true by
looking at the process of filling up the `format` array. This can result
in using wrong font e.g. FONT_SATCOM for the \button character which
should be FONT_BUTTON, hence the `StrUtils::Utf8CharSizeAt` method
complains about unexpected continuation byte via exception, causing
the crash.
Incrementing the index by the UTF-8 length seems to have fixed the
issue. However, I am not sure if I haven't broken anything else by this
and if I analyzed the intended behaviour correctly.
Gfx::CText needs a major refactor IMHO.
Moved list of mods logic to a new CModManager class.
The list of enabled mods is now managed by a flag instead of directory
names of mods.
Mods are now disabled by default.
Also general cleanup, fixing issues from the code review in
https://github.com/colobot/colobot/pull/1191 and fixing linter issues.
Regression: the state of enabled/disabled mods is now not persistent.
The plan is to use some kind of config file for this.
This significantly speeds up text rendering. On my computer, looking at
the program editor with a full screen of text, this commit takes the
framerate from under 30 to 60 (hitting vsync).
Performance could be further improved in the gl33 renderer by using
instancing or glPrimitiveRestartIndex instead of glMultiDrawArrays, but
that would be a more invasive change.
All of the interface rendering could use a unified quad batching system,
instead of it being limited to CText, but that would require some
refactoring in CText as it currently draws using a different coordinate
space to the rest of the interface.
Fixes#1104.
Text rendering is now done in window coordinates corresponding to window
pixels to avoid floating-point rounding errors that show up as rendering
artifacts
TTF fonts will be loaded to memory instead of file pointers
Added new class CSDLMemoryWrapper that loads data from PHYSFS into memory block
and closes file after. This closes issues #519, #708 and #619
Tomasz Kapuściński