This fixes indexing out-of-bounds in the functions that draw all the
special images such as the elephant and teleporters. Let's make sure the
game doesn't segfault.
I tracked down all the functions that took in an entity's drawframe and
made sure that no matter what value an entity's drawframe was, the game
would never segfault.
To deal with using a different image file for Flip Mode, it looks like
copy-paste was used. This isn't exactly maintainable code. So I'm
replacing it with a reference that changes depending on if the game is
in Flip Mode or not, instead.
Removing the player entity has all sorts of nasty effects, such as
softlocking the game because many inputs require there to be a player
present, such as opening the quit menu.
The most infamous glitch to remove the player entity is the Gravitron
Fling, where the game doesn't see a gravity line at a specific
y-position in the current room, and when it moves the bottom gravity
line it moves the player instead. When the gravity line gets outside the
room, it gets destroyed, so if the player gets dragged outside the room,
they get destroyed, too. (Don't misinterpret this as saying anytime the
player gets dragged outside the room, they get destroyed - it's only the
Gravitron logic that destroys them.)
Also, there are many places in the code that use entity-getting
functions that have a fallback value of 0. If it was possible to remove
the player, then it's possible for this fallback value of 0 to index
obj.entities out-of-bounds, which is not good.
To fix this, entityclass::removeentity() is now a bool that signifies if
the entity was successfully removed or not. If the entity given is the
player (meaning it first checks if it's rule 0, just so in 99% of cases
it'll short-circuit and won't do the next check, which is if
entityclass::getplayer() says the indice to be removed is the player),
then it'll refuse to remove the entity, and return false.
This is a change in behavior where callers might expect
entityclass::removeentity() to always succeed, so I changed the
removeentity_iter() macro to only decrement if removing the entity
succeeded. I also changed entityclass::updateentities() from
'removeentity(i); return true;' to 'return removeentity(i);'.
Apparently it results in Undefined Behavior if the argument given isn't
representable as an unsigned char. This means that (potentially) plain
char and signed chars could be unsafe to use as well.
It's rare that this could happen in practice, though. std::isdigit() is
only used by is_positive_num() which is only used by find_tag(), so
someone would have to deliberately put something crazy after an `&#` in
a custom level file in order for this to happen. Still, better to be
safe than sorry and all.
This fixes a compile error that could happen where the compiler doesn't
know what std::isdigit() is, but I'm puzzled as to why this wasn't
happening earlier, 'cause I've only been reported that it happens by
only one person.
Continuing from #280, another potential source of out-of-bounds indexing
(and thus, Undefined Behavior badness) comes from script commands. A
majority of them don't do any input validation at all, which means the
potential for out-of-bounds indexing and segfaulting in custom levels.
So it's always good to add bounds checks to them.
Interesting note, the only existing command that has bounds checks is
the flag() command. That means you can't turn out-of-bounds flags on or
off. But there's no bounds checks for ifflag(), or customifflag(), which
means you CAN index out-of-bounds with those commands! That's a bit bad
to do, so.
Also, I decided to add the bounds checks for playef() at the
musicclass::playef() level, instead of just the level of the playef()
command. I don't know of any other cases outside of the command where
musicclass::playef() will index out of bounds, but musicclass is the one
containing the indexed vector anyway, I wanted to cover more cases, and
it's better to be safe than sorry.
And this the function with the least amount of cases where its sentinel
value is used unchecked. Thankfully. obj.getplayer() was a bit of a slug
to get through.
obj.getplayer() can return -1, which can cause out-of-bounds indexing of
obj.entities, which is really bad. This was by far the most changes, as
obj.getplayer() is the most used entity-getting function that returns
-1, as well as the most-used function whose sentinel value goes
unchecked.
To deal with the usage of obj.getplayer() in mapclass::warpto(), I just
added general bounds checks inside that function instead of changing all
the callers.
A few months ago, I added ghosts to the VVVVVV: Community Edition editor. I was told recently I should think
about upstreaming it, and with Terry saying go ahead I finally ported them into VVVVVV. There's one slight
difference however--you can choose whether you have them or not in the editor's settings menu. They're off by
default, and this is saved to the save file.
Anyway, when you're playtesting, the game saves the players position, color, room coordinates and sprite every 3
frames. The max is 100, where if it tries to add more, the oldest one gets removed.
When you exit playtesting, the saved positions appear one at a time, and you can use the Z key to speed it up.
[Here's a video of them in action.](https://o.lol-sa.me/4H21zCv.mp4)
2.2 and earlier had this god-awful thing where it put the closing tag of
an edentity onto the next line, and then kept the indentation the same.
This requires parsing the XML in an extremely specific way (i.e.
ignoring the whitespace) so the newline and indentation isn't taken as
part of the actual contents of the tag.
2.3 removed this awful whitespace entirely to make it easier on parsers.
When I tested #270, I tested against a 2.3 re-save of Dimension Open and
diffed the two, because I thought testing against the original version
of the level would result in a bunch of noise I didn't want due to the
whitespace change. Well, I did exactly what I intended, and ended up
ignoring the whitespace change so much that levels saved in this stupid
format ended up getting broken.
Luckily, we can just tell TinyXML-2 to parse a document exactly like how
TinyXML-1 would've parsed it, by supplying the COLLAPSE_WHITESPACE enum
to it (by default it's on PRESERVE_WHITESPACE).
This removes the TinyXML source files, removes it from CMakeLists.txt,
removes all the includes, and removes the functions
FILESYSTEM_saveTiXmlDocument() and FILESYSTEM_loadTiXmlDocument() (use
FILESYSTEM_saveTiXml2Document() and FILESYSTEM_loadTiXml2Document()
instead).
Additionally I've cleaned up the tinyxml2.h include in FileSystemUtils.h
so that it doesn't actually include tinyxml2.h unnecessarily, meaning a
change to TinyXML2 shouldn't rebuild all files that include
FileSystemUtils.h.
Seems a bit wasteful to do the whole "parse the XML document thing"
instead of a simple file check. It doesn't even fail if the XML document
is invalid, but whatever.
Ok, so it was a bit of a struggle at first figuring out the new API, but
honestly it wasn't so bad in the end.
I made a copy of my old unlock.vvv before testing this, and checking
with `diff` the only difference is the new `encoding="UTF-8"` in the XML
declaration, which isn't a bad thing.
Surprisingly, I only had to change some names and stuff around at the
top of the function. The rest of the function could be left untouched
and it worked fine.
Some of the file was indented with two spaces and the rest indented with
tabs. It feels like two different people worked on the file, one more
than the other. Since most of it uses two spaces, I'll just replace the
tabs with two spaces.
This is to respect the fact that the top half of the file is indented
with spaces, while the bottom half is indented with tabs.
Graphics::reloadresources() is on the bottom half.
The previous way manually concatenated the first 7 characters of the
string together (and had an std::min() calculation). The new way instead
does std::string::substr(), which is much more snappy.
The actual unindent is done in a separate commit to minimize noise,
because diffs are terrible at clearly conveying unindents (it should put
all the minus lines together and all the plus lines together, too).
The entirety of the rest of scriptclass::loadcustom() is encased in a
block that first checks if the script with the name even exists. Instead
of indenting the rest of the function, just invert the check and reduce
indentation level.
This commit refactors custom level scripts to no longer be stored in one
giant vector containing not only every single script name, but every
single script's contents as well. More specifically,
scriptclass::customscript has been converted to an std::vector<Script>
scriptclass::customscripts (note the extra S), and a Script is just a
struct with an std::string name and std::vector<std::string> contents.
This is an improvement in both performance and maintainability. The game
no longer has to look through script contents in case they're actually
script names, and then manually extract the script contents from there.
Instead, all it has to do is look for script names only. And the
contents are provided for free. This results in a performance gain.
Also, the old system resulted in lots of boilerplate everywhere anytime
scripts had to be handled or parsed. Now, the boilerplate is only done
when saving or loading a custom level. This makes code quality much,
much better.
To be sure I didn't actually change anything, I tested by first saving
Dimension Open in current 2.3 (because current 2.3 gets rid of the
awful edentity whitespace), and then resaved it on this patch. There is
absolutely no difference between the current-2.3-resave and
this-patch-resave.
This resulted in two bugs:
1. Custom assets would not be unmounted when quitting to the menu.
2. Custom assets would be unmounted when playtesting a level.
The solution is to unmount assets in Game::quittomenu() instead.
Main game would retain custom level assets, now fixed. Also, custom fonts load properly. Finally, levels can be stored as a zip and placed in the levels folder, with the .vvvvvv file at the root of the zip and custom asset folders (graphics, sounds etc) also at the root.
Previously, the editor would always say it saved or loaded a level,
even if it was not successful. For example, because a file to load does
not exist, a file to save has illegal characters in its name or the
name is too long to be stored. Now failure is reported. Also, when
quitting the editor and saving before quitting is unsuccessful, the
editor will abort quitting.
I think it's a bit silly to always include the Steam and GOG APIs
whenever we build VVVVVV, since the only time they'll ever be used is in
a live build and not a dev build.
So now Steam and GOG are disabled by default. If you want them, you'll
need to add -DSTEAM=ON or -DGOG=ON respectively at CMake time. They're
also both automatically enabled for release builds.
This commit adds bounds checks to those commands in case say()/reply()
asks for more lines than there are left in the all-script-lines buffer
(not just the current script, so in order for it to segfault your script
has to be last in the all-script-lines vector), and in case text() asks
for more lines than there are in the rest of the rest of the parsed
internal script.
For some reason, scriptclass::loadcustom() sometimes refers to
`customscript` as `script.customscript` instead of `customscript`. The
`script.` is unnecessary.
Although it's not an issue, this should minimize the stack footprint of
calling scriptclass::load(), especially if it goes down to calling
scriptclass::loadcustom() or scriptclass::loadother().
Otherwise, it would end up indexing a vector out-of-bounds, which is UB
and causes a segfault, too. This is used in some constructs like
say(-1)
<internal command>
where the text contents don't matter, only that a text box shows up.
The `speak` command is the exact same as the `speak_active` command,
except without one line of code. So instead of copy-pasting the entire
thing, it's better to just combine them into the same chunk of code.
Instead each line is now held in a const char* array, like it should be.
This results in less work for the compiler, especially with
optimization, since every time the compiler encounters a constant
argument in a function, it has to go off and locate a place to put it.
But if we're upfront and say, hey, here's all the strings we ever need
conveniently packaged into one place, it'll be much more cooperative.
I have the feeling that none of the devs understood what extern did, and
they kind of just sprinkled it everywhere until things started working.
But like all other classes, it should just be one line in the class's
respective header file, and shouldn't be so messy.
When the game loads a room in a custom level, previously it would load
the tilemap of that room into ed.swapmap, and then mapclass::loadlevel()
would manually go through each element in ed.swapmap to set each tile in
`contents`. Why do that, when you can just return the vector from
editorclass::loadlevel() and set it directly? ed.swapmap is really
unnecessary.
It's a bit bad for the compiler if you have lots of function calls with
hardcoded strings in them, because every time the compiler encounters
one, it has to go out of its way to find a dedicated storage location
for the string, which is really inefficient. And it does this
inefficient thing every single time.
There's not much of an impact compiling these lists, but I at least want
to encourage this sort of code style, instead of the push_back(string)
style, in case we ever need a hardcoded array of things later.
Resetting game.activeactivity fixes triggering Undefined Behavior if you
quit to the menu while standing inside an activity zone, and then
re-entered the game. Resetting game.act_fade also fixes the activity
zone prompt fadeout that happens once the above segfault is fixed.
Don't worry, other activity-zone like variables such as teleporter
prompts and trophy text are already covered. obj.trophytext is reset in
scriptclass::hardreset(), too, and game.readytotele is reset every time
mapclass::gotoroom() is called, and mapclass::gotoroom() is called every
time a gamemode is started.
For some reason, the only way to get a cyan crewmate is by cycling
through an already-existing crewmate by keeping left-clicking on it.
This is because when you cycle through crewmate colors, the allowed
colors are 0-5, but when you place down a crewmate, it picks a random
color from 1-5, which seems to be a bit consistent.
So placing and cycling a crewmate now use the same color ranges.
Blackout mode doesn't work properly, because the game doesn't actually
black out the screen, it merely stops drawing things. Oh and only some
things at that, some other things are still drawn. This results in a
freeze-frame effect, which is apparently fixed in the Switch version.
Custom levels utilize this freeze-frame effect, so it's a bit annoying
that the "Game paused" screen draws on top of it and makes it so that
the freeze-frame freezes on "Game paused" until blackout is turned off.
So I'm making it so that "Game paused" doesn't get drawn in blackout
mode. However it will still do graphics.render() because otherwise it'll
just result in a black screen.
This fixes being able to trigger Undefined Behavior by pressing R when
not in-bounds in the Outside Dimension VVVVVV map, usually when you're
falling upwards towards Game Complete.
I also put bounds checks on normal roomdeaths for good measure. You'll
never know when you need it.
Instead of using somewhat-obtuse for-loops to initialize or reset these
vectors, it takes up less lines of code and is clearer if we use
std::vector::resize() and std::vector::clear() instead.
Since translucent roomname backgrounds were introduced in
TerryCavanagh/VVVVVV#122, it exposes one glaring flaw with the game that
until now has been kept hidden: in rooms with room names, the game
cheapens out with the tile data and doesn't have a 30th row, because the
room name would hide the missing row. As a result, rooms with room names
have 29 rows instead of 30 to fill up the entire screen. And it looks
really weird when there's nothing but empty space behind the translucent
room name background.
To remedy this, I added one row to each room with a room name in the level.
First, I had to filter out all the rooms with no room names. However, that's
actually all contained in Otherlevel.cpp, the Overworld, which contains 221
rooms (8 of which are the Secret Lab, 6 more of which are the Ship, so 207 are
the actual Overworld, right? Wrong, 2 of those Overworld no-roomname rooms are
in the Lab, so there are actually 205 Overworld rooms). The remaining level
data files all contain rooms with room names.
But the process wasn't that easy. I noticed a while ago that each room
contains 29 `tmap.push_back()`s, one for each row of the room, and each row is
simply a string containing the 40 tiles for that row, concatenated with
commas.
However, I decided to actually check my intuition by doing a grep on each
level file and counting the number of results, for example `grep 'push_back'
Labclass.cpp | wc -l`. Whatever number comes out should be divisible by 29.
That particular grep on Labclass.cpp returns 1306, which divided by 29 is 45
with a remainder of 1.
So what does that mean? Does that mean there's 45 rooms each, and 1 leftover
row? Well, not exactly. The extra row comes from the fact that Outer Space has
30 rows instead of 29. Outer Space is the room that comes up when the game
finds a room is non-existent, which shouldn't happen with a properly-working
game, except in Outside Dimension VVVVVV. In fact, each level file has their
own Outer Space, and every single Outer Space also has 30 rooms. So really,
this means there are 44 rooms in the Lab and one Outer Space room. (Well, in
reality, there are 46 rooms in the Lab, because 2 of them use the Outside
tileset but have no room names, so they're stored in Otherlevel.cpp instead.)
We find the same result for the Warp Zone. `grep 'push_back' WarpClass.cpp |
wc -l` returns 697, which is 24 remainder 1, meaning 23 rooms of 29 rows and 1
room of 30 rows, which corresponds with 23 rooms in the Warp Zone and one
Outer Space room.
However, Outside Dimension VVVVVV + Tower Hallways and Space Station 1 and 2
are both odd curiosities. Finalclass.cpp contains Outside Dimension VVVVVV,
(which is Intermission 1 and 2 and the Final Level), but also the Tower
Hallway rooms, i.e. the auxiliary Tower rooms that are not a part of the main
tower. Spacestation2.cpp contains both Space Station 1 and 2, so don't be
deceived by the name.
`grep 'push_back' Finalclass.cpp | wc -l` returns 1597, which is actually 55
remainder 2. So... are there two rooms with 30 rows? Yes, in fact, The
Gravitron and Outer Space both contain 30 rows. So there are actually 55 rooms
stored in Finalclass.cpp (not including the minitowers Panic Room and The
Final Challenge), 54 rooms of actual level data and one Outer Space room, and
breaking down the 54 rooms even further, 51 of them are actually in Outside
Dimension VVVVVV and 3 of them are Tower Hallways. Of the 51 Outside Dimension
VVVVVV rooms, 14 of those are Intermission 1, 4 of them are Intermission 2,
and the rest of the 33 rooms are the Final Level (again, not including the
minitowers).
`grep 'push_back' Spacestation2.cpp | wc -l` returns 2148, which is 74
remainder 2. Are there two rooms with 30 rows again? No; one of those counted
2148 rows is a false-positive, because there's an if-else in Prize for the
Reckless that replaces the row with spikes with a row without spikes if you
are in a time trial or in No Death Mode. So there's 73 rooms in Space Station
1 and 2, and one Outer Space room.
With all this in mind, I decided to duplicate the current last row of each
room, the 29th row, to add a 30th row. However, I wasn't going to do this
automatically! But neither was I going to write some kludge-y code to parse
each nightmare of a level file and duplicate the rows that way.
Enter: Vim macros! (Er, well, actually, I use Neovim.) I first did
`/push_back`, so that pressing `n` would keep going to the next `push_back` in
the file. Then I went to the 29th row of the first room in the file, did a
`Yp`, and then started my macro with `qq`. The macro went like this: `30nYp`,
which is simply going to the 29th row of the next room over and duplicating
it. And that's all there was to it. However, I had to make sure that (1) my
cursor was before the `push_back` on the line of the 29th row of the room, and
(2) that I didn't skip rooms, both of which were problems I encountered when
pressing Ctrl+Z a given invocation of the macro (the Ctrl+Z is just a
metaphor, you actually undo by typing `u` in Vim). And also I had to make sure
to be careful around the extra lines of `push_back`s in Prize for the Reckless
and The Gravitron, and make sure I didn't run past the end of the file and
loop back around. Thankfully, all Outer Space rooms are at the end of each
file.
But first, I had to increase the number of rows drawn in Graphics.cpp by 1 in
order to compensate for this, and do the same when reading the tile data in
Map.cpp. I had to change fillcontent(), drawmap(), drawfinalmap(),
drawtowermap(), and drawtowermap_nobackground(). Funnily enough, the tower
functions already used 30 rows, but I guess it's an off-by-one due to the
camera scrolling, so they now draw 31 rows each.
Then, I went in-game to make sure that the row behind each room name looked
fine. I checked EVERY single room with a room name. I turned on invincibility
mode and added a temporary line to hardreset() that always turned on
game.nocutscenes for a smoother playtesting experience. And to make sure that
rooms which have entirely empty bottom rows actually still have 30 rows,
instead of having 29 and the game assuming that the 30th row was empty
(because that sounds like it could lead to Undefined Behavior), I added this
temporary debugging line to the start of mapclass::fillcontent():
printf("(%i,%i) has %i rows\n", game.roomx, game.roomy, (int) tmap.size());
Everywhere I checked - and I made sure to check all rooms - every room had 30
rows and not 29 rows.
Unfortunately, some rooms simply couldn't be left alone with their 29th row
duplicated and had to be manually edited. This was because the 29th row would
contain some edge tiles because the player would be able to walk somewhere on
the 28th, 27th, and 26th rows, and if you duplicated said edge tiles behind
the room name, it would look bad.
Here's a list of rooms whose 30th rows I had to manually edit:
- Comms Relay
- The Yes Men
- Stop and Reflect
- They Call Him Flipper
- Double-slit Experiment
- Square Root
- Brought to you by the letter G
- The Bernoulli Principle
- Purest Unobtainium
- I Smell Ozone
- Conveying a New Idea
- Upstream Downstream
- Give Me A V
- $eeing Dollar $ign$
- Doing Things The Hard Way
- Very Good
- Must I Do Everything For You?
- Now Stay Close To Me...
- ...But Not Too Close
- ...Not as I Do
- Do Try To Keep Up
- Whee Sports
- As you like it
This is actually a strange case where it looked bad because of the 29th
row, instead of the 30th row, and I had to change the 29th row instead of
the 30th row to fix it.
- Maze With No Entrance
- Ascending and Descending
- Mind The Gap
Same strange case as "As you like it" (it's the 29th row I had to change
that was the problem, not the 30th).
- 1950 Silverstone Grand V
- The Villi People
I found that Panic Room and The Final Challenge also looked strange behind the
roomname background, but I can't do much about either because towers' tile
data wrap around at the top and bottom, and if I added another row to either
it would be visible above the room name.
I've considered updating the development editors with these new level tiles,
but I decided against it as the development editors are already pretty
outdated anyway.
This didn't cause any issues on the old push_back(string) system, but it
causes a huge alignment issue now. Don't know why these 7 zeroes are
here, but I'm fixing them.
It's a bit inconsistent how every time you toggle flip mode, it does
this flashing and shaking (and different SFX), regardless of whether or
not you're turning it on or off. To be more consistent with what happens
when you toggle screen effects, only turning on Flip Mode should do the
flashing/shaking/game-saved sound, and turning it off should just play
the Viridian squeak.
It's always personally irked me that the only time you get a Viridian
squeak when pressing ACTION to start fading out and going in-game is
when you start playing a custom level. And that's only if you don't have
a quicksave.
To make things more consistent (and to add more polish to the game), I
made sure there was a music.playef(11) every time game.mainmenu gets
set. I also made sure that this doesn't result in double-squeaking, i.e.
music.playef(11) being called twice, which can be very loud and
ear-hurting.
