PhysFS by default just uses system malloc(), realloc(), and free(); it
provides a way to change them, with a struct named PHYSFS_Allocator and
a function named PHYSFS_setAllocator().
According to PhysFS docs, this function should be called before
PHYSFS_init(), which is why this allocator stuff is handled in
FileSystemUtils.cpp.
Also, I've had to make two "bridge" functions, because PHYSFS_Allocator
wants pointers to functions taking in `PHYSFS_uint64`s, not `size_t`s.
Apparently in C, if you have `void test();`, it's completely okay to do
`test(2);`. The function will take in the argument, but just discard it
and throw it away. It's like a trash can, and a rude one at that. If you
declare it like `void test(void);`, this is prevented.
This is not a problem in C++ - doing `void test();` and `test(2);` is
guaranteed to result in a compile error (this also means that right now,
at least in all `.cpp` files, nobody is ever calling a void parameter
function with arguments and having their arguments be thrown away).
However, we may not be using C++ in the future, so I just want to lay
down the precedent that if a function takes in no arguments, you must
explicitly declare it as such.
I would've added `-Wstrict-prototypes`, but it produces an annoying
warning message saying it doesn't work in C++ mode if you're compiling
in C++ mode. So it can be added later.
One of these days, I need to get around to running Include What You Use
on this codebase. Until then, while I was working on #624, I noticed
these; I'm removing them now.
The recently released SDL 2.0.14 adds a native function for opening URIs
from the host system, superseding the OS-specific implementations of
FILESYSTEM_openDirectory.
If PHYSFS_enumerate() isn't successful, we now print that it wasn't
successful, and print the PhysFS error message. (We should get that
logging thing going sometime...)
Note that level dir listing still uses plenty of STL (including the end
product - the `LevelMetaData` struct - which, for the purposes of 2.3,
is okay enough (2.4 should remove STL usage entirely)); it's just that
the initial act of iterating over the levels directory no longer takes
four or SIX(!!!) heap allocations (not counting reallocations and other
heap allocations this patch does not remove), and no longer does any
data marshalling.
Like text splitting, and binary blob extra indice grabbing, the current
approach that FILESYSTEM_getLevelDirFileNames() uses is a temporary
std::vector of std::strings as a middleman to store all the filenames,
and the game iterates over that std::vector to grab each level metadata.
Except, it's even worse in this case, because PHYSFS_enumerateFiles()
ALREADY does a heap allocation. Oh, and
FILESYSTEM_getLevelDirFileNames() gets called two or three times. Yeah,
let me explain:
1. FILESYSTEM_getLevelDirFileNames() calls PHYSFS_enumerateFiles().
2. PHYSFS_enumerateFiles() allocates an array of pointers to arrays of
chars on the heap. For each filename, it will:
a. Allocate an array of chars for the filename.
b. Reallocate the array of pointers to add the pointer to the above
char array.
(In this step, it also inserts the filename in alphabetically -
without any further allocations, as far as I know - but this is a
COMPLETELY unnecessary step, because we are going to sort the list
of levels by ourselves via the metadata title in the end anyways.)
3. FILESYSTEM_getLevelDirFileNames() iterates over the PhysFS list, and
allocates an std::vector on the heap to shove the list into. Then,
for each filename, it will:
a. Allocate an std::string, initialized to "levels/".
b. Append the filename to the std::string above. This will most
likely require a re-allocation.
c. Duplicate the std::string - which requires allocating more memory
again - to put it into the std::vector.
(Compared to the PhysFS list above, the std::vector does less
reallocations; it however will still end up reallocating a certain
amount of times in the end.)
4. FILESYSTEM_getLevelDirFileNames() will free the PhysFS list.
5. Then to get the std::vector<std::string> back to the caller, we end
up having to reallocate the std::vector again - reallocating every
single std::string inside it, too - to give it back to the caller.
And to top it all off, FILESYSTEM_getLevelDirFileNames() is guaranteed
to either be called two times, or three times. This is because
editorclass::getDirectoryData() will call editorclass::loadZips(), which
will unconditionally call FILESYSTEM_getLevelDirFileNames(), then call
it AGAIN if a zip was found. Then once the function returns,
getDirectoryData() will still unconditionally call
FILESYSTEM_getLevelDirFileNames(). This smells like someone bolting
something on without regard for the whole picture of the system, but
whatever; I can clean up their mess just fine.
So, what do I do about this? Well, just like I did with text splitting
and binary blob extras, make the final for-loop - the one that does the
actual metadata parsing - more immediate.
So how do I do that? Well, PhysFS has a function named
PHYSFS_enumerate(). PHYSFS_enumerateFiles(), in fact, uses this function
internally, and is basically just a wrapper with some allocation and
alphabetization.
PHYSFS_enumerate() takes in a pointer to a function, which it will call
for every single entry that it iterates over. It also lets you pass in
another arbitrary pointer that it leaves alone, which I use to pass
through a function pointer that is the actual callback.
So to clarify, there are two callbacks - one callback is passed through
into another callback that gets passed through to PHYSFS_enumerate().
The callback that gets passed to PHYSFS_enumerate() is always the same,
but the callback that gets passed through the callback can be different
(if you look at the calling code, you can see that one caller passes
through a normal level metadata callback; the other passes through a zip
file callback).
Furthermore, I've also cleaned it up so that if editorclass::loadZips()
finds a zip file, it won't iterate over all the files in the levels
directory a third time. Instead, the level directory only gets iterated
over twice - once to check for zips, and another to load every level
plus all zips; the second time is when all the heap allocations happen.
And with that, level list loading now uses less STL templated stuff and
much less heap allocations.
Also, ed.directoryList basically has no reason to exist other than being
a temporary std::vector, so I've removed it. This further decreases
memory usage, depending on how many levels you have in your levels
folder (I know that I usually have a lot and don't really ever clean it
up, lol).
Lastly, in the callback passed to PhysFS, `builtLocation` is actually no
longer hardcoded to just the `levels` directory, since instead we now
use the `origdir` variable that PhysFS passes us. So that's good, too.
When Ethan added PhysFS to the game, he put in a hardcoded check (marked
with a FIXME) that explicitly removed all filenames that were "data"
returned by PHYSFS_enumerateFiles(). Apparently this was due to a weird
bug with the function putting in "data" strings in its output in PhysFS
2.0.3; however, the game now uses PhysFS 3.0.2, and I could not
reproduce this bug on my system. (I also tested, and this also
straight-up ignores legitimate level filenames that just happen to be
"data" (without the .vvvvvv extension).)
After talking with Ethan in Discord DMs, I asked if we could remove this
check, and he said that we could. So I'm doing it now.
Following Ethan's example of bailing (calling VVV_exit()) if
binaryBlob::unPackBinary() couldn't allocate memory, I've searched
through and found every SDL_malloc(), then made sure that if it returned
NULL, the caller would bail (because you can't do much when you're out
of memory).
There should probably be an error message printed when the process is
out of memory, but unPackBinary() doesn't print an error message for
being out of memory, so this can probably be added later. (Also we don't
really have a logging system, I'd like to have something like that added
in first before adding more messages.)
Also, this doesn't account for any allocators used by STL stuff, but
we're working on removing the STL, and allocation failure just results
in an abort anyway, so there's not really a problem there.
A quick glance at PhysFS source code will show that PhysFS will bail if
PHYSFS_deinit() is called if it's not initialized.
"Bail" here just means setting an error code and returning early, so
it's not that bad. Still, it's the principle of the thing, and I just
want to ensure that FILESYSTEM_deinit() can be safely called no matter
if the filesystem hasn't initialized yet; having an error set by PhysFS
kind of taints the whole safety thing, even if it does nothing wrong,
no?
(although, speaking of which, we should be handling all errors by
PhysFS, but that's for later...)
During 2.3 development, there's been a gradual shift to using SDL stdlib
functions instead of libc functions, but there are still some libc
functions (or the same libc function but from the STL) in the code.
Well, this patch replaces all the rest of them in one fell swoop.
SDL's stdlib can replace most of these, but its SDL_min() and SDL_max()
are inadequate - they aren't really functions, they're more like macros
with a nasty penchant for double-evaluation. So I just made my own
VVV_min() and VVV_max() functions and placed them in Maths.h instead,
then replaced all the previous usages of min(), max(), std::min(),
std::max(), SDL_min(), and SDL_max() with VVV_min() and VVV_max().
Additionally, there's no SDL_isxdigit(), so I just implemented my own
VVV_isxdigit().
SDL has SDL_malloc() and SDL_free(), but they have some refcounting
built in to them, so in order to use them with LodePNG, I have to
replace the malloc() and free() that LodePNG uses. Which isn't too hard,
I did it in a new file called ThirdPartyDeps.c, and LodePNG is now
compiled with the LODEPNG_NO_COMPILE_ALLOCATORS definition.
Lastly, I also refactored the awful strcpy() and strcat() usages in
PLATFORM_migrateSaveData() to use SDL_snprintf() instead. I know save
migration is getting axed in 2.4, but it still bothers me to have
something like that in the codebase otherwise.
Without further ado, here is the full list of functions that the
codebase now uses:
- SDL_strlcpy() instead of strcpy()
- SDL_strlcat() instead of strcat()
- SDL_snprintf() instead of sprintf(), strcpy(), or strcat() (see above)
- VVV_min() instead of min(), std::min(), or SDL_min()
- VVV_max() instead of max(), std::max(), or SDL_max()
- VVV_isxdigit() instead of isxdigit()
- SDL_strcmp() instead of strcmp()
- SDL_strcasecmp() instead of strcasecmp() or Win32 strcmpi()
- SDL_strstr() instead of strstr()
- SDL_strlen() instead of strlen()
- SDL_sscanf() instead of sscanf()
- SDL_getenv() instead of getenv()
- SDL_malloc() instead of malloc() (replacing in LodePNG as well)
- SDL_free() instead of free() (replacing in LodePNG as well)
This patch cleans up unnecessary exports from header files (there were
only a few), as well as adds the static keyword to all symbols that
aren't exported and are specific to a file. This helps the linker out in
not doing any unnecessary work, speeding it up and avoiding silent
symbol conflicts (otherwise two symbols with the same name (and
type/signature in C++) would quietly resolve as okay by the linker).
strcat()s and strcpy()s have been replaced with SDL_snprintf() where
possible, to clearly convey the intent of just building a string that
looks a certain way, instead of spanning it out over multiple lines.
Where there's not really a good way to avoid strcat()/strcpy() (e.g. in
PLATFORM_getOSDirectory()), they will at least be replaced with
SDL_strlcat() and SDL_strlcpy(), which are safer functions and are less
likely to have issues with null termination.
I decided not to bother with PLATFORM_migrateSaveData(), because it's
going to be axed in 2.4 anyways.
There's no need to call a string function and have function call
overhead if you remember how C strings work: they have a null
terminator. So if the first char in a string is a null terminator, then
the string is completely empty. So you don't need to call that function.
The previous check by mwpenny had a few issues:
(a) It was completely overcomplicated for no good reason, and was
basically a Rube Goldberg machine. The original check was...
(1) Creating an std::string of the last char of 'output'...
(2) ...except instead of using the normal std::string constructor, it
was using the one where you pass in a number and a char to create
a string that's just that char repeated N times... except this
was only used to create a 1-length string.
(3) Converted that std::string to a C string.
(4) Then passed it to strcmp(), despite the string at this point
being only one byte and you could just compare the char values
directly.
The original check could've just been:
output[SDL_strlen(output) - 1] == *pathSep
(b) Use of libc strcmp() and strlen() instead of SDL_strcmp() and
SDL_strlen().
Now, actually, PHYSFS_getDirSeparator() happens to be a char array and
not a single char, so mwpenny was going in the right direction by using
strcmp() after all. Except it doesn't seem like he thought about the
fact that PHYSFS_getDirSeparator() could be multiple bytes instead of
one, and so he ended up making the first argument to strcmp() always be
a one-byte char array.
So there's issue (c), which is that it assumes the path separator is one
byte instead of multiple.
This commit fixes all of these issues with the trailing path separator
check.
Okay, so basically here's the include layout that this game now
consistently uses:
[The "main" header file, if any (e.g. Graphics.h for Graphics.cpp)]
[blank line]
[All system includes, such as tinyxml2/physfs/utfcpp/SDL]
[blank line]
[All project includes, such as Game.h/Entity.h/etc.]
And if applicable, another blank line, and then some special-case
include screwy stuff (take a look at editor.cpp or FileSystemUtils.cpp,
for example, they have ifdefs and defines with their includes).
Including a header file inside another header file means a bunch of
files are going to be unnecessarily recompiled whenever that inner
header file is changed. So I minimized the amount of header files
included in a header file, and only included the ones that were
necessary (system includes don't count, I'm only talking about includes
from within this project). Then the includes are only in the .cpp files
themselves.
This also minimizes problems such as a NO_CUSTOM_LEVELS build failing
because some file depended on an include that got included in editor.h,
which is another benefit of removing unnecessary includes from header
files.
This is just in case these values happen to be used without being
initialized or anything. I vaguely recall someone reporting an issue
where they didn't have a "Documents" folder on Windows and their level
folder ended up being a garbage path, so it's good to do this.
Some levels (like Unshackled) have decided to manually re-color the
one-way tiles on their own, and us overriding their re-color is not
something they would want. This does mean custom levels with custom
assets don't get to take advantage of the re-color, but it's the exact
same behavior as before, so it shouldn't really matter that much.
I would've liked to specifically detect if a custom tiles.png or
tiles2.png was in play, rather than simply disabling it if any asset was
mounted, but it seems that detecting if a specific file was mounted from
a specific zip isn't really PHYSFS's strong suit.
The assets mounting code was put directly in editorclass::load(), but
now it's in a neat little function so it can be called from multiple
places without having to call editorclass::load().
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.
Whenever I compile with -O2, GCC gives me a warning that the return
value of fread() is being ignored. Which is a fair point, it means that
we're not doing proper error handling if something goes wrong. But I'm
also going to check the return value of fwrite() for good measure.
I believe that checking that this number is not equal to length is the
way to catch all errors and output an error message accordingly. I
didn't use ferror() or feof() mostly because I think it takes up too
much code. Also an error from fwrite() only says "Warning" because I
don't think there's much we can do if we don't fully write all bytes to
the intended file.
Previously:
- Linux: xdg-open
- Everything else: open
Now:
- macOS and Haiku: open
- Everything else: xdg-open
This is all according to a comment by leo60228 in PR #203.
The problem here is that we're directly using the C stdio library,
instead of using PHYSFS's stuff. So I've added a function
FILESYSTEM_delete() that does exactly that.
The environment variable SteamTenfoot corresponds with the game running
in Steam Big Picture mode or SteamOS if it is defined. There's a
certification process for both full controller support and Big Picture
mode, and being able to launch a file window in Big Picture mode is an
instant cert failure.
Have to add some includes and put these behind some ifdefs, of course.
I'm pretty sure FreeBSD and OpenBSD and Haiku are POSIX enough that the
"open" command will work on them, too.
I would've loved to make FILESYSTEM_openDirectoryEnabled a simple bool
instead of a function, but I ran into issues with putting it in the
FileSystemUtils header file, so I'll just make it a function and call it
a day.
This fixes a bug where levels in the levels list duplicate if there's an
invalid file (such as a folder) in the levels directory.
It looks like it happens because we don't free the memory if
PHYSFS_readBytes() encounters an error, even though we should. Then we
get into Undefined Behavior territory and end up reusing memory, and
here it just happens that previously, parsing the entire XML document
for each level file was enough to make the loaded file pointer point to
garbage that would fail the metadata check, but if we optimize it so we
don't parse the entire XML document, it starts reusing memory instead.
This is useful for distributions, which may not want to put data.zip in
the same directory as the binary. This can't be distribution-specific
due to the license ("Altered source/binary versions must be plainly
marked as such, and must not be misrepresented as being the original
software.").
This uses utfcpp combined with a custom font, in the form of a PNG and text file. By default, the game acts exactly as it did before; custom fonts can be provided by third parties.
* Add a null terminator to loaded TinyXML files
The TinyXML parse() function expect a C-like string, including terminator.
When xml loading was changed, it loaded the file, but included no such thing.
Thus, we load the file, then reallocate the memory so that we can insert a
null terminator to it, before passing it to parse().
* Tweak TinyXML file loading
Instead of first loading the file content into memory, then reallocate it
to add a null pointer, add an argument to the file load function for whether
to append a null terminator or not, defaulting to false. It still returns the
length without the null pointer in case a length ptr is passed.
The TinyXml functions to load and save files don't properly support
unicode file paths on Windows, so in order to support that properly, I
saw no other option than to do the actual loading and saving via PHYSFS
(or to use the Windows API on Windows and retain doc.LoadFile and
doc.SaveFile on other OSes, but that'd be more complicated and
unnecessary, we already have PHYSFS, right?).
There are two new functions in FileSystemUtils:
bool FILESYSTEM_saveTiXmlDocument(const char *name, TiXmlDocument *doc)
bool FILESYSTEM_loadTiXmlDocument(const char *name, TiXmlDocument *doc)
Any instances of doc.SaveFile(<FULL_PATH>) have been replaced by
FILESYSTEM_saveTiXmlDocument(<VVVVVV_FOLDER_PATH>, &doc), where
<FULL_PATH> included the full path to the saves or levels directory,
and <VVVVVV_FOLDER_PATH> only includes the path relative to the VVVVVV
directory.
When loading a document, a TiXmlDocument used to be created with a full
path in its constructor and doc.LoadFile() would then be called, now a
TiXmlDocument is constructed with no path name and
FILESYSTEM_loadTiXmlDocument(<VVVVVV_FOLDER_PATH>, &doc) is called.
There's now a thin layer of UTF-16 around the WinAPI functions to get
the path to the Documents folder and to create a new directory, so that
account usernames with non-ASCII characters do not result in no VVVVVV
folder being created or used.
The next official VVVVVV build removes 32-bit Linux (like all my other games),
and I need to get rid of the shell script on macOS at some point, so this
basically brings it up to what my other games are doing. Plus, this probably
fixes a bug where someone tries to run their executable away from the root...
Follow-up to #19 which did the change for macOS.
It appears to work as expected on Linux too.
Tested on a distro where XDG_DATA_HOME is undefined by default,
and `PHYSFS_getPrefDir` also resolves to `~/.local/share/VVVVVV/`.
The first organization parameter is unused on Linux and macOS.
Instead use PHYSFS_getPrefDir which does the same than the manual
concatenation done before. The organization name argument is required
but is not used on macOS.
