Make sure the VPK header is always initialized to 0 instead of actual values, as otherwise it would cause the reader to succeed on truncated files if at least the first byte matches that of the VPK magic. This also lead to the simplification of the tree header writer; instead of writing the tree size apart from the header, write the whole thing at once.
* Fix multiple issues related to threading, such as not locking the mutex when using thread-unsafe members or methods.
* Netconsole now logs its output to a file on the disk.
* Properly handle close events, to allow netconsole to shutdown properly, so stuff like log buffers could flush properly. Input is now ran in a separate thread so main thread could check if we need to shutdown, and do so if needed. This was needed as previously input was ran in the main thread and the issue is that its blocking until an input is given.
SpdLog_Create() is no longer externed through its header as its a static internal function. Moved SpdLog_Create() above SpdLog_Init() to fix the compile error and rename it to SpdLog_CreateRotatingLoggers(), which is a more suitable name.
* Make sure the workspace path actually exists before attempting to pack it.
* Make sure the VPK directory tree file was parsed correctly before unpacking store.
* Log debug output for each pack operation to a file.
* Fix bug in s_DirFileRegex regex pattern, which would include a trailing '_' in the context part of the directory tree file name.
* Fix bug in 'GetLevelName()' and 'GetDirNameParts()' causing it to parse the path as well, prune the path before running the regex.
* Renamed 'GetLevelName()' to 'PackedStore_GetDirLevelName()'.
* Renamed 'GetDirNameParts()' to 'PackedStore_GetDirNameParts()'.
* Write a front-end enable file when building client VPK's.
Check if the parsing was actually successful, else free and return nullptr. The null check at call site never worked since we never return null on failure.
We do this as if we happen to not have any manifest whatsoever, but already opened the block file, and the block file already existed and was still valid, it would be emptied out. Only open it if we are actually going to write into it!
Make the utility methods statics within the implementation so it could be used by other structs/classes of the VPK builder as well, these utility functions weren't useful outside the class. This also allowed for the removal of the singleton which had no purpose whatsoever.
Should read element count of bytes instead of element count of file size, else this always returns 1 instead of the actual number of bytes read/written.
CCommandLine isn't implemented in tools, thus unresolved external symbol errors will be thrown. HushAsserts() was unused and therefore removed which fixes these errors.
Standalone VPK tool that does everything the VPK tool implemented in the dedicated server does (pack, unpack), but independent from the game's runtime.
Standalone FileSystem implementation, allowing SDK code to be used in tools without adjusting them as the vtables remain identical. NOTE that this reimplementation is partial and only contains the reading and writing of files to/from buffers or disk as that's what we are mainly interested at. We can always implement the other methods when needed.
NOTE: the filesystem singleton accessor (CFileSystem_Stdio* FileSystem()) has been moved to an implementation file instead of being inline as this would allow existing code to link to the tools implementation instead.
The KeyValues class belongs here. Also reimplemented most loading methods for KeyValues, and adjusted the VPK building code to account for it. Pointers to the engine's implementation of KeyValues have been moved to a separate header ('keyvalues_iface.h'), as this allows external tools code to utilize the standalone KeyValues class implementation. Playlist utilities are completely separated from the KeyValues header; these have nothing to do with KeyValues other than manipulating a global KeyValues object for the playlists, and thus have been named as such and moved to rtech/playlists.
Class is slightly modified; it uses 2 CBlob pointers instead of a single CBlob member. CUtlMemoryPool::m_pHeadOfFreeList is also replaced with the new CUtlMemoryPool::m_pPrev pointer. Code has been modified to accommodate this change and aligns with the assembly code of r5.
This is a more reliable solution as pretty much all code patch in CMDLCache checks the ptr for NULL, but none of them do for 0xDEADFEEDDEADFEED (probably a debug only thing that somehow made it into this build). Patched out so the engine could deal with it properly. Anything interesting already gets logged from the SDK so we won't miss out on anything patching this out.
