AuroraRuntime/Source/IO/FS/FS.cpp

/***
    Copyright (C) 2021 J Reece Wilson (a/k/a "Reece"). All rights reserved.

    File: FS.cpp
    Date: 2021-6-16
    Author: Reece
***/
#include <Source/RuntimeInternal.hpp>
#include "FS.hpp"
#include "Source/Locale/Locale.hpp"

namespace Aurora::IO::FS
{
    static bool IsMagicCharacter(char c)
    {
        return (c == '.') ||
               (c == '!') ||
               (c == '~') ||
               (c == '?') ||
               (c == '^');
    }

    static void ResolveAbsolutePath(bool requireMountExpand, bool requireSanitization, const AuString &path, AuString &result)
    {
        AuString buffer;
        bool isResource {};
        buffer.reserve(4096);
        result.reserve(4096);

        /**
            Resolve special character prefixes
        */
        if ((path.size() >= 2) && (requireMountExpand))
        {
            if (path[1] == kPathSplitter)
            {
                // Working directory
                if (path[0] == '.')
                {
                    SysAssert(Process::GetWorkingDirectory(buffer));
                    buffer += kPathSplitter;
                }
                // Binary directory
                else if (path[0] == '^')
                {
                    SysAssert(Process::GetProcDirectory(buffer));
                    buffer += kPathSplitter;
                }
                // Local Aurora profile
                else if (path[0] == '~')
                {
                    SysAssert(FS::GetProfileDomain(buffer));
                }
                // Global Aurora profile
                else if (path[0] == '!')
                {
                    SysAssert(FS::GetSystemDomain(buffer));
                }
                else if (path[0] == '?')
                {
                    // ...except for this one
                    isResource = true;
                }
                else
                {
                    buffer.insert(buffer.begin(), path.begin(), path.begin() + 2); // ???
                }
                
                buffer.insert(buffer.end(), path.begin() + 2, path.end());
            }
            else
            {
                buffer += path;
            }
        }
        else
        {
            buffer += path;
        }

        if (requireSanitization)
        {
            /**
                Quickly handle the edge case in some modern UNIX derived systems, and in 
                 Windows UNC paths, where paths may start with '//' or '\\' respectively
            */
            if (AuStartsWith(buffer, kDoublePathSplitter))
            {
                result += kDoublePathSplitter;
            }
            else if (buffer.size() && buffer[0] == kPathSplitter)
            {
                result += kPathSplitter;
            }

            /**
                Technically, UTF-8 strings may contain the "NUL" byte. 
            
                   o  Character numbers from U+0000 to U+007F (US-ASCII repertoire)
                      correspond to octets 00 to 7F (7 bit US-ASCII paths).  A direct
                      consequence is that a plain ASCII string is also a valid UTF-8
                      string.
                  
                You know what would suck? If we were targeting an esoteric platform, 
                say a kernel, and we we're trying to avoid string exploits by potential 
                attackers. I'm sure that would never happen to poor anticheat devs.
            */
            const auto parts = AuSplitString(buffer, AuString(1, kPathSplitter), true /*ignore empty tokens*/); /// zzzz im going to SLEEP
            for (const auto &ch : parts) // can you tell why FIO shouldn't be in hot paths yet?
            {
                if (ch == "..")
                {
                    auto i = result.size() - 1 - (result[result.size() - 1] == kPathSplitter);
                    while (i > 0 && result[i] != kPathSplitter)
                    {
                        --i;
                    }
                
                    if (i >= 0)
                    {
                        result.resize(i);
                        result += kPathSplitter;
                    }
                
                    continue;
                }
            
                if ((ch.size() == 1) && (IsMagicCharacter(ch[0])))
                {
                    continue;
                }

                result += ch;
                result += kPathSplitter;
            }

            auto i = result.size() - 1;
            if (result[i] == kPathSplitter)
            {
                result.resize(i);
            }
        }
        else // !requireSanitization
        {
            result = buffer;
        }

        if (isResource)
        {
            AuString path;
            if (!FS::GetSystemResourcePath(result, path))
            {
                result.clear();
            }
            else
            {
                result = path;
            }
        }
    }

    void /* internal, local export */ _NormalizePath(AuString &str)
    {
        bool requiresExpanding = false;
        bool requiresMountUpdate = false;
        
        requiresExpanding = str.size() && IsMagicCharacter(str[0]);

        if (str.size() == 1)
        {
            if (str[0] == '.'
            // Win32 has no concept of a rootfs
            // However, all users expect paths in userspace programs to assume CWD
            // Since we are not on a NIX operating system, we can assume these mean CWD
            #if defined(AURORA_IS_MODERNNT_DERIVED)
                ||
                str[0] == '/' ||
                str[1] == '\\'
            #endif
                )
            {
                AuProcess::GetWorkingDirectory(str);
                return;
            }

            if (str[0] == '^')
            {
                AuProcess::GetProcDirectory(str);
                return;
            }

            if (str[0] == '~')
            {
                AuFS::GetProfileDomain(str);
                return;
            }

            if (str[0] == '!')
            {
                AuFS::GetSystemDomain(str);
                return;
            }
        }
        
        // best case -> O(n) wherein we merely check each BYTE with a handful of branch conditions
        int doubleDots = 0;
        int doubleSlash = 0;
        for (auto &character : str)
        {
            if ((character == '\\') || (character == '/'))
            {
                character = kPathSplitter;
                doubleSlash++;
            }
            else
            {
                doubleSlash = 0;
            }
            
            if (character == '.')
            {
                doubleDots++;
            }
            else
            {
                doubleDots = 0;
            }
            
            requiresExpanding |= doubleDots >= 2;
            requiresExpanding |= doubleSlash >= 2;
        }
        
        // plus this, i guess
        if (str.size() >= 2)
        {
            if ((str[1] == '\\') || (str[1] == '/'))
            {
                auto c = str[0];
                if ((c == '.') || (c == '~') || (c == '!') || (c == '?' || (c == '^')))
                {
                    requiresMountUpdate = true;
                }
            }
        }
        
        // worst case -> yea have fun
        if (requiresExpanding || requiresMountUpdate)
        {
            AuString temp;
            ResolveAbsolutePath(requiresMountUpdate, requiresExpanding, str, temp);
            str = temp;
        }
    }

    AUKN_SYM bool ReadString(const AuString &path, AuString &buffer)
    {
        AuByteBuffer fileBuffer;

        if (!ReadFile(path, fileBuffer))
        {
            return false;
        }

        return Locale::Encoding::DecodeUTF8(fileBuffer.data(), fileBuffer.size(), buffer, Locale::ECodePage::eUTF8).first != 0;
    }

    AUKN_SYM bool WriteString(const AuString &path, const AuString &str)
    {
        char bom[3]
        {
            '\xEF', '\xBB', '\xBF'
        };

        auto pStream = FS::OpenWriteUnique(path);
        if (!pStream)
        {
            return false;
        }

        bool ok {};
        ok = pStream->Write(AuMemoryViewStreamRead { bom });
        ok &= pStream->Write(AuMemoryViewStreamRead { str });
        pStream->Flush();
        pStream->WriteEoS();

        return ok;
    }

    AUKN_SYM bool WriteNewString(const AuString &path, const AuString &str)
    {
        char bom[3]
        {
            '\xEF', '\xBB', '\xBF'
        };

        auto pStream = FS::OpenWriteUnique(path);
        if (!pStream)
        {
            return false;
        }

        if (pStream->GetLength())
        {
            SysPushErrorIO("File exists: {}", path);
            return {};
        }

        bool ok {};
        ok = pStream->Write(AuMemoryViewStreamRead { bom });
        ok &= pStream->Write(AuMemoryViewStreamRead { str });
        pStream->Flush();
        pStream->WriteEoS();

        return ok;
    }

    AUKN_SYM bool NormalizePath(AuString &out, const AuString &in)
    {
        try
        {
            out = NormalizePathRet(in);
            return out.size();
        }
        catch (...)
        {
            return false;
        }
    }

    static AuUInt GetLastSplitterIndex(const AuString &path)
    {
        AuUInt indexA {}, indexB {};

        auto a = path.find_last_of('\\');
        if (a != AuString::npos) indexA = a;

        auto b = path.find_last_of('/');
        if (b != AuString::npos) indexB = b;

        return AuMax(indexA, indexB);
    }

    AUKN_SYM bool GetFileFromPath(AuString &out, const AuString &path)
    {
        try
        {
            if (path.empty()) return false;
            if (path[path.size() - 1] == '.') return false;

            AuUInt max = GetLastSplitterIndex(path);
            if (max == path.size()) return false;

            out = path.substr(max + 1);
            return true;
        }
        catch (...)
        {
            return false;
        }
    }

    AUKN_SYM bool GetDirectoryFromPath(AuString &out, const AuString &path)
    {
        try
        {
            if (path.empty()) return false;
            if (path[path.size() - 1] == '.')
            {
                if (path.size() > 2 && (path[path.size() - 2] == '\\' || path[path.size() - 2] == '/'))
                {
                    out = path.substr(0, path.size() - 1);
                }
                else
                {
                    return false;
                }
            }

            AuUInt max = GetLastSplitterIndex(path);
            if (!max)
            {
                return false;
            }

            if (max == path.size())
            {
                out = path;
                return true;
            }

            out = path.substr(0, max + 1);
            return true;
        }
        catch (...)
        {
            out.clear();
            return false;
        }
    }

    AUKN_SYM bool GoUpToSeparator(AuString &out, const AuString &path)
    {
        try
        {
            if (path.empty()) return false;
            if (path[path.size() - 1] == '.')
            {
                if (path.size() > 2 && (path[path.size() - 2] == '\\' || path[path.size() - 2] == '/'))
                {
                    out = path.substr(0, path.size() - 2);
                }
                else
                {
                    return false;
                }
            }

            AuUInt max = GetLastSplitterIndex(path);
            if (!max)
            {
                return false;
            }

            if (max == path.size())
            {
                out = path.substr(0, max - 1);
                return true;
            }

            out = path.substr(0, max);
            return true;
        }
        catch (...)
        {
            out.clear();
            return false;
        }
    }
}