vn/code/vn_string.h

/* date = May 7th 2023 9:01 pm */

#ifndef VN_STRING_H
#define VN_STRING_H

inline b32 IsWhitespace(char C)
{
    b32 Result = ((C == ' ') ||
                  (C == '\n') ||
                  (C == '\t') ||
                  (C == '\r'));
    return(Result);
}

inline s64 StringLength(char *String)
{
    s64 Result = 0;
    while(*String++)
    {
        ++Result;
    }
    return(Result);
}

inline u64 HashString(string String)
{
    u64 Result = 5731;
    for(s64 Index = 0;
        Index < String.Count;
        ++Index)
    {
        Result += String.Data[Index];
        Result ^= Result << 13;
        Result ^= Result >> 7;
        Result ^= Result << 17;
    }
    
    return(Result);
}

inline string MakeStringFromCString(char *Data)
{
    string Result = {StringLength(Data), (u8 *)Data};
    return(Result);
}

inline s64 FirstIndexOf(string String, char Char)
{
    s64 Result = -1;
    for(s64 Index = 0;
        Index < String.Count;
        ++Index)
    {
        if(String.Data[Index] == Char)
        {
            Result = Index;
            break;
        }
    }
    return(Result);
}

inline s64 LastIndexOf(string String, char Char)
{
    s64 Result = -1;
    for(s64 Index = String.Count-1;
        Index >= 0;
        --Index)
    {
        if(String.Data[Index] == Char)
        {
            Result = Index;
            break;
        }
    }
    return(Result);
}

inline b32 AreEqual(string A, string B)
{
    b32 Result = false;
    if(A.Count == B.Count)
    {
        Result = true;
        
        for(s64 Index = 0;
            Index < A.Count;
            ++Index)
        {
            if(A.Data[Index] != B.Data[Index])
            {
                Result = false;
                break;
            }
        }
    }
    
    return(Result);
}

static s64 UTF8FromCodepoint(u8 *Out, u32 Codepoint)
{
    s64 Length = 0;
    if(Codepoint <= 0x7F)
    {
        Out[0] = (u8)Codepoint;
        Length = 1;
    }
    else if(Codepoint <= 0x7FF)
    {
        Out[0] = (0x3 << 6) | ((Codepoint >> 6) & 0x1F);
        Out[1] = 0x80       | ( Codepoint       & 0x3F);
        Length = 2;
    }
    else if(Codepoint <= 0xFFFF)
    {
        Out[0] = (0x7 << 5) | ((Codepoint >> 12) & 0x0F);
        Out[1] = 0x80       | ((Codepoint >> 6) & 0x3F);
        Out[2] = 0x80       | ( Codepoint       & 0x3F);
        Length = 3;
    }
    else if(Codepoint <= 0x10FFFF)
    {
        Out[0] = (0xF << 4) | ((Codepoint >> 12) & 0x07);
        Out[1] = 0x80       | ((Codepoint >> 12) & 0x3F);
        Out[2] = 0x80       | ((Codepoint >> 6)  & 0x3F);
        Out[3] = 0x80       | ( Codepoint        & 0x3F);
        Length = 4;
    }
    else
    {
        Out[0] = '?';
        Length = 1;
    }
    
    return(Length);
}

inline s64 GetCodepointSize(u32 Codepoint)
{
    s64 Result = 0;
    if(Codepoint <= 0x7F)
    {
        Result = 1;
    }
    else if(Codepoint <= 0x7FF)
    {
        Result = 2;
    }
    else if(Codepoint <= 0xFFFF)
    {
        Result = 3;
    }
    else if(Codepoint <= 0x10FFFF)
    {
        Result = 4;
    }
    else
    {
        Result = 1;
    }
    return(Result);
}

// sixten(TODO): Remove this forward decl.
inline string PushCString(struct memory_arena *Arena, char *CString);

inline string StringFromCodepoint(struct memory_arena *Arena, u32 Codepoint)
{
    char Buffer[5] = {};
    UTF8FromCodepoint((u8 *)Buffer, Codepoint);
    
    string Result = PushCString(Arena, Buffer);
    return(Result);
}

struct utf8_iterator
{
    string Data;
    s64 Index;
    
    u32 Codepoint;
};

inline void Advance(utf8_iterator *Iter)
{
    u8 *At = Iter->Data.Data + Iter->Index;
    
    if(Iter->Index < Iter->Data.Count)
    {
        if((At[0] & 0x80) == 0x00)
        {
            Iter->Codepoint = (At[0] & 0x7F);
            Iter->Index += 1;
        }
        else if((At[0] & 0xE0) == 0xC0)
        {
            Iter->Codepoint = ((At[0] & 0x1F) << 6)|(At[1] & 0x3F);
            Iter->Index += 2;
        }
        else if((At[0] & 0xF0) == 0xE0)
        {
            Iter->Codepoint = ((At[0] & 0x0F) << 12)|((At[1] & 0x3F) << 6)|(At[2] & 0x3F);
            Iter->Index += 3;
        }
        else if((Iter->Data.Data[Iter->Index] & 0xF8) == 0xF0)
        {
            Iter->Codepoint = ((At[0] & 0x0F) << 18)|((At[1] & 0x3F) << 12)|((At[2] & 0x3F) << 6)|(At[3] & 0x3F);
            Iter->Index += 4;
        }
    }
    else
    {
        Iter->Codepoint = 0;
    }
}

inline utf8_iterator IterateUTF8String(string String)
{
    utf8_iterator Iter = {};
    Iter.Data = String;
    Advance(&Iter);
    
    return(Iter);
}

#endif //VN_STRING_H