vn/code/core/core_string.cpp

#define STB_SPRINTF_IMPLEMENTATION
#include "third_party/stb_sprintf.h"

//~ sixten: Char funcitons

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

inline b32 IsDigit(char C)
{
    b32 Result = ((C >= '0') && (C <= '9'));
    return(Result);
}

inline b32 IsLetter(char C)
{
    b32 Result = ((C >= 'A') && (C <= 'Z')) || ((C >= 'a') && (C <= 'z'));
    return(Result);
}


//~ sixten: String functions

//- sixten: Basic constructors

inline string MakeString(u8 *Data, s64 Count)
{
    string Result = {Count, Data};
    return(Result);
}

inline string MakeString(char *CString)
{
    string Result = {StringLength(CString), (u8 *)CString};
    return(Result);
}

//- sixten: Equality

static 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);
}

//- sixten: Substring

static string Substring(string String, range1_s64 Range)
{
    string Result = MakeString(String.Data + Range.Min, DimOfRange(Range));
    return(Result);
}

static string Prefix(string String, s64 Count)
{
    range1_s64 Range = Range1S64(0, Count);
    string Result = Substring(String, Range);
    return(Result);
}

static string Suffix(string String, s64 Count)
{
    range1_s64 Range = Range1S64(String.Count - Count, String.Count);
    string Result = Substring(String, Range);
    return(Result);
}

//- sixten: Hashing

static 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);
}

//- sixten: Searching

static 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);
}

static 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);
}

static s64 FirstIndexOf(string String, string Sub)
{
    s64 Result = -1;
    if(String.Count >= Sub.Count)
    {
        for(s64 Index = 0;
            Index < String.Count - Sub.Count;
            ++Index)
        {
            string ToCheck = Substring(String, Range1S64(Index, Index + Sub.Count));
            if(AreEqual(ToCheck, Sub))
            {
                Result = Index;
                break;
            }
        }
    }
    
    return(Result);
}

static s64 LastIndexOf(string String, string Sub)
{
    s64 Result = -1;
    if(String.Count >= Sub.Count)
    {
        for(s64 Index = String.Count - Sub.Count - 1;
            Index >= 0;
            --Index)
        {
            string ToCheck = Substring(String, Range1S64(Index, Index + Sub.Count));
            if(AreEqual(ToCheck, Sub))
            {
                Result = Index;
                break;
            }
        }
    }
    
    return(Result);
}

//- sixten: Allocation

static string PushString(memory_arena *Arena, string String)
{
    string Result;
    Result.Data = PushArray(Arena, u8, String.Count, NoClear());
    Result.Count = String.Count;
    Copy(Result.Data, String.Data, String.Count);
    
    return(Result);
}

static string PushFormatVariadic(memory_arena *Arena, char *Format, va_list Arguments)
{
    va_list ArgumentsCopy;
    va_copy(ArgumentsCopy, Arguments);
    
    string Result;
    Result.Count = stbsp_vsnprintf(0, 0, Format, ArgumentsCopy);
    Result.Data = PushArray(Arena, u8, Result.Count + 1, NoClear());
    Result.Data[Result.Count] = 0;
    
    stbsp_vsnprintf((char *)Result.Data, (s32)Result.Count + 1, Format, Arguments);
    
    return(Result);
}

static string PushFormat(memory_arena *Arena, char *Format, ...)
{
    va_list Arguments;
    va_start(Arguments, Format);
    string Result = PushFormatVariadic(Arena, Format, Arguments);
    va_end(Arguments);
    
    return(Result);
}

static string PushCString(memory_arena *Arena, char *CString)
{
    string String = MakeString(CString);
    string Result = PushString(Arena, String);
    return(Result);
}

//- sixten: Conversion

static s64 ConvertStringToS64(string String)
{
    s64 Result = 0;
    b32 IsNegative = false;
    
    s64 Index = 0;
    if(String.Data[Index] == '-')
    {
        IsNegative = true;
        ++Index;
    }
    
    for(;Index < String.Count; ++Index)
    {
        u8 Char = String.Data[Index];
        Assert(IsDigit(Char));
        Result = Result*10 + (Char-'0');
    }
    
    if(IsNegative)
    {
        Result = -Result;;
    }
    
    return(Result);
}

static string ConvertS64ToString(memory_arena *Arena, s64 Value)
{
    b32 IsNegative = (Value < 0);
    if(IsNegative)
    {
        Value = -Value;
    }
    
    s64 DigitCount = (s64)Floor(Log(Max(Value, 1)) / Log(10)) + 1;
    
    s64 TotalBufferCount = DigitCount + IsNegative;
    
    string String = {TotalBufferCount, PushArray(Arena, u8, TotalBufferCount + 1)};
    String.Data[TotalBufferCount] = 0;
    
    if(IsNegative)
    {
        String.Data[0] = '-';
    }
    
    for(s64 Index = 0;
        Index < DigitCount;
        ++Index)
    {
        String.Data[TotalBufferCount - 1 - Index] = '0' + (Value % 10);
        Value /= 10;
    }
    
    return(String);
}

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

//- sixten: "C Style" strings

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


//~ sixten: String list

static void AppendString(string_list *List, string String, memory_arena *Arena)
{
    string_node *Node = PushStruct(Arena, string_node);
    Node->String = String;
    
    List->TotalCount += String.Count;
    
    DLLInsertLast(List->First, List->Last, Node);
}

static string JoinStringList(string_list *List, memory_arena *Arena)
{
    u8 *Buffer = PushArray(Arena, u8, List->TotalCount + 1);
    Buffer[List->TotalCount] = 0;
    
    s64 GlobalIndex = 0;
    
    for(string_node *Node = List->First;
        Node != 0;
        Node = Node->Next)
    {
        string String = Node->String;
        for(s64 Index = 0;
            Index < String.Count;
            ++Index)
        {
            Buffer[GlobalIndex++] = String.Data[Index];
        }
    }
    
    string Result = MakeString(Buffer, List->TotalCount);
    return(Result);
}


//~ sixten: Unicode

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

static 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;
    }
}

static b32 IsValid(utf8_iterator *Iter)
{
    b32 Result = (Iter->Codepoint != 0);
    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);
}