nob/src/string.c

#include "string.h"

Str str_empty(void)
{
    static const U8 empty_;
    static const Str str = {(U8 *)&empty_, 0};
    return str;
}

Str str_slice(Str str, U32 start, U32 len)
{
    Assert(start <= str.len);
    Assert(len <= str.len - start);

    Str new_str = str_create(&str.str[start], len);
    return new_str;
}

Str str_concat(Allocator *alloc, Str first, Str second)
{
    Str str;
    Assert(first.len <= U32_Max - second.len);
    str.len = first.len + second.len;
    str.str = allocator_push(alloc, str.len, 1);;

    MemCopy(str.str, first.str, first.len);
    MemCopy(&str.str[first.len], second.str, second.len);

    return str;
}

Str str_push(Allocator *alloc, Str str)
{
    U8 *mem = (U8 *)allocator_push(alloc, str.len, 1);
    MemCopy(mem, str.str, str.len);

    return str_create(mem, str.len);
}

static INLINE bool char_is_ascii_whitespace(U8 byte)
{
    return byte == ' ' || byte == '\t' || byte == '\v' || byte == '\f' || byte == '\r' || byte == '\n';

}

Str str_ascii_trim_left(Str str)
{
    U8 *start = str.str;
    U8 *end = str.str + str.len;
    while (start != end) {
        if (!char_is_ascii_whitespace(*start)) {
            break;
        }
        start++;
    }

    return str_create(start, (U32)(end - start));
}

Str str_ascii_trim_right(Str str)
{
    U8 *start = str.str;
    U8 *end = str.str + str.len;
    while (start != end) {
        if (!char_is_ascii_whitespace(*(end - 1))) {
            break;
        }
        end--;
    }

    return str_create(start, (U32)(end - start));
}

Str str_ascii_trim(Str str)
{
    Str new_str = str_ascii_trim_left(str);
    new_str = str_ascii_trim_right(new_str);
    return new_str;
}

S64 str_index(Str haystack, Str needle)
{
    if (needle.len == 0) {
        return 0;
    }

    if (needle.len == 1) {
        return str_find_byte(haystack, needle.str[0]);
    }
    if (needle.len > haystack.len) {
        return -1;
    }

    U8 *found = MemFind(haystack.str, haystack.len, needle.str, needle.len);
    if (found != NULL) {
        return (S64)(found - haystack.str);
    }

    return -1;
}

// split @str on any byte in @separators, place split strings in @vec
void str_split(Str str, Str separators, StrVec *vec)
{
    if (vec->max_len == 0 || str.len == 0) {
        return;
    }

    U32 start_pos = 0;
    for (U32 i = 0; i < str.len; i++) {
        U8 byte = str.str[i];
        if (str_find_byte(separators, byte) != -1) {
            if (vec->len >= vec->max_len) {
                break;
            }

            Str sub_str = str_slice(str, start_pos, i - start_pos);
            vec->strings[vec->len] = sub_str;
            vec->len++;
            start_pos = i + 1;
        }
    }
    if (start_pos <= str.len && vec->len < vec->max_len) {
        Str sub_str = str_slice(str, start_pos, str.len - start_pos);
        vec->strings[vec->len] = sub_str;
        vec->len++;
    }
}

Str str_join(Allocator *allocator, StrVec strings, Str separator)
{

    U32 joined_str_len = 0;

    for (U32 i = 0; i < strings.len; i++) {
        U32 part_len = strings.strings[i].len;
        Assert(joined_str_len <= U32_Max - part_len); // overflow
        joined_str_len += part_len;
    }
    if (strings.len > 1) {
        U32 separator_count = strings.len - 1;
        Assert(separator.len <= U32_Max / separator_count); // overflow
        Assert(joined_str_len <= U32_Max - separator.len * separator_count); // overflow
        joined_str_len += separator.len * separator_count;
    }

    U8 *joined_str = allocator_push(allocator, joined_str_len, 1);
    if (strings.len > 0) {
        U32 pos = 0;
        Str *str = &strings.strings[0];

        MemCopy(&joined_str[pos], str->str, str->len);
        pos += str->len;

        for (U32 i = 1; i < strings.len; i++) {
            str = &strings.strings[i];
            MemCopy(&joined_str[pos], separator.str, separator.len);
            pos += separator.len;
            MemCopy(&joined_str[pos], str->str, str->len);
            pos += str->len;
        }
    }

    return str_create(joined_str, joined_str_len);
}