mirror of
https://github.com/odin-lang/Odin.git
synced 2026-01-04 12:07:45 +00:00
758 lines
19 KiB
Odin
758 lines
19 KiB
Odin
#shared_global_scope;
|
|
|
|
#import "os.odin";
|
|
#import "fmt.odin";
|
|
#import "utf8.odin";
|
|
#import "raw.odin";
|
|
|
|
// Naming Conventions:
|
|
// In general, PascalCase for types and snake_case for values
|
|
//
|
|
// Import Name: snake_case (but prefer single word)
|
|
// Types: PascalCase
|
|
// Union Variants: PascalCase
|
|
// Enum Values: PascalCase
|
|
// Procedures: snake_case
|
|
// Local Variables: snake_case
|
|
// Constant Variables: SCREAMING_SNAKE_CASE
|
|
|
|
// IMPORTANT NOTE(bill): `type_info` & `type_info_val` cannot be used within a
|
|
// #shared_global_scope due to the internals of the compiler.
|
|
// This could change at a later date if the all these data structures are
|
|
// implemented within the compiler rather than in this "preload" file
|
|
|
|
|
|
// IMPORTANT NOTE(bill): Do not change the order of any of this data
|
|
// The compiler relies upon this _exact_ order
|
|
TypeInfoEnumValue :: raw_union {
|
|
f: f64,
|
|
i: i128,
|
|
}
|
|
// NOTE(bill): This must match the compiler's
|
|
CallingConvention :: enum {
|
|
Odin = 0,
|
|
C = 1,
|
|
Std = 2,
|
|
Fast = 3,
|
|
}
|
|
|
|
TypeInfoRecord :: struct #ordered {
|
|
types: []^TypeInfo,
|
|
names: []string,
|
|
offsets: []int, // offsets may not be used in tuples
|
|
usings: []bool, // usings may not be used in tuples
|
|
packed: bool,
|
|
ordered: bool,
|
|
custom_align: bool,
|
|
}
|
|
|
|
TypeInfo :: union {
|
|
size: int,
|
|
align: int,
|
|
|
|
Named{name: string, base: ^TypeInfo},
|
|
Integer{signed: bool},
|
|
Rune{},
|
|
Float{},
|
|
Complex{},
|
|
String{},
|
|
Boolean{},
|
|
Any{},
|
|
Pointer{
|
|
elem: ^TypeInfo, // nil -> rawptr
|
|
},
|
|
Atomic{elem: ^TypeInfo},
|
|
Procedure{
|
|
params: ^TypeInfo, // TypeInfo.Tuple
|
|
results: ^TypeInfo, // TypeInfo.Tuple
|
|
variadic: bool,
|
|
convention: CallingConvention,
|
|
},
|
|
Array{
|
|
elem: ^TypeInfo,
|
|
elem_size: int,
|
|
count: int,
|
|
},
|
|
DynamicArray{elem: ^TypeInfo, elem_size: int},
|
|
Slice {elem: ^TypeInfo, elem_size: int},
|
|
Vector {elem: ^TypeInfo, elem_size, count: int},
|
|
Tuple {using record: TypeInfoRecord}, // Only really used for procedures
|
|
Struct {using record: TypeInfoRecord},
|
|
RawUnion {using record: TypeInfoRecord},
|
|
Union{
|
|
common_fields: struct {
|
|
types: []^TypeInfo,
|
|
names: []string,
|
|
offsets: []int, // offsets may not be used in tuples
|
|
},
|
|
variant_names: []string,
|
|
variant_types: []^TypeInfo,
|
|
},
|
|
Enum{
|
|
base: ^TypeInfo,
|
|
names: []string,
|
|
values: []TypeInfoEnumValue,
|
|
},
|
|
Map{
|
|
key: ^TypeInfo,
|
|
value: ^TypeInfo,
|
|
generated_struct: ^TypeInfo,
|
|
count: int, // == 0 if dynamic
|
|
},
|
|
BitField{
|
|
names: []string,
|
|
bits: []i32,
|
|
offsets: []i32,
|
|
},
|
|
}
|
|
|
|
|
|
// NOTE(bill): only the ones that are needed (not all types)
|
|
// This will be set by the compiler
|
|
__type_table: []TypeInfo;
|
|
|
|
__argv__: ^^u8;
|
|
__argc__: i32;
|
|
|
|
type_info_base :: proc(info: ^TypeInfo) -> ^TypeInfo {
|
|
if info == nil {
|
|
return nil;
|
|
}
|
|
base := info;
|
|
match i in base {
|
|
case TypeInfo.Named:
|
|
base = i.base;
|
|
}
|
|
return base;
|
|
}
|
|
|
|
|
|
type_info_base_without_enum :: proc(info: ^TypeInfo) -> ^TypeInfo {
|
|
if info == nil {
|
|
return nil;
|
|
}
|
|
base := info;
|
|
match i in base {
|
|
case TypeInfo.Named:
|
|
base = i.base;
|
|
case TypeInfo.Enum:
|
|
base = i.base;
|
|
}
|
|
return base;
|
|
}
|
|
|
|
|
|
|
|
assume :: proc(cond: bool) #foreign __llvm_core "llvm.assume";
|
|
|
|
__debug_trap :: proc() #foreign __llvm_core "llvm.debugtrap";
|
|
__trap :: proc() #foreign __llvm_core "llvm.trap";
|
|
read_cycle_counter :: proc() -> u64 #foreign __llvm_core "llvm.readcyclecounter";
|
|
|
|
|
|
// IMPORTANT NOTE(bill): Must be in this order (as the compiler relies upon it)
|
|
AllocatorMode :: enum u8 {
|
|
Alloc,
|
|
Free,
|
|
FreeAll,
|
|
Resize,
|
|
}
|
|
AllocatorProc :: #type proc(allocator_data: rawptr, mode: AllocatorMode,
|
|
size, alignment: int,
|
|
old_memory: rawptr, old_size: int, flags: u64) -> rawptr;
|
|
Allocator :: struct #ordered {
|
|
procedure: AllocatorProc,
|
|
data: rawptr,
|
|
}
|
|
|
|
|
|
Context :: struct #ordered {
|
|
thread_id: int,
|
|
|
|
allocator: Allocator,
|
|
|
|
user_data: rawptr,
|
|
user_index: int,
|
|
}
|
|
|
|
#thread_local __context: Context;
|
|
|
|
|
|
DEFAULT_ALIGNMENT :: align_of([vector 4]f32);
|
|
|
|
|
|
__check_context :: proc() {
|
|
c := &__context;
|
|
|
|
if c.allocator.procedure == nil {
|
|
c.allocator = default_allocator();
|
|
}
|
|
if c.thread_id == 0 {
|
|
c.thread_id = os.current_thread_id();
|
|
}
|
|
}
|
|
|
|
alloc :: proc(size: int) -> rawptr #inline { return alloc_align(size, DEFAULT_ALIGNMENT); }
|
|
|
|
alloc_align :: proc(size, alignment: int) -> rawptr #inline {
|
|
__check_context();
|
|
a := context.allocator;
|
|
return a.procedure(a.data, AllocatorMode.Alloc, size, alignment, nil, 0, 0);
|
|
}
|
|
|
|
free_ptr_with_allocator :: proc(a: Allocator, ptr: rawptr) #inline {
|
|
if ptr == nil {
|
|
return;
|
|
}
|
|
if a.procedure == nil {
|
|
return;
|
|
}
|
|
a.procedure(a.data, AllocatorMode.Free, 0, 0, ptr, 0, 0);
|
|
}
|
|
|
|
free_ptr :: proc(ptr: rawptr) #inline {
|
|
__check_context();
|
|
free_ptr_with_allocator(context.allocator, ptr);
|
|
}
|
|
|
|
free_all :: proc() #inline {
|
|
__check_context();
|
|
a := context.allocator;
|
|
a.procedure(a.data, AllocatorMode.FreeAll, 0, 0, nil, 0, 0);
|
|
}
|
|
|
|
|
|
resize :: proc(ptr: rawptr, old_size, new_size: int) -> rawptr #inline { return resize_align(ptr, old_size, new_size, DEFAULT_ALIGNMENT); }
|
|
resize_align :: proc(ptr: rawptr, old_size, new_size, alignment: int) -> rawptr #inline {
|
|
__check_context();
|
|
a := context.allocator;
|
|
return a.procedure(a.data, AllocatorMode.Resize, new_size, alignment, ptr, old_size, 0);
|
|
}
|
|
|
|
|
|
|
|
default_resize_align :: proc(old_memory: rawptr, old_size, new_size, alignment: int) -> rawptr {
|
|
if old_memory == nil {
|
|
return alloc_align(new_size, alignment);
|
|
}
|
|
|
|
if new_size == 0 {
|
|
free(old_memory);
|
|
return nil;
|
|
}
|
|
|
|
if new_size == old_size {
|
|
return old_memory;
|
|
}
|
|
|
|
new_memory := alloc_align(new_size, alignment);
|
|
if new_memory == nil {
|
|
return nil;
|
|
}
|
|
|
|
__mem_copy(new_memory, old_memory, min(old_size, new_size));;
|
|
free(old_memory);
|
|
return new_memory;
|
|
}
|
|
|
|
|
|
default_allocator_proc :: proc(allocator_data: rawptr, mode: AllocatorMode,
|
|
size, alignment: int,
|
|
old_memory: rawptr, old_size: int, flags: u64) -> rawptr {
|
|
using AllocatorMode;
|
|
|
|
match mode {
|
|
case Alloc:
|
|
return os.heap_alloc(size);
|
|
|
|
case Free:
|
|
os.heap_free(old_memory);
|
|
return nil;
|
|
|
|
case FreeAll:
|
|
// NOTE(bill): Does nothing
|
|
|
|
case Resize:
|
|
ptr := os.heap_resize(old_memory, size);
|
|
assert(ptr != nil);
|
|
return ptr;
|
|
}
|
|
|
|
return nil;
|
|
}
|
|
|
|
default_allocator :: proc() -> Allocator {
|
|
return Allocator{
|
|
procedure = default_allocator_proc,
|
|
data = nil,
|
|
};
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
__string_eq :: proc(a, b: string) -> bool {
|
|
if len(a) != len(b) {
|
|
return false;
|
|
}
|
|
if len(a) == 0 {
|
|
return true;
|
|
}
|
|
if &a[0] == &b[0] {
|
|
return true;
|
|
}
|
|
return __string_cmp(a, b) == 0;
|
|
}
|
|
|
|
__string_cmp :: proc(a, b: string) -> int {
|
|
return __mem_compare(&a[0], &b[0], min(len(a), len(b)));
|
|
}
|
|
|
|
__string_ne :: proc(a, b: string) -> bool #inline { return !__string_eq(a, b); }
|
|
__string_lt :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) < 0; }
|
|
__string_gt :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) > 0; }
|
|
__string_le :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) <= 0; }
|
|
__string_ge :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) >= 0; }
|
|
|
|
|
|
__complex64_eq :: proc(a, b: complex64) -> bool #inline { return real(a) == real(b) && imag(a) == imag(b); }
|
|
__complex64_ne :: proc(a, b: complex64) -> bool #inline { return real(a) != real(b) || imag(a) != imag(b); }
|
|
|
|
__complex128_eq :: proc(a, b: complex128) -> bool #inline { return real(a) == real(b) && imag(a) == imag(b); }
|
|
__complex128_ne :: proc(a, b: complex128) -> bool #inline { return real(a) != real(b) || imag(a) != imag(b); }
|
|
|
|
__assert :: proc(file: string, line, column: int, msg: string) #inline {
|
|
fmt.fprintf(os.stderr, "%s(%d:%d) Runtime assertion: %s\n",
|
|
file, line, column, msg);
|
|
__debug_trap();
|
|
}
|
|
__panic :: proc(file: string, line, column: int, msg: string) #inline {
|
|
fmt.fprintf(os.stderr, "%s(%d:%d) Panic: %s\n",
|
|
file, line, column, msg);
|
|
__debug_trap();
|
|
}
|
|
__bounds_check_error :: proc(file: string, line, column: int, index, count: int) {
|
|
if 0 <= index && index < count {
|
|
return;
|
|
}
|
|
fmt.fprintf(os.stderr, "%s(%d:%d) Index %d is out of bounds range 0..<%d\n",
|
|
file, line, column, index, count);
|
|
__debug_trap();
|
|
}
|
|
|
|
__slice_expr_error :: proc(file: string, line, column: int, low, high, max: int) {
|
|
if 0 <= low && low <= high && high <= max {
|
|
return;
|
|
}
|
|
fmt.fprintf(os.stderr, "%s(%d:%d) Invalid slice indices: [%d..<%d..<%d]\n",
|
|
file, line, column, low, high, max);
|
|
__debug_trap();
|
|
}
|
|
|
|
__substring_expr_error :: proc(file: string, line, column: int, low, high: int) {
|
|
if 0 <= low && low <= high {
|
|
return;
|
|
}
|
|
fmt.fprintf(os.stderr, "%s(%d:%d) Invalid substring indices: [%d..<%d]\n",
|
|
file, line, column, low, high);
|
|
__debug_trap();
|
|
}
|
|
__type_assertion_check :: proc(ok: bool, file: string, line, column: int, from, to: ^TypeInfo) {
|
|
if !ok {
|
|
fmt.fprintf(os.stderr, "%s(%d:%d) Invalid type_assertion from %T to %T\n",
|
|
file, line, column, from, to);
|
|
__debug_trap();
|
|
}
|
|
}
|
|
|
|
__string_decode_rune :: proc(s: string) -> (rune, int) #inline {
|
|
return utf8.decode_rune(s);
|
|
}
|
|
|
|
|
|
__mem_set :: proc(data: rawptr, value: i32, len: int) -> rawptr {
|
|
llvm_memset_64bit :: proc(dst: rawptr, val: u8, len: int, align: i32, is_volatile: bool) #foreign __llvm_core "llvm.memset.p0i8.i64";
|
|
llvm_memset_64bit(data, u8(value), len, 1, false);
|
|
return data;
|
|
}
|
|
__mem_zero :: proc(data: rawptr, len: int) -> rawptr {
|
|
return __mem_set(data, 0, len);
|
|
}
|
|
__mem_copy :: proc(dst, src: rawptr, len: int) -> rawptr {
|
|
// NOTE(bill): This _must_ be implemented like C's memmove
|
|
llvm_memmove_64bit :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) #foreign __llvm_core "llvm.memmove.p0i8.p0i8.i64";
|
|
llvm_memmove_64bit(dst, src, len, 1, false);
|
|
return dst;
|
|
}
|
|
__mem_copy_non_overlapping :: proc(dst, src: rawptr, len: int) -> rawptr {
|
|
// NOTE(bill): This _must_ be implemented like C's memcpy
|
|
llvm_memcpy_64bit :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) #foreign __llvm_core "llvm.memcpy.p0i8.p0i8.i64";
|
|
llvm_memcpy_64bit(dst, src, len, 1, false);
|
|
return dst;
|
|
}
|
|
|
|
__mem_compare :: proc(a, b: ^u8, n: int) -> int {
|
|
for i in 0..<n {
|
|
match {
|
|
case (a+i)^ < (b+i)^:
|
|
return -1;
|
|
case (a+i)^ > (b+i)^:
|
|
return +1;
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
__sqrt_f32 :: proc(x: f32) -> f32 #foreign __llvm_core "llvm.sqrt.f32";
|
|
__sqrt_f64 :: proc(x: f64) -> f64 #foreign __llvm_core "llvm.sqrt.f64";
|
|
__abs_complex64 :: proc(x: complex64) -> f32 #inline {
|
|
r, i := real(x), imag(x);
|
|
return __sqrt_f32(r*r + i*i);
|
|
}
|
|
__abs_complex128 :: proc(x: complex128) -> f64 #inline {
|
|
r, i := real(x), imag(x);
|
|
return __sqrt_f64(r*r + i*i);
|
|
}
|
|
|
|
|
|
|
|
|
|
__dynamic_array_make :: proc(array_: rawptr, elem_size, elem_align: int, len, cap: int) {
|
|
array := ^raw.DynamicArray(array_);
|
|
__check_context();
|
|
array.allocator = context.allocator;
|
|
assert(array.allocator.procedure != nil);
|
|
|
|
if cap > 0 {
|
|
__dynamic_array_reserve(array_, elem_size, elem_align, cap);
|
|
array.len = len;
|
|
}
|
|
}
|
|
|
|
__dynamic_array_reserve :: proc(array_: rawptr, elem_size, elem_align: int, cap: int) -> bool {
|
|
array := ^raw.DynamicArray(array_);
|
|
|
|
if cap <= array.cap {
|
|
return true;
|
|
}
|
|
|
|
__check_context();
|
|
if array.allocator.procedure == nil {
|
|
array.allocator = context.allocator;
|
|
}
|
|
assert(array.allocator.procedure != nil);
|
|
|
|
old_size := array.cap * elem_size;
|
|
new_size := cap * elem_size;
|
|
allocator := array.allocator;
|
|
|
|
new_data := allocator.procedure(allocator.data, AllocatorMode.Resize, new_size, elem_align, array.data, old_size, 0);
|
|
if new_data == nil {
|
|
return false;
|
|
}
|
|
|
|
array.data = new_data;
|
|
array.cap = cap;
|
|
return true;
|
|
}
|
|
|
|
__dynamic_array_resize :: proc(array_: rawptr, elem_size, elem_align: int, len: int) -> bool {
|
|
array := ^raw.DynamicArray(array_);
|
|
|
|
ok := __dynamic_array_reserve(array_, elem_size, elem_align, len);
|
|
if ok {
|
|
array.len = len;
|
|
}
|
|
return ok;
|
|
}
|
|
|
|
|
|
__dynamic_array_append :: proc(array_: rawptr, elem_size, elem_align: int,
|
|
items: rawptr, item_count: int) -> int {
|
|
array := ^raw.DynamicArray(array_);
|
|
|
|
if item_count <= 0 || items == nil {
|
|
return array.len;
|
|
}
|
|
|
|
|
|
ok := true;
|
|
if array.cap <= array.len+item_count {
|
|
cap := 2 * array.cap + max(8, item_count);
|
|
ok = __dynamic_array_reserve(array, elem_size, elem_align, cap);
|
|
}
|
|
if !ok {
|
|
// TODO(bill): Better error handling for failed reservation
|
|
return array.len;
|
|
}
|
|
data := ^u8(array.data);
|
|
assert(data != nil);
|
|
__mem_copy(data + (elem_size*array.len), items, elem_size * item_count);
|
|
array.len += item_count;
|
|
return array.len;
|
|
}
|
|
|
|
__dynamic_array_append_nothing :: proc(array_: rawptr, elem_size, elem_align: int) -> int {
|
|
array := ^raw.DynamicArray(array_);
|
|
|
|
ok := true;
|
|
if array.cap <= array.len+1 {
|
|
cap := 2 * array.cap + max(8, 1);
|
|
ok = __dynamic_array_reserve(array, elem_size, elem_align, cap);
|
|
}
|
|
if !ok {
|
|
// TODO(bill): Better error handling for failed reservation
|
|
return array.len;
|
|
}
|
|
data := ^u8(array.data);
|
|
assert(data != nil);
|
|
__mem_zero(data + (elem_size*array.len), elem_size);
|
|
array.len++;
|
|
return array.len;
|
|
}
|
|
|
|
__slice_append :: proc(slice_: rawptr, elem_size, elem_align: int,
|
|
items: rawptr, item_count: int) -> int {
|
|
slice := ^raw.Slice(slice_);
|
|
|
|
if item_count <= 0 || items == nil {
|
|
return slice.len;
|
|
}
|
|
|
|
item_count = min(slice.cap-slice.len, item_count);
|
|
if item_count > 0 {
|
|
data := ^u8(slice.data);
|
|
assert(data != nil);
|
|
__mem_copy(data + (elem_size*slice.len), items, elem_size * item_count);
|
|
slice.len += item_count;
|
|
}
|
|
return slice.len;
|
|
}
|
|
|
|
|
|
// Map stuff
|
|
|
|
__default_hash :: proc(data: []u8) -> u128 {
|
|
fnv128a :: proc(data: []u8) -> u128 {
|
|
h: u128 = 0x6c62272e07bb014262b821756295c58d;
|
|
for b in data {
|
|
h = (h ~ u128(b)) * 0x1000000000000000000013b;
|
|
}
|
|
return h;
|
|
}
|
|
return fnv128a(data);
|
|
}
|
|
__default_hash_string :: proc(s: string) -> u128 {
|
|
return __default_hash([]u8(s));
|
|
}
|
|
|
|
__INITIAL_MAP_CAP :: 16;
|
|
|
|
__MapKey :: struct #ordered {
|
|
hash: u128,
|
|
str: string,
|
|
}
|
|
|
|
__MapFindResult :: struct #ordered {
|
|
hash_index: int,
|
|
entry_prev: int,
|
|
entry_index: int,
|
|
}
|
|
|
|
__MapEntryHeader :: struct #ordered {
|
|
key: __MapKey,
|
|
next: int,
|
|
/*
|
|
value: Value_Type,
|
|
*/
|
|
}
|
|
|
|
__MapHeader :: struct #ordered {
|
|
m: ^raw.DynamicMap,
|
|
is_key_string: bool,
|
|
entry_size: int,
|
|
entry_align: int,
|
|
value_offset: int,
|
|
value_size: int,
|
|
}
|
|
|
|
__dynamic_map_reserve :: proc(using header: __MapHeader, cap: int) {
|
|
__dynamic_array_reserve(&m.hashes, size_of(int), align_of(int), cap);
|
|
__dynamic_array_reserve(&m.entries, entry_size, entry_align, cap);
|
|
}
|
|
|
|
__dynamic_map_rehash :: proc(using header: __MapHeader, new_count: int) {
|
|
new_header: __MapHeader = header;
|
|
nm: raw.DynamicMap;
|
|
new_header.m = &nm;
|
|
|
|
header_hashes := ^raw.DynamicArray(&header.m.hashes);
|
|
nm_hashes := ^raw.DynamicArray(&nm.hashes);
|
|
|
|
__dynamic_array_resize(nm_hashes, size_of(int), align_of(int), new_count);
|
|
__dynamic_array_reserve(&nm.entries, entry_size, entry_align, m.entries.len);
|
|
for i in 0..<new_count {
|
|
nm.hashes[i] = -1;
|
|
}
|
|
|
|
for i := 0; i < m.entries.len; i++ {
|
|
if len(nm.hashes) == 0 {
|
|
__dynamic_map_grow(new_header);
|
|
}
|
|
|
|
entry_header := __dynamic_map_get_entry(header, i);
|
|
data := ^u8(entry_header);
|
|
|
|
fr := __dynamic_map_find(new_header, entry_header.key);
|
|
j := __dynamic_map_add_entry(new_header, entry_header.key);
|
|
if fr.entry_prev < 0 {
|
|
nm.hashes[fr.hash_index] = j;
|
|
} else {
|
|
e := __dynamic_map_get_entry(new_header, fr.entry_prev);
|
|
e.next = j;
|
|
}
|
|
|
|
e := __dynamic_map_get_entry(new_header, j);
|
|
e.next = fr.entry_index;
|
|
ndata := ^u8(e);
|
|
__mem_copy(ndata+value_offset, data+value_offset, value_size);
|
|
|
|
if __dynamic_map_full(new_header) {
|
|
__dynamic_map_grow(new_header);
|
|
}
|
|
}
|
|
free_ptr_with_allocator(header_hashes.allocator, header_hashes.data);
|
|
free_ptr_with_allocator(header.m.entries.allocator, header.m.entries.data);
|
|
header.m^ = nm;
|
|
}
|
|
|
|
__dynamic_map_get :: proc(h: __MapHeader, key: __MapKey) -> rawptr {
|
|
index := __dynamic_map_find(h, key).entry_index;
|
|
if index >= 0 {
|
|
data := ^u8(__dynamic_map_get_entry(h, index));
|
|
val := data + h.value_offset;
|
|
return val;
|
|
}
|
|
return nil;
|
|
}
|
|
|
|
__dynamic_map_set :: proc(using h: __MapHeader, key: __MapKey, value: rawptr) {
|
|
index: int;
|
|
assert(value != nil);
|
|
|
|
|
|
if len(m.hashes) == 0 {
|
|
__dynamic_map_reserve(h, __INITIAL_MAP_CAP);
|
|
__dynamic_map_grow(h);
|
|
}
|
|
|
|
fr := __dynamic_map_find(h, key);
|
|
if fr.entry_index >= 0 {
|
|
index = fr.entry_index;
|
|
} else {
|
|
index = __dynamic_map_add_entry(h, key);
|
|
if fr.entry_prev >= 0 {
|
|
entry := __dynamic_map_get_entry(h, fr.entry_prev);
|
|
entry.next = index;
|
|
} else {
|
|
m.hashes[fr.hash_index] = index;
|
|
}
|
|
}
|
|
{
|
|
e := __dynamic_map_get_entry(h, index);
|
|
e.key = key;
|
|
val := ^u8(e) + value_offset;
|
|
__mem_copy(val, value, value_size);
|
|
}
|
|
|
|
if __dynamic_map_full(h) {
|
|
__dynamic_map_grow(h);
|
|
}
|
|
}
|
|
|
|
|
|
__dynamic_map_grow :: proc(using h: __MapHeader) {
|
|
new_count := max(2*m.entries.cap + 8, __INITIAL_MAP_CAP);
|
|
__dynamic_map_rehash(h, new_count);
|
|
}
|
|
|
|
__dynamic_map_full :: proc(using h: __MapHeader) -> bool {
|
|
return int(0.75 * f64(len(m.hashes))) <= m.entries.cap;
|
|
}
|
|
|
|
|
|
__dynamic_map_hash_equal :: proc(h: __MapHeader, a, b: __MapKey) -> bool {
|
|
if a.hash == b.hash {
|
|
if h.is_key_string {
|
|
return a.str == b.str;
|
|
}
|
|
return true;
|
|
}
|
|
return false;
|
|
}
|
|
|
|
__dynamic_map_find :: proc(using h: __MapHeader, key: __MapKey) -> __MapFindResult {
|
|
fr := __MapFindResult{-1, -1, -1};
|
|
if len(m.hashes) > 0 {
|
|
fr.hash_index = int(key.hash % u128(len(m.hashes)));
|
|
fr.entry_index = m.hashes[fr.hash_index];
|
|
for fr.entry_index >= 0 {
|
|
entry := __dynamic_map_get_entry(h, fr.entry_index);
|
|
if __dynamic_map_hash_equal(h, entry.key, key) {
|
|
return fr;
|
|
}
|
|
fr.entry_prev = fr.entry_index;
|
|
fr.entry_index = entry.next;
|
|
}
|
|
}
|
|
return fr;
|
|
}
|
|
|
|
__dynamic_map_add_entry :: proc(using h: __MapHeader, key: __MapKey) -> int {
|
|
prev := m.entries.len;
|
|
c := __dynamic_array_append_nothing(&m.entries, entry_size, entry_align);
|
|
if c != prev {
|
|
end := __dynamic_map_get_entry(h, c-1);
|
|
end.key = key;
|
|
end.next = -1;
|
|
}
|
|
return prev;
|
|
}
|
|
|
|
|
|
__dynamic_map_delete :: proc(using h: __MapHeader, key: __MapKey) {
|
|
fr := __dynamic_map_find(h, key);
|
|
if fr.entry_index >= 0 {
|
|
__dynamic_map_erase(h, fr);
|
|
}
|
|
}
|
|
|
|
__dynamic_map_get_entry :: proc(using h: __MapHeader, index: int) -> ^__MapEntryHeader {
|
|
data := ^u8(m.entries.data) + index*entry_size;
|
|
return ^__MapEntryHeader(data);
|
|
}
|
|
|
|
__dynamic_map_erase :: proc(using h: __MapHeader, fr: __MapFindResult) {
|
|
if fr.entry_prev < 0 {
|
|
m.hashes[fr.hash_index] = __dynamic_map_get_entry(h, fr.entry_index).next;
|
|
} else {
|
|
__dynamic_map_get_entry(h, fr.entry_prev).next = __dynamic_map_get_entry(h, fr.entry_index).next;
|
|
}
|
|
|
|
if fr.entry_index == m.entries.len-1 {
|
|
m.entries.len--;
|
|
}
|
|
__mem_copy(__dynamic_map_get_entry(h, fr.entry_index), __dynamic_map_get_entry(h, m.entries.len-1), entry_size);
|
|
last := __dynamic_map_find(h, __dynamic_map_get_entry(h, fr.entry_index).key);
|
|
if last.entry_prev >= 0 {
|
|
__dynamic_map_get_entry(h, last.entry_prev).next = fr.entry_index;
|
|
} else {
|
|
m.hashes[last.hash_index] = fr.entry_index;
|
|
}
|
|
}
|