sync with main

This commit is contained in:
phillvancejr
2022-02-04 13:15:43 -05:00
parent 8f600798ef
commit 42364f2fce
16 changed files with 2597 additions and 0 deletions

View File

@@ -0,0 +1,266 @@
package bindgen
import "core:fmt"
import "core:strconv"
// Evaluates an expression to a i64, without checking.
evaluate_i64 :: proc(data : ^ParserData) -> i64 {
ok : bool;
value : LiteralValue;
value, ok = evaluate(data);
return value.(i64);
}
// Evaluate an expression, returns whether it succeeded.
evaluate :: proc(data : ^ParserData) -> (LiteralValue, bool) {
return evaluate_level_5(data);
}
// @note Evaluate levels numbers are based on
// https://en.cppreference.com/w/c/language/operator_precedence.
// Bitwise shift level.
evaluate_level_5 :: proc(data : ^ParserData) -> (value : LiteralValue, ok : bool) {
value, ok = evaluate_level_4(data);
if !ok do return;
invalid_value : LiteralValue;
token := peek_token(data);
if token == "<<" {
v : LiteralValue;
eat_token(data);
v, ok = evaluate_level_5(data);
if is_i64(v) do value = value.(i64) << cast(u64) v.(i64);
else do invalid_value = v;
} else if token == ">>" {
v : LiteralValue;
eat_token(data);
v, ok = evaluate_level_5(data);
if is_i64(v) do value = value.(i64) >> cast(u64) v.(i64);
else do invalid_value = v;
}
if invalid_value != nil {
print_warning("Invalid operand for bitwise shift ", invalid_value);
}
return;
}
// Additive level.
evaluate_level_4 :: proc(data : ^ParserData) -> (value : LiteralValue, ok : bool) {
value, ok = evaluate_level_3(data);
if !ok do return;
token := peek_token(data);
if token == "+" {
v : LiteralValue;
eat_token(data);
v, ok = evaluate_level_4(data);
if is_i64(v) do value = value.(i64) + v.(i64);
else if is_f64(v) do value = value.(f64) + v.(f64);
}
else if token == "-" {
v : LiteralValue;
eat_token(data);
v, ok = evaluate_level_4(data);
if is_i64(v) do value = value.(i64) - v.(i64);
else if is_f64(v) do value = value.(f64) - v.(f64);
}
return;
}
// Multiplicative level.
evaluate_level_3 :: proc(data : ^ParserData) -> (value : LiteralValue, ok : bool) {
value, ok = evaluate_level_2(data);
if !ok do return;
token := peek_token(data);
if token == "*" {
v : LiteralValue;
eat_token(data);
v, ok = evaluate_level_3(data);
if is_i64(v) do value = value.(i64) * v.(i64);
else if is_f64(v) do value = value.(f64) * v.(f64);
}
else if token == "/" {
v : LiteralValue;
eat_token(data);
v, ok = evaluate_level_3(data);
if is_i64(v) do value = value.(i64) / v.(i64);
else if is_f64(v) do value = value.(f64) / v.(f64);
}
return;
}
// Prefix level.
evaluate_level_2 :: proc(data : ^ParserData) -> (value : LiteralValue, ok : bool) {
token := peek_token(data);
// Bitwise not
if token == "~" {
check_and_eat_token(data, "~");
value, ok = evaluate_level_2(data);
value = ~value.(i64);
}
else {
// @note Should call evaluate_level_1, but we don't have that because we do not dereferenciation.
value, ok = evaluate_level_0(data);
}
return;
}
// Does not try to compose with arithmetics, it just evaluates one single expression.
evaluate_level_0 :: proc(data : ^ParserData) -> (value : LiteralValue, ok : bool) {
ok = true;
value = 0;
token := peek_token(data);
// Parentheses
if token == "(" {
value, ok = evaluate_parentheses(data);
} // Number literal
else if (token[0] == '-') || (token[0] >= '0' && token[0] <= '9') {
value, ok = evaluate_number_literal(data);
} // String literal
else if token[0] == '"' {
value = evaluate_string_literal(data);
} // Function-like
else if token == "sizeof" {
value = evaluate_sizeof(data);
} // Knowned literal
else if token in data.knownedLiterals {
value = evaluate_knowned_literal(data);
} // Custom expression
else if token in data.options.customExpressionHandlers {
value = data.options.customExpressionHandlers[token](data);
}
else {
print_warning("Unknown token ", token, " for expression evaluation.");
ok = false;
}
return;
}
evaluate_sizeof :: proc(data : ^ParserData) -> LiteralValue {
print_warning("Using 'sizeof()'. Currently not able to precompute that. Please check generated code.");
check_and_eat_token(data, "sizeof");
check_and_eat_token(data, "(");
for data.bytes[data.offset] != ')' {
data.offset += 1;
}
check_and_eat_token(data, ")");
return 1;
}
evaluate_parentheses :: proc(data : ^ParserData) -> (value : LiteralValue, ok : bool) {
check_and_eat_token(data, "(");
// Cast to int (via "(int)" syntax)
token := peek_token(data);
if token == "int" {
check_and_eat_token(data, "int");
check_and_eat_token(data, ")");
value, ok = evaluate(data);
return;
} // Cast to enum value (via "(enum XXX)" syntax)
else if token == "enum" {
check_and_eat_token(data, "enum");
eat_token(data);
check_and_eat_token(data, ")");
value, ok = evaluate(data);
return;
}
value, ok = evaluate(data);
check_and_eat_token(data, ")");
return;
}
evaluate_number_literal :: proc(data : ^ParserData, loc := #caller_location) -> (value : LiteralValue, ok : bool) {
token := parse_any(data);
// Unary - before numbers
numberLitteral := token;
for token == "-" {
token = parse_any(data);
numberLitteral = tcat(numberLitteral, token);
}
token = numberLitteral;
// Check if any point or scientific notation in number
foundPointOrExp := false;
for c in token {
if c == '.' || c == 'e' || c == 'E' {
foundPointOrExp = true;
break;
}
}
isHexadecimal := len(token) >= 2 && token[:2] == "0x";
// Computing postfix
tokenLength := len(token);
l := tokenLength - 1;
for l > 0 {
c := token[l];
if c >= '0' && c <= '9' { break; }
if isHexadecimal && ((c >= 'a' && c <= 'f') || (c >= 'A' && c <= 'F')) { break; }
l -= 1;
}
postfix : string;
if l != tokenLength - 1 {
postfix = token[l+1:];
token = token[:l+1];
}
if postfix != "" && (postfix[0] == 'u' || postfix[0] == 'U') {
print_warning("Found number litteral '", token, "' with unsigned postfix, we cast it to an int64 internally.");
}
// Floating point
if !isHexadecimal && (foundPointOrExp || postfix == "f") {
value, ok = strconv.parse_f64(token);
} // Integer
else {
value, ok = strconv.parse_i64(token);
}
if !ok {
print_error(data, loc, "Expected number litteral but got '", token, "'.");
}
return value, ok;
}
evaluate_string_literal :: proc(data : ^ParserData) -> string {
token := parse_any(data);
return token;
}
evaluate_knowned_literal :: proc(data : ^ParserData) -> LiteralValue {
token := parse_any(data);
return data.knownedLiterals[token];
}
is_i64 :: proc(value : LiteralValue) -> (ok : bool) {
v : i64;
v, ok = value.(i64);
return ok;
}
is_f64 :: proc(value : LiteralValue) -> (ok : bool) {
v : f64;
v, ok = value.(f64);
return ok;
}

View File

@@ -0,0 +1,267 @@
package bindgen
import "core:os"
import "core:fmt"
import "core:strings"
import "core:strconv"
// Extract from start (included) to end (excluded) offsets
extract_string :: proc(data : ^ParserData, startOffset : u32, endOffset : u32) -> string {
return strings.string_from_ptr(&data.bytes[startOffset], cast(int) (endOffset - startOffset));
}
// Peek the end offset of the next token
peek_token_end :: proc(data : ^ParserData) -> u32 {
offset : u32;
for true {
eat_whitespaces_and_comments(data);
if data.offset >= data.bytesLength {
return data.bytesLength;
}
offset = data.offset;
// Identifier
if (data.bytes[offset] >= 'a' && data.bytes[offset] <= 'z') ||
(data.bytes[offset] >= 'A' && data.bytes[offset] <= 'Z') ||
(data.bytes[offset] == '_') {
offset += 1;
for (data.bytes[offset] >= 'a' && data.bytes[offset] <= 'z') ||
(data.bytes[offset] >= 'A' && data.bytes[offset] <= 'Z') ||
(data.bytes[offset] >= '0' && data.bytes[offset] <= '9') ||
(data.bytes[offset] == '_') {
offset += 1;
}
}
if offset != data.offset {
// Nothing to do: we found an identifier
} // Number literal
else if (data.bytes[offset] >= '0' && data.bytes[offset] <= '9') {
offset += 1;
// Hexademical literal
if data.bytes[offset - 1] == '0' && data.bytes[offset] == 'x' {
offset += 1;
for (data.bytes[offset] >= '0' && data.bytes[offset] <= '9') ||
(data.bytes[offset] >= 'a' && data.bytes[offset] <= 'f') ||
(data.bytes[offset] >= 'A' && data.bytes[offset] <= 'F') {
offset += 1;
}
} // Basic number literal
else {
for (data.bytes[offset] >= '0' && data.bytes[offset] <= '9') ||
data.bytes[offset] == '.' {
offset += 1;
}
if (data.bytes[offset] == 'e' || data.bytes[offset] == 'E') {
offset += 1;
if data.bytes[offset] == '-' {
offset += 1;
}
}
for (data.bytes[offset] >= '0' && data.bytes[offset] <= '9') {
offset += 1;
}
}
// Number suffix?
for (data.bytes[offset] == 'u' || data.bytes[offset] == 'U') ||
(data.bytes[offset] == 'l' || data.bytes[offset] == 'L') ||
(data.bytes[offset] == 'f') {
offset += 1;
}
} // String literal
else if data.bytes[offset] == '"' {
offset += 1;
for data.bytes[offset-1] == '\\' || data.bytes[offset] != '"' {
offset += 1;
}
offset += 1;
} // Possible shifts
else if data.bytes[offset] == '<' || data.bytes[offset] == '>' {
offset += 1;
if data.bytes[offset] == data.bytes[offset-1] {
offset += 1;
}
} // Single character
else {
offset += 1;
}
token := extract_string(data, data.offset, offset);
// Ignore __attribute__
if token == "__attribute__" {
print_warning("__attribute__ is ignored.");
for data.bytes[offset] != '(' {
offset += 1;
}
parenthesesCount := 1;
for true {
offset += 1;
if data.bytes[offset] == '(' do parenthesesCount += 1;
else if data.bytes[offset] == ')' do parenthesesCount -= 1;
if parenthesesCount == 0 do break;
}
offset += 1;
data.offset = offset;
} // Ignore certain keywords
else if (token == "inline" || token == "__inline" || token == "static"
|| token == "restrict" || token == "__restrict"
|| token == "volatile"
|| token == "__extension__") {
data.offset = offset;
} // Ignore ignored tokens ;)
else {
for ignoredToken in data.options.ignoredTokens {
if token == ignoredToken {
data.offset = offset;
break;
}
}
}
if data.offset != offset {
break;
}
}
return offset;
}
// Peek the next token (just eating whitespaces and comment)
peek_token :: proc(data : ^ParserData) -> string {
tokenEnd := peek_token_end(data);
if tokenEnd == data.bytesLength {
return "EOF";
}
return extract_string(data, data.offset, tokenEnd);
}
// Find the end of the define directive (understanding endline backslashes)
// @note Tricky cases like comments hiding a backslash effect are not handled.
peek_define_end :: proc(data : ^ParserData) -> u32 {
defineEndOffset := data.offset;
for data.bytes[defineEndOffset-1] == '\\' || data.bytes[defineEndOffset] != '\n' {
defineEndOffset += 1;
}
return defineEndOffset;
}
eat_comment :: proc(data : ^ParserData) {
if data.offset >= data.bytesLength || data.bytes[data.offset] != '/' {
return;
}
// Line comment
if data.bytes[data.offset + 1] == '/' {
eat_line(data);
} // Range comment
else if data.bytes[data.offset + 1] == '*' {
data.offset += 2;
for data.bytes[data.offset] != '*' || data.bytes[data.offset + 1] != '/' {
data.offset += 1;
}
data.offset += 2;
}
}
// Eat whitespaces
eat_whitespaces :: proc(data : ^ParserData) {
// Effective whitespace
for data.offset < data.bytesLength &&
(data.bytes[data.offset] == ' ' || data.bytes[data.offset] == '\t' ||
data.bytes[data.offset] == '\r' || data.bytes[data.offset] == '\n') {
if data.bytes[data.offset] == '\n' && data.bytes[data.offset] != '\\' {
data.foundFullReturn = true;
}
data.offset += 1;
}
}
// Removes whitespaces and comments
eat_whitespaces_and_comments :: proc(data : ^ParserData) {
startOffset : u32 = 0xFFFFFFFF;
for startOffset != data.offset {
startOffset = data.offset;
eat_whitespaces(data);
eat_comment(data);
}
}
// Eat full line
eat_line :: proc(data : ^ParserData) {
for ; data.bytes[data.offset] != '\n'; data.offset += 1 {
}
}
// Eat a line, and repeat if it ends with a backslash
eat_define_lines :: proc(data : ^ParserData) {
for data.bytes[data.offset-1] == '\\' || data.bytes[data.offset] != '\n' {
data.offset += 1;
}
}
// Eat next token
eat_token :: proc(data : ^ParserData) {
data.offset = peek_token_end(data);
}
// Eat next token
check_and_eat_token :: proc(data : ^ParserData, expectedToken : string, loc := #caller_location) {
token := peek_token(data);
if token != expectedToken {
print_error(data, loc, "Expected ", expectedToken, " but found ", token, ".");
}
data.offset += cast(u32) len(token);
}
// Check whether the next token is outside #define range
is_define_end :: proc(data : ^ParserData) -> bool {
defineEnd := peek_define_end(data);
tokenEnd := peek_token_end(data);
return (defineEnd < tokenEnd);
}
// Check if the current #define is a macro definition
is_define_macro :: proc(data : ^ParserData) -> bool {
startOffset := data.offset;
defer data.offset = startOffset;
token := parse_any(data);
if token != "(" do return false;
// Find the other parenthesis
parenthesesCount := 1;
for parenthesesCount != 0 {
token = parse_any(data);
if token == "(" do parenthesesCount += 1;
else if token == ")" do parenthesesCount -= 1;
}
// Its a macro if after the parentheses, it's not the end
return !is_define_end(data);
}
// @note Very slow function to get line number,
// use only for errors.
// @todo Well, this does not seem to work properly, UTF-8 problem?
get_line_column :: proc(data : ^ParserData) -> (u32, u32) {
line : u32 = 1;
column : u32 = 0;
for i : u32 = 0; i < data.offset; i += 1 {
if data.bytes[i] == '\n' {
column = 0;
line += 1;
}
else {
column += 1;
}
}
return line, column;
}

View File

@@ -0,0 +1,132 @@
package bindgen
DefineNode :: struct {
name : string,
value : LiteralValue,
}
StructDefinitionNode :: struct {
name : string,
members : [dynamic]StructOrUnionMember,
forwardDeclared : bool,
}
UnionDefinitionNode :: struct {
name : string,
members : [dynamic]StructOrUnionMember,
}
EnumDefinitionNode :: struct {
name : string,
members : [dynamic]EnumMember,
}
FunctionDeclarationNode :: struct {
name : string,
returnType : Type,
parameters : [dynamic]FunctionParameter,
}
TypedefNode :: struct {
name : string,
type : Type,
}
Nodes :: struct {
defines : [dynamic]DefineNode,
enumDefinitions : [dynamic]EnumDefinitionNode,
unionDefinitions : [dynamic]UnionDefinitionNode,
structDefinitions : [dynamic]StructDefinitionNode,
functionDeclarations : [dynamic]FunctionDeclarationNode,
typedefs : [dynamic]TypedefNode,
}
LiteralValue :: union {
i64,
f64,
string,
}
// Type, might be an array
Type :: struct {
base : BaseType,
dimensions : [dynamic]u64, // Array dimensions
}
BaseType :: union {
BuiltinType,
PointerType,
IdentifierType,
FunctionType,
FunctionPointerType,
}
BuiltinType :: enum {
Unknown,
Void,
Int,
UInt,
LongInt,
ULongInt,
LongLongInt,
ULongLongInt,
ShortInt,
UShortInt,
Char,
SChar,
UChar,
Float,
Double,
LongDouble,
// Not defined by C language but in <stdint.h>
Int8,
Int16,
Int32,
Int64,
UInt8,
UInt16,
UInt32,
UInt64,
Size,
SSize,
PtrDiff,
UIntPtr,
IntPtr,
}
PointerType :: struct {
type : ^Type, // Pointer is there to prevent definition cycle. Null means void.
}
IdentifierType :: struct {
name : string,
anonymous : bool, // An anonymous identifier can be hard-given a name in some contexts.
}
FunctionType :: struct {
returnType : ^Type, // Pointer is there to prevent definition cycle. Null means void.
parameters : [dynamic]FunctionParameter,
}
FunctionPointerType :: struct {
name : string,
returnType : ^Type, // Pointer is there to prevent definition cycle. Null means void.
parameters : [dynamic]FunctionParameter,
}
EnumMember :: struct {
name : string,
value : i64,
hasValue : bool,
}
StructOrUnionMember :: struct {
name : string,
type : Type,
}
FunctionParameter :: struct {
name : string,
type : Type,
}

840
core/bindgen/c-parser.odin Normal file
View File

@@ -0,0 +1,840 @@
package bindgen
import "core:os"
import "core:fmt"
import "core:strings"
import "core:strconv"
// Global counters
anonymousStructCount := 0;
anonymousUnionCount := 0;
anonymousEnumCount := 0;
knownTypeAliases : map[string]Type;
CustomHandler :: proc(data : ^ParserData);
CustomExpressionHandler :: proc(data : ^ParserData) -> LiteralValue;
ParserOptions :: struct {
ignoredTokens : []string,
// Handlers
customHandlers : map[string]CustomHandler,
customExpressionHandlers : map[string]CustomExpressionHandler,
}
ParserData :: struct {
bytes : []u8,
bytesLength : u32,
offset : u32,
// References
nodes : Nodes,
options : ^ParserOptions,
// Knowned values
knownedLiterals : map[string]LiteralValue,
// Whether we have eaten a '\n' character that has no backslash just before
foundFullReturn : bool,
}
is_identifier :: proc(token : string) -> bool {
return (token[0] >= 'a' && token[0] <= 'z') ||
(token[0] >= 'A' && token[0] <= 'Z') ||
(token[0] == '_');
}
parse :: proc(bytes : []u8, options : ParserOptions, loc := #caller_location) -> Nodes {
options := options;
data : ParserData;
data.bytes = bytes;
data.bytesLength = cast(u32) len(bytes);
data.options = &options;
for data.offset = 0; data.offset < data.bytesLength; {
token := peek_token(&data);
if data.offset == data.bytesLength do break;
if token in options.customHandlers {
options.customHandlers[token](&data);
}
else if token == "{" || token == "}" || token == ";" {
eat_token(&data);
}
else if token == "extern" {
check_and_eat_token(&data, "extern");
}
else if token == "\"C\"" {
check_and_eat_token(&data, "\"C\"");
}
else if token == "#" {
parse_directive(&data);
}
else if token == "typedef" {
parse_typedef(&data);
}
else if is_identifier(token) {
parse_variable_or_function_declaration(&data);
}
else {
print_error(&data, loc, "Unexpected token: ", token, ".");
return data.nodes;
}
}
return data.nodes;
}
parse_any :: proc(data : ^ParserData) -> string {
offset := peek_token_end(data);
identifier := extract_string(data, data.offset, offset);
data.offset = offset;
return identifier;
}
parse_identifier :: proc(data : ^ParserData, loc := #caller_location) -> string {
identifier := parse_any(data);
if (identifier[0] < 'a' || identifier[0] > 'z') &&
(identifier[0] < 'A' || identifier[0] > 'Z') &&
(identifier[0] != '_') {
print_error(data, loc, "Expected identifier but found ", identifier, ".");
}
return identifier;
}
parse_type_dimensions :: proc(data : ^ParserData, type : ^Type) {
token := peek_token(data);
for token == "[" {
eat_token(data);
token = peek_token(data);
if token == "]" {
pointerType : PointerType;
pointerType.type = new(Type);
pointerType.type^ = type^; // Copy
type.base = pointerType;
delete(type.dimensions);
} else {
dimension := evaluate_i64(data);
append(&type.dimensions, cast(u64) dimension);
}
check_and_eat_token(data, "]");
token = peek_token(data);
}
}
// This will parse anything that look like a type:
// Builtin: char/int/float/...
// Struct-like: struct A/struct { ... }/enum E
// Function pointer: void (*f)(...)
//
// Definition permitted: If a struct-like definition is found, it will generate
// the according Node and return a corresponding type.
parse_type :: proc(data : ^ParserData, definitionPermitted := false) -> Type {
type : Type;
// Eat qualifiers
token := peek_token(data);
if token == "const" {
eat_token(data);
token = peek_token(data);
}
// Parse main type
if token == "struct" {
type.base = parse_struct_type(data, definitionPermitted);
}
else if token == "union" {
type.base = parse_union_type(data);
}
else if token == "enum" {
type.base = parse_enum_type(data);
}
else {
// Test builtin type
type.base = parse_builtin_type(data);
if type.base.(BuiltinType) == BuiltinType.Unknown {
// Basic identifier type
identifierType : IdentifierType;
identifierType.name = parse_identifier(data);
type.base = identifierType;
}
}
// Eat qualifiers
token = peek_token(data);
if token == "const" {
eat_token(data);
token = peek_token(data);
}
// Check if pointer
for token == "*" {
check_and_eat_token(data, "*");
token = peek_token(data);
pointerType : PointerType;
pointerType.type = new(Type);
pointerType.type^ = type; // Copy
type.base = pointerType;
// Eat qualifiers
if token == "const" {
eat_token(data);
token = peek_token(data);
}
}
// Parse array dimensions if any.
parse_type_dimensions(data, &type);
// ----- Function pointer type
if token == "(" {
check_and_eat_token(data, "(");
check_and_eat_token(data, "*");
functionPointerType : FunctionPointerType;
functionPointerType.returnType = new(Type);
functionPointerType.returnType^ = type;
functionPointerType.name = parse_identifier(data);
check_and_eat_token(data, ")");
parse_function_parameters(data, &functionPointerType.parameters);
type.base = functionPointerType;
}
return type;
}
parse_builtin_type :: proc(data : ^ParserData) -> BuiltinType {
previousBuiltinType := BuiltinType.Unknown;
intFound := false;
shortFound := false;
signedFound := false;
unsignedFound := false;
longCount := 0;
for true {
token := peek_token(data);
// Attribute
attributeFound := true;
if token == "long" do longCount += 1;
else if token == "short" do shortFound = true;
else if token == "unsigned" do unsignedFound = true;
else if token == "signed" do signedFound = true;
else do attributeFound = false;
if attributeFound { eat_token(data); continue; }
// Known type alias
if token in knownTypeAliases {
builtinType, ok := knownTypeAliases[token].base.(BuiltinType);
if ok {
eat_token(data);
previousBuiltinType = builtinType;
}
break;
}
// Classic type and standard types
if token == "void" { eat_token(data); return BuiltinType.Void; }
else if token == "int" {
eat_token(data);
intFound = true;
}
else if token == "float" { eat_token(data); return BuiltinType.Float; }
else if token == "double" {
eat_token(data);
if longCount == 0 do return BuiltinType.Double;
else do return BuiltinType.LongDouble;
}
else if token == "char" {
eat_token(data);
if signedFound do return BuiltinType.SChar;
else if unsignedFound do return BuiltinType.UChar;
else do return BuiltinType.Char;
}
else if token == "__int8" {
// @note :MicrosoftDumminess __intX are Microsoft's fixed-size integers
// https://docs.microsoft.com/fr-fr/cpp/cpp/int8-int16-int32-int64
// and for unsigned version, they prefixed it with "unsigned"...
eat_token(data);
if unsignedFound do return BuiltinType.UInt8;
else do return BuiltinType.Int8;
}
else if token == "__int16" {
eat_token(data);
if unsignedFound do return BuiltinType.UInt16;
else do return BuiltinType.Int16;
}
else if token == "__int32" {
eat_token(data);
if unsignedFound do return BuiltinType.UInt32;
else do return BuiltinType.Int32;
}
else if token == "__int64" {
eat_token(data);
if unsignedFound do return BuiltinType.UInt64;
else do return BuiltinType.Int64;
}
else if token == "int8_t" { eat_token(data); return BuiltinType.Int8; }
else if token == "int16_t" { eat_token(data); return BuiltinType.Int16; }
else if token == "int32_t" { eat_token(data); return BuiltinType.Int32; }
else if token == "int64_t" { eat_token(data); return BuiltinType.Int64; }
else if token == "uint8_t" { eat_token(data); return BuiltinType.UInt8; }
else if token == "uint16_t" { eat_token(data); return BuiltinType.UInt16; }
else if token == "uint32_t" { eat_token(data); return BuiltinType.UInt32; }
else if token == "uint64_t" { eat_token(data); return BuiltinType.UInt64; }
else if token == "size_t" { eat_token(data); return BuiltinType.Size; }
else if token == "ssize_t" { eat_token(data); return BuiltinType.SSize; }
else if token == "ptrdiff_t" { eat_token(data); return BuiltinType.PtrDiff; }
else if token == "uintptr_t" { eat_token(data); return BuiltinType.UIntPtr; }
else if token == "intptr_t" { eat_token(data); return BuiltinType.IntPtr; }
break;
}
// Adapt previous builtin type
if previousBuiltinType == BuiltinType.ShortInt {
shortFound = true;
}
else if previousBuiltinType == BuiltinType.Int {
intFound = true;
}
else if previousBuiltinType == BuiltinType.LongInt {
longCount += 1;
}
else if previousBuiltinType == BuiltinType.LongLongInt {
longCount += 2;
}
else if previousBuiltinType == BuiltinType.UShortInt {
unsignedFound = true;
shortFound = true;
}
else if previousBuiltinType == BuiltinType.UInt {
unsignedFound = true;
}
else if previousBuiltinType == BuiltinType.ULongInt {
unsignedFound = true;
longCount += 1;
}
else if previousBuiltinType == BuiltinType.ULongLongInt {
unsignedFound = true;
longCount += 2;
}
else if (previousBuiltinType != BuiltinType.Unknown) {
return previousBuiltinType; // float, void, etc.
}
// Implicit and explicit int
if intFound || shortFound || unsignedFound || signedFound || longCount > 0 {
if unsignedFound {
if shortFound do return BuiltinType.UShortInt;
if longCount == 0 do return BuiltinType.UInt;
if longCount == 1 do return BuiltinType.ULongInt;
if longCount == 2 do return BuiltinType.ULongLongInt;
} else {
if shortFound do return BuiltinType.ShortInt;
if longCount == 0 do return BuiltinType.Int;
if longCount == 1 do return BuiltinType.LongInt;
if longCount == 2 do return BuiltinType.LongLongInt;
}
}
return BuiltinType.Unknown;
}
parse_struct_type :: proc(data : ^ParserData, definitionPermitted : bool) -> IdentifierType {
check_and_eat_token(data, "struct");
type : IdentifierType;
token := peek_token(data);
if !definitionPermitted || token != "{" {
type.name = parse_identifier(data);
token = peek_token(data);
} else {
type.name = tcat("AnonymousStruct", anonymousStructCount);
type.anonymous = true;
anonymousStructCount += 1;
}
if token == "{" {
node := parse_struct_definition(data);
node.name = type.name;
} else if definitionPermitted {
// @note Whatever happens, we create a definition of the struct,
// as it might be used to forward declare it and then use it only with a pointer.
// This for instance the pattern for xcb_connection_t which definition
// is never known from user API.
node : StructDefinitionNode;
node.forwardDeclared = false;
node.name = type.name;
append(&data.nodes.structDefinitions, node);
}
return type;
}
parse_union_type :: proc(data : ^ParserData) -> IdentifierType {
check_and_eat_token(data, "union");
type : IdentifierType;
token := peek_token(data);
if token != "{" {
type.name = parse_identifier(data);
token = peek_token(data);
} else {
type.name = tcat("AnonymousUnion", anonymousUnionCount);
type.anonymous = true;
anonymousUnionCount += 1;
}
if token == "{" {
node := parse_union_definition(data);
node.name = type.name;
}
return type;
}
parse_enum_type :: proc(data : ^ParserData) -> IdentifierType {
check_and_eat_token(data, "enum");
type : IdentifierType;
token := peek_token(data);
if token != "{" {
type.name = parse_identifier(data);
token = peek_token(data);
} else {
type.name = tcat("AnonymousEnum", anonymousEnumCount);
type.anonymous = true;
anonymousEnumCount += 1;
}
if token == "{" {
node := parse_enum_definition(data);
node.name = type.name;
}
return type;
}
/**
* We only care about defines of some value
*/
parse_directive :: proc(data : ^ParserData) {
check_and_eat_token(data, "#");
token := peek_token(data);
if token == "define" {
parse_define(data);
} // We ignore all other directives
else {
eat_line(data);
}
}
parse_define :: proc(data : ^ParserData) {
check_and_eat_token(data, "define");
data.foundFullReturn = false;
node : DefineNode;
node.name = parse_identifier(data);
// Does it look like end? It might be a #define with no expression
if is_define_end(data) {
node.value = 1;
append(&data.nodes.defines, node);
data.knownedLiterals[node.name] = node.value;
} // Macros are ignored
else if is_define_macro(data) {
print_warning("Ignoring define macro for ", node.name, ".");
}
else {
literalValue, ok := evaluate(data);
if ok {
node.value = literalValue;
append(&data.nodes.defines, node);
data.knownedLiterals[node.name] = node.value;
}
else {
print_warning("Ignoring define expression for ", node.name, ".");
}
}
// Evaluating the expression, we might have already eaten a full return,
// if so, do nothing.
if !data.foundFullReturn {
eat_define_lines(data);
}
}
// @fixme Move
change_anonymous_node_name :: proc (data : ^ParserData, oldName : string, newName : string) -> bool {
for i := 0; i < len(data.nodes.structDefinitions); i += 1 {
if data.nodes.structDefinitions[i].name == oldName {
data.nodes.structDefinitions[i].name = newName;
return true;
}
}
for i := 0; i < len(data.nodes.enumDefinitions); i += 1 {
if data.nodes.enumDefinitions[i].name == oldName {
data.nodes.enumDefinitions[i].name = newName;
return true;
}
}
for i := 0; i < len(data.nodes.unionDefinitions); i += 1 {
if data.nodes.unionDefinitions[i].name == oldName {
data.nodes.unionDefinitions[i].name = newName;
return true;
}
}
return false;
}
/**
* Type aliasing.
* typedef <sourceType> <name>;
*/
parse_typedef :: proc(data : ^ParserData) {
check_and_eat_token(data, "typedef");
// @note Struct-like definitions (and such)
// are generated within type parsing.
//
// So that typedef struct { int foo; }* Ap; is valid.
// Parsing type
node : TypedefNode;
node.type = parse_type(data, true);
if sourceType, ok := node.type.base.(FunctionPointerType); ok {
node.name = sourceType.name;
} else {
node.name = parse_identifier(data);
}
// Checking if function type
token := peek_token(data);
if token == "(" {
functionType : FunctionType;
functionType.returnType = new(Type);
functionType.returnType^ = node.type;
parse_function_parameters(data, &functionType.parameters);
node.type.base = functionType;
}
// Checking if array
parse_type_dimensions(data, &node.type);
// If the underlying type is anonymous,
// we just affect it the name.
addTypedefNode := true;
if identifierType, ok := node.type.base.(IdentifierType); ok {
if identifierType.anonymous {
addTypedefNode = !change_anonymous_node_name(data, identifierType.name, node.name);
}
}
if addTypedefNode {
knownTypeAliases[node.name] = node.type;
append(&data.nodes.typedefs, node);
}
check_and_eat_token(data, ";");
// @note Commented tool for debug
// fmt.println("Typedef: ", node.type, node.name);
}
parse_struct_definition :: proc(data : ^ParserData) -> ^StructDefinitionNode {
node : StructDefinitionNode;
node.forwardDeclared = false;
parse_struct_or_union_members(data, &node.members);
append(&data.nodes.structDefinitions, node);
return &data.nodes.structDefinitions[len(data.nodes.structDefinitions) - 1];
}
parse_union_definition :: proc(data : ^ParserData) -> ^UnionDefinitionNode {
node : UnionDefinitionNode;
parse_struct_or_union_members(data, &node.members);
append(&data.nodes.unionDefinitions, node);
return &data.nodes.unionDefinitions[len(data.nodes.unionDefinitions) - 1];
}
parse_enum_definition :: proc(data : ^ParserData) -> ^EnumDefinitionNode {
node : EnumDefinitionNode;
parse_enum_members(data, &node.members);
append(&data.nodes.enumDefinitions, node);
return &data.nodes.enumDefinitions[len(data.nodes.enumDefinitions) - 1];
}
/**
* {
* <name> = <value>,
* <name>,
* }
*/
parse_enum_members :: proc(data : ^ParserData, members : ^[dynamic]EnumMember) {
check_and_eat_token(data, "{");
nextMemberValue : i64 = 0;
token := peek_token(data);
for token != "}" {
member : EnumMember;
member.name = parse_identifier(data);
member.hasValue = false;
token = peek_token(data);
if token == "=" {
check_and_eat_token(data, "=");
member.hasValue = true;
member.value = evaluate_i64(data);
nextMemberValue = member.value;
token = peek_token(data);
} else {
member.value = nextMemberValue;
}
data.knownedLiterals[member.name] = member.value;
nextMemberValue += 1;
// Eat until end, as this might be a complex expression that we couldn't understand
if token != "," && token != "}" {
print_warning("Parser cannot understand fully the expression of enum member ", member.name, ".");
for token != "," && token != "}" {
eat_token(data);
token = peek_token(data);
}
}
if token == "," {
check_and_eat_token(data, ",");
token = peek_token(data);
}
append(members, member);
}
check_and_eat_token(data, "}");
}
/**
* {
* <type> <name>;
* <type> <name1>, <name2>;
* <type> <name>[<dimension>];
* }
*/
parse_struct_or_union_members :: proc(data : ^ParserData, structOrUnionMembers : ^[dynamic]StructOrUnionMember) {
check_and_eat_token(data, "{");
// To ensure unique id
unamedCount := 0;
token := peek_token(data);
for token != "}" {
member : StructOrUnionMember;
member.type = parse_type(data, true);
for true {
// In the case of function pointer types, the name has been parsed
// during type inspection.
if type, ok := member.type.base.(FunctionPointerType); ok {
member.name = type.name;
}
else {
// Unamed (struct or union)
token = peek_token(data);
if !is_identifier(token) {
member.name = tcat("unamed", unamedCount);
unamedCount += 1;
}
else {
member.name = parse_identifier(data);
}
}
parse_type_dimensions(data, &member.type);
token = peek_token(data);
if token == ":" {
check_and_eat_token(data, ":");
print_warning("Found bitfield in struct, which is not handled correctly.");
evaluate_i64(data);
token = peek_token(data);
}
append(structOrUnionMembers, member);
// Multiple declarations on one line
if token == "," {
check_and_eat_token(data, ",");
continue;
}
break;
}
check_and_eat_token(data, ";");
token = peek_token(data);
}
check_and_eat_token(data, "}");
}
parse_variable_or_function_declaration :: proc(data : ^ParserData) {
type := parse_type(data, true);
// If it's just a type, it might be a struct definition
token := peek_token(data);
if token == ";" {
check_and_eat_token(data, ";");
return;
}
// Eat array declaration if any
// @fixme The return type of a function declaration will be wrong!
for data.bytes[data.offset] == '[' {
for data.bytes[data.offset] != ']' {
data.offset += 1;
}
data.offset += 1;
}
name := parse_identifier(data);
token = peek_token(data);
if token == "(" {
functionDeclarationNode := parse_function_declaration(data);
functionDeclarationNode.returnType = type;
functionDeclarationNode.name = name;
return;
} else if token == "[" {
// Eat whole array declaration
for data.bytes[data.offset] == '[' {
for data.bytes[data.offset] != ']' {
data.offset += 1;
}
data.offset += 1;
}
}
// Global variable declaration (with possible multiple declarations)
token = peek_token(data);
for true {
if token == "," {
print_warning("Found global variable declaration '", name, "', we won't generated any binding for it.");
check_and_eat_token(data, ",");
name = parse_identifier(data);
token = peek_token(data);
continue;
}
else if token == ";" {
if name != "" {
print_warning("Found global variable declaration '", name, "', we won't generated any binding for it.");
}
check_and_eat_token(data, ";");
break;
}
// Global variable assignment, considered as constant define.
node : DefineNode;
check_and_eat_token(data, "=");
literalValue, ok := evaluate(data);
if ok {
node.name = name;
node.value = literalValue;
append(&data.nodes.defines, node);
}
else {
print_warning("Ignoring global variable expression for '", name, "'.");
}
name = "";
token = peek_token(data);
}
}
parse_function_declaration :: proc(data : ^ParserData) -> ^FunctionDeclarationNode {
node : FunctionDeclarationNode;
parse_function_parameters(data, &node.parameters);
// Function definition? Ignore it.
token := peek_token(data);
if token == "{" {
bracesCount := 1;
for true {
data.offset += 1;
if data.bytes[data.offset] == '{' do bracesCount += 1;
else if data.bytes[data.offset] == '}' do bracesCount -= 1;
if bracesCount == 0 do break;
}
data.offset += 1;
} // Function declaration
else {
check_and_eat_token(data, ";");
}
append(&data.nodes.functionDeclarations, node);
return &data.nodes.functionDeclarations[len(data.nodes.functionDeclarations) - 1];
}
parse_function_parameters :: proc(data : ^ParserData, parameters : ^[dynamic]FunctionParameter) {
check_and_eat_token(data, "(");
token := peek_token(data);
for token != ")" {
parameter : FunctionParameter;
token = peek_token(data);
if token == "." {
print_warning("A function accepts variadic arguments, this is currently not handled within generated code.");
check_and_eat_token(data, ".");
check_and_eat_token(data, ".");
check_and_eat_token(data, ".");
break;
} else {
parameter.type = parse_type(data);
}
// Check if named parameter
token = peek_token(data);
if token != ")" && token != "," {
parameter.name = parse_identifier(data);
parse_type_dimensions(data, &parameter.type);
token = peek_token(data);
}
if token == "," {
eat_token(data);
token = peek_token(data);
}
append(parameters, parameter);
}
check_and_eat_token(data, ")");
}

44
core/bindgen/errors.odin Normal file
View File

@@ -0,0 +1,44 @@
package bindgen
import "core:fmt"
import "core:os"
seenWarnings : map[string]bool;
print_warning :: proc(args : ..any) {
message := tcat(..args);
if !seenWarnings[message] {
fmt.eprint("[bindgen] Warning: ", message, "\n");
seenWarnings[message] = true;
}
}
print_error :: proc(data : ^ParserData, loc := #caller_location, args : ..any) {
message := tcat(..args);
min : u32 = 0;
for i := data.offset - 1; i > 0; i -= 1 {
if data.bytes[i] == '\n' {
min = i + 1;
break;
}
}
max := min + 200;
for i := min + 1; i < max; i += 1 {
if data.bytes[i] == '\n' {
max = i;
break;
}
}
line, _ := get_line_column(data);
fmt.eprint("[bindgen] Error: ", message, "\n");
fmt.eprint("[bindgen] ... from ", loc.procedure, "\n");
fmt.eprint("[bindgen] ... at line ", line, " within this context:\n");
fmt.eprint("> ", extract_string(data, min, max), "\n");
os.exit(1);
}

View File

@@ -0,0 +1,284 @@
package bindgen
import "core:fmt"
// Prevent keywords clashes and other tricky cases
clean_identifier :: proc(name : string) -> string {
name := name;
if name == "" {
return name;
}
// Starting with _? Try removing that.
for true {
if name[0] == '_' {
name = name[1:];
}
else {
break;
}
}
// Number
if name[0] >= '0' && name[0] <= '9' {
return tcat("_", name);
} // Keywords clash
else if name == "map" || name == "proc" || name == "opaque" || name == "in" {
return tcat("_", name);
} // Jai keywords clash
else if name == "context" ||
name == "float32" || name == "float64" ||
name == "s8" || name == "s16" || name == "s32" || name == "s64" ||
name == "u8" || name == "u16" || name == "u32" || name == "u64" {
return tcat("_", name);
}
return name;
}
clean_variable_name :: proc(name : string, options : ^GeneratorOptions) -> string {
name := name;
name = change_case(name, options.variableCase);
return clean_identifier(name);
}
clean_pseudo_type_name :: proc(structName : string, options : ^GeneratorOptions) -> string {
structName := structName;
structName = remove_postfixes(structName, options.pseudoTypePostfixes, options.pseudoTypeTransparentPostfixes);
structName = remove_prefixes(structName, options.pseudoTypePrefixes, options.pseudoTypeTransparentPrefixes);
structName = change_case(structName, options.pseudoTypeCase);
return structName;
}
// Clean up the enum name so that it can be used to remove the prefix from enum values.
clean_enum_name_for_prefix_removal :: proc(enumName : string, options : ^GeneratorOptions) -> (string, [dynamic]string) {
enumName := enumName;
if !options.enumValueNameRemove {
return enumName, nil;
}
// Remove postfix and use same case convention as the enum values
removedPostfixes : [dynamic]string;
enumName, removedPostfixes = remove_postfixes_with_removed(enumName, options.enumValueNameRemovePostfixes);
enumName = change_case(enumName, options.enumValueCase);
return enumName, removedPostfixes;
}
clean_enum_value_name :: proc(valueName : string, enumName : string, postfixes : []string, options : ^GeneratorOptions) -> string {
valueName := valueName;
valueName = remove_prefixes(valueName, options.enumValuePrefixes, options.enumValueTransparentPrefixes);
valueName = remove_postfixes(valueName, postfixes, options.enumValueTransparentPostfixes);
if options.enumValueNameRemove {
valueName = remove_prefixes(valueName, []string{enumName});
}
valueName = change_case(valueName, options.enumValueCase);
return clean_identifier(valueName);
}
clean_function_name :: proc(functionName : string, options : ^GeneratorOptions) -> string {
functionName := functionName;
functionName = remove_prefixes(functionName, options.functionPrefixes, options.functionTransparentPrefixes);
functionName = remove_postfixes(functionName, options.definePostfixes, options.defineTransparentPostfixes);
functionName = change_case(functionName, options.functionCase);
return functionName;
}
clean_define_name :: proc(defineName : string, options : ^GeneratorOptions) -> string {
defineName := defineName;
defineName = remove_prefixes(defineName, options.definePrefixes, options.defineTransparentPrefixes);
defineName = remove_postfixes(defineName, options.definePostfixes, options.defineTransparentPostfixes);
defineName = change_case(defineName, options.defineCase);
return defineName;
}
// Convert to Odin's types
clean_type :: proc(data : ^GeneratorData, type : Type, baseTab : string = "", explicitSharpType := true) -> string {
output := "";
for dimension in type.dimensions {
output = tcat(output, "[", dimension, "]");
}
output = tcat(output, clean_base_type(data, type.base, baseTab, explicitSharpType));
return output;
}
clean_base_type :: proc(data : ^GeneratorData, baseType : BaseType, baseTab : string = "", explicitSharpType := true) -> string {
options := data.options;
if _type, ok := baseType.(BuiltinType); ok {
if _type == BuiltinType.Void do return options.mode == "jai" ? "void" : "";
else if _type == BuiltinType.Int do return options.mode == "jai" ? "s64" : "_c.int";
else if _type == BuiltinType.UInt do return options.mode == "jai" ? "u64" :"_c.uint";
else if _type == BuiltinType.LongInt do return options.mode == "jai" ? "s64" :"_c.long";
else if _type == BuiltinType.ULongInt do return options.mode == "jai" ? "u64" :"_c.ulong";
else if _type == BuiltinType.LongLongInt do return options.mode == "jai" ? "s64" :"_c.longlong";
else if _type == BuiltinType.ULongLongInt do return options.mode == "jai" ? "u64" :"_c.ulonglong";
else if _type == BuiltinType.ShortInt do return options.mode == "jai" ? "s16" :"_c.short";
else if _type == BuiltinType.UShortInt do return options.mode == "jai" ? "u16" :"_c.ushort";
else if _type == BuiltinType.Char do return options.mode == "jai" ? "u8" :"_c.char";
else if _type == BuiltinType.SChar do return options.mode == "jai" ? "s8" :"_c.schar";
else if _type == BuiltinType.UChar do return options.mode == "jai" ? "u8" :"_c.uchar";
else if _type == BuiltinType.Float do return options.mode == "jai" ? "float32" :"_c.float";
else if _type == BuiltinType.Double do return options.mode == "jai" ? "float64" :"_c.double";
else if _type == BuiltinType.LongDouble {
print_warning("Found long double which is currently not supported. Fallback to double in generated code.");
return options.mode == "jai" ? "double" :"_c.double";
}
else if _type == BuiltinType.Int8 do return options.mode == "jai" ? "s8" :"i8";
else if _type == BuiltinType.Int16 do return options.mode == "jai" ? "s16" :"i16";
else if _type == BuiltinType.Int32 do return options.mode == "jai" ? "s32" :"i32";
else if _type == BuiltinType.Int64 do return options.mode == "jai" ? "s64" :"i64";
else if _type == BuiltinType.UInt8 do return options.mode == "jai" ? "u8" :"u8";
else if _type == BuiltinType.UInt16 do return options.mode == "jai" ? "u16" :"u16";
else if _type == BuiltinType.UInt32 do return options.mode == "jai" ? "u32" :"u32";
else if _type == BuiltinType.UInt64 do return options.mode == "jai" ? "u64" :"u64";
else if _type == BuiltinType.Size do return options.mode == "jai" ? "u64" :"_c.size_t";
else if _type == BuiltinType.SSize do return options.mode == "jai" ? "u64" :"_c.ssize_t";
else if _type == BuiltinType.PtrDiff do return options.mode == "jai" ? "s64" :"_c.ptrdiff_t";
else if _type == BuiltinType.UIntPtr do return options.mode == "jai" ? "u64" :"_c.uintptr_t";
else if _type == BuiltinType.IntPtr do return options.mode == "jai" ? "s64" :"_c.intptr_t";
}
else if _type, ok := baseType.(PointerType); ok {
if options.mode == "jai" {
// Hide pointers to types that were not declared.
if !is_known_base_type(data, _type.type.base) {
print_warning("*", _type.type.base.(IdentifierType).name, " replaced by *void as the pointed type is unknown.");
return "*void";
}
} else {
if __type, ok := _type.type.base.(BuiltinType); ok {
if __type == BuiltinType.Void do return "rawptr";
else if __type == BuiltinType.Char do return "cstring";
}
}
name := clean_type(data, _type.type^, baseTab);
return tcat(options.mode == "jai" ? "*" :"^", name);
}
else if _type, ok := baseType.(IdentifierType); ok {
return clean_pseudo_type_name(_type.name, options);
}
else if _type, ok := baseType.(FunctionType); ok {
output : string;
if explicitSharpType {
output = "#type ";
}
output = tcat(output, options.mode == "jai" ? "(" :"proc(");
parameters := clean_function_parameters(data, _type.parameters, baseTab);
output = tcat(output, parameters, ")");
returnType := clean_type(data, _type.returnType^);
if len(returnType) > 0 && returnType != "void" {
output = tcat(output, " -> ", returnType);
}
return output;
}
else if _type, ok := baseType.(FunctionPointerType); ok {
output : string;
if explicitSharpType {
output = "#type ";
}
output = tcat(output, options.mode == "jai" ? "(" :"proc(");
parameters := clean_function_parameters(data, _type.parameters, baseTab);
output = tcat(output, parameters, ")");
returnType := clean_type(data, _type.returnType^);
if len(returnType) > 0 && returnType != "void" {
output = tcat(output, " -> ", returnType);
}
if options.mode == "jai" {
output = tcat(output, " #foreign");
}
return output;
}
return "<niy>";
}
clean_function_parameters :: proc(data : ^GeneratorData, parameters : [dynamic]FunctionParameter, baseTab : string) -> string {
output := "";
options := data.options;
// Special case: function(void) does not really have a parameter
if len(parameters) == 1 {
if _type, ok := parameters[0].type.base.(BuiltinType); ok {
if _type == BuiltinType.Void {
return "";
}
}
}
tab := "";
if options.mode == "jai" { // @note :OdinCodingStyle Odin forces a coding style, now. Ugh.
if (len(parameters) > 1) {
output = tcat(output, "\n");
tab = tcat(baseTab, " ");
}
}
unamedParametersCount := 0;
for parameter, i in parameters {
type := clean_type(data, parameter.type);
name : string;
if len(parameter.name) != 0 {
name = clean_variable_name(parameter.name, options);
} else {
name = tcat("unamed", unamedParametersCount);
unamedParametersCount += 1;
}
output = tcat(output, tab, name, " : ", type);
if i != len(parameters) - 1 {
if options.mode == "jai" { // @note :OdinCodingStyle
output = tcat(output, ",\n");
} else {
output = tcat(output, ", ");
}
}
}
if (len(parameters) > 1) {
if options.mode == "jai" { // @note :OdinCodingStyle
output = tcat(output, "\n", baseTab);
}
}
return output;
}
is_known_base_type :: proc(data : ^GeneratorData, baseType : BaseType) -> bool {
if _type, ok := baseType.(IdentifierType); ok {
for it in data.nodes.typedefs {
if _type.name == it.name {
return true;
}
}
for it in data.nodes.structDefinitions {
if _type.name == it.name {
return true;
}
}
for it in data.nodes.enumDefinitions {
if _type.name == it.name {
return true;
}
}
for it in data.nodes.unionDefinitions {
if _type.name == it.name {
return true;
}
}
return false;
}
return true;
}

View File

@@ -0,0 +1,166 @@
package bindgen
import "core:os"
import "core:fmt"
export_defines :: proc(data : ^GeneratorData) {
for node in data.nodes.defines {
defineName := clean_define_name(node.name, data.options);
// @fixme fprint of float numbers are pretty badly handled,
// just has a 10^-3 precision.
fcat(data.handle, defineName, " :: ", node.value, ";\n");
}
fcat(data.handle, "\n");
}
export_typedefs :: proc(data : ^GeneratorData) {
for node in data.nodes.typedefs {
name := clean_pseudo_type_name(node.name, data.options);
type := clean_type(data, node.type, "", true);
if name == type do continue;
fcat(data.handle, name, " :: ", type, ";\n");
}
fcat(data.handle, "\n");
}
export_enums :: proc(data : ^GeneratorData) {
for node in data.nodes.enumDefinitions {
enumName := clean_pseudo_type_name(node.name, data.options);
if data.options.mode == "jai" {
consideredFlags := false;
for postfix in data.options.enumConsideredFlagsPostfixes {
if ends_with(node.name, postfix) {
consideredFlags = true;
break;
}
}
if consideredFlags {
fcat(data.handle, enumName, " :: enum_flags u32 {");
} else {
fcat(data.handle, enumName, " :: enum s32 {");
}
} else {
fcat(data.handle, enumName, " :: enum i32 {");
}
postfixes : [dynamic]string;
enumName, postfixes = clean_enum_name_for_prefix_removal(enumName, data.options);
// Changing the case of postfixes to the enum value one,
// so that they can be removed.
enumValueCase := find_case(node.members[0].name);
for postfix, i in postfixes {
postfixes[i] = change_case(postfix, enumValueCase);
}
// And changing the case of enumName to the enum value one
enumName = change_case(enumName, enumValueCase);
// Merging enum value postfixes with postfixes that have been removed from the enum name.
for postfix in data.options.enumValuePostfixes {
append(&postfixes, postfix);
}
export_enum_members(data, node.members, enumName, postfixes[:]);
fcat(data.handle, data.options.mode == "jai" ? "}\n" : "};\n");
fcat(data.handle, "\n");
}
}
export_structs :: proc(data : ^GeneratorData) {
for node in data.nodes.structDefinitions {
structName := clean_pseudo_type_name(node.name, data.options);
fcat(data.handle, structName, " :: struct {");
export_struct_or_union_members(data, node.members);
fcat(data.handle, data.options.mode == "jai" ? "}\n" : "};\n");
fcat(data.handle, "\n");
}
}
export_unions :: proc(data : ^GeneratorData) {
for node in data.nodes.unionDefinitions {
unionName := clean_pseudo_type_name(node.name, data.options);
fcat(data.handle, unionName, data.options.mode == "jai" ? " :: union {" : " :: struct #raw_union {");
export_struct_or_union_members(data, node.members);
fcat(data.handle, data.options.mode == "jai" ? "}\n" : "};\n");
fcat(data.handle, "\n");
}
}
export_functions :: proc(data : ^GeneratorData) {
for node in data.nodes.functionDeclarations {
functionName := clean_function_name(node.name, data.options);
if data.options.mode == "jai" {
fcat(data.handle, functionName, " :: (");
} else {
fcat(data.handle, " @(link_name=\"", node.name, "\")\n");
fcat(data.handle, " ", functionName, " :: proc(");
}
parameters := clean_function_parameters(data, node.parameters, data.options.mode == "jai" ? "" : " ");
fcat(data.handle, parameters, ")");
returnType := clean_type(data, node.returnType);
if len(returnType) > 0 {
fcat(data.handle, " -> ", returnType);
}
if data.options.mode == "jai" {
fcat(data.handle, " #foreign ", data.foreignLibrary, " \"", node.name ,"\";\n");
} else {
fcat(data.handle, " ---;\n");
}
fcat(data.handle, "\n");
}
}
export_enum_members :: proc(data : ^GeneratorData, members : [dynamic]EnumMember, enumName : string, postfixes : []string) {
if (len(members) > 0) {
fcat(data.handle, "\n");
}
cleanedMembers : [dynamic]EnumMember;
for member in members {
cleanedMember : EnumMember;
cleanedMember.hasValue = member.hasValue;
cleanedMember.value = member.value;
cleanedMember.name = clean_enum_value_name(member.name, enumName, postfixes, data.options);
if len(cleanedMember.name) == 0 {
// print_warning("Enum member ", member.name, " resolves to an empty name. Ignoring it.");
continue;
}
// Ensuring that we don't collide with an other enum member.
foundCopy := false;
for existingCleanedMember in cleanedMembers {
if cleanedMember.name == existingCleanedMember.name &&
cleanedMember.hasValue == existingCleanedMember.hasValue &&
cleanedMember.value == existingCleanedMember.value {
print_warning("Enum member ", member.name, " is duplicated once cleaned. Keeping only one copy.");
foundCopy = true;
break;
}
}
if foundCopy do continue;
fcat(data.handle, " ", cleanedMember.name);
if member.hasValue {
fcat(data.handle, data.options.mode == "jai" ? " :: " : " = ", member.value);
}
fcat(data.handle, data.options.mode == "jai" ? ";\n" : ",\n");
append(&cleanedMembers, cleanedMember);
}
}
export_struct_or_union_members :: proc(data : ^GeneratorData, members : [dynamic]StructOrUnionMember) {
if (len(members) > 0) {
fcat(data.handle, "\n");
}
for member in members {
type := clean_type(data, member.type, " ");
name := clean_variable_name(member.name, data.options);
fcat(data.handle, " ", name, " : ", type, data.options.mode == "jai" ? ";\n" : ",\n");
}
}

View File

@@ -0,0 +1,392 @@
package bindgen
import "core:fmt"
import "core:os"
import "core:io"
import "core:strings"
import "core:unicode/utf8"
Case :: enum {
Unknown,
Camel,
Constant,
Kebab,
Pascal,
Snake,
}
WordCase :: enum {
Unknown,
Up,
Low,
FirstUp,
// When first upping, numbers are followed always by a capital
FirstUpNumberReset,
}
// Change a character to a capital.
to_uppercase :: proc(c : rune) -> rune {
c := c;
if c >= 'a' && c <= 'z' {
c = c - 'a' + 'A';
}
return c;
}
// Change a character to lowercase.
to_lowercase :: proc(c : rune) -> rune {
c := c;
if c >= 'A' && c <= 'Z' {
c = c - 'A' + 'a';
}
return c;
}
// @note Stolen tprint and fprint from fmt package, because it was confusing due to args: ..any and sep default parameter.
tcat :: proc(args: ..any) -> string {
return fmt.tprint(args=args, sep="");
}
fcat :: proc(fd: os.Handle, args: ..any) -> int {
return fmt.fprint(fd=fd, args=args, sep="");
}
// Change the case convention of a word.
change_word_case :: proc(str : string, targetCase : WordCase) -> string {
newStr : string;
if targetCase == WordCase.Up {
for c in str {
newStr = tcat(newStr, to_uppercase(c));
}
}
else if targetCase == WordCase.Low {
for c in str {
newStr = tcat(newStr, to_lowercase(c));
}
}
else if targetCase == WordCase.FirstUp {
for c, i in str {
if i == 0 {
newStr = tcat(newStr, to_uppercase(c));
} else {
newStr = tcat(newStr, to_lowercase(c));
}
}
}
else if targetCase == WordCase.FirstUpNumberReset {
for c, i in str {
if i == 0 || (str[i - 1] >= '0' && str[i - 1] <= '9') {
newStr = tcat(newStr, to_uppercase(c));
} else {
newStr = tcat(newStr, to_lowercase(c));
}
}
}
return newStr;
}
// Change the case convention of a string by detecting original convention,
// then splitting it into words.
change_case :: proc(str : string, targetCase : Case) -> string {
if targetCase == Case.Unknown {
return str;
}
// Split
parts := autosplit_string(str);
// Join
newStr : string;
if targetCase == Case.Pascal {
for part, i in parts {
newStr = tcat(newStr, change_word_case(part, WordCase.FirstUpNumberReset));
}
}
else if targetCase == Case.Snake {
for part, i in parts {
newStr = tcat(newStr, change_word_case(part, WordCase.Low), (i != len(parts) - 1) ? "_" : "");
}
}
else if targetCase == Case.Kebab {
for part, i in parts {
newStr = tcat(newStr, change_word_case(part, WordCase.Low), (i != len(parts) - 1) ? "-" : "");
}
}
else if targetCase == Case.Camel {
for part, i in parts {
if i == 0 {
newStr = tcat(newStr, change_word_case(part, WordCase.Low));
} else {
newStr = tcat(newStr, change_word_case(part, WordCase.FirstUpNumberReset));
}
}
}
else if targetCase == Case.Constant {
for part, i in parts {
newStr = tcat(newStr, change_word_case(part, WordCase.Up), (i != len(parts) - 1) ? "_" : "");
}
}
return newStr;
}
// Identify the case of the provided string.
// Full lowercase with no separator is identified as camelCase.
find_case :: proc(str : string) -> Case {
refuted : bool;
// CONSTANT_CASE
refuted = false;
for c in str {
if (c != '_') && (c < 'A' || c > 'Z') && (c < '0' || c > '9') {
refuted = true;
break;
}
}
if !refuted do return Case.Constant;
for c in str {
// snake_case
if c == '_' {
return Case.Snake;
} // kebab-case
else if c == '-' {
return Case.Kebab;
}
}
// PascalCase
if str[0] >= 'A' && str[0] <= 'Z' {
return Case.Pascal;
}
// camelCase
return Case.Camel;
}
// Splits the string according to detected case.
// HeyBuddy -> {"Hey", "Buddy"}
// hey-buddy -> {"hey", "buddy"}
// _hey_buddy -> {"", "hey", "buddy"}
// and such...
autosplit_string :: proc(str : string) -> [dynamic]string {
lowCount := 0;
upCount := 0;
for c in str {
// If any '_', split according to that (CONSTANT_CASE or snake_case)
if c == '_' {
return split_from_separator(str, '_');
} // If any '-', split according to that (kebab-case)
else if c == '-' {
return split_from_separator(str, '-');
}
else if c >= 'a' && c <= 'z' {
lowCount += 1;
}
else if c >= 'A' && c <= 'Z' {
upCount += 1;
}
}
// If it seems to be only one word
if lowCount == 0 || upCount == 0 {
parts : [dynamic]string;
append(&parts, str);
return parts;
}
// Split at each uppercase letter (PascalCase or camelCase)
return split_from_capital(str);
}
split_from_separator :: proc(str : string, sep : rune) -> [dynamic]string {
parts : [dynamic]string;
lastI := 0;
// Empty strings for starting separators in string
for c in str {
if c == sep {
append(&parts, "");
lastI += 1;
} else {
break;
}
}
// Ignore non letter prefix
if lastI == 0 {
for c in str {
if (c < 'a' || c > 'z') && (c < 'A' || c > 'Z') {
lastI += 1;
}
else {
break;
}
}
}
for c, i in str {
if i > lastI + 1 && c == sep {
append(&parts, str[lastI:i]);
lastI = i + 1;
}
}
append(&parts, str[lastI:]);
return parts;
}
split_from_capital :: proc(str : string) -> [dynamic]string {
parts : [dynamic]string;
// Ignore non letter prefix
lastI := 0;
for c in str {
if (c < 'a' || c > 'z') && (c < 'A' || c > 'Z') {
lastI += 1;
}
else {
break;
}
}
// We want to handle:
// myBrainIsCRAZY -> my Brain Is Crazy
// myCRAZYBrain -> my CRAZY Brain
// SOLO -> SOLO
// Do split
for i := 1; i < len(str); i += 1 {
if str[i] >= 'A' && str[i] <= 'Z' {
// Do not split too much if it seems to be a capitalized word
if (lastI == i - 1) && (str[lastI] >= 'A' && str[lastI] <= 'Z') {
for ; i + 1 < len(str); i += 1 {
if str[i + 1] < 'A' || str[i + 1] > 'Z' {
break;
}
}
if (i + 1 == len(str)) && (str[i] >= 'A' && str[i] <= 'Z') {
i += 1;
}
}
append(&parts, str[lastI:i]);
lastI = i;
}
}
if lastI != len(str) {
append(&parts, str[lastI:]);
}
return parts;
}
// Check if str if prefixed with any of the provided strings,
// even combinaisons of those, and remove them.
remove_prefixes :: proc(str : string, prefixes : []string, transparentPrefixes : []string = nil) -> string {
str := str;
transparentStr := "";
found := true;
for found {
found = false;
// Remove effective prefixes
for prefix in prefixes {
if len(str) >= len(prefix) &&
str[:len(prefix)] == prefix {
str = str[len(prefix):];
if len(str) != 0 && (str[0] == '_' || str[0] == '-') {
str = str[1:];
}
found = true;
break;
}
}
if found do continue;
// Remove transparent ones, only one by one,
// as we want effective ones to be fully removed.
for prefix in transparentPrefixes {
if len(str) >= len(prefix) &&
str[:len(prefix)] == prefix {
str = str[len(prefix):];
transparentStr = tcat(transparentStr, prefix);
if len(str) != 0 && (str[0] == '_' || str[0] == '-') {
str = str[1:];
transparentStr = tcat(transparentStr, '_');
}
found = true;
break;
}
}
}
return tcat(transparentStr, str);
}
// Check if str if postfixes with any of the provided strings,
// even combinaisons of those, and remove them.
remove_postfixes_with_removed :: proc(
str : string,
postfixes : []string,
transparentPostfixes : []string = nil) -> (string, [dynamic]string) {
str := str;
removedPostfixes : [dynamic]string;
transparentStr := "";
found := true;
for found {
found = false;
// Remove effective postfixes
for postfix in postfixes {
if ends_with(str, postfix) {
str = str[:len(str) - len(postfix)];
if len(str) != 0 && (str[len(str)-1] == '_' || str[len(str)-1] == '-') {
str = str[:len(str)-1];
}
append(&removedPostfixes, postfix);
found = true;
break;
}
}
if found do continue;
// Remove transparent ones, only one by one,
// as we want effective ones to be fully removed.
for postfix in transparentPostfixes {
if ends_with(str, postfix) {
str = str[:len(str) - len(postfix)];
transparentStr = tcat(postfix, transparentStr);
if len(str) != 0 && (str[len(str)-1] == '_' || str[len(str)-1] == '-') {
str = str[:len(str)-1];
transparentStr = tcat('_', transparentStr);
}
found = true;
break;
}
}
}
return tcat(str, transparentStr), removedPostfixes;
}
remove_postfixes :: proc(
str : string,
postfixes : []string,
transparentPostfixes : []string = nil) -> string {
str := str;
removedPostfixes : [dynamic]string;
str, removedPostfixes = remove_postfixes_with_removed(str, postfixes, transparentPostfixes);
return str;
}
ends_with :: proc(str : string, postfix : string) -> bool {
return len(str) >= len(postfix) && str[len(str) - len(postfix):] == postfix;
}

205
core/bindgen/generator.odin Normal file
View File

@@ -0,0 +1,205 @@
/**
* Odin binding generator from C header data.
*/
package bindgen
import "core:os"
import "core:fmt"
import "core:runtime"
GeneratorOptions :: struct {
mode : string, // "odin" or "jai"
// Variable
variableCase : Case,
// Defines
definePrefixes : []string,
defineTransparentPrefixes : []string,
definePostfixes : []string,
defineTransparentPostfixes : []string,
defineCase : Case,
// Pseudo-types
pseudoTypePrefixes : []string,
pseudoTypeTransparentPrefixes : []string,
pseudoTypePostfixes : []string,
pseudoTypeTransparentPostfixes : []string,
pseudoTypeCase : Case,
// Enums
enumConsideredFlagsPostfixes : []string,
// Functions
functionPrefixes : []string,
functionTransparentPrefixes : []string,
functionPostfixes : []string,
functionTransparentPostfixes : []string,
functionCase : Case,
// Enum values
enumValuePrefixes : []string,
enumValueTransparentPrefixes : []string,
enumValuePostfixes : []string,
enumValueTransparentPostfixes : []string,
enumValueCase : Case,
enumValueNameRemove : bool,
enumValueNameRemovePostfixes : []string,
parserOptions : ParserOptions,
}
GeneratorData :: struct {
handle : os.Handle,
nodes : Nodes,
// References
foreignLibrary : string,
options : ^GeneratorOptions,
}
generate :: proc(
packageName : string,
foreignLibrary : string,
outputFile : string,
headerFiles : []string,
options : GeneratorOptions,
) {
options := options;
data : GeneratorData;
data.options = &options;
data.foreignLibrary = foreignLibrary;
if options.mode == "" {
options.mode = "odin";
}
// Outputing odin file
errno : os.Errno;
// chmod 664 when creating file
mode: int = 0;
when os.OS == "linux" || os.OS == "darwin" {
mode = os.S_IRUSR | os.S_IWUSR | os.S_IRGRP | os.S_IWGRP | os.S_IROTH;
}
data.handle, errno = os.open(outputFile, os.O_WRONLY | os.O_CREATE | os.O_TRUNC, mode);
if errno != 0 {
fmt.eprint("[bindgen] Unable to write to output file ", outputFile, " (", errno ,")\n");
return;
}
defer os.close(data.handle);
if options.mode == "jai" {
fcat(data.handle, foreignLibrary, " :: #foreign_library \"", foreignLibrary, "\";\n");
fcat(data.handle, "\n");
} else {
fcat(data.handle, "package ", packageName, "\n");
fcat(data.handle, "\n");
fcat(data.handle, "foreign import \"", foreignLibrary, "\"\n");
fcat(data.handle, "\n");
fcat(data.handle, "import _c \"core:c\"\n");
fcat(data.handle, "\n");
}
// Parsing header files
anonymousStructCount = 0;
anonymousUnionCount = 0;
anonymousEnumCount = 0;
for headerFile in headerFiles {
bytes, ok := os.read_entire_file(headerFile);
if !ok {
fmt.eprint("[bindgen] Unable to read file ", headerFile, "\n");
return;
}
// We fuse the SOAs
headerNodes := parse(bytes, options.parserOptions);
merge_generic_nodes(&data.nodes.defines, &headerNodes.defines);
merge_generic_nodes(&data.nodes.enumDefinitions, &headerNodes.enumDefinitions);
merge_generic_nodes(&data.nodes.unionDefinitions, &headerNodes.unionDefinitions);
merge_forward_declared_nodes(&data.nodes.structDefinitions, &headerNodes.structDefinitions);
merge_generic_nodes(&data.nodes.functionDeclarations, &headerNodes.functionDeclarations);
merge_generic_nodes(&data.nodes.typedefs, &headerNodes.typedefs);
}
// Exporting
export_defines(&data);
export_typedefs(&data);
export_enums(&data);
export_structs(&data);
export_unions(&data);
// Foreign block for functions
if options.mode != "jai" {
foreignLibrarySimple := simplify_library_name(foreignLibrary);
fcat(data.handle, "@(default_calling_convention=\"c\")\n");
fcat(data.handle, "foreign ", foreignLibrarySimple, " {\n");
fcat(data.handle, "\n");
}
export_functions(&data);
if options.mode != "jai" {
fcat(data.handle, "}\n");
}
}
// system:foo.lib -> foo
simplify_library_name :: proc(libraryName : string) -> string {
startOffset := 0;
endOffset := len(libraryName);
for c, i in libraryName {
if startOffset == 0 && c == ':' {
startOffset = i + 1;
}
else if c == '.' {
endOffset = i;
break;
}
}
return libraryName[startOffset:endOffset];
}
merge_generic_nodes :: proc(nodes : ^$T, headerNodes : ^T) {
for headerNode in headerNodes {
// Check that there are no duplicated nodes (due to forward declaration or such)
duplicatedIndex := -1;
for i := 0; i < len(nodes); i += 1 {
node := nodes[i];
if node.name == headerNode.name {
duplicatedIndex = i;
break;
}
}
if duplicatedIndex < 0 {
append(nodes, headerNode);
}
}
}
merge_forward_declared_nodes :: proc(nodes : ^$T, headerNodes : ^T) {
for headerNode in headerNodes {
// Check that there are no duplicated nodes (due to forward declaration or such)
duplicatedIndex := -1;
for i := 0; i < len(nodes); i += 1 {
node := nodes[i];
if node.name == headerNode.name {
duplicatedIndex = i;
break;
}
}
if duplicatedIndex < 0 {
append(nodes, headerNode);
}
else if !headerNode.forwardDeclared && len(headerNode.members) > 0 {
nodes[duplicatedIndex] = headerNode;
}
}
}

BIN
vendor/stb/lib/darwin/libstb_image.a vendored Normal file

Binary file not shown.

BIN
vendor/stb/lib/darwin/stb_image.a vendored Normal file

Binary file not shown.

BIN
vendor/stb/lib/darwin/stb_image_resize.a vendored Normal file

Binary file not shown.

BIN
vendor/stb/lib/darwin/stb_image_write.a vendored Normal file

Binary file not shown.

BIN
vendor/stb/lib/darwin/stb_rect_pack.a vendored Normal file

Binary file not shown.

BIN
vendor/stb/lib/darwin/stb_truetype.a vendored Normal file

Binary file not shown.

1
wasm-ld Symbolic link
View File

@@ -0,0 +1 @@
/Volumes/Phill_Backup/pers/programming/sdk/emsdk/upstream/bin/lld