add mark for serializer

This commit is contained in:
blueloveTH 2026-01-06 13:53:26 +08:00
parent 07a65a8f21
commit 0bdbf17d46
3 changed files with 58 additions and 1 deletions

View File

@ -12,6 +12,7 @@ typedef struct c11_serializer {
void c11_serializer__ctor(c11_serializer* self, int16_t magic, int8_t major_ver, int8_t minor_ver); void c11_serializer__ctor(c11_serializer* self, int16_t magic, int8_t major_ver, int8_t minor_ver);
void c11_serializer__dtor(c11_serializer* self); void c11_serializer__dtor(c11_serializer* self);
void c11_serializer__write_mark(c11_serializer* self, char mark);
void c11_serializer__write_cstr(c11_serializer* self, const char*); void c11_serializer__write_cstr(c11_serializer* self, const char*);
void c11_serializer__write_bytes(c11_serializer* self, const void* data, int size); void c11_serializer__write_bytes(c11_serializer* self, const void* data, int size);
void* c11_serializer__submit(c11_serializer* self, int* size); void* c11_serializer__submit(c11_serializer* self, int* size);
@ -27,6 +28,7 @@ typedef struct c11_deserializer {
void c11_deserializer__ctor(c11_deserializer* self, const void* data, int size); void c11_deserializer__ctor(c11_deserializer* self, const void* data, int size);
void c11_deserializer__dtor(c11_deserializer* self); void c11_deserializer__dtor(c11_deserializer* self);
void c11_deserializer__consume_mark(c11_deserializer* self, char expected);
bool c11_deserializer__check_header(c11_deserializer* self, int16_t magic, int8_t major_ver, int8_t minor_ver_min); bool c11_deserializer__check_header(c11_deserializer* self, int16_t magic, int8_t major_ver, int8_t minor_ver_min);
const char* c11_deserializer__read_cstr(c11_deserializer* self); const char* c11_deserializer__read_cstr(c11_deserializer* self);
void* c11_deserializer__read_bytes(c11_deserializer* self, int size); void* c11_deserializer__read_bytes(c11_deserializer* self, int size);

View File

@ -25,6 +25,10 @@ void c11_serializer__dtor(c11_serializer* self){
c11_vector__dtor(&self->data); c11_vector__dtor(&self->data);
} }
void c11_serializer__write_mark(c11_serializer *self, char mark) {
c11_serializer__write_i8(self, (int8_t)mark);
}
void c11_serializer__write_cstr(c11_serializer *self, const char* cstr) { void c11_serializer__write_cstr(c11_serializer *self, const char* cstr) {
int len = (int)strlen(cstr); int len = (int)strlen(cstr);
c11_serializer__write_bytes(self, cstr, len + 1); c11_serializer__write_bytes(self, cstr, len + 1);
@ -56,6 +60,13 @@ bool c11_deserializer__error(c11_deserializer* self, const char* msg){
return false; return false;
} }
void c11_deserializer__consume_mark(c11_deserializer* self, char expected) {
int8_t mark = c11_deserializer__read_i8(self);
if (mark != (int8_t)expected) {
c11__abort("internal error: deserialize mark mismatch: expected %c but got %c", expected, (char)mark);
}
}
bool c11_deserializer__check_header(c11_deserializer* self, int16_t magic, int8_t major_ver, int8_t minor_ver_min){ bool c11_deserializer__check_header(c11_deserializer* self, int16_t magic, int8_t major_ver, int8_t minor_ver_min){
if(self->size < 8){ if(self->size < 8){
return c11_deserializer__error(self, "bad header: size < 8"); return c11_deserializer__error(self, "bad header: size < 8");

View File

@ -115,33 +115,43 @@ static void CodeObject__serialize(c11_serializer* s,
// codes // codes
_Static_assert(sizeof(Bytecode) == sizeof(uint16_t) * 2, ""); _Static_assert(sizeof(Bytecode) == sizeof(uint16_t) * 2, "");
c11_serializer__write_i32(s, co->codes.length); c11_serializer__write_i32(s, co->codes.length);
c11_serializer__write_mark(s, '[');
c11_serializer__write_bytes(s, co->codes.data, co->codes.length * sizeof(Bytecode)); c11_serializer__write_bytes(s, co->codes.data, co->codes.length * sizeof(Bytecode));
c11_serializer__write_mark(s, ']');
// codes_ex // codes_ex
_Static_assert(sizeof(BytecodeEx) == sizeof(int32_t) * 2, ""); _Static_assert(sizeof(BytecodeEx) == sizeof(int32_t) * 2, "");
c11_serializer__write_i32(s, co->codes_ex.length); c11_serializer__write_i32(s, co->codes_ex.length);
c11_serializer__write_mark(s, '[');
c11_serializer__write_bytes(s, co->codes_ex.data, co->codes_ex.length * sizeof(BytecodeEx)); c11_serializer__write_bytes(s, co->codes_ex.data, co->codes_ex.length * sizeof(BytecodeEx));
c11_serializer__write_mark(s, ']');
// consts // consts
c11_serializer__write_i32(s, co->consts.length); c11_serializer__write_i32(s, co->consts.length);
c11_serializer__write_mark(s, '[');
for(int i = 0; i < co->consts.length; i++) { for(int i = 0; i < co->consts.length; i++) {
py_Ref val = c11__at(py_TValue, &co->consts, i); py_Ref val = c11__at(py_TValue, &co->consts, i);
TValue__serialize(s, val); TValue__serialize(s, val);
} }
c11_serializer__write_mark(s, ']');
// varnames (as cstr via py_name2str) // varnames (as cstr via py_name2str)
c11_serializer__write_i32(s, co->varnames.length); c11_serializer__write_i32(s, co->varnames.length);
c11_serializer__write_mark(s, '[');
for(int i = 0; i < co->varnames.length; i++) { for(int i = 0; i < co->varnames.length; i++) {
py_Name name = c11__getitem(py_Name, &co->varnames, i); py_Name name = c11__getitem(py_Name, &co->varnames, i);
c11_serializer__write_cstr(s, py_name2str(name)); c11_serializer__write_cstr(s, py_name2str(name));
} }
c11_serializer__write_mark(s, ']');
// names (as cstr via py_name2str) // names (as cstr via py_name2str)
c11_serializer__write_i32(s, co->names.length); c11_serializer__write_i32(s, co->names.length);
c11_serializer__write_mark(s, '[');
for(int i = 0; i < co->names.length; i++) { for(int i = 0; i < co->names.length; i++) {
py_Name name = c11__getitem(py_Name, &co->names, i); py_Name name = c11__getitem(py_Name, &co->names, i);
c11_serializer__write_cstr(s, py_name2str(name)); c11_serializer__write_cstr(s, py_name2str(name));
} }
c11_serializer__write_mark(s, ']');
// nlocals // nlocals
c11_serializer__write_i32(s, co->nlocals); c11_serializer__write_i32(s, co->nlocals);
@ -149,14 +159,19 @@ static void CodeObject__serialize(c11_serializer* s,
// blocks // blocks
_Static_assert(sizeof(CodeBlock) == sizeof(int32_t) * 5, ""); _Static_assert(sizeof(CodeBlock) == sizeof(int32_t) * 5, "");
c11_serializer__write_i32(s, co->blocks.length); c11_serializer__write_i32(s, co->blocks.length);
c11_serializer__write_mark(s, '[');
c11_serializer__write_bytes(s, co->blocks.data, co->blocks.length * sizeof(CodeBlock)); c11_serializer__write_bytes(s, co->blocks.data, co->blocks.length * sizeof(CodeBlock));
c11_serializer__write_mark(s, ']');
// func_decls // func_decls
c11_serializer__write_i32(s, co->func_decls.length); c11_serializer__write_i32(s, co->func_decls.length);
c11_serializer__write_mark(s, '[');
for(int i = 0; i < co->func_decls.length; i++) { for(int i = 0; i < co->func_decls.length; i++) {
const FuncDecl* decl = c11__getitem(FuncDecl_, &co->func_decls, i); const FuncDecl* decl = c11__getitem(FuncDecl_, &co->func_decls, i);
FuncDecl__serialize(s, decl, co->src); FuncDecl__serialize(s, decl, co->src);
c11_serializer__write_mark(s, '|');
} }
c11_serializer__write_mark(s, ']');
// start_line, end_line // start_line, end_line
c11_serializer__write_i32(s, co->start_line); c11_serializer__write_i32(s, co->start_line);
@ -190,54 +205,70 @@ static CodeObject CodeObject__deserialize(c11_deserializer* d, const char* filen
// codes // codes
int codes_len = c11_deserializer__read_i32(d); int codes_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
c11_vector__extend(&co.codes, c11_vector__extend(&co.codes,
c11_deserializer__read_bytes(d, codes_len * sizeof(Bytecode)), c11_deserializer__read_bytes(d, codes_len * sizeof(Bytecode)),
codes_len); codes_len);
c11_deserializer__consume_mark(d, ']');
// codes_ex // codes_ex
int codes_ex_len = c11_deserializer__read_i32(d); int codes_ex_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
c11_vector__extend(&co.codes_ex, c11_vector__extend(&co.codes_ex,
c11_deserializer__read_bytes(d, codes_ex_len * sizeof(BytecodeEx)), c11_deserializer__read_bytes(d, codes_ex_len * sizeof(BytecodeEx)),
codes_ex_len); codes_ex_len);
c11_deserializer__consume_mark(d, ']');
// consts // consts
int consts_len = c11_deserializer__read_i32(d); int consts_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
for(int i = 0; i < consts_len; i++) { for(int i = 0; i < consts_len; i++) {
py_Ref p_val = c11_vector__emplace(&co.consts); py_Ref p_val = c11_vector__emplace(&co.consts);
TValue__deserialize(d, p_val); TValue__deserialize(d, p_val);
} }
c11_deserializer__consume_mark(d, ']');
// varnames // varnames
int varnames_len = c11_deserializer__read_i32(d); int varnames_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
for(int i = 0; i < varnames_len; i++) { for(int i = 0; i < varnames_len; i++) {
const char* s = c11_deserializer__read_cstr(d); const char* s = c11_deserializer__read_cstr(d);
py_Name n = py_name(s); py_Name n = py_name(s);
c11_vector__push(py_Name, &co.varnames, n); c11_vector__push(py_Name, &co.varnames, n);
c11_smallmap_n2d__set(&co.varnames_inv, n, i); c11_smallmap_n2d__set(&co.varnames_inv, n, i);
} }
c11_deserializer__consume_mark(d, ']');
// names // names
int names_len = c11_deserializer__read_i32(d); int names_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
for(int i = 0; i < names_len; i++) { for(int i = 0; i < names_len; i++) {
const char* s = c11_deserializer__read_cstr(d); const char* s = c11_deserializer__read_cstr(d);
py_Name n = py_name(s); py_Name n = py_name(s);
c11_vector__push(py_Name, &co.names, n); c11_vector__push(py_Name, &co.names, n);
c11_smallmap_n2d__set(&co.names_inv, n, i); c11_smallmap_n2d__set(&co.names_inv, n, i);
} }
c11_deserializer__consume_mark(d, ']');
// nlocals // nlocals
co.nlocals = c11_deserializer__read_i32(d); co.nlocals = c11_deserializer__read_i32(d);
// blocks // blocks
int blocks_len = c11_deserializer__read_i32(d); int blocks_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
c11_vector__extend(&co.blocks, c11_vector__extend(&co.blocks,
c11_deserializer__read_bytes(d, blocks_len * sizeof(CodeBlock)), c11_deserializer__read_bytes(d, blocks_len * sizeof(CodeBlock)),
blocks_len); blocks_len);
c11_deserializer__consume_mark(d, ']');
// func_decls // func_decls
int func_decls_len = c11_deserializer__read_i32(d); int func_decls_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
for(int i = 0; i < func_decls_len; i++) { for(int i = 0; i < func_decls_len; i++) {
FuncDecl_ decl = FuncDecl__deserialize(d, src); FuncDecl_ decl = FuncDecl__deserialize(d, src);
c11_vector__push(FuncDecl_, &co.func_decls, decl); c11_vector__push(FuncDecl_, &co.func_decls, decl);
c11_deserializer__consume_mark(d, '|');
} }
c11_deserializer__consume_mark(d, ']');
// start_line, end_line // start_line, end_line
co.start_line = c11_deserializer__read_i32(d); co.start_line = c11_deserializer__read_i32(d);
@ -251,20 +282,26 @@ static void FuncDecl__serialize(c11_serializer* s,
const FuncDecl* decl, const FuncDecl* decl,
const struct SourceData* parent_src) { const struct SourceData* parent_src) {
// CodeObject (embedded) // CodeObject (embedded)
c11_serializer__write_mark(s, '{');
CodeObject__serialize(s, &decl->code, parent_src); CodeObject__serialize(s, &decl->code, parent_src);
c11_serializer__write_mark(s, '}');
// args // args
c11_serializer__write_i32(s, decl->args.length); c11_serializer__write_i32(s, decl->args.length);
c11_serializer__write_mark(s, '[');
c11_serializer__write_bytes(s, decl->args.data, decl->args.length * sizeof(int32_t)); c11_serializer__write_bytes(s, decl->args.data, decl->args.length * sizeof(int32_t));
c11_serializer__write_mark(s, ']');
// kwargs // kwargs
c11_serializer__write_i32(s, decl->kwargs.length); c11_serializer__write_i32(s, decl->kwargs.length);
c11_serializer__write_mark(s, '[');
for(int i = 0; i < decl->kwargs.length; i++) { for(int i = 0; i < decl->kwargs.length; i++) {
FuncDeclKwArg* kw = c11__at(FuncDeclKwArg, &decl->kwargs, i); FuncDeclKwArg* kw = c11__at(FuncDeclKwArg, &decl->kwargs, i);
c11_serializer__write_i32(s, kw->index); c11_serializer__write_i32(s, kw->index);
c11_serializer__write_cstr(s, py_name2str(kw->key)); c11_serializer__write_cstr(s, py_name2str(kw->key));
TValue__serialize(s, &kw->value); TValue__serialize(s, &kw->value);
} }
c11_serializer__write_mark(s, ']');
// starred_arg, starred_kwarg // starred_arg, starred_kwarg
c11_serializer__write_i32(s, decl->starred_arg); c11_serializer__write_i32(s, decl->starred_arg);
@ -293,16 +330,21 @@ static FuncDecl_ FuncDecl__deserialize(c11_deserializer* d, SourceData_ embedded
c11_smallmap_n2d__ctor(&self->kw_to_index); c11_smallmap_n2d__ctor(&self->kw_to_index);
// CodeObject (embedded) // CodeObject (embedded)
c11_deserializer__consume_mark(d, '{');
self->code = CodeObject__deserialize(d, NULL, embedded_src); self->code = CodeObject__deserialize(d, NULL, embedded_src);
c11_deserializer__consume_mark(d, '}');
// args // args
int args_len = c11_deserializer__read_i32(d); int args_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
c11_vector__extend(&self->args, c11_vector__extend(&self->args,
c11_deserializer__read_bytes(d, args_len * sizeof(int32_t)), c11_deserializer__read_bytes(d, args_len * sizeof(int32_t)),
args_len); args_len);
c11_deserializer__consume_mark(d, ']');
// kwargs // kwargs
int kwargs_len = c11_deserializer__read_i32(d); int kwargs_len = c11_deserializer__read_i32(d);
c11_deserializer__consume_mark(d, '[');
for(int i = 0; i < kwargs_len; i++) { for(int i = 0; i < kwargs_len; i++) {
FuncDeclKwArg* kw = c11_vector__emplace(&self->kwargs); FuncDeclKwArg* kw = c11_vector__emplace(&self->kwargs);
kw->index = c11_deserializer__read_i32(d); kw->index = c11_deserializer__read_i32(d);
@ -311,6 +353,8 @@ static FuncDecl_ FuncDecl__deserialize(c11_deserializer* d, SourceData_ embedded
TValue__deserialize(d, &kw->value); TValue__deserialize(d, &kw->value);
c11_smallmap_n2d__set(&self->kw_to_index, kw->key, kw->index); c11_smallmap_n2d__set(&self->kw_to_index, kw->key, kw->index);
} }
c11_deserializer__consume_mark(d, ']');
// starred_arg // starred_arg
self->starred_arg = c11_deserializer__read_i32(d); self->starred_arg = c11_deserializer__read_i32(d);
// starred_kwarg // starred_kwarg