Protocol Buffers - Google's data interchange format (grpc依赖) https://developers.google.com/protocol-buffers/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

785 lines
24 KiB

/*
* upb - a minimalist implementation of protocol buffers.
*
* Copyright (c) 2008-2009 Google Inc. See LICENSE for details.
* Author: Josh Haberman <jhaberman@gmail.com>
*/
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include "upb/def.h"
#define alignof(t) offsetof(struct { char c; t x; }, x)
void upb_deflist_init(upb_deflist *l) {
l->size = 8;
l->defs = malloc(l->size * sizeof(void*));
l->len = 0;
}
void upb_deflist_uninit(upb_deflist *l) {
for(uint32_t i = 0; i < l->len; i++) upb_def_unref(l->defs[i]);
free(l->defs);
}
void upb_deflist_push(upb_deflist *l, upb_def *d) {
if(l->len == l->size) {
l->size *= 2;
l->defs = realloc(l->defs, l->size * sizeof(void*));
}
l->defs[l->len++] = d;
}
/* upb_def ********************************************************************/
static void upb_msgdef_free(upb_msgdef *m);
static void upb_enumdef_free(upb_enumdef *e);
static void upb_unresolveddef_free(struct _upb_unresolveddef *u);
bool upb_def_ismutable(upb_def *def) { return def->symtab == NULL; }
bool upb_def_setfqname(upb_def *def, const char *fqname) {
assert(upb_def_ismutable(def));
free(def->fqname);
def->fqname = strdup(fqname);
return true; // TODO: check for acceptable characters.
}
static void upb_def_free(upb_def *def) {
switch (def->type) {
case UPB_DEF_MSG: upb_msgdef_free(upb_downcast_msgdef(def)); break;
case UPB_DEF_ENUM: upb_enumdef_free(upb_downcast_enumdef(def)); break;
case UPB_DEF_UNRESOLVED:
upb_unresolveddef_free(upb_downcast_unresolveddef(def)); break;
default:
assert(false);
}
}
upb_def *upb_def_dup(upb_def *def) {
switch (def->type) {
case UPB_DEF_MSG: return UPB_UPCAST(upb_msgdef_dup(upb_downcast_msgdef(def)));
case UPB_DEF_ENUM: return UPB_UPCAST(upb_enumdef_dup(upb_downcast_enumdef(def)));
default: assert(false); return NULL;
}
}
// Prior to being in a symtab, the def's refcount controls the lifetime of the
// def itself. If the refcount falls to zero, the def is deleted. Once the
// def belongs to a symtab, the def is owned by the symtab and its refcount
// determines whether the def owns a ref on the symtab or not.
void upb_def_ref(upb_def *def) {
if (upb_atomic_ref(&def->refcount) && def->symtab)
upb_symtab_ref(def->symtab);
}
static void upb_def_movetosymtab(upb_def *d, upb_symtab *s) {
assert(upb_atomic_read(&d->refcount) > 0);
d->symtab = s;
upb_symtab_ref(s);
upb_msgdef *m = upb_dyncast_msgdef(d);
if (m) upb_inttable_compact(&m->itof);
}
void upb_def_unref(upb_def *def) {
if (!def) return;
if (upb_atomic_unref(&def->refcount)) {
if (def->symtab) {
upb_symtab_unref(def->symtab);
// Def might be deleted now.
} else {
upb_def_free(def);
}
}
}
static void upb_def_init(upb_def *def, upb_deftype_t type) {
def->type = type;
def->fqname = NULL;
def->symtab = NULL;
upb_atomic_init(&def->refcount, 1);
upb_stream: all callbacks registered ahead-of-time. This is a significant change to the upb_stream protocol, and should hopefully be the last significant change. All callbacks are now registered ahead-of-time instead of having delegated callbacks registered at runtime, which makes it much easier to aggressively optimize ahead-of-time (like with a JIT). Other impacts of this change: - You no longer need to have loaded descriptor.proto as a upb_def to load other descriptors! This means the special-case code we used for bootstrapping is no longer necessary, and we no longer need to link the descriptor for descriptor.proto into upb. - A client can now register any upb_value as what will be delivered to their value callback, not just a upb_fielddef*. This should allow for other clients to get more bang out of the streaming decoder. This change unfortunately causes a bit of a performance regression -- I think largely due to highly suboptimal code that GCC generates when structs are returned by value. See: http://blog.reverberate.org/2011/03/19/when-a-compilers-slow-code-actually-bites-you/ On the other hand, once we have a JIT this should no longer matter. Performance numbers: plain.parsestream_googlemessage1.upb_table: 374 -> 396 (5.88) plain.parsestream_googlemessage2.upb_table: 616 -> 449 (-27.11) plain.parsetostruct_googlemessage1.upb_table_byref: 268 -> 269 (0.37) plain.parsetostruct_googlemessage1.upb_table_byval: 215 -> 204 (-5.12) plain.parsetostruct_googlemessage2.upb_table_byref: 307 -> 281 (-8.47) plain.parsetostruct_googlemessage2.upb_table_byval: 297 -> 272 (-8.42) omitfp.parsestream_googlemessage1.upb_table: 423 -> 410 (-3.07) omitfp.parsestream_googlemessage2.upb_table: 679 -> 483 (-28.87) omitfp.parsetostruct_googlemessage1.upb_table_byref: 287 -> 282 (-1.74) omitfp.parsetostruct_googlemessage1.upb_table_byval: 226 -> 219 (-3.10) omitfp.parsetostruct_googlemessage2.upb_table_byref: 315 -> 298 (-5.40) omitfp.parsetostruct_googlemessage2.upb_table_byval: 297 -> 287 (-3.37)
14 years ago
}
static void upb_def_uninit(upb_def *def) {
free(def->fqname);
}
/* upb_unresolveddef **********************************************************/
// Unresolved defs are used as temporary placeholders for a def whose name has
// not been resolved yet. During the name resolution step, all unresolved defs
// are replaced with pointers to the actual def being referenced.
typedef struct _upb_unresolveddef {
upb_def base;
} upb_unresolveddef;
// Is passed a ref on the string.
static upb_unresolveddef *upb_unresolveddef_new(const char *str) {
upb_unresolveddef *def = malloc(sizeof(*def));
upb_def_init(&def->base, UPB_DEF_UNRESOLVED);
def->base.fqname = strdup(str);
return def;
}
static void upb_unresolveddef_free(struct _upb_unresolveddef *def) {
upb_def_uninit(&def->base);
free(def);
}
/* upb_enumdef ****************************************************************/
upb_enumdef *upb_enumdef_new() {
upb_enumdef *e = malloc(sizeof(*e));
upb_def_init(&e->base, UPB_DEF_ENUM);
upb_strtable_init(&e->ntoi, 0, sizeof(upb_ntoi_ent));
upb_inttable_init(&e->iton, 0, sizeof(upb_iton_ent));
return e;
}
static void upb_enumdef_free(upb_enumdef *e) {
upb_enum_iter i;
for(i = upb_enum_begin(e); !upb_enum_done(i); i = upb_enum_next(e, i)) {
// Frees the ref taken when the string was parsed.
free(upb_enum_iter_name(i));
}
upb_strtable_free(&e->ntoi);
upb_inttable_free(&e->iton);
upb_def_uninit(&e->base);
free(e);
}
upb_enumdef *upb_enumdef_dup(upb_enumdef *e) {
upb_enumdef *new_e = upb_enumdef_new();
upb_enum_iter i;
for(i = upb_enum_begin(e); !upb_enum_done(i); i = upb_enum_next(e, i)) {
assert(upb_enumdef_addval(new_e, upb_enum_iter_name(i),
upb_enum_iter_number(i)));
}
return new_e;
}
bool upb_enumdef_addval(upb_enumdef *e, char *name, int32_t num) {
if (upb_enumdef_iton(e, num) || upb_enumdef_ntoi(e, name, NULL))
return false;
upb_iton_ent ent = {0, strdup(name)};
upb_strtable_insert(&e->ntoi, name, &num);
upb_inttable_insert(&e->iton, num, &ent);
return true;
}
void upb_enumdef_setdefault(upb_enumdef *e, int32_t val) {
assert(upb_def_ismutable(UPB_UPCAST(e)));
e->defaultval = val;
}
upb_enum_iter upb_enum_begin(upb_enumdef *e) {
// We could iterate over either table here; the choice is arbitrary.
return upb_inttable_begin(&e->iton);
}
upb_enum_iter upb_enum_next(upb_enumdef *e, upb_enum_iter iter) {
return upb_inttable_next(&e->iton, iter);
}
const char *upb_enumdef_iton(upb_enumdef *def, int32_t num) {
upb_iton_ent *e = upb_inttable_fastlookup(&def->iton, num, sizeof(*e));
return e ? e->str : NULL;
}
bool upb_enumdef_ntoil(upb_enumdef *def, char *name, size_t len, int32_t *num) {
upb_ntoi_ent *e = upb_strtable_lookupl(&def->ntoi, name, len);
if (!e) return false;
if (num) *num = e->value;
return true;
}
bool upb_enumdef_ntoi(upb_enumdef *e, char *name, int32_t *num) {
return upb_enumdef_ntoil(e, name, strlen(name), num);
}
/* upb_fielddef ***************************************************************/
upb_fielddef *upb_fielddef_new() {
upb_fielddef *f = malloc(sizeof(*f));
f->msgdef = NULL;
f->def = NULL;
upb_atomic_init(&f->refcount, 1);
f->finalized = false;
f->type = 0;
f->label = UPB_LABEL(OPTIONAL);
f->hasbit = 0;
f->offset = 0;
f->number = 0; // not a valid field number.
f->hasdefault = false;
f->name = NULL;
f->accessor = NULL;
upb_value_setfielddef(&f->fval, f);
return f;
}
static void upb_fielddef_free(upb_fielddef *f) {
if (upb_isstring(f)) {
free(upb_value_getptr(f->defaultval));
}
if (f->def) {
// We own a ref on the subdef iff we are not part of a msgdef.
assert((f->msgdef == NULL) == (upb_dyncast_unresolveddef(f->def) != NULL));
if (f->msgdef == NULL) upb_def_unref(f->def);
}
free(f->name);
free(f);
}
void upb_fielddef_ref(upb_fielddef *f) {
// TODO.
(void)f;
}
void upb_fielddef_unref(upb_fielddef *f) {
// TODO.
(void)f;
if (!f) return;
if (upb_atomic_unref(&f->refcount)) {
if (f->msgdef) {
upb_msgdef_unref(f->msgdef);
// fielddef might be deleted now.
} else {
upb_fielddef_free(f);
}
}
}
upb_fielddef *upb_fielddef_dup(upb_fielddef *f) {
upb_fielddef *newf = upb_fielddef_new();
newf->msgdef = f->msgdef;
newf->type = f->type;
newf->label = f->label;
newf->number = f->number;
newf->name = f->name;
upb_fielddef_settypename(newf, f->def->fqname);
return f;
}
bool upb_fielddef_ismutable(upb_fielddef *f) {
return !f->msgdef || upb_def_ismutable(UPB_UPCAST(f->msgdef));
}
upb_def *upb_fielddef_subdef(upb_fielddef *f) {
if (upb_hassubdef(f) && !upb_fielddef_ismutable(f))
return f->def;
else
return NULL;
}
static bool upb_fielddef_resolve(upb_fielddef *f, upb_def *def, upb_status *s) {
assert(upb_dyncast_unresolveddef(f->def));
upb_def_unref(f->def);
f->def = def;
if (f->type == UPB_TYPE(ENUM)) {
// Resolve the enum's default from a string to an integer.
char *str = upb_value_getptr(f->defaultval);
assert(str); // Should point to either a real default or the empty string.
upb_enumdef *e = upb_downcast_enumdef(f->def);
int32_t val = 0;
if (str[0] == '\0') {
upb_value_setint32(&f->defaultval, e->defaultval);
} else {
bool success = upb_enumdef_ntoi(e, str, &val);
free(str);
if (!success) {
upb_status_setf(s, UPB_ERROR, "Default enum value (%s) is not a "
"member of the enum", str);
return false;
}
upb_value_setint32(&f->defaultval, val);
}
}
return true;
}
bool upb_fielddef_setnumber(upb_fielddef *f, int32_t number) {
assert(f->msgdef == NULL);
f->number = number;
return true;
}
bool upb_fielddef_setname(upb_fielddef *f, const char *name) {
assert(f->msgdef == NULL);
free(f->name);
f->name = strdup(name);
return true;
}
bool upb_fielddef_settype(upb_fielddef *f, uint8_t type) {
assert(!f->finalized);
f->type = type;
return true;
}
bool upb_fielddef_setlabel(upb_fielddef *f, uint8_t label) {
assert(!f->finalized);
f->label = label;
return true;
}
void upb_fielddef_setdefault(upb_fielddef *f, upb_value value) {
assert(!f->finalized);
// TODO: string ownership?
f->defaultval = value;
upb_stream: all callbacks registered ahead-of-time. This is a significant change to the upb_stream protocol, and should hopefully be the last significant change. All callbacks are now registered ahead-of-time instead of having delegated callbacks registered at runtime, which makes it much easier to aggressively optimize ahead-of-time (like with a JIT). Other impacts of this change: - You no longer need to have loaded descriptor.proto as a upb_def to load other descriptors! This means the special-case code we used for bootstrapping is no longer necessary, and we no longer need to link the descriptor for descriptor.proto into upb. - A client can now register any upb_value as what will be delivered to their value callback, not just a upb_fielddef*. This should allow for other clients to get more bang out of the streaming decoder. This change unfortunately causes a bit of a performance regression -- I think largely due to highly suboptimal code that GCC generates when structs are returned by value. See: http://blog.reverberate.org/2011/03/19/when-a-compilers-slow-code-actually-bites-you/ On the other hand, once we have a JIT this should no longer matter. Performance numbers: plain.parsestream_googlemessage1.upb_table: 374 -> 396 (5.88) plain.parsestream_googlemessage2.upb_table: 616 -> 449 (-27.11) plain.parsetostruct_googlemessage1.upb_table_byref: 268 -> 269 (0.37) plain.parsetostruct_googlemessage1.upb_table_byval: 215 -> 204 (-5.12) plain.parsetostruct_googlemessage2.upb_table_byref: 307 -> 281 (-8.47) plain.parsetostruct_googlemessage2.upb_table_byval: 297 -> 272 (-8.42) omitfp.parsestream_googlemessage1.upb_table: 423 -> 410 (-3.07) omitfp.parsestream_googlemessage2.upb_table: 679 -> 483 (-28.87) omitfp.parsetostruct_googlemessage1.upb_table_byref: 287 -> 282 (-1.74) omitfp.parsetostruct_googlemessage1.upb_table_byval: 226 -> 219 (-3.10) omitfp.parsetostruct_googlemessage2.upb_table_byref: 315 -> 298 (-5.40) omitfp.parsetostruct_googlemessage2.upb_table_byval: 297 -> 287 (-3.37)
14 years ago
}
void upb_fielddef_setfval(upb_fielddef *f, upb_value fval) {
assert(!f->finalized);
// TODO: string ownership?
f->fval = fval;
}
void upb_fielddef_setaccessor(upb_fielddef *f, struct _upb_accessor_vtbl *vtbl) {
assert(!f->finalized);
f->accessor = vtbl;
}
bool upb_fielddef_settypename(upb_fielddef *f, const char *name) {
upb_def_unref(f->def);
f->def = UPB_UPCAST(upb_unresolveddef_new(name));
return true;
}
// Returns an ordering of fields based on:
// 1. value size (small to large).
// 2. field number.
static int upb_fielddef_cmpval(const void *_f1, const void *_f2) {
upb_fielddef *f1 = *(void**)_f1;
upb_fielddef *f2 = *(void**)_f2;
size_t size1 = upb_types[f1->type].size;
size_t size2 = upb_types[f2->type].size;
if (size1 != size2) return size1 - size2;
// Otherwise return in number order.
return f1->number - f2->number;
}
// Returns an ordering of all fields based on:
// 1. required/optional (required fields first).
// 2. field number
static int upb_fielddef_cmphasbit(const void *_f1, const void *_f2) {
upb_fielddef *f1 = *(void**)_f1;
upb_fielddef *f2 = *(void**)_f2;
size_t req1 = f1->label == UPB_LABEL(REQUIRED);
size_t req2 = f2->label == UPB_LABEL(REQUIRED);
if (req1 != req2) return req1 - req2;
// Otherwise return in number order.
return f1->number - f2->number;
}
/* upb_msgdef *****************************************************************/
upb_msgdef *upb_msgdef_new() {
upb_msgdef *m = malloc(sizeof(*m));
upb_def_init(&m->base, UPB_DEF_MSG);
upb_inttable_init(&m->itof, 4, sizeof(upb_itof_ent));
upb_strtable_init(&m->ntof, 4, sizeof(upb_ntof_ent));
m->size = 0;
m->hasbit_bytes = 0;
m->extstart = 0;
m->extend = 0;
return m;
}
static void upb_msgdef_free(upb_msgdef *m) {
upb_msg_iter i;
for(i = upb_msg_begin(m); !upb_msg_done(i); i = upb_msg_next(m, i))
upb_fielddef_free(upb_msg_iter_field(i));
upb_strtable_free(&m->ntof);
upb_inttable_free(&m->itof);
upb_def_uninit(&m->base);
free(m);
}
upb_msgdef *upb_msgdef_dup(upb_msgdef *m) {
upb_msgdef *newm = upb_msgdef_new();
newm->size = m->size;
newm->hasbit_bytes = m->hasbit_bytes;
newm->extstart = m->extstart;
newm->extend = m->extend;
upb_msg_iter i;
for(i = upb_msg_begin(m); !upb_msg_done(i); i = upb_msg_next(m, i)) {
upb_msgdef_addfield(newm, upb_fielddef_dup(upb_msg_iter_field(i)));
}
return newm;
}
void upb_msgdef_setsize(upb_msgdef *m, uint16_t size) {
assert(upb_def_ismutable(UPB_UPCAST(m)));
m->size = size;
}
void upb_msgdef_sethasbit_bytes(upb_msgdef *m, uint16_t bytes) {
assert(upb_def_ismutable(UPB_UPCAST(m)));
m->hasbit_bytes = bytes;
}
bool upb_msgdef_setextrange(upb_msgdef *m, uint32_t start, uint32_t end) {
assert(upb_def_ismutable(UPB_UPCAST(m)));
if (start == 0 && end == 0) {
// Clearing the extension range -- ok to fall through.
} else if (start >= end || start < 1 || end > UPB_MAX_FIELDNUMBER) {
return false;
}
m->extstart = start;
m->extend = start;
return true;
}
bool upb_msgdef_addfields(upb_msgdef *m, upb_fielddef **fields, int n) {
// Check constraints for all fields before performing any action.
for (int i = 0; i < n; i++) {
upb_fielddef *f = fields[i];
assert(upb_atomic_read(&f->refcount) > 0);
if (f->name == NULL || f->number == 0 ||
upb_msgdef_itof(m, f->number) || upb_msgdef_ntof(m, f->name))
return false;
}
// Constraint checks ok, perform the action.
for (int i = 0; i < n; i++) {
upb_fielddef *f = fields[i];
upb_msgdef_ref(m);
assert(f->msgdef == NULL);
f->msgdef = m;
upb_itof_ent itof_ent = {0, f};
upb_inttable_insert(&m->itof, f->number, &itof_ent);
upb_strtable_insert(&m->ntof, f->name, &f);
}
return true;
}
static int upb_div_round_up(int numerator, int denominator) {
/* cf. http://stackoverflow.com/questions/17944/how-to-round-up-the-result-of-integer-division */
return numerator > 0 ? (numerator - 1) / denominator + 1 : 0;
}
void upb_msgdef_layout(upb_msgdef *m) {
// Create an ordering over the fields, but only include fields with accessors.
upb_fielddef **sorted_fields =
malloc(sizeof(upb_fielddef*) * upb_msgdef_numfields(m));
int n = 0;
upb_msg_iter i;
for (i = upb_msg_begin(m); !upb_msg_done(i); i = upb_msg_next(m, i)) {
upb_fielddef *f = upb_msg_iter_field(i);
if (f->accessor) sorted_fields[n++] = f;
}
m->hasbit_bytes = upb_div_round_up(n, 8);
m->size = m->hasbit_bytes; // + header_size?
// Assign hasbits.
qsort(sorted_fields, n, sizeof(*sorted_fields), upb_fielddef_cmphasbit);
for (int i = 0; i < n; i++) {
upb_fielddef *f = sorted_fields[i];
f->hasbit = i;
}
// Assign value offsets.
qsort(sorted_fields, n, sizeof(*sorted_fields), upb_fielddef_cmpval);
size_t max_align = 0;
for (int i = 0; i < n; i++) {
upb_fielddef *f = sorted_fields[i];
const upb_type_info *type_info = &upb_types[f->type];
size_t size = type_info->size;
size_t align = type_info->align;
if (upb_isseq(f)) {
size = sizeof(void*);
align = alignof(void*);
}
// General alignment rules are: each member must be at an address that is a
// multiple of that type's alignment. Also, the size of the structure as a
// whole must be a multiple of the greatest alignment of any member.
f->offset = upb_align_up(m->size, align);
m->size = f->offset + size;
max_align = UPB_MAX(max_align, align);
}
if (max_align > 0) m->size = upb_align_up(m->size, max_align);
free(sorted_fields);
}
upb_msg_iter upb_msg_begin(upb_msgdef *m) {
return upb_inttable_begin(&m->itof);
}
upb_msg_iter upb_msg_next(upb_msgdef *m, upb_msg_iter iter) {
return upb_inttable_next(&m->itof, iter);
}
/* upb_symtab *****************************************************************/
typedef struct {
upb_def *def;
} upb_symtab_ent;
// Given a symbol and the base symbol inside which it is defined, find the
// symbol's definition in t.
static upb_symtab_ent *upb_resolve(upb_strtable *t,
const char *base, const char *sym) {
if(strlen(sym) == 0) return NULL;
if(sym[0] == UPB_SYMBOL_SEPARATOR) {
// Symbols starting with '.' are absolute, so we do a single lookup.
// Slice to omit the leading '.'
return upb_strtable_lookup(t, sym + 1);
} else {
// Remove components from base until we find an entry or run out.
// TODO: This branch is totally broken, but currently not used.
(void)base;
assert(false);
return NULL;
}
}
static void _upb_symtab_free(upb_strtable *t) {
upb_strtable_iter i;
upb_strtable_begin(&i, t);
for (; !upb_strtable_done(&i); upb_strtable_next(&i)) {
const upb_symtab_ent *e = upb_strtable_iter_value(&i);
assert(upb_atomic_read(&e->def->refcount) == 0);
upb_def_free(e->def);
}
upb_strtable_free(t);
}
static void upb_symtab_free(upb_symtab *s) {
_upb_symtab_free(&s->symtab);
for (uint32_t i = 0; i < s->olddefs.len; i++) {
upb_def *d = s->olddefs.defs[i];
assert(upb_atomic_read(&d->refcount) == 0);
upb_def_free(d);
}
upb_rwlock_destroy(&s->lock);
upb_deflist_uninit(&s->olddefs);
free(s);
}
void upb_symtab_unref(upb_symtab *s) {
if(s && upb_atomic_unref(&s->refcount)) {
upb_symtab_free(s);
}
}
upb_symtab *upb_symtab_new() {
upb_symtab *s = malloc(sizeof(*s));
upb_atomic_init(&s->refcount, 1);
upb_rwlock_init(&s->lock);
upb_strtable_init(&s->symtab, 16, sizeof(upb_symtab_ent));
upb_deflist_init(&s->olddefs);
return s;
}
upb_def **upb_symtab_getdefs(upb_symtab *s, int *count, upb_deftype_t type) {
upb_rwlock_rdlock(&s->lock);
int total = upb_strtable_count(&s->symtab);
// We may only use part of this, depending on how many symbols are of the
// correct type.
upb_def **defs = malloc(sizeof(*defs) * total);
upb_strtable_iter iter;
upb_strtable_begin(&iter, &s->symtab);
int i = 0;
for(; !upb_strtable_done(&iter); upb_strtable_next(&iter)) {
const upb_symtab_ent *e = upb_strtable_iter_value(&iter);
upb_def *def = e->def;
assert(def);
if(type == UPB_DEF_ANY || def->type == type)
defs[i++] = def;
}
upb_rwlock_unlock(&s->lock);
*count = i;
for(i = 0; i < *count; i++) upb_def_ref(defs[i]);
return defs;
}
upb_def *upb_symtab_lookup(upb_symtab *s, const char *sym) {
upb_rwlock_rdlock(&s->lock);
upb_symtab_ent *e = upb_strtable_lookup(&s->symtab, sym);
upb_def *ret = NULL;
if(e) {
ret = e->def;
upb_def_ref(ret);
}
upb_rwlock_unlock(&s->lock);
return ret;
}
upb_def *upb_symtab_resolve(upb_symtab *s, const char *base, const char *sym) {
upb_rwlock_rdlock(&s->lock);
upb_symtab_ent *e = upb_resolve(&s->symtab, base, sym);
upb_def *ret = NULL;
if(e) {
ret = e->def;
upb_def_ref(ret);
}
upb_rwlock_unlock(&s->lock);
return ret;
}
bool upb_symtab_dfs(upb_def *def, upb_def **open_defs, int n,
upb_strtable *addtab) {
// This linear search makes the DFS O(n^2) in the length of the paths.
// Could make this O(n) with a hash table, but n is small.
for (int i = 0; i < n; i++) {
if (def == open_defs[i]) return false;
}
bool needcopy = false;
upb_msgdef *m = upb_dyncast_msgdef(def);
if (m) {
upb_msg_iter i;
open_defs[n++] = def;
for(i = upb_msg_begin(m); !upb_msg_done(i); i = upb_msg_next(m, i)) {
upb_fielddef *f = upb_msg_iter_field(i);
if (!upb_hassubdef(f)) continue;
needcopy |= upb_symtab_dfs(f->def, open_defs, n, addtab);
}
}
bool replacing = (upb_strtable_lookup(addtab, m->base.fqname) != NULL);
if (needcopy && !replacing) {
upb_symtab_ent e = {upb_def_dup(def)};
//fprintf(stderr, "Replacing def: %p\n", e.def);
upb_strtable_insert(addtab, def->fqname, &e);
replacing = true;
}
return replacing;
}
bool upb_symtab_add(upb_symtab *s, upb_def **defs, int n, upb_status *status) {
upb_rwlock_wrlock(&s->lock);
// Add all defs to a table for resolution.
upb_strtable addtab;
upb_strtable_init(&addtab, n, sizeof(upb_symtab_ent));
for (int i = 0; i < n; i++) {
upb_def *def = defs[i];
if (upb_strtable_lookup(&addtab, def->fqname)) {
upb_status_setf(status, UPB_ERROR,
"Conflicting defs named '%s'", def->fqname);
upb_strtable_free(&addtab);
return false;
}
upb_strtable_insert(&addtab, def->fqname, &def);
}
// All existing defs that can reach defs that are being replaced must
// themselves be replaced with versions that will point to the new defs.
// Do a DFS -- any path that finds a new def must replace all ancestors.
upb_strtable *symtab = &s->symtab;
upb_strtable_iter i;
upb_strtable_begin(&i, symtab);
for(; !upb_strtable_done(&i); upb_strtable_next(&i)) {
upb_def *open_defs[UPB_MAX_TYPE_DEPTH];
const upb_symtab_ent *e = upb_strtable_iter_value(&i);
upb_symtab_dfs(e->def, open_defs, 0, &addtab);
}
// Resolve all refs.
upb_strtable_begin(&i, &addtab);
for(; !upb_strtable_done(&i); upb_strtable_next(&i)) {
const upb_symtab_ent *e = upb_strtable_iter_value(&i);
upb_msgdef *m = upb_dyncast_msgdef(e->def);
if(!m) continue;
// Type names are resolved relative to the message in which they appear.
const char *base = m->base.fqname;
upb_msg_iter j;
for(j = upb_msg_begin(m); !upb_msg_done(j); j = upb_msg_next(m, j)) {
upb_fielddef *f = upb_msg_iter_field(j);
if (f->type == 0) {
upb_status_setf(status, UPB_ERROR, "Field type was not set.");
return false;
}
// Set default default if none was set explicitly.
if (!f->hasdefault) {
switch (upb_fielddef_type(f)) {
case UPB_TYPE(DOUBLE): upb_value_setdouble(&f->defaultval, 0); break;
case UPB_TYPE(FLOAT): upb_value_setfloat(&f->defaultval, 0); break;
case UPB_TYPE(UINT64):
case UPB_TYPE(FIXED64): upb_value_setuint64(&f->defaultval, 0); break;
case UPB_TYPE(INT64):
case UPB_TYPE(SFIXED64):
case UPB_TYPE(SINT64): upb_value_setint64(&f->defaultval, 0); break;
case UPB_TYPE(INT32):
case UPB_TYPE(SINT32):
case UPB_TYPE(SFIXED32): upb_value_setint32(&f->defaultval, 0); break;
case UPB_TYPE(UINT32):
case UPB_TYPE(FIXED32): upb_value_setuint32(&f->defaultval, 0); break;
case UPB_TYPE(BOOL): upb_value_setbool(&f->defaultval, false); break;
case UPB_TYPE(ENUM): // Will be resolved by upb_resolve().
case UPB_TYPE(STRING):
case UPB_TYPE(BYTES):
case UPB_TYPE(GROUP):
case UPB_TYPE(MESSAGE): break; // do nothing for now.
}
}
if (!upb_hassubdef(f)) continue; // No resolving necessary.
upb_downcast_unresolveddef(f->def); // Type check.
const char *name = f->def->fqname;
// Resolve from either the addtab (pending adds) or symtab (existing
// defs). If both exist, prefer the pending add, because it will be
// overwriting the existing def.
upb_symtab_ent *found;
if(!(found = upb_resolve(&addtab, base, name)) &&
!(found = upb_resolve(symtab, base, name))) {
upb_status_setf(status, UPB_ERROR, "could not resolve symbol '%s' "
"in context '%s'", name, base);
return false;
}
// Check the type of the found def.
upb_fieldtype_t expected = upb_issubmsg(f) ? UPB_DEF_MSG : UPB_DEF_ENUM;
//fprintf(stderr, "found: %p\n", found);
//fprintf(stderr, "found->def: %p\n", found->def);
//fprintf(stderr, "found->def->type: %d\n", found->def->type);
if(found->def->type != expected) {
upb_status_setf(status, UPB_ERROR, "Unexpected type");
return false;
}
if (!upb_fielddef_resolve(f, found->def, status)) return false;
}
}
// The defs in the transaction have been vetted, and can be moved to the
// symtab without causing errors.
upb_strtable_begin(&i, &addtab);
for(; !upb_strtable_done(&i); upb_strtable_next(&i)) {
const upb_symtab_ent *tmptab_e = upb_strtable_iter_value(&i);
upb_def_movetosymtab(tmptab_e->def, s);
upb_symtab_ent *symtab_e =
upb_strtable_lookup(&s->symtab, tmptab_e->def->fqname);
if(symtab_e) {
upb_deflist_push(&s->olddefs, symtab_e->def);
symtab_e->def = tmptab_e->def;
} else {
//fprintf(stderr, "Inserting def: %p\n", tmptab_e->def);
upb_strtable_insert(&s->symtab, tmptab_e->def->fqname, tmptab_e);
}
}
upb_strtable_free(&addtab);
upb_rwlock_unlock(&s->lock);
upb_symtab_gc(s);
return true;
}
void upb_symtab_gc(upb_symtab *s) {
(void)s;
// TODO.
}