Added support for unknown fields to upb_msg.

After this CL, upb passes all existing proto3 conformance tests.
However the conformance suite is missing a lot of cases and should
be fleshed out.

4 files changed, 45 insertions, 6 deletions

diff --git a/upb/decode.c b/upb/decode.c
index cd13125..fc0e644 100644
--- a/upb/decode.c
+++ b/upb/decode.c
@@ -130,9 +130,7 @@ static void upb_set32(void *msg, size_t ofs, uint32_t val) {
 
 static bool upb_append_unknown(upb_decstate *d, upb_decframe *frame,
                                const char *start) {
-  UPB_UNUSED(d);
-  UPB_UNUSED(frame);
-  UPB_UNUSED(start);
+  upb_msg_addunknown(frame->msg, start, d->ptr - start);
   return true;
 }
 
@@ -545,7 +543,9 @@ static bool upb_decode_field(upb_decstate *d, upb_decframe *frame) {
     }
   } else {
     CHK(field_number != 0);
-    return upb_skip_unknownfielddata(d, frame, field_number, wire_type);
+    CHK(upb_skip_unknownfielddata(d, frame, field_number, wire_type));
+    CHK(upb_append_unknown(d, frame, field_start));
+    return true;
   }
 }
 
diff --git a/upb/encode.c b/upb/encode.c
index a9f2c0d..bff8262 100644
--- a/upb/encode.c
+++ b/upb/encode.c
@@ -331,6 +331,8 @@ bool upb_encode_message(upb_encstate *e, const char *msg,
                         const upb_msglayout *m, size_t *size) {
   int i;
   size_t pre_len = e->limit - e->ptr;
+  const char *unknown;
+  size_t unknown_size;
 
   for (i = m->field_count - 1; i >= 0; i--) {
     const upb_msglayout_field *f = &m->fields[i];
@@ -357,6 +359,12 @@ bool upb_encode_message(upb_encstate *e, const char *msg,
     }
   }
 
+  unknown = upb_msg_getunknown(msg, &unknown_size);
+
+  if (unknown) {
+    upb_put_bytes(e, unknown, unknown_size);
+  }
+
   *size = (e->limit - e->ptr) - pre_len;
   return true;
 }
diff --git a/upb/msg.c b/upb/msg.c
index b70fe40..4125d4c 100644
--- a/upb/msg.c
+++ b/upb/msg.c
@@ -113,8 +113,12 @@ static upb_ctype_t upb_fieldtotabtype(upb_fieldtype_t type) {
 
 /* Used when a message is not extendable. */
 typedef struct {
-  /* TODO(haberman): add unknown fields. */
+  /* TODO(haberman): use pointer tagging so we we are slim when known unknown
+   * fields are not present. */
   upb_arena *arena;
+  char *unknown;
+  size_t unknown_len;
+  size_t unknown_size;
 } upb_msg_internal;
 
 /* Used when a message is extendable. */
@@ -141,6 +145,25 @@ static upb_msg_internal_withext *upb_msg_getinternalwithext(
   return VOIDPTR_AT(msg, -sizeof(upb_msg_internal_withext));
 }
 
+void upb_msg_addunknown(upb_msg *msg, const char *data, size_t len) {
+  upb_msg_internal* in = upb_msg_getinternal(msg);
+  if (len > in->unknown_size - in->unknown_len) {
+    upb_alloc *alloc = upb_arena_alloc(in->arena);
+    size_t need = in->unknown_size + len;
+    size_t newsize = UPB_MAX(in->unknown_size * 2, need);
+    in->unknown = upb_realloc(alloc, in->unknown, in->unknown_size, newsize);
+    in->unknown_size = newsize;
+  }
+  memcpy(in->unknown + in->unknown_len, data, len);
+  in->unknown_len += len;
+}
+
+const char *upb_msg_getunknown(upb_msg *msg, size_t *len) {
+  upb_msg_internal* in = upb_msg_getinternal(msg);
+  *len = in->unknown_len;
+  return in->unknown;
+}
+
 static const upb_msglayout_field *upb_msg_checkfield(int field_index,
                                                      const upb_msglayout *l) {
   UPB_ASSERT(field_index >= 0 && field_index < l->field_count);
@@ -165,6 +188,7 @@ static size_t upb_msg_sizeof(const upb_msglayout *l) {
 upb_msg *upb_msg_new(const upb_msglayout *l, upb_arena *a) {
   upb_alloc *alloc = upb_arena_alloc(a);
   void *mem = upb_malloc(alloc, upb_msg_sizeof(l));
+  upb_msg_internal *in;
   upb_msg *msg;
 
   if (!mem) {
@@ -177,7 +201,11 @@ upb_msg *upb_msg_new(const upb_msglayout *l, upb_arena *a) {
   memset(msg, 0, l->size);
 
   /* Initialize internal members. */
-  upb_msg_getinternal(msg)->arena = a;
+  in = upb_msg_getinternal(msg);
+  in->arena = a;
+  in->unknown = NULL;
+  in->unknown_len = 0;
+  in->unknown_size = 0;
 
   if (l->extendable) {
     upb_msg_getinternalwithext(msg, l)->extdict = NULL;
diff --git a/upb/msg.h b/upb/msg.h
index 8236799..3de5bab 100644
--- a/upb/msg.h
+++ b/upb/msg.h
@@ -154,6 +154,9 @@ upb_msg *upb_msg_new(const upb_msglayout *l, upb_arena *a);
 /* Returns the arena for the given message. */
 upb_arena *upb_msg_arena(const upb_msg *msg);
 
+void upb_msg_addunknown(upb_msg *msg, const char *data, size_t len);
+const char *upb_msg_getunknown(upb_msg *msg, size_t *len);
+
 /* Read-only message API.  Can be safely called by anyone. */
 
 /* Returns the value associated with this field: