Merge pull request #145 from haberman/arrayapi

Generated code API changes for array and hazzers.

19 files changed, 315 insertions, 250 deletions

diff --git a/upb/bindings/lua/upb/pb.c b/upb/bindings/lua/upb/pb.c
index bca2ee8..b5284df 100644
--- a/upb/bindings/lua/upb/pb.c
+++ b/upb/bindings/lua/upb/pb.c
@@ -16,7 +16,7 @@ static int lupb_pb_decode(lua_State *L) {
   const upb_msglayout *layout;
   upb_msg *msg = lupb_msg_checkmsg2(L, 1, &layout);
   const char *pb = lua_tolstring(L, 2, &len);
-  upb_stringview buf = upb_stringview_make(pb, len);
+  upb_strview buf = upb_strview_make(pb, len);
 
   upb_decode(buf, msg, layout);
   /* TODO(haberman): check for error. */
diff --git a/upb/bindings/lua/upb/table.c b/upb/bindings/lua/upb/table.c
index 79120c7..3a9319d 100644
--- a/upb/bindings/lua/upb/table.c
+++ b/upb/bindings/lua/upb/table.c
@@ -21,7 +21,6 @@
 #include "lauxlib.h"
 #include "upb/bindings/lua/upb.h"
 #include "upb/def.h"
-#include "upb/structdefs.int.h"
 #include "upb/table.int.h"
 
 static void lupbtable_setnum(lua_State *L, int tab, const char *key,
diff --git a/upb/decode.c b/upb/decode.c
index fc0e644..c54ee3c 100644
--- a/upb/decode.c
+++ b/upb/decode.c
@@ -112,14 +112,14 @@ static int64_t upb_zzdecode_64(uint64_t n) {
 }
 
 static bool upb_decode_string(const char **ptr, const char *limit,
-                              upb_stringview *val) {
+                              upb_strview *val) {
   uint32_t len;
 
   CHK(upb_decode_varint32(ptr, limit, &len) &&
       len < INT32_MAX &&
       limit - *ptr >= (int32_t)len);
 
-  *val = upb_stringview_make(*ptr, len);
+  *val = upb_strview_make(*ptr, len);
   *ptr += len;
   return true;
 }
@@ -150,7 +150,7 @@ static bool upb_skip_unknownfielddata(upb_decstate *d, upb_decframe *frame,
       return upb_decode_64bit(&d->ptr, frame->limit, &val);
     }
     case UPB_WIRE_TYPE_DELIMITED: {
-      upb_stringview val;
+      upb_strview val;
       return upb_decode_string(&d->ptr, frame->limit, &val);
     }
     case UPB_WIRE_TYPE_START_GROUP:
@@ -374,7 +374,7 @@ static bool upb_decode_32bitfield(upb_decstate *d, upb_decframe *frame,
   return true;
 }
 
-static bool upb_decode_fixedpacked(upb_array *arr, upb_stringview data,
+static bool upb_decode_fixedpacked(upb_array *arr, upb_strview data,
                                    int elem_size) {
   int elements = data.size / elem_size;
   void *field_mem;
@@ -389,7 +389,7 @@ static bool upb_decode_fixedpacked(upb_array *arr, upb_stringview data,
 static bool upb_decode_toarray(upb_decstate *d, upb_decframe *frame,
                                const char *field_start,
                                const upb_msglayout_field *field,
-                               upb_stringview val) {
+                               upb_strview val) {
   upb_array *arr = upb_getorcreatearr(frame, field);
 
 #define VARINT_CASE(ctype, decode) { \
@@ -470,7 +470,7 @@ static bool upb_decode_toarray(upb_decstate *d, upb_decframe *frame,
 static bool upb_decode_delimitedfield(upb_decstate *d, upb_decframe *frame,
                                       const char *field_start,
                                       const upb_msglayout_field *field) {
-  upb_stringview val;
+  upb_strview val;
 
   CHK(upb_decode_string(&d->ptr, frame->limit, &val));
 
@@ -584,7 +584,7 @@ static bool upb_decode_message(upb_decstate *d, const char *limit,
   return true;
 }
 
-bool upb_decode(upb_stringview buf, void *msg, const upb_msglayout *l) {
+bool upb_decode(upb_strview buf, void *msg, const upb_msglayout *l) {
   upb_decstate state;
   state.ptr = buf.data;
 
diff --git a/upb/decode.h b/upb/decode.h
index 79774ed..dee4c0f 100644
--- a/upb/decode.h
+++ b/upb/decode.h
@@ -9,7 +9,7 @@
 
 UPB_BEGIN_EXTERN_C
 
-bool upb_decode(upb_stringview buf, upb_msg *msg, const upb_msglayout *l);
+bool upb_decode(upb_strview buf, upb_msg *msg, const upb_msglayout *l);
 
 UPB_END_EXTERN_C
 
diff --git a/upb/def.c b/upb/def.c
index 223fb1d..8adf8c1 100644
--- a/upb/def.c
+++ b/upb/def.c
@@ -4,7 +4,6 @@
 #include <ctype.h>
 #include <stdlib.h>
 #include <string.h>
-#include "upb/structdefs.int.h"
 #include "upb/handlers.h"
 
 typedef struct {
@@ -665,6 +664,8 @@ static void freefield(upb_refcounted *r) {
   upb_fielddef_uninit_default(f);
   if (f->subdef_is_symbolic)
     upb_gfree(f->sub.name);
+  if (f->msg_is_symbolic)
+    upb_gfree(f->msg.name);
   upb_def_uninit(upb_fielddef_upcast_mutable(f));
   upb_gfree(f);
 }
@@ -1022,7 +1023,7 @@ bool upb_fielddef_setnumber(upb_fielddef *f, uint32_t number, upb_status *s) {
         s, "cannot change field number after adding to a message");
     return false;
   }
-  if (number == 0 || number > UPB_MAX_FIELDNUMBER) {
+  if (number == 0) {
     upb_status_seterrf(s, "invalid field number (%u)", number);
     return false;
   }
diff --git a/upb/def.h b/upb/def.h
index 19b2a20..ea22e40 100644
--- a/upb/def.h
+++ b/upb/def.h
@@ -118,9 +118,24 @@ class upb::Def {
 
  private:
   UPB_DISALLOW_POD_OPS(Def, upb::Def)
+#else
+struct upb_def {
+  upb_refcounted base;
+
+  const char *fullname;
+  const upb_filedef* file;
+  char type;  /* A upb_deftype_t (char to save space) */
+
+  /* Used as a flag during the def's mutable stage.  Must be false unless
+   * it is currently being used by a function on the stack.  This allows
+   * us to easily determine which defs were passed into the function's
+   * current invocation. */
+  bool came_from_user;
+#endif
 };
 
-#endif  /* __cplusplus */
+#define UPB_DEF_INIT(name, type, vtbl, refs, ref2s) \
+    { UPB_REFCOUNT_INIT(vtbl, refs, ref2s), name, NULL, type, false }
 
 UPB_BEGIN_EXTERN_C
 
@@ -583,12 +598,57 @@ class upb::FieldDef {
 
  private:
   UPB_DISALLOW_POD_OPS(FieldDef, upb::FieldDef)
-};
-
+#else
+struct upb_fielddef {
+  upb_def base;
+
+  union {
+    int64_t sint;
+    uint64_t uint;
+    double dbl;
+    float flt;
+    void *bytes;
+  } defaultval;
+  union {
+    const upb_msgdef *def;  /* If !msg_is_symbolic. */
+    char *name;             /* If msg_is_symbolic. */
+  } msg;
+  union {
+    const upb_def *def;  /* If !subdef_is_symbolic. */
+    char *name;          /* If subdef_is_symbolic. */
+  } sub;  /* The msgdef or enumdef for this field, if upb_hassubdef(f). */
+  bool subdef_is_symbolic;
+  bool msg_is_symbolic;
+  const upb_oneofdef *oneof;
+  bool default_is_string;
+  bool type_is_set_;     /* False until type is explicitly set. */
+  bool is_extension_;
+  bool lazy_;
+  bool packed_;
+  upb_intfmt_t intfmt;
+  bool tagdelim;
+  upb_fieldtype_t type_;
+  upb_label_t label_;
+  uint32_t number_;
+  uint32_t selector_base;  /* Used to index into a upb::Handlers table. */
+  uint32_t index_;
 # endif  /* defined(__cplusplus) */
+};
 
 UPB_BEGIN_EXTERN_C
 
+extern const struct upb_refcounted_vtbl upb_fielddef_vtbl;
+
+#define UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, lazy,   \
+                          packed, name, num, msgdef, subdef, selector_base,    \
+                          index, defaultval, refs, ref2s)                      \
+  {                                                                            \
+    UPB_DEF_INIT(name, UPB_DEF_FIELD, &upb_fielddef_vtbl, refs, ref2s),        \
+        defaultval, {msgdef}, {subdef}, NULL, false, false,                    \
+        type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES, true, is_extension, \
+        lazy, packed, intfmt, tagdelim, type, label, num, selector_base, index \
+  }
+
 /* Native C API. */
 upb_fielddef *upb_fielddef_new(const void *owner);
 
@@ -938,12 +998,45 @@ class upb::MessageDef {
 
  private:
   UPB_DISALLOW_POD_OPS(MessageDef, upb::MessageDef)
-};
+#else
+struct upb_msgdef {
+  upb_def base;
+
+  size_t selector_count;
+  uint32_t submsg_field_count;
+
+  /* Tables for looking up fields by number and name. */
+  upb_inttable itof;  /* int to field */
+  upb_strtable ntof;  /* name to field/oneof */
+
+  /* Is this a map-entry message? */
+  bool map_entry;
+
+  /* Whether this message has proto2 or proto3 semantics. */
+  upb_syntax_t syntax;
 
+  /* Type of well known type message. UPB_WELLKNOWN_UNSPECIFIED for
+   * non-well-known message. */
+  upb_wellknowntype_t well_known_type;
+
+  /* TODO(haberman): proper extension ranges (there can be multiple). */
 #endif  /* __cplusplus */
+};
 
 UPB_BEGIN_EXTERN_C
 
+extern const struct upb_refcounted_vtbl upb_msgdef_vtbl;
+
+/* TODO: also support static initialization of the oneofs table. This will be
+ * needed if we compile in descriptors that contain oneofs. */
+#define UPB_MSGDEF_INIT(name, selector_count, submsg_field_count, itof, ntof, \
+                        map_entry, syntax, well_known_type, refs, ref2s)      \
+  {                                                                           \
+    UPB_DEF_INIT(name, UPB_DEF_MSG, &upb_fielddef_vtbl, refs, ref2s),         \
+        selector_count, submsg_field_count, itof, ntof, map_entry, syntax,    \
+        well_known_type                                                       \
+  }
+
 /* Returns NULL if memory allocation failed. */
 upb_msgdef *upb_msgdef_new(const void *owner);
 
@@ -1122,12 +1215,24 @@ class upb::EnumDef {
 
  private:
   UPB_DISALLOW_POD_OPS(EnumDef, upb::EnumDef)
-};
+#else
+struct upb_enumdef {
+  upb_def base;
 
+  upb_strtable ntoi;
+  upb_inttable iton;
+  int32_t defaultval;
 #endif  /* __cplusplus */
+};
 
 UPB_BEGIN_EXTERN_C
 
+extern const struct upb_refcounted_vtbl upb_enumdef_vtbl;
+
+#define UPB_ENUMDEF_INIT(name, ntoi, iton, defaultval, refs, ref2s) \
+  { UPB_DEF_INIT(name, UPB_DEF_ENUM, &upb_enumdef_vtbl, refs, ref2s), ntoi,    \
+    iton, defaultval }
+
 /* Native C API. */
 upb_enumdef *upb_enumdef_new(const void *owner);
 
@@ -1276,12 +1381,25 @@ class upb::OneofDef {
 
  private:
   UPB_DISALLOW_POD_OPS(OneofDef, upb::OneofDef)
-};
-
+#else
+struct upb_oneofdef {
+  upb_refcounted base;
+
+  uint32_t index;  /* Index within oneofs. */
+  const char *name;
+  upb_strtable ntof;
+  upb_inttable itof;
+  const upb_msgdef *parent;
 #endif  /* __cplusplus */
+};
 
 UPB_BEGIN_EXTERN_C
 
+extern const struct upb_refcounted_vtbl upb_oneofdef_vtbl;
+
+#define UPB_ONEOFDEF_INIT(name, ntof, itof, refs, ref2s) \
+  { UPB_REFCOUNT_INIT(&upb_oneofdef_vtbl, refs, ref2s), 0, name, ntof, itof }
+
 /* Native C API. */
 upb_oneofdef *upb_oneofdef_new(const void *owner);
 
@@ -1400,12 +1518,25 @@ class upb::FileDef {
 
  private:
   UPB_DISALLOW_POD_OPS(FileDef, upb::FileDef)
-};
+#else
+struct upb_filedef {
+  upb_refcounted base;
+
+  const char *name;
+  const char *package;
+  const char *phpprefix;
+  const char *phpnamespace;
+  upb_syntax_t syntax;
 
+  upb_inttable defs;
+  upb_inttable deps;
 #endif
+};
 
 UPB_BEGIN_EXTERN_C
 
+extern const struct upb_refcounted_vtbl upb_filedef_vtbl;
+
 upb_filedef *upb_filedef_new(const void *owner);
 
 /* Include upb_refcounted methods like upb_msgdef_ref(). */
@@ -1535,9 +1666,13 @@ class upb::SymbolTable {
 
  private:
   UPB_DISALLOW_POD_OPS(SymbolTable, upb::SymbolTable)
-};
+#else
+struct upb_symtab {
+  upb_refcounted base;
 
+  upb_strtable symtab;
 #endif  /* __cplusplus */
+};
 
 UPB_BEGIN_EXTERN_C
 
diff --git a/upb/descriptor/descriptor.upbdefs.c b/upb/descriptor/descriptor.upbdefs.c
index fa5bb3b..d91f674 100644
--- a/upb/descriptor/descriptor.upbdefs.c
+++ b/upb/descriptor/descriptor.upbdefs.c
@@ -7,8 +7,6 @@
  * regenerated. */
 
 #include "upb/def.h"
-#include "upb/structdefs.int.h"
-
 static const upb_msgdef msgs[22];
 static const upb_fielddef fields[107];
 static const upb_enumdef enums[5];
diff --git a/upb/descriptor/reader.c b/upb/descriptor/reader.c
index 054ebd2..fd219fe 100644
--- a/upb/descriptor/reader.c
+++ b/upb/descriptor/reader.c
@@ -321,14 +321,21 @@ static void *file_startenum(void *closure, const void *hd) {
 
 static void *file_startext(void *closure, const void *hd) {
   upb_descreader *r = closure;
-  bool ok;
   r->f = upb_fielddef_new(r);
-  ok = upb_filedef_addext(r->file, r->f, r, NULL);
   UPB_UNUSED(hd);
-  UPB_ASSERT(ok);
   return r;
 }
 
+static bool file_endext(void *closure, const void *hd) {
+  /* The current symtab code can't handle extensions, so we just discard
+   * them for now. */
+  upb_descreader *r = closure;
+  upb_fielddef_unref(r->f, r);
+  UPB_UNUSED(hd);
+  r->f = NULL;
+  return true;
+}
+
 static size_t file_ondep(void *closure, const void *hd, const char *buf,
                          size_t n, const upb_bufhandle *handle) {
   upb_descreader *r = closure;
@@ -702,13 +709,21 @@ static void *msg_startmsg(void *closure, const void *hd) {
 
 static void *msg_startext(void *closure, const void *hd) {
   upb_descreader *r = closure;
-  upb_fielddef *f = upb_fielddef_new(&f);
-  bool ok = upb_filedef_addext(r->file, f, &f, NULL);
+  r->f = upb_fielddef_new(r);
   UPB_UNUSED(hd);
-  UPB_ASSERT(ok);
   return r;
 }
 
+static bool msg_endext(void *closure, const void *hd) {
+  /* The current symtab code can't handle extensions, so we just discard
+   * them for now. */
+  upb_descreader *r = closure;
+  upb_fielddef_unref(r->f, r);
+  UPB_UNUSED(hd);
+  r->f = NULL;
+  return true;
+}
+
 static void *msg_startfield(void *closure, const void *hd) {
   upb_descreader *r = closure;
   r->f = upb_fielddef_new(&r->f);
@@ -763,6 +778,8 @@ static void reghandlers(const void *closure, upb_handlers *h) {
     upb_handlers_setstring(h, F(DescriptorProto, name), &msg_name, NULL);
     upb_handlers_setstartsubmsg(h, F(DescriptorProto, extension), &msg_startext,
                                 NULL);
+    upb_handlers_setendsubmsg(h, F(DescriptorProto, extension), &msg_endext,
+                              NULL);
     upb_handlers_setstartsubmsg(h, F(DescriptorProto, nested_type),
                                 &msg_startmsg, NULL);
     upb_handlers_setstartsubmsg(h, F(DescriptorProto, field),
@@ -786,6 +803,8 @@ static void reghandlers(const void *closure, upb_handlers *h) {
                                 &file_startenum, NULL);
     upb_handlers_setstartsubmsg(h, F(FileDescriptorProto, extension),
                                 &file_startext, NULL);
+    upb_handlers_setendsubmsg(h, F(FileDescriptorProto, extension),
+                              &file_endext, NULL);
     upb_handlers_setstring(h, F(FileDescriptorProto, dependency),
                            &file_ondep, NULL);
   } else if (upbdefs_google_protobuf_EnumValueDescriptorProto_is(m)) {
diff --git a/upb/encode.c b/upb/encode.c
index bff8262..b728353 100644
--- a/upb/encode.c
+++ b/upb/encode.c
@@ -207,8 +207,8 @@ do { ; } while(0)
       VARINT_CASE(int64_t, upb_zzencode_64(*ptr));
     case UPB_DESCRIPTOR_TYPE_STRING:
     case UPB_DESCRIPTOR_TYPE_BYTES: {
-      upb_stringview *start = arr->data;
-      upb_stringview *ptr = start + arr->len;
+      upb_strview *start = arr->data;
+      upb_strview *ptr = start + arr->len;
       do {
         ptr--;
         CHK(upb_put_bytes(e, ptr->data, ptr->size) &&
@@ -292,7 +292,7 @@ static bool upb_encode_scalarfield(upb_encstate *e, const char *field_mem,
       CASE(int64_t, varint, UPB_WIRE_TYPE_VARINT, upb_zzencode_64(val));
     case UPB_DESCRIPTOR_TYPE_STRING:
     case UPB_DESCRIPTOR_TYPE_BYTES: {
-      upb_stringview view = *(upb_stringview*)field_mem;
+      upb_strview view = *(upb_strview*)field_mem;
       if (skip_zero_value && view.size == 0) {
         return true;
       }
diff --git a/upb/generated_util.h b/upb/generated_util.h
new file mode 100644
index 0000000..bd47389
--- /dev/null
+++ b/upb/generated_util.h
@@ -0,0 +1,102 @@
+/*
+** Functions for use by generated code.  These are not public and users must
+** not call them directly.
+*/
+
+#ifndef UPB_GENERATED_UTIL_H_
+#define UPB_GENERATED_UTIL_H_
+
+#include <stdint.h>
+#include "upb/structs.int.h"
+
+#define PTR_AT(msg, ofs, type) (type*)((const char*)msg + ofs)
+
+UPB_INLINE const void *_upb_array_accessor(const void *msg, size_t ofs,
+                                           size_t *size) {
+  const upb_array *arr = *PTR_AT(msg, ofs, const upb_array*);
+  if (arr) {
+    if (size) *size = arr->size;
+    return arr->data;
+  } else {
+    if (size) *size = 0;
+    return NULL;
+  }
+}
+
+UPB_INLINE void *_upb_array_mutable_accessor(void *msg, size_t ofs,
+                                             size_t *size) {
+  upb_array *arr = *PTR_AT(msg, ofs, upb_array*);
+  if (arr) {
+    if (size) *size = arr->size;
+    return arr->data;
+  } else {
+    if (size) *size = 0;
+    return NULL;
+  }
+}
+
+/* TODO(haberman): this is a mess.  It will improve when upb_array no longer
+ * carries reflective state (type, elem_size). */
+UPB_INLINE void *_upb_array_resize_accessor(void *msg, size_t ofs, size_t size,
+                                            size_t elem_size,
+                                            upb_fieldtype_t type,
+                                            upb_arena *arena) {
+  upb_array *arr = *PTR_AT(msg, ofs, upb_array*);
+
+  if (!arr) {
+    arr = upb_array_new(type, arena);
+    if (!arr) return NULL;
+    *PTR_AT(msg, ofs, upb_array*) = arr;
+  }
+
+  if (size > arr->size) {
+    size_t new_size = UPB_MAX(arr->size, 4);
+    size_t old_bytes = arr->size * elem_size;
+    size_t new_bytes;
+    upb_alloc *alloc = upb_arena_alloc(arr->arena);
+    while (new_size < size) new_size *= 2;
+    new_bytes = new_size * elem_size;
+    arr->data = upb_realloc(alloc, arr->data, old_bytes, new_bytes);
+    if (!arr->data) {
+      return NULL;
+    }
+    arr->size = new_size;
+  }
+
+  arr->len = size;
+  return arr->data;
+}
+
+UPB_INLINE bool _upb_array_append_accessor(void *msg, size_t ofs,
+                                           size_t elem_size,
+                                           upb_fieldtype_t type,
+                                           const void *value,
+                                           upb_arena *arena) {
+  upb_array *arr = *PTR_AT(msg, ofs, upb_array*);
+  size_t i = arr ? arr->len : 0;
+  void *data =
+      _upb_array_resize_accessor(msg, ofs, i + 1, elem_size, type, arena);
+  if (!data) return false;
+  memcpy(PTR_AT(data, i * elem_size, char), value, elem_size);
+  return true;
+}
+
+UPB_INLINE bool _upb_has_field(const void *msg, size_t idx) {
+  return (*PTR_AT(msg, idx / 8, const char) & (idx % 8)) != 0;
+}
+
+UPB_INLINE bool _upb_sethas(const void *msg, size_t idx) {
+  return (*PTR_AT(msg, idx / 8, char)) |= (1 << (idx % 8));
+}
+
+UPB_INLINE bool _upb_clearhas(const void *msg, size_t idx) {
+  return (*PTR_AT(msg, idx / 8, char)) &= ~(1 << (idx % 8));
+}
+
+UPB_INLINE bool _upb_has_oneof_field(const void *msg, size_t case_ofs, int32_t num) {
+  return *PTR_AT(msg, case_ofs, int32_t) == num;
+}
+
+#undef PTR_AT
+
+#endif  /* UPB_GENERATED_UTIL_H_ */
diff --git a/upb/handlers-inl.h b/upb/handlers-inl.h
index afc1382..4a0494c 100644
--- a/upb/handlers-inl.h
+++ b/upb/handlers-inl.h
@@ -7,6 +7,8 @@
 #define UPB_HANDLERS_INL_H_
 
 #include <limits.h>
+#include "upb/upb.h"
+#include "upb/handlers.h"
 
 /* C inline methods. */
 
diff --git a/upb/handlers.c b/upb/handlers.c
index 0022c32..33c43c3 100644
--- a/upb/handlers.c
+++ b/upb/handlers.c
@@ -4,7 +4,6 @@
 */
 
 #include "upb/handlers.h"
-#include "upb/structdefs.int.h"
 
 #include <string.h>
 
diff --git a/upb/msg.c b/upb/msg.c
index 29ba2e1..c6f0c17 100644
--- a/upb/msg.c
+++ b/upb/msg.c
@@ -51,7 +51,7 @@ static size_t upb_msgval_sizeof(upb_fieldtype_t type) {
       return sizeof(void*);
     case UPB_TYPE_BYTES:
     case UPB_TYPE_STRING:
-      return sizeof(upb_stringview);
+      return sizeof(upb_strview);
   }
   UPB_UNREACHABLE();
 }
diff --git a/upb/msg.h b/upb/msg.h
index c72e9f0..058f4a9 100644
--- a/upb/msg.h
+++ b/upb/msg.h
@@ -36,10 +36,12 @@ class MessageLayout;
 
 #endif
 
-UPB_DECLARE_TYPE(upb::Array, upb_array)
 UPB_DECLARE_TYPE(upb::Map, upb_map)
 UPB_DECLARE_TYPE(upb::MapIterator, upb_mapiter)
 
+struct upb_array;
+typedef struct upb_array upb_array;
+
 /* TODO(haberman): C++ accessors */
 
 UPB_BEGIN_EXTERN_C
@@ -73,21 +75,21 @@ typedef struct upb_msglayout {
 } upb_msglayout;
 
 
-/** upb_stringview ************************************************************/
+/** upb_strview ************************************************************/
 
 typedef struct {
   const char *data;
   size_t size;
-} upb_stringview;
+} upb_strview;
 
-UPB_INLINE upb_stringview upb_stringview_make(const char *data, size_t size) {
-  upb_stringview ret;
+UPB_INLINE upb_strview upb_strview_make(const char *data, size_t size) {
+  upb_strview ret;
   ret.data = data;
   ret.size = size;
   return ret;
 }
 
-#define UPB_STRINGVIEW_INIT(ptr, len) {ptr, len}
+#define UPB_STRVIEW_INIT(ptr, len) {ptr, len}
 
 
 /** upb_msgval ****************************************************************/
@@ -107,7 +109,7 @@ typedef union {
   const upb_msg* msg;
   const upb_array* arr;
   const void* ptr;
-  upb_stringview str;
+  upb_strview str;
 } upb_msgval;
 
 #define ACCESSORS(name, membername, ctype) \
@@ -134,12 +136,12 @@ ACCESSORS(map,    map, const upb_map*)
 ACCESSORS(msg,    msg, const upb_msg*)
 ACCESSORS(ptr,    ptr, const void*)
 ACCESSORS(arr,    arr, const upb_array*)
-ACCESSORS(str,    str, upb_stringview)
+ACCESSORS(str,    str, upb_strview)
 
 #undef ACCESSORS
 
 UPB_INLINE upb_msgval upb_msgval_makestr(const char *data, size_t size) {
-  return upb_msgval_str(upb_stringview_make(data, size));
+  return upb_msgval_str(upb_strview_make(data, size));
 }
 
 
diff --git a/upb/msgfactory.c b/upb/msgfactory.c
index 593c9dc..05f5975 100644
--- a/upb/msgfactory.c
+++ b/upb/msgfactory.c
@@ -32,7 +32,7 @@ static size_t upb_msgval_sizeof2(upb_fieldtype_t type) {
       return sizeof(void*);
     case UPB_TYPE_BYTES:
     case UPB_TYPE_STRING:
-      return sizeof(upb_stringview);
+      return sizeof(upb_strview);
   }
   UPB_UNREACHABLE();
 }
diff --git a/upb/msgfactory.h b/upb/msgfactory.h
index 73a26ba..8b12461 100644
--- a/upb/msgfactory.h
+++ b/upb/msgfactory.h
@@ -5,6 +5,12 @@
 #ifndef UPB_MSGFACTORY_H_
 #define UPB_MSGFACTORY_H_
 
+#ifdef __cplusplus
+namespace upb {
+class MessageFactory;
+}
+#endif
+
 UPB_DECLARE_TYPE(upb::MessageFactory, upb_msgfactory)
 
 /** upb_msgfactory ************************************************************/
diff --git a/upb/pb/decoder.int.h b/upb/pb/decoder.int.h
index 4032570..c9ad551 100644
--- a/upb/pb/decoder.int.h
+++ b/upb/pb/decoder.int.h
@@ -9,17 +9,10 @@
 #include "upb/handlers.h"
 #include "upb/pb/decoder.h"
 #include "upb/sink.h"
-#include "upb/structdefs.int.h"
 #include "upb/table.int.h"
 
-/* C++ names are not actually used since this type isn't exposed to users. */
-#ifdef __cplusplus
-namespace upb {
-namespace pb {
-class MessageGroup;
-}  /* namespace pb */
-}  /* namespace upb */
-#endif
+#ifndef __cplusplus
+
 UPB_DECLARE_DERIVED_TYPE(upb::pb::MessageGroup, upb::RefCounted,
                          mgroup, upb_refcounted)
 
@@ -82,7 +75,7 @@ typedef enum {
 
 #define OP_MAX OP_HALT
 
-UPB_INLINE opcode getop(uint32_t instr) { return instr & 0xff; }
+UPB_INLINE opcode getop(uint32_t instr) { return (opcode)(instr & 0xff); }
 
 /* Method group; represents a set of decoder methods that had their code
  * emitted together, and must therefore be freed together.  Immutable once
@@ -329,4 +322,6 @@ UPB_INLINE void upb_pbdecoder_unpackdispatch(uint64_t dispatch, uint64_t *ofs,
 
 #define CHECK_RETURN(x) { int32_t ret = x; if (ret >= 0) return ret; }
 
+#endif  /* __cplusplus */
+
 #endif  /* UPB_DECODER_INT_H_ */
diff --git a/upb/structdefs.int.h b/upb/structdefs.int.h
deleted file mode 100644
index cf8bd1d..0000000
--- a/upb/structdefs.int.h
+++ /dev/null
@@ -1,196 +0,0 @@
-/*
-** This file contains definitions of structs that should be considered private
-** and NOT stable across versions of upb.
-**
-** The only reason they are declared here and not in .c files is to allow upb
-** and the application (if desired) to embed statically-initialized instances
-** of structures like defs.
-**
-** If you include this file, all guarantees of ABI compatibility go out the
-** window!  Any code that includes this file needs to recompile against the
-** exact same version of upb that they are linking against.
-**
-** You also need to recompile if you change the value of the UPB_DEBUG_REFS
-** flag.
-*/
-
-#include "upb/def.h"
-
-#ifndef UPB_STATICINIT_H_
-#define UPB_STATICINIT_H_
-
-#ifdef __cplusplus
-/* Because of how we do our typedefs, this header can't be included from C++. */
-#error This file cannot be included from C++
-#endif
-
-/* upb_refcounted *************************************************************/
-
-
-/* upb_def ********************************************************************/
-
-struct upb_def {
-  upb_refcounted base;
-
-  const char *fullname;
-  const upb_filedef* file;
-  char type;  /* A upb_deftype_t (char to save space) */
-
-  /* Used as a flag during the def's mutable stage.  Must be false unless
-   * it is currently being used by a function on the stack.  This allows
-   * us to easily determine which defs were passed into the function's
-   * current invocation. */
-  bool came_from_user;
-};
-
-#define UPB_DEF_INIT(name, type, vtbl, refs, ref2s) \
-    { UPB_REFCOUNT_INIT(vtbl, refs, ref2s), name, NULL, type, false }
-
-
-/* upb_fielddef ***************************************************************/
-
-struct upb_fielddef {
-  upb_def base;
-
-  union {
-    int64_t sint;
-    uint64_t uint;
-    double dbl;
-    float flt;
-    void *bytes;
-  } defaultval;
-  union {
-    const upb_msgdef *def;  /* If !msg_is_symbolic. */
-    char *name;             /* If msg_is_symbolic. */
-  } msg;
-  union {
-    const upb_def *def;  /* If !subdef_is_symbolic. */
-    char *name;          /* If subdef_is_symbolic. */
-  } sub;  /* The msgdef or enumdef for this field, if upb_hassubdef(f). */
-  bool subdef_is_symbolic;
-  bool msg_is_symbolic;
-  const upb_oneofdef *oneof;
-  bool default_is_string;
-  bool type_is_set_;     /* False until type is explicitly set. */
-  bool is_extension_;
-  bool lazy_;
-  bool packed_;
-  upb_intfmt_t intfmt;
-  bool tagdelim;
-  upb_fieldtype_t type_;
-  upb_label_t label_;
-  uint32_t number_;
-  uint32_t selector_base;  /* Used to index into a upb::Handlers table. */
-  uint32_t index_;
-};
-
-extern const struct upb_refcounted_vtbl upb_fielddef_vtbl;
-
-#define UPB_FIELDDEF_INIT(label, type, intfmt, tagdelim, is_extension, lazy,   \
-                          packed, name, num, msgdef, subdef, selector_base,    \
-                          index, defaultval, refs, ref2s)                      \
-  {                                                                            \
-    UPB_DEF_INIT(name, UPB_DEF_FIELD, &upb_fielddef_vtbl, refs, ref2s),        \
-        defaultval, {msgdef}, {subdef}, NULL, false, false,                    \
-        type == UPB_TYPE_STRING || type == UPB_TYPE_BYTES, true, is_extension, \
-        lazy, packed, intfmt, tagdelim, type, label, num, selector_base, index \
-  }
-
-
-/* upb_msgdef *****************************************************************/
-
-struct upb_msgdef {
-  upb_def base;
-
-  size_t selector_count;
-  uint32_t submsg_field_count;
-
-  /* Tables for looking up fields by number and name. */
-  upb_inttable itof;  /* int to field */
-  upb_strtable ntof;  /* name to field/oneof */
-
-  /* Is this a map-entry message? */
-  bool map_entry;
-
-  /* Whether this message has proto2 or proto3 semantics. */
-  upb_syntax_t syntax;
-
-  /* Type of well known type message. UPB_WELLKNOWN_UNSPECIFIED for
-   * non-well-known message. */
-  upb_wellknowntype_t well_known_type;
-
-  /* TODO(haberman): proper extension ranges (there can be multiple). */
-};
-
-extern const struct upb_refcounted_vtbl upb_msgdef_vtbl;
-
-/* TODO: also support static initialization of the oneofs table. This will be
- * needed if we compile in descriptors that contain oneofs. */
-#define UPB_MSGDEF_INIT(name, selector_count, submsg_field_count, itof, ntof, \
-                        map_entry, syntax, well_known_type, refs, ref2s)      \
-  {                                                                           \
-    UPB_DEF_INIT(name, UPB_DEF_MSG, &upb_fielddef_vtbl, refs, ref2s),         \
-        selector_count, submsg_field_count, itof, ntof, map_entry, syntax,    \
-        well_known_type                                                       \
-  }
-
-
-/* upb_enumdef ****************************************************************/
-
-struct upb_enumdef {
-  upb_def base;
-
-  upb_strtable ntoi;
-  upb_inttable iton;
-  int32_t defaultval;
-};
-
-extern const struct upb_refcounted_vtbl upb_enumdef_vtbl;
-
-#define UPB_ENUMDEF_INIT(name, ntoi, iton, defaultval, refs, ref2s) \
-  { UPB_DEF_INIT(name, UPB_DEF_ENUM, &upb_enumdef_vtbl, refs, ref2s), ntoi,    \
-    iton, defaultval }
-
-
-/* upb_oneofdef ***************************************************************/
-
-struct upb_oneofdef {
-  upb_refcounted base;
-
-  uint32_t index;  /* Index within oneofs. */
-  const char *name;
-  upb_strtable ntof;
-  upb_inttable itof;
-  const upb_msgdef *parent;
-};
-
-extern const struct upb_refcounted_vtbl upb_oneofdef_vtbl;
-
-#define UPB_ONEOFDEF_INIT(name, ntof, itof, refs, ref2s) \
-  { UPB_REFCOUNT_INIT(&upb_oneofdef_vtbl, refs, ref2s), 0, name, ntof, itof }
-
-
-/* upb_symtab *****************************************************************/
-
-struct upb_symtab {
-  upb_refcounted base;
-
-  upb_strtable symtab;
-};
-
-struct upb_filedef {
-  upb_refcounted base;
-
-  const char *name;
-  const char *package;
-  const char *phpprefix;
-  const char *phpnamespace;
-  upb_syntax_t syntax;
-
-  upb_inttable defs;
-  upb_inttable deps;
-};
-
-extern const struct upb_refcounted_vtbl upb_filedef_vtbl;
-
-#endif  /* UPB_STATICINIT_H_ */
diff --git a/upb/structs.int.h b/upb/structs.int.h
index 6bc502e..1394dd4 100644
--- a/upb/structs.int.h
+++ b/upb/structs.int.h
@@ -5,6 +5,9 @@
 #ifndef UPB_STRUCTS_H_
 #define UPB_STRUCTS_H_
 
+#include "upb/def.h"
+#include "upb/msg.h"
+
 struct upb_array {
   upb_fieldtype_t type;
   uint8_t element_size;