upbc compiles and links! But probably doesn't work yet.

5 files changed, 201 insertions, 126 deletions

diff --git a/src/upb.h b/src/upb.h
index 158d631..e130dd8 100644
--- a/src/upb.h
+++ b/src/upb.h
@@ -13,6 +13,7 @@
 #include <stdint.h>
 #include <stdio.h>  // only for size_t.
 #include "descriptor_const.h"
+#include "upb_atomic.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -124,13 +125,14 @@ struct upb_tag {
 
 // INTERNAL-ONLY: never refer to these types with a tag ("union", "struct").
 // Always use the typedefs.
-union upb_string;
-union upb_array;
-struct upb_msg;
+union _upb_string;
+union _upb_array;
+struct _upb_msg;
 
-typedef union upb_string upb_string;
-typedef union upb_array upb_array;
-typedef struct upb_msg upb_msg;
+typedef union _upb_string upb_string;
+typedef union _upb_array upb_array;
+typedef struct _upb_msg upb_msg;
+typedef upb_atomic_refcount_t upb_data;
 
 // A single .proto value.  The owner must have an out-of-band way of knowing
 // the type, so that it knows which union member to use.
@@ -142,9 +144,10 @@ union upb_value {
   uint32_t uint32;
   uint64_t uint64;
   bool _bool;
-  union upb_string *str;
-  union upb_array *arr;
-  struct upb_msg *msg;
+  upb_string *str;
+  upb_array *arr;
+  upb_msg *msg;
+  upb_data *data;
 };
 
 // A pointer to a .proto value.  The owner must have an out-of-band way of
@@ -158,9 +161,10 @@ union upb_value_ptr {
   uint32_t *uint32;
   uint64_t *uint64;
   bool *_bool;
-  union upb_string **str;
-  union upb_array **arr;
-  struct upb_msg **msg;
+  upb_string **str;
+  upb_array **arr;
+  upb_msg **msg;
+  upb_data **data;
   void *_void;
 };
 
diff --git a/src/upb_data.c b/src/upb_data.c
index 595015f..084a03c 100644
--- a/src/upb_data.c
+++ b/src/upb_data.c
@@ -7,6 +7,7 @@
 #include <stdlib.h>
 #include "upb_data.h"
 #include "upb_def.h"
+#include "upb_parse.h"
 
 static uint32_t round_up_to_pow2(uint32_t v)
 {
@@ -51,6 +52,18 @@ static void check_not_frozen(upb_data *d) {
   if(upb_data_hasflag(d, UPB_DATA_FROZEN)) abort();
 }
 
+
+/* upb_string *******************************************************************/
+
+upb_string *upb_string_new() {
+  upb_string *s = malloc(sizeof(upb_refcounted_string));
+  data_init(&s->common.base, UPB_DATA_HEAPALLOCATED | UPB_DATA_REFCOUNTED);
+  s->refcounted.byte_size = 0;
+  s->common.byte_len = 0;
+  s->common.ptr = NULL;
+  return s;
+}
+
 static upb_strlen_t string_get_bytesize(upb_string *s) {
   if(upb_data_hasflag(&s->common.base, UPB_DATA_REFCOUNTED)) {
     return s->refcounted.byte_size;
@@ -68,18 +81,6 @@ static void string_set_bytesize(upb_string *s, upb_strlen_t newsize) {
   }
 }
 
-
-/* upb_string *******************************************************************/
-
-upb_string *upb_string_new() {
-  upb_string *s = malloc(sizeof(upb_refcounted_string));
-  data_init(&s->common.base, UPB_DATA_HEAPALLOCATED | UPB_DATA_REFCOUNTED);
-  s->refcounted.byte_size = 0;
-  s->common.byte_len = 0;
-  s->common.ptr = NULL;
-  return s;
-}
-
 void _upb_string_free(upb_string *s)
 {
   if(string_get_bytesize(s) != 0) free(s->common.ptr);
@@ -136,11 +137,6 @@ void upb_strcat(upb_string *s, upb_string *append) {
          upb_string_getrobuf(append), append_len);
 }
 
-void upb_strcpy(upb_string *dest, upb_string *src) {
-  upb_strlen_t src_len = upb_strlen(src);
-  memcpy(upb_string_getrwbuf(dest, src_len), upb_string_getrobuf(src), src_len);
-}
-
 upb_string *upb_strslice(upb_string *s, int offset, int len) {
   upb_string *slice = upb_string_new();
   len = UPB_MIN((upb_strlen_t)len, upb_strlen(s) - (upb_strlen_t)offset);
@@ -165,6 +161,15 @@ int upb_strcmp(upb_string *s1, upb_string *s2) {
 
 /* upb_array ******************************************************************/
 
+upb_array *upb_array_new() {
+  upb_array *a = malloc(sizeof(upb_refcounted_array));
+  data_init(&a->common.base, UPB_DATA_HEAPALLOCATED | UPB_DATA_REFCOUNTED);
+  a->refcounted.size = 0;
+  a->common.len = 0;
+  a->common.elements._void = NULL;
+  return a;
+}
+
 // ONLY handles refcounted arrays for the moment.
 void _upb_array_free(upb_array *a, struct upb_fielddef *f)
 {
@@ -178,9 +183,42 @@ void _upb_array_free(upb_array *a, struct upb_fielddef *f)
   free(a);
 }
 
+static upb_arraylen_t array_get_size(upb_array *a) {
+  if(upb_data_hasflag(&a->common.base, UPB_DATA_REFCOUNTED)) {
+    return a->refcounted.size;
+  } else {
+    return (a->norefcount.size_and_flags & 0xFFFFFFF8) >> 3;
+  }
+}
+
+static void array_set_size(upb_array *a, upb_arraylen_t newsize) {
+  if(upb_data_hasflag(&a->common.base, UPB_DATA_REFCOUNTED)) {
+    a->refcounted.size = newsize;
+  } else {
+    a->norefcount.size_and_flags &= 0x7;
+    a->norefcount.size_and_flags |= (newsize << 3);
+  }
+}
+
+void upb_array_resize(upb_array *a, upb_strlen_t len) {
+  check_not_frozen(&a->common.base);
+  if(array_get_size(a) < len) {
+    // Need to resize.
+    size_t new_size = round_up_to_pow2(len);
+    a->common.elements._void = realloc(a->common.elements._void, new_size);
+    array_set_size(a, new_size);
+  }
+  a->common.len = len;
+}
+
+
 
 /* upb_msg ********************************************************************/
 
+static void upb_msg_sethas(upb_msg *msg, struct upb_fielddef *f) {
+  msg->data[f->field_index/8] |= (1 << (f->field_index % 8));
+}
+
 upb_msg *upb_msg_new(struct upb_msgdef *md) {
   upb_msg *msg = malloc(md->size);
   memset(msg, 0, md->size);
@@ -219,74 +257,75 @@ struct upb_msgparser {
 /* Helper function that returns a pointer to where the next value for field "f"
  * should be stored, taking into account whether f is an array that may need to
  * be allocated or resized. */
-static union upb_value_ptr get_value_ptr(struct upb_msg *msg,
-                                         struct upb_fielddef *f)
+static union upb_value_ptr get_value_ptr(upb_msg *msg, struct upb_fielddef *f)
 {
-  union upb_value_ptr p = upb_msg_getptr(msg, f);
+  union upb_value_ptr p = _upb_msg_getptr(msg, f);
   if(upb_isarray(f)) {
-    if(!upb_msg_isset(msg, f)) {
-      if(!*p.arr || !upb_mmhead_only(&((*p.arr)->mmhead))) {
+    if(!upb_msg_has(msg, f)) {
+      if(!*p.arr || !upb_data_only(*p.data)) {
         if(*p.arr)
-          upb_array_unref(*p.arr);
-        *p.arr = upb_array_new(f);
+          upb_array_unref(*p.arr, f);
+        *p.arr = upb_array_new();
       }
       upb_array_truncate(*p.arr);
-      upb_msg_set(msg, f);
+      upb_msg_sethas(msg, f);
     }
-    p = upb_array_append(*p.arr);
+    upb_arraylen_t oldlen = upb_array_len(*p.arr);
+    upb_array_resize(*p.arr, oldlen + 1);
+    p = _upb_array_getptr(*p.arr, f, oldlen);
   }
   return p;
 }
 
-/* Callbacks for the stream parser. */
+// Callbacks for the stream parser.
+// TODO: implement these in terms of public interfaces.
 
 static bool value_cb(void *udata, struct upb_msgdef *msgdef,
                      struct upb_fielddef *f, union upb_value val)
 {
   (void)msgdef;
   struct upb_msgparser *mp = udata;
-  struct upb_msg *msg = mp->top->msg;
+  upb_msg *msg = mp->top->msg;
   union upb_value_ptr p = get_value_ptr(msg, f);
-  upb_msg_set(msg, f);
+  upb_msg_sethas(msg, f);
   upb_value_write(p, val, f->type);
   return true;
 }
 
 static bool str_cb(void *udata, struct upb_msgdef *msgdef,
-                   struct upb_fielddef *f, uint8_t *str, size_t avail_len,
+                   struct upb_fielddef *f, const uint8_t *str, size_t avail_len,
                    size_t total_len)
 {
   (void)msgdef;
   struct upb_msgparser *mp = udata;
-  struct upb_msg *msg = mp->top->msg;
+  upb_msg *msg = mp->top->msg;
   union upb_value_ptr p = get_value_ptr(msg, f);
-  upb_msg_set(msg, f);
+  upb_msg_sethas(msg, f);
   if(avail_len != total_len) abort();  /* TODO: support streaming. */
-  if(!*p.str || !upb_mmhead_only(&((*p.str)->mmhead))) {
+  if(!*p.str || !upb_data_only(*p.data)) {
     if(*p.str)
       upb_string_unref(*p.str);
     *p.str = upb_string_new();
   }
-  upb_string_resize(*p.str, total_len);
-  memcpy((*p.str)->ptr, str, avail_len);
-  (*p.str)->byte_len = avail_len;
+  upb_strcpylen(*p.str, str, avail_len);
   return true;
 }
 
 static void start_cb(void *udata, struct upb_fielddef *f)
 {
   struct upb_msgparser *mp = udata;
-  struct upb_msg *oldmsg = mp->top->msg;
+  upb_msg *oldmsg = mp->top->msg;
   union upb_value_ptr p = get_value_ptr(oldmsg, f);
 
-  if(upb_isarray(f) || !upb_msg_isset(oldmsg, f)) {
-    if(!*p.msg || !upb_mmhead_only(&((*p.msg)->mmhead))) {
+  if(upb_isarray(f) || !upb_msg_has(oldmsg, f)) {
+    struct upb_msgdef *md = upb_downcast_msgdef(f->def);
+    if(!*p.msg || !upb_data_only(*p.data)) {
       if(*p.msg)
-        upb_msg_unref(*p.msg);
-      *p.msg = upb_msg_new(upb_downcast_msgdef(f->def));
+        upb_msg_unref(*p.msg, md);
+      *p.msg = upb_msg_new(md);
     }
-    upb_msg_clear(*p.msg);
-    upb_msg_set(oldmsg, f);
+    upb_msg_clear(*p.msg, md);
+    upb_msg_sethas(oldmsg, f);
   }
 
   mp->top++;
@@ -308,10 +347,9 @@ struct upb_msgparser *upb_msgparser_new(struct upb_msgdef *def)
   return mp;
 }
 
-void upb_msgparser_reset(struct upb_msgparser *s, struct upb_msg *msg, bool byref)
+void upb_msgparser_reset(struct upb_msgparser *s, upb_msg *msg)
 {
   upb_cbparser_reset(s->s, s);
-  s->byref = byref;
   s->top = s->stack;
   s->top->msg = msg;
 }
@@ -322,13 +360,13 @@ void upb_msgparser_free(struct upb_msgparser *s)
   free(s);
 }
 
-void upb_msg_parsestr(struct upb_msg *msg, void *buf, size_t len,
+void upb_msg_parsestr(upb_msg *msg, struct upb_msgdef *md, upb_string *str,
                       struct upb_status *status)
 {
-  struct upb_msgparser *mp = upb_msgparser_new(msg->def);
-  upb_msgparser_reset(mp, msg, false);
-  upb_msg_clear(msg);
-  upb_msgparser_parse(mp, buf, len, status);
+  struct upb_msgparser *mp = upb_msgparser_new(md);
+  upb_msgparser_reset(mp, msg);
+  upb_msg_clear(msg, md);
+  upb_msgparser_parse(mp, str, status);
   upb_msgparser_free(mp);
 }
 
diff --git a/src/upb_data.h b/src/upb_data.h
index b3bbb3c..872f237 100644
--- a/src/upb_data.h
+++ b/src/upb_data.h
@@ -35,7 +35,8 @@ struct upb_fielddef;
 // The "base class" of strings, arrays, and messages.  Contains a few flags and
 // possibly a reference count.  None of the functions for upb_data are public,
 // but some of the constants are.
-typedef upb_atomic_refcount_t upb_data;
+
+// typedef upb_atomic_refcount_t upb_data;
 
 // The flags in upb_data.
 typedef enum {
@@ -196,7 +197,7 @@ typedef struct {
   uint32_t byte_size;
 } upb_refcounted_string;
 
-union upb_string {
+union _upb_string {
   upb_norefcount_string norefcount;
   upb_string_common common;
   upb_refcounted_string refcounted;
@@ -265,8 +266,22 @@ INLINE bool upb_streql(upb_string *s1, upb_string *s2) {
 // Like strcmp().
 int upb_strcmp(upb_string *s1, upb_string *s2);
 
+// Like upb_strcpy, but copies from a buffer and length.
+INLINE void upb_strcpylen(upb_string *dest, const void *src, upb_strlen_t len) {
+  memcpy(upb_string_getrwbuf(dest, len), src, len);
+}
+
 // Replaces the contents of "dest" with the contents of "src".
-void upb_strcpy(upb_string *dest, upb_string *src);
+INLINE void upb_strcpy(upb_string *dest, upb_string *src) {
+  upb_strcpylen(dest, upb_string_getrobuf(src), upb_strlen(src));
+}
+
+// Like upb_strcpy, but copies from a NULL-terminated string.
+INLINE void upb_strcpyc(upb_string *dest, const char *src) {
+  // This does two passes over src, but that is necessary unless we want to
+  // repeatedly re-allocate dst, which seems worse.
+  upb_strcpylen(dest, src, strlen(src));
+}
 
 // Returns a new string whose contents are a copy of s.
 upb_string *upb_strdup(upb_string *s);
@@ -323,7 +338,7 @@ typedef struct {
   upb_arraylen_t size;
 } upb_refcounted_array;
 
-union upb_array {
+union _upb_array {
   upb_norefcount_array norefcount;
   upb_array_common common;
   upb_refcounted_array refcounted;
@@ -395,12 +410,17 @@ INLINE size_t upb_array_len(upb_array *a) {
   return a->common.len;
 }
 
+INLINE void upb_array_truncate(upb_array *a) {
+  a->common.len = 0;
+}
+
+
 /* upb_msg ********************************************************************/
 
 // Note that some inline functions for upb_msg are defined in upb_def.h since
 // they rely on the defs.
 
-struct upb_msg {
+struct _upb_msg {
   upb_data base;
   uint8_t data[4];  // We allocate the appropriate amount per message.
 };
@@ -444,7 +464,9 @@ INLINE union upb_value upb_msg_get(upb_msg *msg, struct upb_fielddef *f) {
 // and will drop a ref on whatever was there before.
 void upb_msg_set(upb_msg *msg, struct upb_fielddef *f, union upb_value val);
 
-void upb_msg_clear(upb_msg *msg, struct upb_msgdef *md);
+INLINE void upb_msg_clear(upb_msg *msg, struct upb_msgdef *md) {
+  memset(msg->data, 0, md->set_flags_bytes);
+}
 
 /* Parsing ********************************************************************/
 
diff --git a/src/upb_parse.c b/src/upb_parse.c
index b9cad5a..32332d9 100644
--- a/src/upb_parse.c
+++ b/src/upb_parse.c
@@ -18,12 +18,14 @@
 // generation, because the generated code would want to inline these functions.
 // The same applies to the functions to read .proto values below.
 
-uint8_t *upb_get_v_uint64_t_full(uint8_t *buf, uint8_t *end, uint64_t *val,
-                                 struct upb_status *status);
+const uint8_t *upb_get_v_uint64_t_full(const uint8_t *buf, const uint8_t *end,
+                                       uint64_t *val,
+                                       struct upb_status *status);
 
 // Gets a varint (wire type: UPB_WIRE_TYPE_VARINT).
-INLINE uint8_t *upb_get_v_uint64_t(uint8_t *buf, uint8_t *end, uint64_t *val,
-                                   struct upb_status *status)
+INLINE const uint8_t *upb_get_v_uint64_t(const uint8_t *buf, const uint8_t *end,
+                                         uint64_t *val,
+                                         struct upb_status *status)
 {
   // We inline this common case (1-byte varints), if that fails we dispatch to
   // the full (non-inlined) version.
@@ -36,20 +38,21 @@ INLINE uint8_t *upb_get_v_uint64_t(uint8_t *buf, uint8_t *end, uint64_t *val,
 }
 
 // Gets a varint -- called when we only need 32 bits of it.
-INLINE uint8_t *upb_get_v_uint32_t(uint8_t *buf, uint8_t *end,
-                                   uint32_t *val, struct upb_status *status)
+INLINE const uint8_t *upb_get_v_uint32_t(const uint8_t *buf, const uint8_t *end,
+                                         uint32_t *val,
+                                         struct upb_status *status)
 {
   uint64_t val64;
-  uint8_t *ret = upb_get_v_uint64_t(buf, end, &val64, status);
+  const uint8_t *ret = upb_get_v_uint64_t(buf, end, &val64, status);
   *val = (uint32_t)val64;  // Discard the high bits.
   return ret;
 }
 
 // Gets a fixed-length 32-bit integer (wire type: UPB_WIRE_TYPE_32BIT).
-INLINE uint8_t *upb_get_f_uint32_t(uint8_t *buf, uint8_t *end,
-                                   uint32_t *val, struct upb_status *status)
+INLINE const uint8_t *upb_get_f_uint32_t(const uint8_t *buf, const uint8_t *end,
+                                         uint32_t *val, struct upb_status *status)
 {
-  uint8_t *uint32_end = buf + sizeof(uint32_t);
+  const uint8_t *uint32_end = buf + sizeof(uint32_t);
   if(uint32_end > end) {
     status->code = UPB_STATUS_NEED_MORE_DATA;
     return end;
@@ -65,10 +68,11 @@ INLINE uint8_t *upb_get_f_uint32_t(uint8_t *buf, uint8_t *end,
 }
 
 // Gets a fixed-length 64-bit integer (wire type: UPB_WIRE_TYPE_64BIT).
-INLINE uint8_t *upb_get_f_uint64_t(uint8_t *buf, uint8_t *end,
-                                   uint64_t *val, struct upb_status *status)
+INLINE const uint8_t *upb_get_f_uint64_t(const uint8_t *buf, const uint8_t *end,
+                                         uint64_t *val,
+                                         struct upb_status *status)
 {
-  uint8_t *uint64_end = buf + sizeof(uint64_t);
+  const uint8_t *uint64_end = buf + sizeof(uint64_t);
   if(uint64_end > end) {
     status->code = UPB_STATUS_NEED_MORE_DATA;
     return end;
@@ -84,10 +88,11 @@ INLINE uint8_t *upb_get_f_uint64_t(uint8_t *buf, uint8_t *end,
   return uint64_end;
 }
 
-INLINE uint8_t *upb_skip_v_uint64_t(uint8_t *buf, uint8_t *end,
-                                    struct upb_status *status)
+INLINE const uint8_t *upb_skip_v_uint64_t(const uint8_t *buf,
+                                          const uint8_t *end,
+                                          struct upb_status *status)
 {
-  uint8_t *const maxend = buf + 10;
+  const uint8_t *const maxend = buf + 10;
   uint8_t last = 0x80;
   for(; buf < (uint8_t*)end && (last & 0x80); buf++)
     last = *buf;
@@ -102,10 +107,11 @@ INLINE uint8_t *upb_skip_v_uint64_t(uint8_t *buf, uint8_t *end,
   return buf;
 }
 
-INLINE uint8_t *upb_skip_f_uint32_t(uint8_t *buf, uint8_t *end,
-                                    struct upb_status *status)
+INLINE const uint8_t *upb_skip_f_uint32_t(const uint8_t *buf,
+                                          const uint8_t *end,
+                                          struct upb_status *status)
 {
-  uint8_t *uint32_end = buf + sizeof(uint32_t);
+  const uint8_t *uint32_end = buf + sizeof(uint32_t);
   if(uint32_end > end) {
     status->code = UPB_STATUS_NEED_MORE_DATA;
     return end;
@@ -113,10 +119,11 @@ INLINE uint8_t *upb_skip_f_uint32_t(uint8_t *buf, uint8_t *end,
   return uint32_end;
 }
 
-INLINE uint8_t *upb_skip_f_uint64_t(uint8_t *buf, uint8_t *end,
-                                    struct upb_status *status)
+INLINE const uint8_t *upb_skip_f_uint64_t(const uint8_t *buf,
+                                          const uint8_t *end,
+                                          struct upb_status *status)
 {
-  uint8_t *uint64_end = buf + sizeof(uint64_t);
+  const uint8_t *uint64_end = buf + sizeof(uint64_t);
   if(uint64_end > end) {
     status->code = UPB_STATUS_NEED_MORE_DATA;
     return end;
@@ -137,8 +144,8 @@ INLINE int64_t upb_zzdec_64(uint64_t n) { return (n >> 1) ^ -(int64_t)(n & 1); }
 //  // not contain the entire value UPB_STATUS_NEED_MORE_DATA is returned).
 //  // On success, a pointer will be returned to the first byte that was
 //  // not consumed.
-//  uint8_t *upb_get_INT32(uint8_t *buf, uint8_t *end, int32_t *d,
-//                         struct upb_status *status);
+//  const uint8_t *upb_get_INT32(const uint8_t *buf, const uint8_t *end,
+//                               int32_t *d, struct upb_status *status);
 //
 //  // Given an already read wire value s (source), convert it to a .proto
 //  // value and return it.
@@ -151,10 +158,10 @@ INLINE int64_t upb_zzdec_64(uint64_t n) { return (n >> 1) ^ -(int64_t)(n & 1); }
   INLINE val_t upb_wvtov_ ## type(wire_t s)
 
 #define GET(type, v_or_f, wire_t, val_t, member_name) \
-  INLINE uint8_t *upb_get_ ## type(uint8_t *buf, uint8_t *end, val_t *d, \
-                                   struct upb_status *status) { \
+  INLINE const uint8_t *upb_get_ ## type(const uint8_t *buf, const uint8_t *end, \
+                                         val_t *d, struct upb_status *status) { \
     wire_t tmp = 0; \
-    uint8_t *ret = upb_get_ ## v_or_f ## _ ## wire_t(buf, end, &tmp, status); \
+    const uint8_t *ret = upb_get_ ## v_or_f ## _ ## wire_t(buf, end, &tmp, status); \
     *d = upb_wvtov_ ## type(tmp); \
     return ret; \
   }
@@ -192,11 +199,11 @@ T(FLOAT,    f, uint32_t, float,    _float)  {
 #undef T
 
 // Parses a tag, places the result in *tag.
-INLINE uint8_t *parse_tag(uint8_t *buf, uint8_t *end, struct upb_tag *tag,
-                          struct upb_status *status)
+INLINE const uint8_t *parse_tag(const uint8_t *buf, const uint8_t *end,
+                                struct upb_tag *tag, struct upb_status *status)
 {
   uint32_t tag_int;
-  uint8_t *ret = upb_get_v_uint32_t(buf, end, &tag_int, status);
+  const uint8_t *ret = upb_get_v_uint32_t(buf, end, &tag_int, status);
   tag->wire_type    = (upb_wire_type_t)(tag_int & 0x07);
   tag->field_number = tag_int >> 3;
   return ret;
@@ -207,10 +214,10 @@ INLINE uint8_t *parse_tag(uint8_t *buf, uint8_t *end, struct upb_tag *tag,
  * Parses a 64-bit varint that is known to be >= 2 bytes (the inline version
  * handles 1 and 2 byte varints).
  */
-uint8_t *upb_get_v_uint64_t_full(uint8_t *buf, uint8_t *end, uint64_t *val,
-                                 struct upb_status *status)
+const uint8_t *upb_get_v_uint64_t_full(const uint8_t *buf, const uint8_t *end,
+                                       uint64_t *val, struct upb_status *status)
 {
-  uint8_t *const maxend = buf + 10;
+  const uint8_t *const maxend = buf + 10;
   uint8_t last = 0x80;
   *val = 0;
   int bitpos;
@@ -231,9 +238,10 @@ uint8_t *upb_get_v_uint64_t_full(uint8_t *buf, uint8_t *end, uint64_t *val,
   return buf;
 }
 
-uint8_t *upb_parse_wire_value(uint8_t *buf, uint8_t *end, upb_wire_type_t wt,
-                              union upb_wire_value *wv,
-                              struct upb_status *status)
+const uint8_t *upb_parse_wire_value(uint8_t *buf, uint8_t *end,
+                                    upb_wire_type_t wt,
+                                    union upb_wire_value *wv,
+                                    struct upb_status *status)
 {
   switch(wt) {
     case UPB_WIRE_TYPE_VARINT:
@@ -252,8 +260,9 @@ uint8_t *upb_parse_wire_value(uint8_t *buf, uint8_t *end, upb_wire_type_t wt,
  * Advances buf past the current wire value (of type wt), saving the result in
  * outbuf.
  */
-static uint8_t *skip_wire_value(uint8_t *buf, uint8_t *end, upb_wire_type_t wt,
-                                struct upb_status *status)
+static const uint8_t *skip_wire_value(const uint8_t *buf, const uint8_t *end,
+                                      upb_wire_type_t wt,
+                                      struct upb_status *status)
 {
   switch(wt) {
     case UPB_WIRE_TYPE_VARINT:
@@ -272,8 +281,10 @@ static uint8_t *skip_wire_value(uint8_t *buf, uint8_t *end, upb_wire_type_t wt,
   }
 }
 
-static uint8_t *upb_parse_value(uint8_t *buf, uint8_t *end, upb_field_type_t ft,
-                                union upb_value_ptr v, struct upb_status *status)
+static const uint8_t *upb_parse_value(const uint8_t *buf, const uint8_t *end,
+                                      upb_field_type_t ft,
+                                      union upb_value_ptr v,
+                                      struct upb_status *status)
 {
 #define CASE(t, member_name) \
   case UPB_TYPE(t): return upb_get_ ## t(buf, end, v.member_name, status);
@@ -349,7 +360,7 @@ void upb_cbparser_reset(struct upb_cbparser *p, void *udata)
   p->top->end_offset = 0;
 }
 
-static void *get_msgend(struct upb_cbparser *p, uint8_t *start)
+static const void *get_msgend(struct upb_cbparser *p, const uint8_t *start)
 {
   if(p->top->end_offset > 0)
     return start + (p->top->end_offset - p->completed_offset);
@@ -357,7 +368,7 @@ static void *get_msgend(struct upb_cbparser *p, uint8_t *start)
     return (void*)UINTPTR_MAX;  // group.
 }
 
-static bool isgroup(void *submsg_end)
+static bool isgroup(const void *submsg_end)
 {
   return submsg_end == (void*)UINTPTR_MAX;
 }
@@ -374,9 +385,9 @@ INLINE bool upb_check_type(upb_wire_type_t wt, upb_field_type_t ft) {
  * Pushes a new stack frame for a submessage with the given len (which will
  * be zero if the submessage is a group).
  */
-static uint8_t *push(struct upb_cbparser *p, uint8_t *start,
-                     uint32_t submsg_len, struct upb_fielddef *f,
-                     struct upb_status *status)
+static const uint8_t *push(struct upb_cbparser *p, const uint8_t *start,
+                           uint32_t submsg_len, struct upb_fielddef *f,
+                           struct upb_status *status)
 {
   p->top++;
   if(p->top >= p->limit) {
@@ -397,7 +408,7 @@ static uint8_t *push(struct upb_cbparser *p, uint8_t *start,
  * Pops a stack frame, returning a pointer for where the next submsg should
  * end (or a pointer that is out of range for a group).
  */
-static void *pop(struct upb_cbparser *p, uint8_t *start)
+static const void *pop(struct upb_cbparser *p, const uint8_t *start)
 {
   if(p->end_cb) p->end_cb(p->udata);
   p->top--;
@@ -405,18 +416,18 @@ static void *pop(struct upb_cbparser *p, uint8_t *start)
 }
 
 
-size_t upb_cbparser_parse(struct upb_cbparser *p, void *_buf, size_t len,
+size_t upb_cbparser_parse(struct upb_cbparser *p, upb_string *str,
                           struct upb_status *status)
 {
   // buf is our current offset, moves from start to end.
-  uint8_t *buf = _buf;
-  uint8_t *const start = buf;  // ptr equivalent of p->completed_offset
-  uint8_t *end = buf + len;
+  const uint8_t *buf = (uint8_t*)upb_string_getrobuf(str);
+  const uint8_t *const start = buf;  // ptr equivalent of p->completed_offset
+  const uint8_t *const end = buf + upb_strlen(str);
 
   // When we have fully parsed a tag/value pair, we advance this.
-  uint8_t *completed = buf;
+  const uint8_t *completed = buf;
 
-  uint8_t *submsg_end = get_msgend(p, start);
+  const uint8_t *submsg_end = get_msgend(p, start);
   struct upb_msgdef *msgdef = p->top->msgdef;
   bool keep_going = true;
 
@@ -453,7 +464,7 @@ size_t upb_cbparser_parse(struct upb_cbparser *p, void *_buf, size_t len,
       int32_t delim_len;
       buf = upb_get_INT32(buf, end, &delim_len, status);
       CHECK_STATUS();
-      uint8_t *delim_end = buf + delim_len;
+      const uint8_t *delim_end = buf + delim_len;
       if(f && f->type == UPB_TYPE(MESSAGE)) {
         submsg_end = push(p, start, delim_end - start, f, status);
         msgdef = p->top->msgdef;
diff --git a/src/upb_parse.h b/src/upb_parse.h
index 12e04e1..7086204 100644
--- a/src/upb_parse.h
+++ b/src/upb_parse.h
@@ -40,7 +40,7 @@ typedef bool (*upb_value_cb)(void *udata, struct upb_msgdef *msgdef,
 // available at str.  If the client is streaming and the current buffer ends in
 // the middle of the string, this number could be less than total_len.
 typedef bool (*upb_str_cb)(void *udata, struct upb_msgdef *msgdef,
-                           struct upb_fielddef *f, uint8_t *str,
+                           struct upb_fielddef *f, const uint8_t *str,
                            size_t avail_len, size_t total_len);
 
 // The start and end callbacks are called when a submessage begins and ends,
@@ -80,7 +80,7 @@ void upb_cbparser_reset(struct upb_cbparser *p, void *udata);
 //
 // TODO: see if we can provide the following guarantee efficiently:
 //   retval will always be >= len. */
-size_t upb_cbparser_parse(struct upb_cbparser *p, void *buf, size_t len,
+size_t upb_cbparser_parse(struct upb_cbparser *p, upb_string *str,
                           struct upb_status *status);
 
 /* Pick parser interface. ************************************************/
@@ -117,10 +117,10 @@ struct upb_pickparser *upb_pickparser_new(struct upb_msgdef *msgdef,
                                           char *fields[],
                                           upb_pp_value_cb value_cb,
                                           upb_pp_str_cb str_cb);
-void upb_pickparser_free(struct upb_cbparser *p);
+void upb_pickparser_free(struct upb_pickparser *p);
 void upb_pickparser_reset(struct upb_pickparser *p, void *udata);
-size_t upb_cbparser_parse(struct upb_cbparser *p, void *buf, size_t len,
-                          struct upb_status *status);
+size_t upb_pickparser_parse(struct upb_pickparser *p, upb_string *str,
+                            struct upb_status *status);
 
 #ifdef __cplusplus
 }  /* extern "C" */