Added an example, constified some more methods.

10 files changed, 173 insertions, 66 deletions

diff --git a/Makefile b/Makefile
index 6acb930..143673c 100644
--- a/Makefile
+++ b/Makefile
@@ -185,7 +185,11 @@ descriptorgen: upb/descriptor.pb tools/upbc
 
 tools/upbc: tools/upbc.c $(LIBUPB)
 	$(E) CC $<
-	$(Q) $(CC) $(CFLAGS) $(CPPFLAGS) $(DEF_OPT) -o $@ $< $(LIBUPB)
+	$(Q) $(CC) $(CFLAGS) $(CPPFLAGS) -o $@ $< $(LIBUPB)
+
+examples/msg: examples/msg.c $(LIBUPB)
+	$(E) CC $<
+	$(Q) $(CC) $(CFLAGS) $(CPPFLAGS) -o $@ $< $(LIBUPB)
 
 # Tests. #######################################################################
 
diff --git a/examples/example.proto b/examples/example.proto
new file mode 100644
index 0000000..231c455
--- /dev/null
+++ b/examples/example.proto
@@ -0,0 +1,15 @@
+//
+// upb - a minimalist implementation of protocol buffers.
+//
+// Copyright (c) 2011 Google Inc.  See LICENSE for details.
+// Author: Josh Haberman <jhaberman@gmail.com>
+//
+// A .proto file for the examples in this directory.
+
+package example;
+
+message SampleMessage {
+  optional string name = 1;
+  optional int32 id = 2;
+  optional string email = 3;
+}
diff --git a/examples/msg.c b/examples/msg.c
new file mode 100644
index 0000000..487e743
--- /dev/null
+++ b/examples/msg.c
@@ -0,0 +1,59 @@
+/*
+ * upb - a minimalist implementation of protocol buffers.
+ *
+ * Copyright (c) 2011 Google Inc.  See LICENSE for details.
+ * Author: Josh Haberman <jhaberman@gmail.com>
+ *
+ * A simple example that demonstrates creating a standard message object
+ * and parsing into it, using a dynamic reflection-based approach.
+ *
+ * Note that with this approach there are no strongly-typed struct or class
+ * definitions to use from C -- this is essentially a reflection-based
+ * interface.  Note that parsing and serializing are still very fast since
+ * they are JIT-based.
+ *
+ * If this seems a bit verbose, you may prefer an approach that generates
+ * strongly-typed struct definitions (upb will likely provide this, but it is
+ * not yet implemented).
+ *
+ * TODO: make this file compiled as part of "make examples"
+ * TODO: test that this actually works (WARNING: UNTESTED!)
+ */
+
+#include "upb/msg.h"
+#include "upb/pb/glue.h"
+
+const char *descfile = "example.proto.pb";
+const char *type = "example.SampleMessage";
+const char *msgfile = "sample_message.pb";
+
+int main() {
+  // First we load the descriptor that describes the message into a upb_msgdef.
+  // This could come from a string that is compiled into the program or from a
+  // separate file as we do here.  Since defs always live in a symtab, we
+  // create one of those also.
+  upb_symtab *s = upb_symtab_new();
+  upb_status status = UPB_STATUS_INIT;
+  if (!upb_load_descriptor_file_into_symtab(s, descfile, &status)) {
+    fprintf(stderr, "Couldn't load descriptor file '%s': %s\n", descfile,
+            upb_status_getstr(&status));
+    return -1;
+  }
+
+  const upb_msgdef *md = upb_symtab_lookupmsg(s, type);
+  if (!md) {
+    fprintf(stderr, "Descriptor did not contain type '%s'\n", type);
+    return -1;
+  }
+
+  // Parse a file into a new message object.
+  void *msg = upb_filetonewmsg(msgfile, md, &status);
+  if (!msg) {
+    fprintf(stderr, "Error parsing message file '%s': %s\n", msgfile,
+            upb_status_getstr(&status));
+    return -1;
+  }
+
+  // TODO: Inspect some fields.
+  return 0;
+}
diff --git a/tests/test_vs_proto2.cc b/tests/test_vs_proto2.cc
index 22aa2e2..222bcdb 100644
--- a/tests/test_vs_proto2.cc
+++ b/tests/test_vs_proto2.cc
@@ -30,8 +30,8 @@ void compare_arrays(const google::protobuf::Reflection *r,
 {
   ASSERT(upb_msg_has(upb_msg, upb_f));
   ASSERT(upb_isseq(upb_f));
-  void *arr = upb_value_getptr(upb_msg_getseq(upb_msg, upb_f));
-  void *iter = upb_seq_begin(arr, upb_f);
+  const void *arr = upb_value_getptr(upb_msg_getseq(upb_msg, upb_f));
+  const void *iter = upb_seq_begin(arr, upb_f);
   for(int i = 0;
       i < r->FieldSize(proto2_msg, proto2_f);
       i++, iter = upb_seq_next(arr, iter, upb_f)) {
diff --git a/upb/def.h b/upb/def.h
index d3c7c07..ab84376 100644
--- a/upb/def.h
+++ b/upb/def.h
@@ -294,13 +294,13 @@ void upb_msgdef_layout(upb_msgdef *m);
 // Looks up a field by name or number.  While these are written to be as fast
 // as possible, it will still be faster to cache the results of this lookup if
 // possible.  These return NULL if no such field is found.
-INLINE upb_fielddef *upb_msgdef_itof(upb_msgdef *m, uint32_t i) {
+INLINE upb_fielddef *upb_msgdef_itof(const upb_msgdef *m, uint32_t i) {
   upb_itof_ent *e = (upb_itof_ent*)
       upb_inttable_fastlookup(&m->itof, i, sizeof(upb_itof_ent));
   return e ? e->f : NULL;
 }
 
-INLINE upb_fielddef *upb_msgdef_ntof(upb_msgdef *m, const char *name) {
+INLINE upb_fielddef *upb_msgdef_ntof(const upb_msgdef *m, const char *name) {
   upb_ntof_ent *e = (upb_ntof_ent*)upb_strtable_lookup(&m->ntof, name);
   return e ? e->f : NULL;
 }
diff --git a/upb/msg.c b/upb/msg.c
index ece22ab..87bb61b 100644
--- a/upb/msg.c
+++ b/upb/msg.c
@@ -89,9 +89,9 @@ void upb_stdmsg_sethas(void *_m, upb_value fval) {
   if (f->hasbit >= 0) m[f->hasbit / 8] |= (1 << (f->hasbit % 8));
 }
 
-bool upb_stdmsg_has(void *_m, upb_value fval) {
+bool upb_stdmsg_has(const void *_m, upb_value fval) {
   assert(_m != NULL);
-  char *m = _m;
+  const char *m = _m;
   const upb_fielddef *f = upb_value_getfielddef(fval);
   return f->hasbit < 0 || (m[f->hasbit / 8] & (1 << (f->hasbit % 8)));
 }
@@ -116,15 +116,15 @@ bool upb_stdmsg_has(void *_m, upb_value fval) {
     return UPB_CONTINUE;                                                      \
   }                                                                           \
                                                                               \
-  upb_value upb_stdmsg_get ## type(void *_m, upb_value fval) {                \
+  upb_value upb_stdmsg_get ## type(const void *_m, upb_value fval) {          \
     assert(_m != NULL);                                                       \
-    uint8_t *m = _m;                                                          \
+    const uint8_t *m = _m;                                                    \
     const upb_fielddef *f = upb_value_getfielddef(fval);                      \
     upb_value ret;                                                            \
     upb_value_set ## type(&ret, *(ctype*)&m[f->offset]);                      \
     return ret;                                                               \
   }                                                                           \
-  upb_value upb_stdmsg_seqget ## type(void *i) {                              \
+  upb_value upb_stdmsg_seqget ## type(const void *i) {                        \
     assert(i != NULL);                                                        \
     upb_value val;                                                            \
     upb_value_set ## type(&val, *(ctype*)i);                                  \
@@ -176,12 +176,12 @@ upb_flow_t upb_stdmsg_setstr_r(void *a, upb_value fval, upb_value val) {
   return UPB_CONTINUE;
 }
 
-upb_value upb_stdmsg_getstr(void *m, upb_value fval) {
+upb_value upb_stdmsg_getstr(const void *m, upb_value fval) {
   assert(m != NULL);
   return upb_stdmsg_getptr(m, fval);
 }
 
-upb_value upb_stdmsg_seqgetstr(void *i) {
+upb_value upb_stdmsg_seqgetstr(const void *i) {
   assert(i != NULL);
   return upb_stdmsg_seqgetptr(i);
 }
@@ -275,15 +275,15 @@ upb_sflow_t upb_stdmsg_startsubmsg_r(void *a, upb_value fval) {
   return UPB_CONTINUE_WITH(*subm);
 }
 
-void *upb_stdmsg_seqbegin(void *_a) {
-  upb_stdarray *a = _a;
+const void *upb_stdmsg_seqbegin(const void *_a) {
+  const upb_stdarray *a = _a;
   return a->len > 0 ? a->ptr : NULL;
 }
 
 #define NEXTFUNC(size) \
-  void *upb_stdmsg_ ## size ## byte_seqnext(void *_a, void *iter) {      \
-    upb_stdarray *a = _a;                                                \
-    void *next = (char*)iter + size;                                     \
+  const void *upb_stdmsg_ ## size ## byte_seqnext(const void *_a, const void *iter) {\
+    const upb_stdarray *a = _a;                                          \
+    const void *next = (char*)iter + size;                               \
     return (char*)next < (char*)a->ptr + (a->len * size) ? next : NULL;  \
   }
 
diff --git a/upb/msg.h b/upb/msg.h
index fa53056..67903d0 100644
--- a/upb/msg.h
+++ b/upb/msg.h
@@ -40,13 +40,13 @@ extern "C" {
 // for one specific upb_fielddef.  Each field has a separate accessor, which
 // lives in the fielddef.
 
-typedef bool upb_has_reader(void *m, upb_value fval);
-typedef upb_value upb_value_reader(void *m, upb_value fval);
+typedef bool upb_has_reader(const void *m, upb_value fval);
+typedef upb_value upb_value_reader(const void *m, upb_value fval);
 
-typedef void *upb_seqbegin_handler(void *s);
-typedef void *upb_seqnext_handler(void *s, void *iter);
-typedef upb_value upb_seqget_handler(void *iter);
-INLINE bool upb_seq_done(void *iter) { return iter == NULL; }
+typedef const void *upb_seqbegin_handler(const void *s);
+typedef const void *upb_seqnext_handler(const void *s, const void *iter);
+typedef upb_value upb_seqget_handler(const void *iter);
+INLINE bool upb_seq_done(const void *iter) { return iter == NULL; }
 
 typedef struct _upb_accessor_vtbl {
   // Writers.  These take an fval as a parameter because the callbacks are used
@@ -97,18 +97,18 @@ INLINE void upb_msg_clearbit(void *msg, const upb_fielddef *f) {
 // or arrays.  Also this could be desired to provide proto2 operations on
 // generated messages.
 
-INLINE bool upb_msg_has(void *m, const upb_fielddef *f) {
+INLINE bool upb_msg_has(const void *m, const upb_fielddef *f) {
   return f->accessor && f->accessor->has(m, f->fval);
 }
 
 // May only be called for fields that have accessors.
-INLINE upb_value upb_msg_get(void *m, const upb_fielddef *f) {
+INLINE upb_value upb_msg_get(const void *m, const upb_fielddef *f) {
   assert(f->accessor && !upb_isseq(f));
   return f->accessor->get(m, f->fval);
 }
 
 // May only be called for fields that have accessors.
-INLINE upb_value upb_msg_getseq(void *m, const upb_fielddef *f) {
+INLINE upb_value upb_msg_getseq(const void *m, const upb_fielddef *f) {
   assert(f->accessor && upb_isseq(f));
   return f->accessor->getseq(m, f->fval);
 }
@@ -118,21 +118,36 @@ INLINE void upb_msg_set(void *m, const upb_fielddef *f, upb_value val) {
   f->accessor->set(m, f->fval, val);
 }
 
-INLINE void *upb_seq_begin(void *s, const upb_fielddef *f) {
+INLINE const void *upb_seq_begin(const void *s, const upb_fielddef *f) {
   assert(f->accessor);
   return f->accessor->seqbegin(s);
 }
-INLINE void *upb_seq_next(void *s, void *iter, const upb_fielddef *f) {
+INLINE const void *upb_seq_next(const void *s, const void *iter,
+                                const upb_fielddef *f) {
   assert(f->accessor);
   assert(!upb_seq_done(iter));
   return f->accessor->seqnext(s, iter);
 }
-INLINE upb_value upb_seq_get(void *iter, const upb_fielddef *f) {
+INLINE upb_value upb_seq_get(const void *iter, const upb_fielddef *f) {
   assert(f->accessor);
   assert(!upb_seq_done(iter));
   return f->accessor->seqget(iter);
 }
 
+INLINE bool upb_msg_has_named(const void *m, const upb_msgdef *md,
+                              const char *field_name) {
+  const upb_fielddef *f = upb_msgdef_ntof(md, field_name);
+  return f && upb_msg_has(m, f);
+}
+
+INLINE bool upb_msg_get_named(const void *m, const upb_msgdef *md,
+                                   const char *field_name, upb_value *val) {
+  const upb_fielddef *f = upb_msgdef_ntof(md, field_name);
+  if (!f) return false;
+  *val = upb_msg_get(m, f);
+  return true;
+}
+
 
 /* upb_msgvisitor *************************************************************/
 
@@ -264,30 +279,30 @@ upb_sflow_t upb_stdmsg_startsubmsg_r(void *c, upb_value fval);
 
 /* Standard readers. **********************************************************/
 
-bool upb_stdmsg_has(void *c, upb_value fval);
-void *upb_stdmsg_seqbegin(void *c);
-
-upb_value upb_stdmsg_getint64(void *c, upb_value fval);
-upb_value upb_stdmsg_getint32(void *c, upb_value fval);
-upb_value upb_stdmsg_getuint64(void *c, upb_value fval);
-upb_value upb_stdmsg_getuint32(void *c, upb_value fval);
-upb_value upb_stdmsg_getdouble(void *c, upb_value fval);
-upb_value upb_stdmsg_getfloat(void *c, upb_value fval);
-upb_value upb_stdmsg_getbool(void *c, upb_value fval);
-upb_value upb_stdmsg_getptr(void *c, upb_value fval);
-
-void *upb_stdmsg_8byte_seqnext(void *c, void *iter);
-void *upb_stdmsg_4byte_seqnext(void *c, void *iter);
-void *upb_stdmsg_1byte_seqnext(void *c, void *iter);
-
-upb_value upb_stdmsg_seqgetint64(void *c);
-upb_value upb_stdmsg_seqgetint32(void *c);
-upb_value upb_stdmsg_seqgetuint64(void *c);
-upb_value upb_stdmsg_seqgetuint32(void *c);
-upb_value upb_stdmsg_seqgetdouble(void *c);
-upb_value upb_stdmsg_seqgetfloat(void *c);
-upb_value upb_stdmsg_seqgetbool(void *c);
-upb_value upb_stdmsg_seqgetptr(void *c);
+bool upb_stdmsg_has(const void *c, upb_value fval);
+const void *upb_stdmsg_seqbegin(const void *c);
+
+upb_value upb_stdmsg_getint64(const void *c, upb_value fval);
+upb_value upb_stdmsg_getint32(const void *c, upb_value fval);
+upb_value upb_stdmsg_getuint64(const void *c, upb_value fval);
+upb_value upb_stdmsg_getuint32(const void *c, upb_value fval);
+upb_value upb_stdmsg_getdouble(const void *c, upb_value fval);
+upb_value upb_stdmsg_getfloat(const void *c, upb_value fval);
+upb_value upb_stdmsg_getbool(const void *c, upb_value fval);
+upb_value upb_stdmsg_getptr(const void *c, upb_value fval);
+
+const void *upb_stdmsg_8byte_seqnext(const void *c, const void *iter);
+const void *upb_stdmsg_4byte_seqnext(const void *c, const void *iter);
+const void *upb_stdmsg_1byte_seqnext(const void *c, const void *iter);
+
+upb_value upb_stdmsg_seqgetint64(const void *c);
+upb_value upb_stdmsg_seqgetint32(const void *c);
+upb_value upb_stdmsg_seqgetuint64(const void *c);
+upb_value upb_stdmsg_seqgetuint32(const void *c);
+upb_value upb_stdmsg_seqgetdouble(const void *c);
+upb_value upb_stdmsg_seqgetfloat(const void *c);
+upb_value upb_stdmsg_seqgetbool(const void *c);
+upb_value upb_stdmsg_seqgetptr(const void *c);
 
 #ifdef __cplusplus
 }  /* extern "C" */
diff --git a/upb/pb/glue.c b/upb/pb/glue.c
index b364a6d..37b86d9 100644
--- a/upb/pb/glue.c
+++ b/upb/pb/glue.c
@@ -30,6 +30,20 @@ void upb_strtomsg(const char *str, size_t len, void *msg, const upb_msgdef *md,
   upb_decoder_uninit(&d);
 }
 
+void *upb_filetonewmsg(const char *fname, const upb_msgdef *md, upb_status *s) {
+  void *msg = upb_stdmsg_new(md);
+  size_t len;
+  char *data = upb_readfile(fname, &len);
+  if (!data) goto err;
+  upb_strtomsg(data, len, msg, md, s);
+  if (!upb_ok(s)) goto err;
+  return msg;
+
+err:
+  upb_stdmsg_free(msg, md);
+  return NULL;
+}
+
 #if 0
 void upb_msgtotext(upb_string *str, upb_msg *msg, upb_msgdef *md,
                    bool single_line) {
diff --git a/upb/pb/glue.h b/upb/pb/glue.h
index 7aa4a5f..38e8d8e 100644
--- a/upb/pb/glue.h
+++ b/upb/pb/glue.h
@@ -28,21 +28,20 @@
 
 #include <stdbool.h>
 #include "upb/upb.h"
+#include "upb/def.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-// Forward-declares so we don't have to include everything in this .h file.
-// Clients should use the regular, typedef'd names (eg. upb_string).
-struct _upb_msg;
-struct _upb_msgdef;
-struct _upb_symtab;
-
 // Decodes the given string, which must be in protobuf binary format, to the
 // given upb_msg with msgdef "md", storing the status of the operation in "s".
 void upb_strtomsg(const char *str, size_t len, void *msg,
-                  const struct _upb_msgdef *md, upb_status *s);
+                  const upb_msgdef *md, upb_status *s);
+
+// Parses the given file into a new message of the given type.  Caller owns
+// the returned message (or NULL if an error occurred).
+void *upb_filetonewmsg(const char *fname, const upb_msgdef *md, upb_status *s);
 
 //void upb_msgtotext(struct _upb_string *str, void *msg,
 //                   struct _upb_msgdef *md, bool single_line);
@@ -56,12 +55,12 @@ upb_def **upb_load_defs_from_descriptor(const char *str, size_t len, int *n,
                                         upb_status *status);
 
 // Like the previous but also adds the loaded defs to the given symtab.
-bool upb_load_descriptor_into_symtab(struct _upb_symtab *symtab, const char *str,
+bool upb_load_descriptor_into_symtab(upb_symtab *symtab, const char *str,
                                      size_t len, upb_status *status);
 
 // Like the previous but also reads the descriptor from the given filename.
-bool upb_load_descriptor_file_into_symtab(struct _upb_symtab *symtab,
-                                          const char *fname, upb_status *status);
+bool upb_load_descriptor_file_into_symtab(upb_symtab *symtab, const char *fname,
+                                          upb_status *status);
 
 // Reads the given filename into a character string, returning NULL if there
 // was an error.
diff --git a/upb/table.h b/upb/table.h
index a8c2015..f410457 100644
--- a/upb/table.h
+++ b/upb/table.h
@@ -149,9 +149,10 @@ INLINE size_t _upb_inttable_entrysize(size_t value_size) {
   return upb_align_up(sizeof(upb_inttable_header) + value_size, 8);
 }
 
-INLINE void *upb_inttable_fastlookup(upb_inttable *t, uint32_t key,
-                                      uint32_t value_size) {
-  return _upb_inttable_fastlookup(t, key, _upb_inttable_entrysize(value_size), value_size);
+INLINE void *upb_inttable_fastlookup(const upb_inttable *t, uint32_t key,
+                                     uint32_t value_size) {
+  return _upb_inttable_fastlookup(
+      t, key, _upb_inttable_entrysize(value_size), value_size);
 }
 
 INLINE void *upb_inttable_lookup(upb_inttable *t, uint32_t key) {