Make the compiler correctly define the top-level extern symbol.

1 files changed, 62 insertions, 31 deletions

diff --git a/tools/upbc.c b/tools/upbc.c
index b0f6d6c..b3759be 100644
--- a/tools/upbc.c
+++ b/tools/upbc.c
@@ -20,6 +20,8 @@
 
 /* These are in-place string transformations that do not change the length of
  * the string (and thus never need to re-allocate). */
+
+/* Convert to C identifier: foo.bar.Baz -> foo_bar_Baz. */
 static void to_cident(struct upb_string str)
 {
   for(uint32_t i = 0; i < str.byte_len; i++)
@@ -27,6 +29,7 @@ static void to_cident(struct upb_string str)
       str.ptr[i] = '_';
 }
 
+/* Convert to C proprocessor identifier: foo.bar.Baz -> FOO_BAR_BAZ. */
 static void to_preproc(struct upb_string str)
 {
   to_cident(str);
@@ -58,7 +61,7 @@ void *strtable_to_array(struct upb_strtable *t, int *size)
  *
  * Assumes that d has been validated. */
 static void write_h(struct upb_symtab_entry *entries[], int num_entries,
-                    char *outfile_name, FILE *stream)
+                    char *outfile_name, char *descriptor_cident, FILE *stream)
 {
   /* Header file prologue. */
   struct upb_string include_guard_name = upb_strdupc(outfile_name);
@@ -73,6 +76,12 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
   fputs("extern \"C\" {\n", stream);
   fputs("#endif\n\n", stream);
 
+  if(descriptor_cident) {
+    fputs("struct google_protobuf_FileDescriptorSet;\n", stream);
+    fprintf(stream, "extern struct google_protobuf_FileDescriptorSet *%s;\n\n",
+            descriptor_cident);
+  }
+
   /* Enums. */
   fprintf(stream, "/* Enums. */\n\n");
   for(int i = 0; i < num_entries; i++) {  /* Foreach enum */
@@ -221,7 +230,7 @@ struct strtable_entry {
 struct typetable_entry {
   struct upb_strtable_entry e;
   struct upb_msg_field *field;
-  struct upb_string c_ident;  /* Type name converted with to_cident(). */
+  struct upb_string cident;  /* Type name converted with to_cident(). */
   /* A list of all values of this type, in an established order. */
   union upb_value *values;
   int values_size, values_len;
@@ -297,11 +306,11 @@ struct typetable_entry *get_or_insert_typeentry(struct upb_strtable *t,
   struct typetable_entry *type_e = upb_strtable_lookup(t, &type_name);
   if(type_e == NULL) {
     struct typetable_entry new_type_e = {
-      .e = {.key = type_name}, .field = f, .c_ident = upb_strdup(type_name),
+      .e = {.key = type_name}, .field = f, .cident = upb_strdup(type_name),
       .values = NULL, .values_size = 0, .values_len = 0,
       .arrays = NULL, .arrays_size = 0, .arrays_len = 0
     };
-    to_cident(new_type_e.c_ident);
+    to_cident(new_type_e.cident);
     upb_strtable_insert(t, &new_type_e.e);
     type_e = upb_strtable_lookup(t, &type_name);
     assert(type_e);
@@ -362,8 +371,8 @@ static void add_submsgs(void *data, struct upb_msg *m, struct upb_strtable *t)
 
 /* write_messages_c emits a .c file that contains the data of a protobuf,
  * serialized as C structures. */
-static void write_messages_c(void *data, struct upb_msg *m,
-                             char *hfile_name, FILE *stream)
+static void write_message_c(void *data, struct upb_msg *m,
+                            char *cident, char *hfile_name, FILE *stream)
 {
   fputs("/* This file was generated by upbc (the upb compiler).  "
         "Do not edit. */\n\n", stream),
@@ -424,24 +433,24 @@ static void write_messages_c(void *data, struct upb_msg *m,
   struct typetable_entry *e = upb_strtable_begin(&types);
   for(; e; e = upb_strtable_next(&types, &e->e)) {
     fprintf(stream, "static " UPB_STRFMT " " UPB_STRFMT "_values[%d];\n\n",
-            UPB_STRARG(e->c_ident), UPB_STRARG(e->c_ident), e->values_len);
+            UPB_STRARG(e->cident), UPB_STRARG(e->cident), e->values_len);
     if(e->arrays_len > 0) {
       fprintf(stream, "static " UPB_STRFMT " *" UPB_STRFMT "_array_elems[] = {\n",
-              UPB_STRARG(e->c_ident), UPB_STRARG(e->c_ident));
+              UPB_STRARG(e->cident), UPB_STRARG(e->cident));
       for(int i = 0; i < e->arrays_len; i++) {
         struct array *arr = &e->arrays[i];
         for(int j = 0; j < arr->len; j++)
-          fprintf(stream, "    &" UPB_STRFMT "_values[%d],\n", UPB_STRARG(e->c_ident), arr->offset + j);
+          fprintf(stream, "    &" UPB_STRFMT "_values[%d],\n", UPB_STRARG(e->cident), arr->offset + j);
       }
       fprintf(stream, "};\n");
 
       int cum_offset = 0;
       fprintf(stream, "static UPB_MSG_ARRAY(" UPB_STRFMT ") " UPB_STRFMT "_arrays[%d] = {\n",
-              UPB_STRARG(e->c_ident), UPB_STRARG(e->c_ident), e->arrays_len);
+              UPB_STRARG(e->cident), UPB_STRARG(e->cident), e->arrays_len);
       for(int i = 0; i < e->arrays_len; i++) {
         struct array *arr = &e->arrays[i];
         fprintf(stream, "  {.elements = &" UPB_STRFMT "_array_elems[%d], .len=%d},\n",
-                UPB_STRARG(e->c_ident), cum_offset, arr->len);
+                UPB_STRARG(e->cident), cum_offset, arr->len);
         cum_offset += arr->len;
       }
       fprintf(stream, "};\n");
@@ -451,19 +460,19 @@ static void write_messages_c(void *data, struct upb_msg *m,
   /* Emit definitions. */
   for(e = upb_strtable_begin(&types); e; e = upb_strtable_next(&types, &e->e)) {
     fprintf(stream, "static " UPB_STRFMT " " UPB_STRFMT "_values[%d] = {\n\n",
-            UPB_STRARG(e->c_ident), UPB_STRARG(e->c_ident), e->values_len);
+            UPB_STRARG(e->cident), UPB_STRARG(e->cident), e->values_len);
     for(int i = 0; i < e->values_len; i++) {
       union upb_value val = e->values[i];
       if(upb_issubmsg(e->field)) {
         struct upb_msg *m = e->field->ref.msg;
         void *msgdata = val.msg;
         /* Print set flags. */
-        fprintf(stream, "  {.set_flags = {.bytes = {");
+        fputs("  {.set_flags = {.bytes = {", stream);
         for(unsigned int j = 0; j < m->set_flags_bytes; j++) {
           fprintf(stream, "0x%02hhx", *(uint8_t*)(val.msg + j));
-          if(j < m->set_flags_bytes - 1) fprintf(stream, ", ");
+          if(j < m->set_flags_bytes - 1) fputs(", ", stream);
         }
-        fprintf(stream, "}},\n");
+        fputs("}},\n", stream);
         /* Print msg data. */
         for(unsigned int j = 0; j < m->num_fields; j++) {
           struct upb_msg_field *f = &m->fields[j];
@@ -471,10 +480,10 @@ static void write_messages_c(void *data, struct upb_msg *m,
           union upb_value val = upb_msg_get(msgdata, f);
           fprintf(stream, "    ." UPB_STRFMT " = ", UPB_STRARG(*fd->name));
           if(!upb_msg_isset(msgdata, f)) {
-            fprintf(stream, "0,   /* Not set. */");
+            fputs("0,   /* Not set. */", stream);
           } else if(upb_isstring(f)) {
             if(upb_isarray(f)) {
-              fprintf(stderr, "Ack, string arrays are not supported yet!\n");
+              fputs("Ack, string arrays are not supported yet!\n", stderr);
               exit(1);
             } else {
               struct strtable_entry *str_e = upb_strtable_lookup(&strings, val.str);
@@ -494,7 +503,7 @@ static void write_messages_c(void *data, struct upb_msg *m,
               }
             }
             assert(arr_num != -1);
-            fprintf(stream, "&" UPB_STRFMT "_arrays[%d],", UPB_STRARG(type_e->c_ident), arr_num);
+            fprintf(stream, "&" UPB_STRFMT "_arrays[%d],", UPB_STRARG(type_e->cident), arr_num);
           } else if(upb_issubmsg(f)) {
             /* Find this submessage in the list of msgs for that type. */
             struct typetable_entry  *type_e = get_or_insert_typeentry(&types, f);
@@ -507,14 +516,14 @@ static void write_messages_c(void *data, struct upb_msg *m,
               }
             }
             assert(msg_num != -1);
-            fprintf(stream, "&" UPB_STRFMT "_values[%d],", UPB_STRARG(type_e->c_ident), msg_num);
+            fprintf(stream, "&" UPB_STRFMT "_values[%d],", UPB_STRARG(type_e->cident), msg_num);
           } else {
             upb_text_printval(f->type, val, stream);
-            fprintf(stream, ",");
+            fputs(",", stream);
           }
-          fprintf(stream, "\n");
+          fputs("\n", stream);
         }
-        fprintf(stream, "  },\n");
+        fputs("  },\n", stream);
       } else if(upb_isstring(e->field)) {
 
       } else {
@@ -522,15 +531,28 @@ static void write_messages_c(void *data, struct upb_msg *m,
         upb_text_printval(e->field->type, val, stream);
       }
     }
-    fprintf(stream, "};\n");
+    fputs("};\n", stream);
   }
+
+  struct typetable_entry *toplevel_type =
+      get_or_insert_typeentry(&types, &fake_field);
+  assert(toplevel_type);
+  fputs("/* The externally-visible definition. */\n", stream);
+  /* It is always at offset zero, because we add it first. */
+  fprintf(stream, UPB_STRFMT " *%s = &" UPB_STRFMT "_values[0];\n",
+          UPB_STRARG(toplevel_type->cident), cident,
+          UPB_STRARG(toplevel_type->cident));
 }
 
 const char usage[] =
   "upbc -- upb compiler.\n"
   "upb v0.1  http://blog.reverberate.org/upb/\n"
   "\n"
-  "Usage: upbc [options] input-file\n"
+  "Usage: upbc [options] descriptor-file\n"
+  "\n"
+  "  -i C-IDENFITER     Output the descriptor as a C data structure with the\n"
+  "                     given identifier (otherwise only a header will be\n"
+  "                     generated\n"
   "\n"
   "  -o OUTFILE-BASE    Write to OUTFILE-BASE.h and OUTFILE-BASE.c instead\n"
   "                     of using the input file as a basename.\n"
@@ -552,7 +574,7 @@ void error(char *err)
 int main(int argc, char *argv[])
 {
   /* Parse arguments. */
-  char *outfile_base = NULL, *input_file = NULL;
+  char *outfile_base = NULL, *input_file = NULL, *cident = NULL;
   for(int i = 1; i < argc; i++) {
     if(strcmp(argv[i], "-o") == 0) {
       if(++i == argc)
@@ -560,6 +582,12 @@ int main(int argc, char *argv[])
       else if(outfile_base)
         usage_err("-o was specified multiple times.");
       outfile_base = argv[i];
+    } else if(strcmp(argv[i], "-i") == 0) {
+      if(++i == argc)
+        usage_err("-i must be followed by a C-IDENTIFIER.");
+      else if(cident)
+        usage_err("-i was specified multiple times.");
+      cident = argv[i];
     } else {
       if(input_file)
         usage_err("You can only specify one input file.");
@@ -591,18 +619,21 @@ int main(int argc, char *argv[])
      snprintf(c_filename, maxsize, "%s.c", outfile_base) >= maxsize)
     error("File base too long.\n");
 
-  FILE *h_file = fopen(h_filename, "w"), *c_file = fopen(c_filename, "w");
-  if(!h_file || !c_file)
-    error("Failed to open output file(s)");
+  FILE *h_file = fopen(h_filename, "w");
+  if(!h_file) error("Failed to open .h output file");
 
   int symcount;
   struct upb_symtab_entry **entries = strtable_to_array(&c.symtab, &symcount);
-  write_h(entries, symcount, h_filename, h_file);
-  write_messages_c(fds, c.fds_msg, h_filename, c_file);
+  write_h(entries, symcount, h_filename, cident, h_file);
+  if(cident) {
+    FILE *c_file = fopen(c_filename, "w");
+    if(!c_file) error("Failed to open .h output file");
+    write_message_c(fds, c.fds_msg, cident, h_filename, c_file);
+    fclose(c_file);
+  }
   upb_context_free(&c);
   upb_strfree(descriptor);
   fclose(h_file);
-  fclose(c_file);
 
   return 0;
 }