Significant memory-management refactoring any Python extension.

1 files changed, 81 insertions, 43 deletions

diff --git a/tools/upbc.c b/tools/upbc.c
index a324971..885a259 100644
--- a/tools/upbc.c
+++ b/tools/upbc.c
@@ -15,6 +15,8 @@
 #include "upb_enum.h"
 #include "upb_msg.h"
 #include "upb_text.h"
+#include "upb_array.h"
+#include "upb_mm.h"
 
 /* These are in-place string transformations that do not change the length of
  * the string (and thus never need to re-allocate). */
@@ -54,32 +56,24 @@ void *strtable_to_array(struct upb_strtable *t, int *size)
   return array;
 }
 
-/* The .h file defines structs for the types defined in the .proto file.  It
- * also defines constants for the enum values.
- *
- * Assumes that d has been validated. */
-static void write_h(struct upb_symtab_entry *entries[], int num_entries,
-                    char *outfile_name, char *descriptor_cident, FILE *stream)
+/* The _const.h file defines the constants (enums) defined in the .proto
+ * file. */
+static void write_const_h(struct upb_symtab_entry *entries[], int num_entries,
+                          char *outfile_name, FILE *stream)
 {
   /* Header file prologue. */
   struct upb_string *include_guard_name = upb_strdupc(outfile_name);
   to_preproc(include_guard_name);
+  /* A bit cheesy, but will do the job. */
+  include_guard_name->ptr[include_guard_name->byte_len-1] = 'C';
   fputs("/* This file was generated by upbc (the upb compiler).  "
         "Do not edit. */\n\n", stream),
   fprintf(stream, "#ifndef " UPB_STRFMT "\n", UPB_STRARG(include_guard_name));
   fprintf(stream, "#define " UPB_STRFMT "\n\n", UPB_STRARG(include_guard_name));
-  fputs("#include <upb_string.h>\n\n", stream);
-  fputs("#include <upb_array.h>\n\n", stream);
   fputs("#ifdef __cplusplus\n", stream);
   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 */
@@ -109,12 +103,45 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
                 UPB_STRARG(enum_val_prefix), UPB_STRARG(value_name), v->number);
         if(j != ed->value->len-1) fputc(',', stream);
         fputc('\n', stream);
-        upb_strfree(value_name);
+        upb_string_unref(value_name);
       }
     }
     fprintf(stream, "} " UPB_STRFMT ";\n\n", UPB_STRARG(enum_name));
-    upb_strfree(enum_name);
-    upb_strfree(enum_val_prefix);
+    upb_string_unref(enum_name);
+    upb_string_unref(enum_val_prefix);
+  }
+
+  /* Epilogue. */
+  fputs("#ifdef __cplusplus\n", stream);
+  fputs("}  /* extern \"C\" */\n", stream);
+  fputs("#endif\n\n", stream);
+  fprintf(stream, "#endif  /* " UPB_STRFMT " */\n", UPB_STRARG(include_guard_name));
+  upb_string_unref(include_guard_name);
+}
+
+/* The .h file defines structs for the types defined in the .proto file.  It
+ * also defines constants for the enum values.
+ *
+ * Assumes that d has been validated. */
+static void write_h(struct upb_symtab_entry *entries[], int num_entries,
+                    char *outfile_name, char *descriptor_cident, FILE *stream)
+{
+  /* Header file prologue. */
+  struct upb_string *include_guard_name = upb_strdupc(outfile_name);
+  to_preproc(include_guard_name);
+  fputs("/* This file was generated by upbc (the upb compiler).  "
+        "Do not edit. */\n\n", stream),
+  fprintf(stream, "#ifndef " UPB_STRFMT "\n", UPB_STRARG(include_guard_name));
+  fprintf(stream, "#define " UPB_STRFMT "\n\n", UPB_STRARG(include_guard_name));
+  fputs("#include <upb_struct.h>\n\n", stream);
+  fputs("#ifdef __cplusplus\n", stream);
+  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);
   }
 
   /* Forward declarations. */
@@ -131,7 +158,7 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
     fprintf(stream, "struct " UPB_STRFMT ";\n", UPB_STRARG(msg_name));
     fprintf(stream, "typedef struct " UPB_STRFMT "\n    " UPB_STRFMT ";\n\n",
             UPB_STRARG(msg_name), UPB_STRARG(msg_name));
-    upb_strfree(msg_name);
+    upb_string_unref(msg_name);
   }
 
   /* Message Declarations. */
@@ -144,8 +171,8 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
     struct upb_string *msg_name = upb_strdup(&entry->e.key);
     to_cident(msg_name);
     fprintf(stream, "struct " UPB_STRFMT " {\n", UPB_STRARG(msg_name));
+    fputs("  struct upb_mmhead mmhead;\n", stream);
     fputs("  struct upb_msgdef *def;\n", stream);
-    fputs("  void *gptr;\n", stream);
     fputs("  union {\n", stream);
     fprintf(stream, "    uint8_t bytes[%" PRIu32 "];\n", m->set_flags_bytes);
     fputs("    struct {\n", stream);
@@ -179,7 +206,7 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
           fprintf(stream, "  " UPB_STRFMT "* " UPB_STRFMT ";\n",
                   UPB_STRARG(type_name), UPB_STRARG(fd->name));
         }
-        upb_strfree(type_name);
+        upb_string_unref(type_name);
       } else if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) {
         static char* c_types[] = {
           "", "struct upb_double_array*", "struct upb_float_array*",
@@ -208,7 +235,7 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
     fputs("};\n", stream);
     fprintf(stream, "UPB_DEFINE_MSG_ARRAY(" UPB_STRFMT ")\n\n",
             UPB_STRARG(msg_name));
-    upb_strfree(msg_name);
+    upb_string_unref(msg_name);
   }
 
   /* Epilogue. */
@@ -216,7 +243,7 @@ static void write_h(struct upb_symtab_entry *entries[], int num_entries,
   fputs("}  /* extern \"C\" */\n", stream);
   fputs("#endif\n\n", stream);
   fprintf(stream, "#endif  /* " UPB_STRFMT " */\n", UPB_STRARG(include_guard_name));
-  upb_strfree(include_guard_name);
+  upb_string_unref(include_guard_name);
 }
 
 /* Format of table entries that we use when analyzing data structures for
@@ -300,24 +327,28 @@ static void add_strings_from_msg(void *data, struct upb_msgdef *m,
 struct typetable_entry *get_or_insert_typeentry(struct upb_strtable *t,
                                                 struct upb_msg_fielddef *f)
 {
-  struct upb_string type_name = upb_issubmsg(f) ? f->ref.msg->fqname :
-                                                  upb_type_info[f->type].ctype;
-  struct typetable_entry *type_e = upb_strtable_lookup(t, &type_name);
+  struct upb_string *type_name = upb_issubmsg(f) ? upb_strdup(&f->ref.msg->fqname) :
+                                                   upb_strdupc(upb_type_info[f->type].ctype);
+  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, .cident = 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.cident);
+    assert(upb_strtable_lookup(t, type_name) == NULL);
+    assert(upb_strtable_lookup(t, &new_type_e.e.key) == NULL);
     upb_strtable_insert(t, &new_type_e.e);
-    type_e = upb_strtable_lookup(t, &type_name);
+    type_e = upb_strtable_lookup(t, type_name);
     assert(type_e);
+  } else {
+    upb_string_unref(type_name);
   }
   return type_e;
 }
 
-static void add_value(union upb_value value, struct upb_msg_fielddef *f,
+static void add_value(union upb_value_ptr p, struct upb_msg_fielddef *f,
                       struct upb_strtable *t)
 {
   struct typetable_entry *type_e = get_or_insert_typeentry(t, f);
@@ -325,7 +356,7 @@ static void add_value(union upb_value value, struct upb_msg_fielddef *f,
     type_e->values_size = UPB_MAX(type_e->values_size * 2, 4);
     type_e->values = realloc(type_e->values, sizeof(*type_e->values) * type_e->values_size);
   }
-  type_e->values[type_e->values_len++] = value;
+  type_e->values[type_e->values_len++] = upb_value_read(p, f->type);
 }
 
 static void add_submsgs(void *data, struct upb_msgdef *m, struct upb_strtable *t)
@@ -353,7 +384,7 @@ static void add_submsgs(void *data, struct upb_msgdef *m, struct upb_strtable *t
 
       /* Add the individual values in the array. */
       for(uint32_t j = 0; j < arr->len; j++)
-        add_value(upb_array_getelement(arr, j, f->type), f, t);
+        add_value(upb_array_getelementptr(arr, j, f->type), f, t);
 
       /* Add submsgs.  We must do this separately so that the msgs in this
        * array are contiguous (and don't have submsgs of the same type
@@ -362,7 +393,7 @@ static void add_submsgs(void *data, struct upb_msgdef *m, struct upb_strtable *t
         add_submsgs(*upb_array_getelementptr(arr, j, f->type).msg, f->ref.msg, t);
     } else {
       if(!upb_issubmsg(f)) continue;
-      add_value(upb_deref(p, f->type), f, t);
+      add_value(p, f, t);
       add_submsgs(*p.msg, f->ref.msg, t);
     }
   }
@@ -445,7 +476,8 @@ static void write_message_c(void *data, struct upb_msgdef *m,
       .type = GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_MESSAGE,
       .ref = {.msg = m}
   };
-  add_value(val, &fake_field, &types);
+  union upb_value_ptr p = UPB_VALUE_ADDROF(val);
+  add_value(p, &fake_field, &types);
   add_submsgs(data, m, &types);
 
   /* Emit foward declarations for all msgs of all types, and define arrays. */
@@ -503,7 +535,7 @@ static void write_message_c(void *data, struct upb_msgdef *m,
         for(unsigned int j = 0; j < m->num_fields; j++) {
           struct upb_msg_fielddef *f = &m->fields[j];
           google_protobuf_FieldDescriptorProto *fd = m->field_descriptors[j];
-          union upb_value val = upb_msg_get(msgdata, f);
+          union upb_value val = upb_value_read(upb_msg_getptr(msgdata, f), f->type);
           fprintf(stream, "    ." UPB_STRFMT " = ", UPB_STRARG(fd->name));
           if(!upb_msg_isset(msgdata, f)) {
             fputs("0,   /* Not set. */", stream);
@@ -571,7 +603,7 @@ static void write_message_c(void *data, struct upb_msgdef *m,
 
   /* Free tables. */
   for(e = upb_strtable_begin(&types); e; e = upb_strtable_next(&types, &e->e)) {
-    upb_strfree(e->cident);
+    upb_string_unref(e->cident);
     free(e->values);
     free(e->arrays);
   }
@@ -641,16 +673,17 @@ int main(int argc, char *argv[])
   if(!outfile_base) outfile_base = input_file;
 
   /* Read input file. */
-  struct upb_string *descriptor = upb_strnew();
-  if(!upb_strreadfile(input_file, descriptor))
+  struct upb_string *descriptor = upb_strreadfile(input_file);
+  if(!descriptor)
     error("Couldn't read input file.");
 
   /* Parse input file. */
   struct upb_context *c = upb_context_new();
-  struct upb_msg *fds_msg = upb_msg_parsenew(c->fds_msg, descriptor);
-  google_protobuf_FileDescriptorSet *fds = (void*)fds_msg;
-  if(!fds)
+  struct upb_msg *fds_msg = upb_msg_new(c->fds_msg);
+  if(upb_msg_parsestr(fds_msg, descriptor->ptr, descriptor->byte_len) != UPB_STATUS_OK)
     error("Failed to parse input file descriptor.");
+  //upb_msg_print(fds_msg, false, stderr);
+  google_protobuf_FileDescriptorSet *fds = (void*)fds_msg;
   if(!upb_context_addfds(c, fds))
     error("Failed to resolve symbols in descriptor.\n");
 
@@ -666,17 +699,21 @@ int main(int argc, char *argv[])
 
   /* Emit output files. */
   const int maxsize = 256;
-  char h_filename[maxsize], c_filename[maxsize];
+  char h_filename[maxsize], h_const_filename[maxsize], c_filename[maxsize];
   if(snprintf(h_filename, maxsize, "%s.h", outfile_base) >= maxsize ||
-     snprintf(c_filename, maxsize, "%s.c", outfile_base) >= maxsize)
+     snprintf(c_filename, maxsize, "%s.c", outfile_base) >= maxsize ||
+     snprintf(h_const_filename, maxsize, "%s_const.h", outfile_base) >= maxsize)
     error("File base too long.\n");
 
   FILE *h_file = fopen(h_filename, "w");
   if(!h_file) error("Failed to open .h output file");
+  FILE *h_const_file = fopen(h_const_filename, "w");
+  if(!h_const_file) error("Failed to open _const.h output file");
 
   int symcount;
   struct upb_symtab_entry **entries = strtable_to_array(&c->symtab, &symcount);
   write_h(entries, symcount, h_filename, cident, h_file);
+  write_const_h(entries, symcount, h_filename, h_const_file);
   free(entries);
   if(cident) {
     FILE *c_file = fopen(c_filename, "w");
@@ -684,10 +721,11 @@ int main(int argc, char *argv[])
     write_message_c(fds, c->fds_msg, cident, h_filename, argc, argv, input_file, c_file);
     fclose(c_file);
   }
-  upb_msg_free(fds_msg);
+  upb_msg_unref(fds_msg);
   upb_context_unref(c);
-  upb_strfree(descriptor);
+  upb_string_unref(descriptor);
   fclose(h_file);
+  fclose(h_const_file);
 
   return 0;
 }