More incremental work; ported some of upbc.

1 files changed, 71 insertions, 63 deletions

diff --git a/tools/upbc.c b/tools/upbc.c
index a6ab289..8e89435 100644
--- a/tools/upbc.c
+++ b/tools/upbc.c
@@ -11,33 +11,36 @@
 #include <ctype.h>
 #include <inttypes.h>
 #include <stdarg.h>
+#include <stdlib.h>
 #include "descriptor.h"
-#include "upb_array.h"
+#include "upb_data.h"
 #include "upb_def.h"
-#include "upb_mm.h"
-#include "upb_msg.h"
 #include "upb_text.h"
 
 /* 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)
+// Convert to C identifier: foo.bar.Baz -> foo_bar_Baz.
+static void to_cident(upb_string *str)
 {
-  for(uint32_t i = 0; i < str->byte_len; i++)
-    if(str->ptr[i] == '.' || str->ptr[i] == '/')
-      str->ptr[i] = '_';
+  upb_strlen_t len = upb_strlen(str);
+  char *buf = upb_string_getrwbuf(str, len);
+  for(uint32_t i = 0; i < len; i++)
+    if(buf[i] == '.' || buf[i] == '/')
+      buf[i] = '_';
 }
 
-/* Convert to C proprocessor identifier: foo.bar.Baz -> FOO_BAR_BAZ. */
-static void to_preproc(struct upb_string *str)
+// Convert to C proprocessor identifier: foo.bar.Baz -> FOO_BAR_BAZ.
+static void to_preproc(upb_string *str)
 {
   to_cident(str);
-  for(uint32_t i = 0; i < str->byte_len; i++)
-    str->ptr[i] = toupper(str->ptr[i]);
+  upb_strlen_t len = upb_strlen(str);
+  char *buf = upb_string_getrwbuf(str, len);
+  for(uint32_t i = 0; i < len; i++)
+    buf[i] = toupper(buf[i]);
 }
 
-static int my_memrchr(char *data, char c, size_t len)
+static int my_memrchr(const char *data, char c, size_t len)
 {
   int off = len-1;
   while(off > 0 && data[off] != c) --off;
@@ -62,10 +65,13 @@ static void write_const_h(struct upb_def *defs[], int num_entries,
                           char *outfile_name, FILE *stream)
 {
   /* Header file prologue. */
-  struct upb_string *include_guard_name = upb_strdupc(outfile_name);
+  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';
+  upb_strlen_t len = upb_strlen(include_guard_name);
+  char *buf = upb_string_getrwbuf(include_guard_name, len);
+  buf[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));
@@ -79,14 +85,14 @@ static void write_const_h(struct upb_def *defs[], int num_entries,
   for(int i = 0; i < num_entries; i++) {  /* Foreach enum */
     if(defs[i]->type != UPB_DEF_ENUM) continue;
     struct upb_enumdef *enumdef = upb_downcast_enumdef(defs[i]);
-    struct upb_string *enum_name = upb_strdup(UPB_UPCAST(enumdef)->fqname);
-    struct upb_string *enum_val_prefix = upb_strdup(enum_name);
+    upb_string *enum_name = upb_strdup(UPB_UPCAST(enumdef)->fqname);
+    upb_string *enum_val_prefix = upb_strdup(enum_name);
     to_cident(enum_name);
 
-    enum_val_prefix->byte_len = my_memrchr(enum_val_prefix->ptr,
-                                           UPB_SYMBOL_SEPARATOR,
-                                           enum_val_prefix->byte_len);
-    enum_val_prefix->byte_len++;
+    const char *data = upb_string_getrobuf(enum_val_prefix);
+    upb_strlen_t len = upb_strlen(enum_val_prefix);
+    upb_strlen_t lastsep = my_memrchr(data, UPB_SYMBOL_SEPARATOR, len);
+    upb_string_resize(enum_val_prefix, lastsep + 1);
     to_preproc(enum_val_prefix);
 
     fprintf(stream, "typedef enum " UPB_STRFMT " {\n", UPB_STRARG(enum_name));
@@ -94,7 +100,7 @@ static void write_const_h(struct upb_def *defs[], int num_entries,
     bool first = true;
     /* Foreach enum value. */
     for(upb_enum_begin(&iter, enumdef); !upb_enum_done(&iter); upb_enum_next(&iter)) {
-      struct upb_string *value_name = upb_strdup(iter.name);
+      upb_string *value_name = upb_strdup(iter.name);
       to_preproc(value_name);
       /* "  GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_UINT32 = 13," */
       if (!first) fputs(",\n", stream);
@@ -124,7 +130,7 @@ static void write_h(struct upb_def *defs[], int num_defs, char *outfile_name,
                     char *descriptor_cident, FILE *stream)
 {
   /* Header file prologue. */
-  struct upb_string *include_guard_name = upb_strdupc(outfile_name);
+  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),
@@ -149,7 +155,7 @@ static void write_h(struct upb_def *defs[], int num_defs, char *outfile_name,
   for(int i = 0; i < num_defs; i++) {  /* Foreach message */
     struct upb_msgdef *m = upb_dyncast_msgdef(defs[i]);
     if(!m) continue;
-    struct upb_string *msg_name = upb_strdup(UPB_UPCAST(m)->fqname);
+    upb_string *msg_name = upb_strdup(UPB_UPCAST(m)->fqname);
     to_cident(msg_name);
     fprintf(stream, "struct " UPB_STRFMT ";\n", UPB_STRARG(msg_name));
     fprintf(stream, "typedef struct " UPB_STRFMT "\n    " UPB_STRFMT ";\n\n",
@@ -162,7 +168,7 @@ static void write_h(struct upb_def *defs[], int num_defs, char *outfile_name,
   for(int i = 0; i < num_defs; i++) {  /* Foreach message */
     struct upb_msgdef *m = upb_dyncast_msgdef(defs[i]);
     if(!m) continue;
-    struct upb_string *msg_name = upb_strdup(UPB_UPCAST(m)->fqname);
+    upb_string *msg_name = upb_strdup(UPB_UPCAST(m)->fqname);
     to_cident(msg_name);
     fprintf(stream, "struct " UPB_STRFMT " {\n", UPB_STRARG(msg_name));
     fputs("  struct upb_mmhead mmhead;\n", stream);
@@ -181,7 +187,7 @@ static void write_h(struct upb_def *defs[], int num_defs, char *outfile_name,
     for(upb_field_count_t j = 0; j < m->num_fields; j++) {
       struct upb_fielddef *f = &m->fields[j];
       if(upb_issubmsg(f)) {
-        struct upb_string *type_name = upb_strdup(f->def->fqname);
+        upb_string *type_name = upb_strdup(f->def->fqname);
         to_cident(type_name);
         if(f->label == GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_LABEL_REPEATED) {
           fprintf(stream, "  UPB_MSG_ARRAY(" UPB_STRFMT ")* " UPB_STRFMT ";\n",
@@ -241,14 +247,14 @@ struct strtable_entry {
 struct typetable_entry {
   struct upb_strtable_entry e;
   struct upb_fielddef *field;
-  struct upb_string *cident;  /* Type name converted with to_cident(). */
+  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;
   struct array {
     int offset;
     int len;
-    struct upb_array *ptr;  /* So we can find it later. */
+    upb_array *ptr;  /* So we can find it later. */
   } *arrays;
   int arrays_size, arrays_len;
 };
@@ -270,32 +276,33 @@ int compare_entries(const void *_e1, const void *_e2)
  *
  * TODO: make these use a generic msg visitor. */
 
-static void add_strings_from_msg(struct upb_msg *msg, struct upb_strtable *t);
+static void add_strings_from_msg(upb_msg *msg, struct upb_msgdef *md,
+                                 struct upb_strtable *t);
 
-static void add_strings_from_value(union upb_value_ptr p,
+static void add_strings_from_value(union upb_value p,
                                    struct upb_fielddef *f,
                                    struct upb_strtable *t)
 {
   if(upb_isstringtype(f->type)) {
-    struct strtable_entry e = {.e = {.key = *p.str}};
+    struct strtable_entry e = {.e = {.key = p.str}};
     if(upb_strtable_lookup(t, e.e.key) == NULL)
       upb_strtable_insert(t, &e.e);
   } else if(upb_issubmsg(f)) {
-    add_strings_from_msg(*p.msg, t);
+    add_strings_from_msg(p.msg, upb_downcast_msgdef(f->def), t);
   }
 }
 
-static void add_strings_from_msg(struct upb_msg *msg, struct upb_strtable *t)
+static void add_strings_from_msg(upb_msg *msg, struct upb_msgdef *md,
+                                 struct upb_strtable *t)
 {
-  struct upb_msgdef *m = msg->def;
-  for(upb_field_count_t i = 0; i < m->num_fields; i++) {
-    struct upb_fielddef *f = &m->fields[i];
-    if(!upb_msg_isset(msg, f)) continue;
-    union upb_value_ptr p = upb_msg_getptr(msg, f);
+  for(upb_field_count_t i = 0; i < md->num_fields; i++) {
+    struct upb_fielddef *f = &md->fields[i];
+    if(!upb_msg_has(msg, f)) continue;
+    union upb_value p = upb_msg_get(msg, f);
     if(upb_isarray(f)) {
-      struct upb_array *arr = *p.arr;
-      for(uint32_t j = 0; j < arr->len; j++)
-        add_strings_from_value(upb_array_getelementptr(arr, j), f, t);
+      upb_array *arr = p.arr;
+      for(uint32_t j = 0; j < upb_array_len(arr); j++)
+        add_strings_from_value(upb_array_get(arr, f, j), f, t);
     } else {
       add_strings_from_value(p, f, t);
     }
@@ -310,8 +317,8 @@ static void add_strings_from_msg(struct upb_msg *msg, struct upb_strtable *t)
 struct typetable_entry *get_or_insert_typeentry(struct upb_strtable *t,
                                                 struct upb_fielddef *f)
 {
-  struct upb_string *type_name = upb_issubmsg(f) ? upb_strdup(f->def->fqname) :
-                                                   upb_strdupc(upb_type_info[f->type].ctype);
+  upb_string *type_name = upb_issubmsg(f) ? upb_strdup(f->def->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 = {
@@ -331,7 +338,7 @@ struct typetable_entry *get_or_insert_typeentry(struct upb_strtable *t,
   return type_e;
 }
 
-static void add_value(union upb_value_ptr p, struct upb_fielddef *f,
+static void add_value(union upb_value v, struct upb_fielddef *f,
                       struct upb_strtable *t)
 {
   struct typetable_entry *type_e = get_or_insert_typeentry(t, f);
@@ -339,19 +346,19 @@ static void add_value(union upb_value_ptr p, struct upb_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++] = upb_value_read(p, f->type);
+  type_e->values[type_e->values_len++] = v;
 }
 
-static void add_submsgs(struct upb_msg *msg, struct upb_strtable *t)
+static void add_submsgs(upb_msg *msg, struct upb_msgdef *md,
+                        struct upb_strtable *t)
 {
-  struct upb_msgdef *m = msg->def;
-  for(upb_field_count_t i = 0; i < m->num_fields; i++) {
-    struct upb_fielddef *f = &m->fields[i];
-    if(!upb_msg_isset(msg, f)) continue;
-    union upb_value_ptr p = upb_msg_getptr(msg, f);
+  for(upb_field_count_t i = 0; i < md->num_fields; i++) {
+    struct upb_fielddef *f = &md->fields[i];
+    if(!upb_msg_has(msg, f)) continue;
+    union upb_value v = upb_msg_get(msg, f);
     if(upb_isarray(f)) {
       if(upb_isstring(f)) continue;  /* Handled by a different code-path. */
-      struct upb_array *arr = *p.arr;
+      upb_array *arr = v.arr;
 
       /* Add to our list of arrays for this type. */
       struct typetable_entry *arr_type_e =
@@ -362,30 +369,31 @@ static void add_submsgs(struct upb_msg *msg, struct upb_strtable *t)
                                      sizeof(*arr_type_e->arrays)*arr_type_e->arrays_size);
       }
       arr_type_e->arrays[arr_type_e->arrays_len].offset = arr_type_e->values_len;
-      arr_type_e->arrays[arr_type_e->arrays_len].len = arr->len;
-      arr_type_e->arrays[arr_type_e->arrays_len].ptr = *p.arr;
+      arr_type_e->arrays[arr_type_e->arrays_len].len = upb_array_len(arr);
+      arr_type_e->arrays[arr_type_e->arrays_len].ptr = v.arr;
       arr_type_e->arrays_len++;
 
       /* Add the individual values in the array. */
-      for(uint32_t j = 0; j < arr->len; j++)
-        add_value(upb_array_getelementptr(arr, j), f, t);
+      for(uint32_t j = 0; j < upb_array_len(arr); j++)
+        add_value(upb_array_get(arr, f, j), 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
        * interleaved). */
-      for(uint32_t j = 0; j < arr->len; j++)
-        add_submsgs(*upb_array_getelementptr(arr, j).msg, t);
+      for(uint32_t j = 0; j < upb_array_len(arr); j++)
+        add_submsgs(upb_array_get(arr, f, j).msg, upb_downcast_msgdef(f->def), t);
     } else {
       if(!upb_issubmsg(f)) continue;
-      add_value(p, f, t);
-      add_submsgs(*p.msg, t);
+      add_value(v, f, t);
+      add_submsgs(v.msg, upb_downcast_msgdef(f->def), t);
     }
   }
 }
 
 /* write_messages_c emits a .c file that contains the data of a protobuf,
  * serialized as C structures. */
-static void write_message_c(struct upb_msg *msg, char *cident, char *hfile_name,
+static void write_message_c(upb_msg *msg, struct upb_msgdef *md,
+                            char *cident, char *hfile_name,
                             int argc, char *argv[], char *infile_name,
                             FILE *stream)
 {
@@ -415,7 +423,7 @@ static void write_message_c(struct upb_msg *msg, char *cident, char *hfile_name,
    * same string more than once. */
   struct upb_strtable strings;
   upb_strtable_init(&strings, 16, sizeof(struct strtable_entry));
-  add_strings_from_msg(msg, &strings);
+  add_strings_from_msg(msg, md, &strings);
 
   int size;
   struct strtable_entry **str_entries = strtable_to_array(&strings, &size);
@@ -427,7 +435,7 @@ static void write_message_c(struct upb_msg *msg, char *cident, char *hfile_name,
   int col = 2;
   int offset = 0;
   for(int i = 0; i < size; i++) {
-    struct upb_string *s = str_entries[i]->e.key;
+    upb_string *s = str_entries[i]->e.key;
     str_entries[i]->offset = offset;
     str_entries[i]->num = i;
     for(uint32_t j = 0; j < s->byte_len; j++) {