path: root/src/upbc.c

/*
 * upb - a minimalist implementation of protocol buffers.
 *
 * upbc is the upb compiler.  This is some deep code that I wish could be
 * easier to understand, but by its nature it is doing some very "meta"
 * kinds of things.
 *
 * Copyright (c) 2009 Joshua Haberman.  See LICENSE for details.
 */

#include <ctype.h>
#include <inttypes.h>
#include <stdarg.h>
#include <stdlib.h>
#include "descriptor.h"
#include "upb_def.h"
#include "upb_msg.h"
#include "upb_glue.h"
#include "upb_strstream.h"
#include "upb_decoder.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(upb_string *str)
{
  upb_strlen_t len = upb_string_len(str);
  char *buf = upb_string_getrwbuf(str, len);
  for(int32_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(upb_string *str)
{
  to_cident(str);
  upb_strlen_t len = upb_string_len(str);
  char *buf = upb_string_getrwbuf(str, len);
  for(int32_t i = 0; i < len; i++)
    buf[i] = toupper(buf[i]);
}

static int my_memrchr(const char *data, char c, size_t len)
{
  int off = len-1;
  while(off > 0 && data[off] != c) --off;
  return off;
}

/* The _const.h file defines the constants (enums) defined in the .proto
 * file. */
static void write_const_h(upb_def *defs[], int num_entries, char *outfile_name,
                          FILE *stream)
{
  /* Header file prologue. */
  upb_string *include_guard_name = upb_strdupc(outfile_name);
  to_preproc(include_guard_name);
  /* A bit cheesy, but will do the job. */
  upb_strlen_t len = upb_string_len(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));
  fprintf(stream, "#define " UPB_STRFMT "\n\n", UPB_STRARG(include_guard_name));
  fputs("#ifdef __cplusplus\n", stream);
  fputs("extern \"C\" {\n", stream);
  fputs("#endif\n\n", stream);

  /* Enums. */
  fprintf(stream, "/* Enums. */\n\n");
  for(int i = 0; i < num_entries; i++) {  /* Foreach enum */
    if(defs[i]->type != UPB_DEF_ENUM) continue;
    upb_enumdef *enumdef = upb_downcast_enumdef(defs[i]);
    upb_string *enum_name = upb_strdup(UPB_UPCAST(enumdef)->fqname);
    upb_string *enum_val_prefix = upb_strdup(enum_name);
    to_cident(enum_name);

    const char *data = upb_string_getrobuf(enum_val_prefix);
    upb_strlen_t len = upb_string_len(enum_val_prefix);
    upb_strlen_t lastsep = my_memrchr(data, UPB_SYMBOL_SEPARATOR, len);
    upb_string_getrwbuf(enum_val_prefix, lastsep + 1);
    to_preproc(enum_val_prefix);

    fprintf(stream, "typedef enum " UPB_STRFMT " {\n", UPB_STRARG(enum_name));
    bool first = true;
    /* Foreach enum value. */
    for (upb_enum_iter iter = upb_enum_begin(enumdef);
         !upb_enum_done(iter);
         iter = upb_enum_next(enumdef, iter)) {
      upb_string *value_name = upb_strdup(upb_enum_iter_name(iter));
      to_preproc(value_name);
      /* "  GOOGLE_PROTOBUF_FIELDDESCRIPTORPROTO_TYPE_UINT32 = 13," */
      if (!first) fputs(",\n", stream);
      first = false;
      fprintf(stream, "  " UPB_STRFMT UPB_STRFMT " = %" PRIu32,
              UPB_STRARG(enum_val_prefix), UPB_STRARG(value_name), upb_enum_iter_number(iter));
      upb_string_unref(value_name);
    }
    fprintf(stream, "\n} " UPB_STRFMT ";\n\n", UPB_STRARG(enum_name));
    upb_string_unref(enum_name);
    upb_string_unref(enum_val_prefix);
  }

  /* Constants for field names and numbers. */
  fprintf(stream, "/* Constants for field names and numbers. */\n\n");
  for(int i = 0; i < num_entries; i++) {  /* Foreach enum */
    upb_msgdef *m = upb_dyncast_msgdef(defs[i]);
    if(!m) continue;
    upb_string *msg_name = upb_strdup(UPB_UPCAST(m)->fqname);
    upb_string *msg_val_prefix = upb_strdup(msg_name);
    to_preproc(msg_val_prefix);
    upb_msg_iter i;
    for(i = upb_msg_begin(m); !upb_msg_done(i); i = upb_msg_next(m, i)) {
      upb_fielddef *f = upb_msg_iter_field(i);
      upb_string *preproc_field_name = upb_strdup(f->name);
      to_preproc(preproc_field_name);
      fprintf(stream, "#define " UPB_STRFMT "_" UPB_STRFMT "_FIELDNUM %d\n",
              UPB_STRARG(msg_val_prefix), UPB_STRARG(preproc_field_name), f->number);
      fprintf(stream, "#define " UPB_STRFMT "_" UPB_STRFMT "_FIELDNAME \""
              UPB_STRFMT "\"\n", UPB_STRARG(msg_val_prefix), UPB_STRARG(preproc_field_name), UPB_STRARG(f->name));
      upb_string_unref(preproc_field_name);
    }
    upb_string_unref(msg_val_prefix);
    upb_string_unref(msg_name);
  }

  /* 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);
}

const char usage[] =
  "upbc -- upb compiler.\n"
  "upb v0.1  http://blog.reverberate.org/upb/\n"
  "\n"
  "Usage: upbc [options] descriptor-file\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"
;

void usage_err(char *err)
{
  fprintf(stderr, "upbc: %s\n\n", err);
  fputs(usage, stderr);
  exit(1);
}

void error(char *err, ...)
{
  va_list args;
  va_start(args, err);
  fprintf(stderr, "upbc: ");
  vfprintf(stderr, err, args);
  va_end(args);
  exit(1);
}

int main(int argc, char *argv[])
{
  /* Parse arguments. */
  char *outfile_base = NULL, *input_file = NULL;
  for(int i = 1; i < argc; i++) {
    if(strcmp(argv[i], "-o") == 0) {
      if(++i == argc)
        usage_err("-o must be followed by a FILE-BASE.");
      else if(outfile_base)
        usage_err("-o was specified multiple times.");
      outfile_base = argv[i];
    } else {
      if(input_file)
        usage_err("You can only specify one input file.");
      input_file = argv[i];
    }
  }
  if(!input_file) usage_err("You must specify an input file.");
  if(!outfile_base) outfile_base = input_file;

  // Read and parse input file.
  upb_string *descriptor = upb_strreadfile(input_file);
  if(!descriptor)
    error("Couldn't read input file.");

  // TODO: make upb_parsedesc use a separate symtab, so we can use it here when
  // importing descriptor.proto.
  upb_symtab *s = upb_symtab_new();
  upb_symtab_add_descriptorproto(s);
  upb_symtab *s2 = upb_symtab_new();
  upb_status status = UPB_STATUS_INIT;

  upb_stringsrc strsrc;
  upb_stringsrc_init(&strsrc);
  upb_stringsrc_reset(&strsrc, descriptor);

  upb_decoder d;
  upb_msgdef *fds_msgdef = upb_symtab_fds_def(s);
  upb_decoder_init(&d, fds_msgdef);
  upb_decoder_reset(&d, upb_stringsrc_bytesrc(&strsrc));

  upb_symtab_addfds(s2, upb_decoder_src(&d), &status);
  upb_stringsrc_uninit(&strsrc);
  upb_decoder_uninit(&d);
  upb_def_unref(UPB_UPCAST(fds_msgdef));

  if(!upb_ok(&status)) {
    upb_printerr(&status);
    error("Failed to parse input file descriptor\n");
  }

  upb_status_uninit(&status);

  /* Emit output files. */
  const int maxsize = 256;
  char h_const_filename[maxsize];
  if(snprintf(h_const_filename, maxsize, "%s_const.h", outfile_base) >= maxsize)
    error("File base too long.\n");

  FILE *h_const_file = fopen(h_const_filename, "w");
  if(!h_const_file) error("Failed to open _const.h output file\n");

  int symcount;
  upb_def **defs = upb_symtab_getdefs(s, &symcount, UPB_DEF_ANY);
  write_const_h(defs, symcount, h_const_filename, h_const_file);
  for (int i = 0; i < symcount; i++) upb_def_unref(defs[i]);
  free(defs);
  upb_string_unref(descriptor);
  upb_symtab_unref(s);
  upb_symtab_unref(s2);
  fclose(h_const_file);

  return 0;
}