Can now serialize osh.asdl to oheap!

C++ code to print more of the osh.asdl tree.  This could be generated.

asdl/encode.py

@@ -74,6 +74,8 @@ def Tag(self, i, chunk):
    chunk.append(i & 0xFF)

  def Int(self, n, chunk):
    if n < 0:
      raise Error('ASDL currently supports unsigned integers, got %d' % n)
    if n > self.max_int:
      raise Error('%d is too big to fit in %d bytes' % (n, self.int_width))

@@ -195,17 +197,12 @@ def EncodeObj(obj, enc, out):

  for name in obj.FIELDS:  # encode in order
    desc = obj.DESCRIPTOR_LOOKUP[name]
    #print('\n\n------------')
    #print('field DESC', name, desc)

    field_val = getattr(obj, name)
    #print('VALUE', field_val)

    # TODO:
    # - Float would be inline, etc.
    # - Repeated value: write them all adjacent to each other?


    is_maybe = False
    if isinstance(desc, asdl.MaybeType):
      is_maybe = True

asdl/osh_demo.cc

@@ -10,9 +10,84 @@ int GetRootRef(uint8_t* image) {
  if (image[0] != 'O') return -1;
  if (image[1] != 'H') return -1;
  if (image[2] != 'P') return -1;
  if (image[3] != 4) return -1;
  if (image[3] != 1) return -1;  // version 1
  if (image[4] != 4) return -1;  // alignment 4

  return image[4] + (image[5] << 8) + (image[6] << 16) + (image[7] << 24);
  return image[5] + (image[6] << 8) + (image[7] << 16);
}

void PrintToken(const uint32_t* base, const token_t& e, int indent) {
  for (int i = 0; i < indent; ++i) {
    putchar('\t');
  }
  printf("Token %hhu %s\n", e.id(), e.val(base));
}

void PrintWordPart(const uint32_t* base, const word_part_t& e, int indent) {
  for (int i = 0; i < indent; ++i) {
    putchar('\t');
  }
  printf("t%hhu ", e.tag());
  switch (e.tag()) {
  case word_part_e::LiteralPart: {
    auto& e2 = static_cast<const LiteralPart&>(e);
    printf("LiteralPart\n");
    PrintToken(base, e2.token(base), indent+1);
    break;
  }
  default:
    printf("OTHER word_part\n");
    break;
  }
}

void PrintWord(const uint32_t* base, const word_t& e, int indent) {
  for (int i = 0; i < indent; ++i) {
    putchar('\t');
  }
  printf("t%hhu ", e.tag());
  switch (e.tag()) {
  case word_e::CompoundWord: {
    auto& e2 = static_cast<const CompoundWord&>(e);
    printf("CompoundWord %d\n", e2.parts_size(base));
    for (int i = 0; i < e2.parts_size(base); ++i) {
      PrintWordPart(base, e2.parts(base, i), indent+1);
    }
    break;
  }
  default:
    printf("OTHER word\n");
    break;
  }
}

void PrintCommand(const uint32_t* base, const command_t& e, int indent) {
  for (int i = 0; i < indent; ++i) {
    putchar('\t');
  }
  printf("t%hhu ", e.tag());

  switch (e.tag()) {
  case command_e::SimpleCommand: {
    auto& e2 = static_cast<const SimpleCommand&>(e);
    printf("SimpleCommand %d\n", e2.words_size(base));
    for (int i = 0; i < e2.words_size(base); ++i) {
      PrintWord(base, e2.words(base, i), indent+1);
    }
    break;
  }
  case command_e::CommandList: {
    auto& e2 = static_cast<const CommandList&>(e);
    printf("CommandList %d\n", e2.children_size(base));
    for (int i = 0; i < e2.children_size(base); ++i) {
      PrintCommand(base, e2.children(base, i), indent+1);
    }
    break;
  }
  default:
    printf("OTHER\n");
    break;
  }
}

int main(int argc, char **argv) {
@@ -48,6 +123,6 @@ int main(int argc, char **argv) {
  auto base = reinterpret_cast<uint32_t*>(image);

  size_t offset = alignment * root_ref;
  auto expr = reinterpret_cast<arith_expr_t*>(image + offset);
  //PrintExpr(base, *expr, 0);
  auto expr = reinterpret_cast<command_t*>(image + offset);
  PrintCommand(base, *expr, 0);
}

asdl/run.sh

@@ -98,7 +98,7 @@ arith-demo() {
  local data=_tmp/${name}.bin

  # Write a binary
  asdl-arith-encode '7 + 9' $data
  asdl-arith-encode '7 * 9' $data

  local bin=_tmp/${name}_demo 

@@ -115,8 +115,10 @@ osh-demo() {
  local name=osh
  local data=_tmp/${name}.bin

  # TODO: Fix bug -- this is only 3 bytes?
  bin/osh -n --ast-format oheap -c 'echo hi  # comment' > $data
  local code='echo hi; echo bye  # comment' 
  #local code='echo $(( 2 + 3 ))'
  #local code='echo $(( -2 * -3 ))'  # test negative integers
  bin/osh -n --ast-format oheap -c "$code" > $data

  ls -l $data