input.cc

早期freebsd实现

static symbol read_escape_name()
{
  int c = get_char_for_escape_name();
  if (c == 0)
    return NULL_SYMBOL;
  if (c == '(')
    return read_two_char_escape_name();
  if (c == '[' && !compatible_flag)
    return read_long_escape_name();
  char buf[2];
  buf[0] = c;
  buf[1] = '\0';
  return symbol(buf);
}

static symbol read_increment_and_escape_name(int *incp)
{
  int c = get_char_for_escape_name();
  switch (c) {
  case 0:
    *incp = 0;
    return NULL_SYMBOL;
  case '(':
    *incp = 0;
    return read_two_char_escape_name();
  case '+':
    *incp = 1;
    return read_escape_name();
  case '-':
    *incp = -1;
    return read_escape_name();
  case '[':
    if (!compatible_flag) {
      *incp = 0;
      return read_long_escape_name();
    }
    break;
  }
  *incp = 0;
  char buf[2];
  buf[0] = c;
  buf[1] = '\0';
  return symbol(buf);
}

static int get_copy(node **nd, int defining)
{
  for (;;) {
    int c = input_stack::get(nd);
    if (c == ESCAPE_NEWLINE) {
      if (defining)
	return c;
      do {
	c = input_stack::get(nd);
      } while (c == ESCAPE_NEWLINE);
    }
    if (c != escape_char || escape_char <= 0)
      return c;
    c = input_stack::peek();
    switch(c) {
    case 0:
      return escape_char;
    case '"':
      (void)input_stack::get(NULL);
      while ((c = input_stack::get(NULL)) != '\n' && c != EOF)
	;
      return c;
    case '#':			// Like \" but newline is ignored.
      (void)input_stack::get(NULL);
      while ((c = input_stack::get(NULL)) != '\n')
	if (c == EOF)
	  return EOF;
      break;
    case '$':
      {
	(void)input_stack::get(NULL);
	symbol s = read_escape_name();
	if (!s.is_null())
	  interpolate_arg(s);
	break;
      }
    case '*':
      {
	(void)input_stack::get(NULL);
	symbol s = read_escape_name();
	if (!s.is_null())
	  interpolate_string(s);
	break;
      }
    case 'a':
      (void)input_stack::get(NULL);
      return '\001';
    case 'e':
      (void)input_stack::get(NULL);
      return ESCAPE_e;
    case 'E':
      (void)input_stack::get(NULL);
      return ESCAPE_E;
    case 'n':
      {
	(void)input_stack::get(NULL);
	int inc;
	symbol s = read_increment_and_escape_name(&inc);
	if (!s.is_null())
	  interpolate_number_reg(s, inc);
	break;
      }
    case 'g':
      {
        (void)input_stack::get(NULL);
        symbol s = read_escape_name();
	if (!s.is_null())
	  interpolate_number_format(s);
        break;
      }
    case 't':
      (void)input_stack::get(NULL);
      return '\t';
    case 'V':
      {
        (void)input_stack::get(NULL);
	symbol s = read_escape_name();
	if (!s.is_null())
	  interpolate_environment_variable(s);
	break;
      }
    case '\n':
      (void)input_stack::get(NULL);
      if (defining)
	return ESCAPE_NEWLINE;
      break;
    case ' ':
      (void)input_stack::get(NULL);
      return ESCAPE_SPACE;
    case '|':
      (void)input_stack::get(NULL);
      return ESCAPE_BAR;
    case '^': 
      (void)input_stack::get(NULL);
      return ESCAPE_CIRCUMFLEX;
    case '{':
      (void)input_stack::get(NULL);
      return ESCAPE_LEFT_BRACE;
    case '}':
      (void)input_stack::get(NULL);
      return ESCAPE_RIGHT_BRACE;
    case '`':
      (void)input_stack::get(NULL);
      return ESCAPE_LEFT_QUOTE;
    case '\'':
      (void)input_stack::get(NULL);
      return ESCAPE_RIGHT_QUOTE;
    case '-':
      (void)input_stack::get(NULL);
      return ESCAPE_HYPHEN;
    case '_':
      (void)input_stack::get(NULL);
      return ESCAPE_UNDERSCORE;
    case 'c':
      (void)input_stack::get(NULL);
      return ESCAPE_c;
    case '!':
      (void)input_stack::get(NULL);
      return ESCAPE_BANG;
    case '?':
      (void)input_stack::get(NULL);
      return ESCAPE_QUESTION;
    case '&':
      (void)input_stack::get(NULL);
      return ESCAPE_AMPERSAND;
    case ')':
      (void)input_stack::get(NULL);
      return ESCAPE_RIGHT_PARENTHESIS;
    case '.':
      (void)input_stack::get(NULL);
      return c;			
    case '%':
      (void)input_stack::get(NULL);
      return ESCAPE_PERCENT;
    default:
      if (c == escape_char) {
	(void)input_stack::get(NULL);
	return c;
      }
      else
	return escape_char;
    }
  }
}

class non_interpreted_char_node : public node {
  unsigned char c;
public:
  non_interpreted_char_node(unsigned char);
  node *copy();
  int interpret(macro *);
  int same(node *);
  const char *type();
};

int non_interpreted_char_node::same(node *nd)
{
  return c == ((non_interpreted_char_node *)nd)->c;
}

const char *non_interpreted_char_node::type()
{
  return "non_interpreted_char_node";
}

non_interpreted_char_node::non_interpreted_char_node(unsigned char n) : c(n)
{
  assert(n != 0);
}

node *non_interpreted_char_node::copy()
{
  return new non_interpreted_char_node(c);
}

int non_interpreted_char_node::interpret(macro *mac)
{
  mac->append(c);
  return 1;
}

static void do_width();
static node *do_non_interpreted();
static node *do_special();
static void do_register();

static node *do_overstrike()
{
  token start;
  overstrike_node *on = new overstrike_node;
  start.next();
  tok.next();
  while (tok != start) {
    if (tok.newline() || tok.eof()) {
      warning(WARN_DELIM, "missing closing delimiter");
      break;
    }
    charinfo *ci = tok.get_char(1);
    if (ci) {
      node *n = curenv->make_char_node(ci);
      if (n)
	on->overstrike(n);
    }
    tok.next();
  }
  return on;
}

static node *do_bracket()
{
  token start;
  bracket_node *bn = new bracket_node;
  start.next();
  tok.next();
  while (tok != start) {
    if (tok.eof()) {
      warning(WARN_DELIM, "missing closing delimiter");
      break;
    }
    if (tok.newline()) {
      warning(WARN_DELIM, "missing closing delimiter");
      input_stack::push(make_temp_iterator("\n"));
      break;
    }
    charinfo *ci = tok.get_char(1);
    if (ci) {
      node *n = curenv->make_char_node(ci);
      if (n)
	bn->bracket(n);
    }
    tok.next();
  }
  return bn;
}

static int do_name_test()
{
  token start;
  start.next();
  int start_level = input_stack::get_level();
  int bad_char = 0;
  int some_char = 0;
  for (;;) {
    tok.next();
    if (tok.newline() || tok.eof()) {
      warning(WARN_DELIM, "missing closing delimiter");
      break;
    }
    if (tok == start
	&& (compatible_flag || input_stack::get_level() == start_level))
      break;
    if (!tok.ch())
      bad_char = 1;
    some_char = 1;
  }
  return some_char && !bad_char;
}

#if 0
static node *do_zero_width()
{
  token start;
  start.next();
  int start_level = input_stack::get_level();
  environment env(curenv);
  environment *oldenv = curenv;
  curenv = &env;
  for (;;) {
    tok.next();
    if (tok.newline() || tok.eof()) {
      error("missing closing delimiter");
      break;
    }
    if (tok == start
	&& (compatible_flag || input_stack::get_level() == start_level))
      break;
    tok.process();
  }
  curenv = oldenv;
  node *rev = env.extract_output_line();
  node *n = 0;
  while (rev) {
    node *tem = rev;
    rev = rev->next;
    tem->next = n;
    n = tem;
  }
  return new zero_width_node(n);
}  

#else

// It's undesirable for \Z to change environments, because then
// \n(.w won't work as expected.

static node *do_zero_width()
{
  node *rev = new dummy_node;
  token start;
  start.next();
  int start_level = input_stack::get_level();
  for (;;) {
    tok.next();
    if (tok.newline() || tok.eof()) {
      warning(WARN_DELIM, "missing closing delimiter");
      break;
    }
    if (tok == start
	&& (compatible_flag || input_stack::get_level() == start_level))
      break;
    if (!tok.add_to_node_list(&rev))
      error("illegal token in argument to \\Z");
  }
  node *n = 0;
  while (rev) {
    node *tem = rev;
    rev = rev->next;
    tem->next = n;
    n = tem;
  }
  return new zero_width_node(n);
}

#endif

token_node *node::get_token_node()
{
  return 0;
}

class token_node : public node {
public:
  token tk;
  token_node(const token &t);
  node *copy();
  token_node *get_token_node();
  int same(node *);
  const char *type();
};

token_node::token_node(const token &t) : tk(t)
{
}

node *token_node::copy()
{
  return new token_node(tk);
}

token_node *token_node::get_token_node()
{
  return this;
}

int token_node::same(node *nd)
{
  return tk == ((token_node *)nd)->tk;
}

const char *token_node::type()
{
  return "token_node";
}

token::token() : nd(0), type(TOKEN_EMPTY)
{
}

token::~token()
{
  delete nd;
}

token::token(const token &t)
: nm(t.nm), c(t.c), val(t.val), dim(t.dim), type(t.type)
{
  // Use two statements to work around bug in SGI C++.
  node *tem = t.nd;
  nd = tem ? tem->copy() : 0;
}

void token::operator=(const token &t)
{
  delete nd;
  nm = t.nm;
  // Use two statements to work around bug in SGI C++.
  node *tem = t.nd;
  nd = tem ? tem->copy() : 0;
  c = t.c;
  val = t.val;
  dim = t.dim;
  type = t.type;
}

void token::skip()
{
  while (space())
    next();
}

int has_arg()
{
  while (tok.space())
    tok.next();
  return !tok.newline();
}

void token::make_space()
{
  type = TOKEN_SPACE;
}

void token::make_newline()
{
  type = TOKEN_NEWLINE;
}

void token::next()
{
  if (nd) {
    delete nd;
    nd = 0;
  }
  units x;
  for (;;) {
    node *n;
    int cc = input_stack::get(&n);
    if (cc != escape_char || escape_char == 0) {
    handle_normal_char:
      switch(cc) {
      case EOF:
	type = TOKEN_EOF;
	return;
      case TRANSPARENT_FILE_REQUEST:
      case TITLE_REQUEST:
      case COPY_FILE_REQUEST:
#ifdef COLUMN
      case VJUSTIFY_REQUEST:
#endif /* COLUMN */
	type = TOKEN_REQUEST;
	c = cc;
	return;
      case BEGIN_TRAP:
	type = TOKEN_BEGIN_TRAP;
	return;
      case END_TRAP:
	type = TOKEN_END_TRAP;
	return;
      case LAST_PAGE_EJECTOR:
	seen_last_page_ejector = 1;
	// fall through
      case PAGE_EJECTOR:
	type = TOKEN_PAGE_EJECTOR;
	return;
      case ESCAPE_PERCENT:
      ESCAPE_PERCENT:
	type = TOKEN_HYPHEN_INDICATOR;
	return;
      case ESCAPE_SPACE:
      ESCAPE_SPACE:
	type = TOKEN_NODE;
	nd = new space_char_hmotion_node(curenv->get_space_width());
	return;
      case ESCAPE_e:
      ESCAPE_e:
	type = TOKEN_ESCAPE;
	return;
      case ESCAPE_E:
	goto handle_escape_char;
      case ESCAPE_BAR:
      ESCAPE_BAR:
	type = TOKEN_NODE;
	nd = new hmotion_node(curenv->get_narrow_space_width());
	return;
      case ESCAPE_CIRCUMFLEX:
      ESCAPE_CIRCUMFLEX:
	type = TOKEN_NODE;
	nd = new hmotion_node(curenv->get_half_narrow_space_width());
	return;
      case ESCAPE_NEWLINE:
	break;
      case ESCAPE_LEFT_BRACE:
      ESCAPE_LEFT_BRACE:
	type = TOKEN_LEFT_BRACE;
	return;
      case ESCAPE_RIGHT_BRACE:
      ESCAPE_RIGHT_BRACE:
	type = TOKEN_RIGHT_BRACE;
	return;
      case ESCAPE_LEFT_QUOTE:
      ESCAPE_LEFT_QUOTE:
	type = TOKEN_SPECIAL;
	nm = symbol("ga");
	return;
      case ESCAPE_RIGHT_QUOTE:
      ESCAPE_RIGHT_QUOTE:
	type = TOKEN_SPECIAL;
	nm = symbol("aa");
	return;
      case ESCAPE_HYPHEN:
      ESCAPE_HYPHEN:
	type = TOKEN_SPECIAL;
	nm = symbol("-");
	return;
      case ESCAPE_UNDERSCORE:
      ESCAPE_UNDERSCORE:
	type = TOKEN_SPECIAL;
	nm = symbol("ul");
	return;
      case ESCAPE_c:
      ESCAPE_c:
	type = TOKEN_INTERRUPT;
	return;
      case ESCAPE_BANG:
      ESCAPE_BANG:
	type = TOKEN_TRANSPARENT;
	return;
      case ESCAPE_QUESTION:
      ESCAPE_QUESTION:
	nd = do_non_interpreted();
	if (nd) {
	  type = TOKEN_NODE;
	  return;
	}
	break;
      case ESCAPE_AMPERSAND:
      ESCAPE_AMPERSAND:
	type = TOKEN_DUMMY;
	return;
      case ESCAPE_RIGHT_PARENTHESIS:
      ESCAPE_RIGHT_PARENTHESIS:
	type = TOKEN_NODE;
	nd = new transparent_dummy_node;
	return;
      case '\b':
	type = TOKEN_BACKSPACE;
	return;
      case ' ':
	type = TOKEN_SPACE;
	return;
      case '\t':
	type = TOKEN_TAB;
	return;
      case '\n':
	type = TOKEN_NEWLINE;
	return;
      case '\001':
	type = TOKEN_LEADER;
	return;
      case 0:
	{
	  assert(n != 0);
	  token_node *tn = n->get_token_node();
	  if (tn) {
	    *this = tn->tk;
	    delete tn;
	  }
	  else {
	    nd = n;
	    type = TOKEN_NODE;
	  }
	}
	return;
      default:
	type = TOKEN_CHAR;
	c = cc;
	return;
      }
    }
    else {
    handle_escape_char:
      cc = input_stack::get(NULL);
      switch(cc) {
      case '(':
	nm = read_two_char_escape_name();
	type = TOKEN_SPECIAL;
	return;
      case EOF:
	type = TOKEN_EOF;
	error("end of input after escape character");
	return;
      case '`':
	goto ESCAPE_LEFT_QUOTE;
      case '\'':
	goto ESCAPE_RIGHT_QUOTE;
      case '-':
	goto ESCAPE_HYPHEN;
      case '_':
	goto ESCAPE_UNDERSCORE;
      case '%':
	goto ESCAPE_PERCENT;
      case ' ':
	goto ESCAPE_SPACE;
      case '0':
	nd = new hmotion_node(curenv->get_digit_width());
	type = TOKEN_NODE;
	return;
      case '|':
	goto ESCAPE_BAR;
      case '^':
	goto ESCAPE_CIRCUMFLEX;
      case '/':
	type = TOKEN_ITALIC_CORRECTION;
	return;
      case ',':
	type = TOKEN_NODE;
	nd = new left_italic_corrected_node;
	return;
      case '&':
	goto ESCAPE_AMPERSAND;
      case ')':
	goto ESCAPE_RIGHT_PARENTHESIS;
      case '!':
	goto ESCAPE_BANG;
      case '?':
	goto ESCAPE_QUESTION;
      case '~':
	nd = new unbreakable_space_node(curenv->get_space_width());
	type = TOKEN_NODE;
	return;
      case '"':
	while ((cc = input_stack::get(NULL)) != '\n' && cc != EOF)