scan-gram.l

bison 2.0 主要可以用来做语法分析用的

		     token_name (token_type));
	obstack_sgrow (&obstack_for_string, "{}");
	STRING_FINISH;
	val->chars = last_string;
	BEGIN INITIAL;
	return token_type;
      }
  }

  <<EOF>>  unexpected_eof (scanner_cursor, "{}"); BEGIN INITIAL;
}


  /*---------------------------------------------------------------.
  | Scanning some code in braces (%union and actions). The initial |
  | "{" is already eaten.                                          |
  `---------------------------------------------------------------*/

<SC_BRACED_CODE>
{
  "{"|"<"{splice}"%"  STRING_GROW; braces_level++;
  "%"{splice}">"      STRING_GROW; braces_level--;
  "}" {
    bool outer_brace = --braces_level < 0;

    /* As an undocumented Bison extension, append `;' before the last
       brace in braced code, so that the user code can omit trailing
       `;'.  But do not append `;' if emulating Yacc, since Yacc does
       not append one.

       FIXME: Bison should warn if a semicolon seems to be necessary
       here, and should omit the semicolon if it seems unnecessary
       (e.g., after ';', '{', or '}', each followed by comments or
       white space).  Such a warning shouldn't depend on --yacc; it
       should depend on a new --pedantic option, which would cause
       Bison to warn if it detects an extension to POSIX.  --pedantic
       should also diagnose other Bison extensions like %yacc.
       Perhaps there should also be a GCC-style --pedantic-errors
       option, so that such warnings are diagnosed as errors.  */
    if (outer_brace && token_type == BRACED_CODE && ! yacc_flag)
      obstack_1grow (&obstack_for_string, ';');

    obstack_1grow (&obstack_for_string, '}');

    if (outer_brace)
      {
	STRING_FINISH;
	rule_length++;
	loc->start = code_start;
	val->chars = last_string;
	BEGIN INITIAL;
	return token_type;
      }
  }

  /* Tokenize `<<%' correctly (as `<<' `%') rather than incorrrectly
     (as `<' `<%').  */
  "<"{splice}"<"  STRING_GROW;

  "$"("<"{tag}">")?(-?[0-9]+|"$")  handle_dollar (token_type, yytext, *loc);
  "@"(-?[0-9]+|"$")		   handle_at (token_type, yytext, *loc);

  <<EOF>>  unexpected_eof (code_start, "}"); BEGIN INITIAL;
}


  /*--------------------------------------------------------------.
  | Scanning some prologue: from "%{" (already scanned) to "%}".  |
  `--------------------------------------------------------------*/

<SC_PROLOGUE>
{
  "%}" {
    STRING_FINISH;
    loc->start = code_start;
    val->chars = last_string;
    BEGIN INITIAL;
    return PROLOGUE;
  }

  <<EOF>>  unexpected_eof (code_start, "%}"); BEGIN INITIAL;
}


  /*---------------------------------------------------------------.
  | Scanning the epilogue (everything after the second "%%", which |
  | has already been eaten).                                       |
  `---------------------------------------------------------------*/

<SC_EPILOGUE>
{
  <<EOF>> {
    STRING_FINISH;
    loc->start = code_start;
    val->chars = last_string;
    BEGIN INITIAL;
    return EPILOGUE;
  }
}


  /*-----------------------------------------.
  | Escape M4 quoting characters in C code.  |
  `-----------------------------------------*/

<SC_COMMENT,SC_LINE_COMMENT,SC_STRING,SC_CHARACTER,SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE>
{
  \$	obstack_sgrow (&obstack_for_string, "$][");
  \@	obstack_sgrow (&obstack_for_string, "@@");
  \[	obstack_sgrow (&obstack_for_string, "@{");
  \]	obstack_sgrow (&obstack_for_string, "@}");
}


  /*-----------------------------------------------------.
  | By default, grow the string obstack with the input.  |
  `-----------------------------------------------------*/

<SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE,SC_STRING,SC_CHARACTER,SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>.	|
<SC_COMMENT,SC_LINE_COMMENT,SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE>\n	STRING_GROW;

%%

/* Keeps track of the maximum number of semantic values to the left of
   a handle (those referenced by $0, $-1, etc.) are required by the
   semantic actions of this grammar. */
int max_left_semantic_context = 0;

/* Set *LOC and adjust scanner cursor to account for token TOKEN of
   size SIZE.  */

static void
adjust_location (location *loc, char const *token, size_t size)
{
  int line = scanner_cursor.line;
  int column = scanner_cursor.column;
  char const *p0 = token;
  char const *p = token;
  char const *lim = token + size;

  loc->start = scanner_cursor;

  for (p = token; p < lim; p++)
    switch (*p)
      {
      case '\n':
	line++;
	column = 1;
	p0 = p + 1;
	break;

      case '\t':
	column += mbsnwidth (p0, p - p0, 0);
	column += 8 - ((column - 1) & 7);
	p0 = p + 1;
	break;
      }

  scanner_cursor.line = line;
  scanner_cursor.column = column + mbsnwidth (p0, p - p0, 0);

  loc->end = scanner_cursor;
}


/* Read bytes from FP into buffer BUF of size SIZE.  Return the
   number of bytes read.  Remove '\r' from input, treating \r\n
   and isolated \r as \n.  */

static size_t
no_cr_read (FILE *fp, char *buf, size_t size)
{
  size_t bytes_read = fread (buf, 1, size, fp);
  if (bytes_read)
    {
      char *w = memchr (buf, '\r', bytes_read);
      if (w)
	{
	  char const *r = ++w;
	  char const *lim = buf + bytes_read;

	  for (;;)
	    {
	      /* Found an '\r'.  Treat it like '\n', but ignore any
		 '\n' that immediately follows.  */
	      w[-1] = '\n';
	      if (r == lim)
		{
		  int ch = getc (fp);
		  if (ch != '\n' && ungetc (ch, fp) != ch)
		    break;
		}
	      else if (*r == '\n')
		r++;

	      /* Copy until the next '\r'.  */
	      do
		{
		  if (r == lim)
		    return w - buf;
		}
	      while ((*w++ = *r++) != '\r');
	    }

	  return w - buf;
	}
    }

  return bytes_read;
}


/*------------------------------------------------------------------.
| TEXT is pointing to a wannabee semantic value (i.e., a `$').      |
|                                                                   |
| Possible inputs: $[<TYPENAME>]($|integer)                         |
|                                                                   |
| Output to OBSTACK_FOR_STRING a reference to this semantic value.  |
`------------------------------------------------------------------*/

static inline bool
handle_action_dollar (char *text, location loc)
{
  const char *type_name = NULL;
  char *cp = text + 1;

  if (! current_rule)
    return false;

  /* Get the type name if explicit. */
  if (*cp == '<')
    {
      type_name = ++cp;
      while (*cp != '>')
	++cp;
      *cp = '\0';
      ++cp;
    }

  if (*cp == '$')
    {
      if (!type_name)
	type_name = symbol_list_n_type_name_get (current_rule, loc, 0);
      if (!type_name && typed)
	complain_at (loc, _("$$ of `%s' has no declared type"),
		     current_rule->sym->tag);
      if (!type_name)
	type_name = "";
      obstack_fgrow1 (&obstack_for_string,
		      "]b4_lhs_value([%s])[", type_name);
    }
  else
    {
      long int num;
      set_errno (0);
      num = strtol (cp, 0, 10);

      if (INT_MIN <= num && num <= rule_length && ! get_errno ())
	{
	  int n = num;
	  if (1-n > max_left_semantic_context)
	    max_left_semantic_context = 1-n;
	  if (!type_name && n > 0)
	    type_name = symbol_list_n_type_name_get (current_rule, loc, n);
	  if (!type_name && typed)
	    complain_at (loc, _("$%d of `%s' has no declared type"),
			 n, current_rule->sym->tag);
	  if (!type_name)
	    type_name = "";
	  obstack_fgrow3 (&obstack_for_string,
			  "]b4_rhs_value([%d], [%d], [%s])[",
			  rule_length, n, type_name);
	}
      else
	complain_at (loc, _("integer out of range: %s"), quote (text));
    }

  return true;
}


/*----------------------------------------------------------------.
| Map `$?' onto the proper M4 symbol, depending on its TOKEN_TYPE |
| (are we in an action?).                                         |
`----------------------------------------------------------------*/

static void
handle_dollar (int token_type, char *text, location loc)
{
  switch (token_type)
    {
    case BRACED_CODE:
      if (handle_action_dollar (text, loc))
	return;
      break;

    case PERCENT_DESTRUCTOR:
    case PERCENT_INITIAL_ACTION:
    case PERCENT_PRINTER:
      if (text[1] == '$')
	{
	  obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
	  return;
	}
      break;

    default:
      break;
    }

  complain_at (loc, _("invalid value: %s"), quote (text));
}


/*------------------------------------------------------.
| TEXT is a location token (i.e., a `@...').  Output to |
| OBSTACK_FOR_STRING a reference to this location.      |
`------------------------------------------------------*/

static inline bool
handle_action_at (char *text, location loc)
{
  char *cp = text + 1;
  locations_flag = true;

  if (! current_rule)
    return false;

  if (*cp == '$')
    obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
  else
    {
      long int num;
      set_errno (0);
      num = strtol (cp, 0, 10);

      if (INT_MIN <= num && num <= rule_length && ! get_errno ())
	{
	  int n = num;
	  obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location([%d], [%d])[",
			  rule_length, n);
	}
      else
	complain_at (loc, _("integer out of range: %s"), quote (text));
    }

  return true;
}


/*----------------------------------------------------------------.
| Map `@?' onto the proper M4 symbol, depending on its TOKEN_TYPE |
| (are we in an action?).                                         |
`----------------------------------------------------------------*/

static void
handle_at (int token_type, char *text, location loc)
{
  switch (token_type)
    {
    case BRACED_CODE:
      handle_action_at (text, loc);
      return;

    case PERCENT_INITIAL_ACTION:
    case PERCENT_DESTRUCTOR:
    case PERCENT_PRINTER:
      if (text[1] == '$')
	{
	  obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
	  return;
	}
      break;

    default:
      break;
    }

  complain_at (loc, _("invalid value: %s"), quote (text));
}


/*------------------------------------------------------.
| Scan NUMBER for a base-BASE integer at location LOC.  |
`------------------------------------------------------*/

static unsigned long int
scan_integer (char const *number, int base, location loc)
{
  unsigned long int num;
  set_errno (0);
  num = strtoul (number, 0, base);
  if (INT_MAX < num || get_errno ())
    {
      complain_at (loc, _("integer out of range: %s"), quote (number));
      num = INT_MAX;
    }
  return num;
}


/*------------------------------------------------------------------.
| Convert universal character name UCN to a single-byte character,  |
| and return that character.  Return -1 if UCN does not correspond  |
| to a single-byte character.					    |
`------------------------------------------------------------------*/

static int
convert_ucn_to_byte (char const *ucn)
{
  unsigned long int code = strtoul (ucn + 2, 0, 16);

  /* FIXME: Currently we assume Unicode-compatible unibyte characters
     on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes).  On
     non-ASCII hosts we support only the portable C character set.
     These limitations should be removed once we add support for
     multibyte characters.  */

  if (UCHAR_MAX < code)
    return -1;

#if ! ('$' == 0x24 && '@' == 0x40 && '`' == 0x60 && '~' == 0x7e)
  {
    /* A non-ASCII host.  Use CODE to index into a table of the C
       basic execution character set, which is guaranteed to exist on
       all Standard C platforms.  This table also includes '$', '@',
       and '`', which are not in the basic execution character set but
       which are unibyte characters on all the platforms that we know
       about.  */
    static signed char const table[] =
      {
	'\0',   -1,   -1,   -1,   -1,   -1,   -1, '\a',
	'\b', '\t', '\n', '\v', '\f', '\r',   -1,   -1,
	  -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
	  -1,   -1,   -1,   -1,   -1,   -1,   -1,   -1,
	 ' ',  '!',  '"',  '#',  '$',  '%',  '&', '\'',
	 '(',  ')',  '*',  '+',  ',',  '-',  '.',  '/',
	 '0',  '1',  '2',  '3',  '4',  '5',  '6',  '7',
	 '8',  '9',  ':',  ';',  '<',  '=',  '>',  '?',
	 '@',  'A',  'B',  'C',  'D',  'E',  'F',  'G',
	 'H',  'I',  'J',  'K',  'L',  'M',  'N',  'O',
	 'P',  'Q',  'R',  'S',  'T',  'U',  'V',  'W',
	 'X',  'Y',  'Z',  '[', '\\',  ']',  '^',  '_',
	 '`',  'a',  'b',  'c',  'd',  'e',  'f',  'g',
	 'h',  'i',  'j',  'k',  'l',  'm',  'n',  'o',
	 'p',  'q',  'r',  's',  't',  'u',  'v',  'w',
	 'x',  'y',  'z',  '{',  '|',  '}',  '~'
      };

    code = code < sizeof table ? table[code] : -1;
  }
#endif

  return code;
}


/*----------------------------------------------------------------.
| Handle `#line INT "FILE"'.  ARGS has already skipped `#line '.  |
`----------------------------------------------------------------*/

static void
handle_syncline (char *args)
{
  int lineno = strtol (args, &args, 10);
  const char *file = NULL;
  file = strchr (args, '"') + 1;
  *strchr (file, '"') = 0;
  scanner_cursor.file = current_file = uniqstr_new (file);
  scanner_cursor.line = lineno;
  scanner_cursor.column = 1;
}


/*----------------------------------------------------------------.
| For a token or comment starting at START, report message MSGID, |
| which should say that an end marker was found before		  |
| the expected TOKEN_END.					  |
`----------------------------------------------------------------*/

static void
unexpected_end (boundary start, char const *msgid, char const *token_end)
{
  location loc;
  loc.start = start;
  loc.end = scanner_cursor;
  complain_at (loc, _(msgid), token_end);
}


/*------------------------------------------------------------------------.
| Report an unexpected EOF in a token or comment starting at START.       |
| An end of file was encountered and the expected TOKEN_END was missing.  |
`------------------------------------------------------------------------*/

static void
unexpected_eof (boundary start, char const *token_end)
{
  unexpected_end (start, N_("missing `%s' at end of file"), token_end);
}


/*----------------------------------------.
| Likewise, but for unexpected newlines.  |
`----------------------------------------*/

static void
unexpected_newline (boundary start, char const *token_end)
{
  unexpected_end (start, N_("missing `%s' at end of line"), token_end);
}


/*-------------------------.
| Initialize the scanner.  |
`-------------------------*/

void
scanner_initialize (void)
{
  obstack_init (&obstack_for_string);
}


/*-----------------------------------------------.
| Free all the memory allocated to the scanner.  |
`-----------------------------------------------*/

void
scanner_free (void)
{
  obstack_free (&obstack_for_string, 0);
  /* Reclaim Flex's buffers.  */
  yy_delete_buffer (YY_CURRENT_BUFFER);
}