comparison lib-src/etags.c @ 35297:e268b7b500f0

Changes for makefile support in etags.

35296:13441490c2f3

 *	Regexp tags by Tom Tromey.
 *
 *	Francesco Potorti` ([e-mail]) is the current maintainer.
 */

char pot_etags_version[] = "@(#) pot revision number is 13.44";

#define	TRUE	1
#define	FALSE	0

#ifndef DEBUG

#define streq(s,t)	((DEBUG && (s) == NULL && (t) == NULL	\
			  && (abort (), 1)) || !strcmp (s, t))
#define strneq(s,t,n)	((DEBUG && (s) == NULL && (t) == NULL	\
			  && (abort (), 1)) || !strncmp (s, t, n))

#define lowcase(c)	tolower ((unsigned char)(c))
#define UPCASE(c)	toupper ((unsigned char)(c))

#define CHARS 256		/* 2^sizeof(char) */
#define CHAR(x)		((unsigned int)x & (CHARS - 1))
#define	iswhite(c)	(_wht[CHAR(c)]) /* c is white */
#define notinname(c)	(_nin[CHAR(c)]) /* c is not in a name */
#define	begtoken(c)	(_btk[CHAR(c)]) /* c can start token */
#define	intoken(c)	(_itk[CHAR(c)]) /* c can be in token */
#define	endtoken(c)	(_etk[CHAR(c)]) /* c ends tokens */

#define ISALNUM(c)	isalnum ((unsigned char) (c))
#define ISALPHA(c)	isalpha ((unsigned char) (c))
#define ISDIGIT(c)	isdigit ((unsigned char) (c))
#define ISLOWER(c)	islower ((unsigned char) (c))


/*
 *	xnew, xrnew -- allocate, reallocate storage
 *

typedef struct
{
  char *name;
  Lang_function *function;
  char **suffixes;
  char **interpreters;
} language;

typedef struct node_st

static void Cstar_entries P_((FILE *));
static void Erlang_functions P_((FILE *));
static void Fortran_functions P_((FILE *));
static void Yacc_entries P_((FILE *));
static void Lisp_functions P_((FILE *));
static void Pascal_functions P_((FILE *));
static void Perl_functions P_((FILE *));
static void Postscript_functions P_((FILE *));
static void Prolog_functions P_((FILE *));
static void Python_functions P_((FILE *));
static void Scheme_functions P_((FILE *));
static void TeX_functions P_((FILE *));
static void Texinfo_functions P_ ((FILE *));
static void just_read_file P_((FILE *));

static void print_language_names P_((void));
static void print_version P_((void));
static void print_help P_((void));
int main P_((int, char **));
static int number_len P_((long));

static compressor *get_compressor_from_suffix P_((char *, char **));
static language *get_language_from_name P_((char *));
static language *get_language_from_interpreter P_((char *));
static language *get_language_from_suffix P_((char *));
static int total_size_of_entries P_((node *));
static long readline P_((linebuffer *, FILE *));
static long readline_internal P_((linebuffer *, FILE *));
static void get_tag P_((char *));

   given.  That is why default_C_entries is called here. */
char *default_C_suffixes [] =
  { "c", "h", NULL };

char *Cplusplus_suffixes [] =
  { "C", "H", "c++", "cc", "cpp", "cxx", "h++", "hh", "hpp", "hxx",
    "M",			/* Objective C++ */
    "pdb",			/* Postscript with C syntax */
    NULL };

char *Cjava_suffixes [] =

char *Fortran_suffixes [] =
  { "F", "f", "f90", "for", NULL };

char *Lisp_suffixes [] =
  { "cl", "clisp", "el", "l", "lisp", "lsp", "ml", "LSP", NULL };

char *Pascal_suffixes [] =
  { "p", "pas", NULL };

char *Perl_suffixes [] =
  { "pl", "pm", NULL };
char *Perl_interpreters [] =
  { "perl", "@PERL@", NULL };

char *plain_C_suffixes [] =
  { "pc",			/* Pro*C file */
    "m",			/* Objective C file */
    "lm",			/* Objective lex file */
     NULL };

char *Postscript_suffixes [] =
  { "ps", "psw", NULL };	/* .psw is for PSWrap */

char *Python_suffixes [] =
  { "py", NULL };

/* Can't do the `SCM' or `scm' prefix with a version number. */
char *Scheme_suffixes [] =
  { "SCM", "SM", "oak", "sch", "scheme", "scm", "sm", "ss", "t", NULL };

char *TeX_suffixes [] =
  { "TeX", "bib", "clo", "cls", "ltx", "sty", "tex", NULL };

char *Texinfo_suffixes [] =
  { "texi", "txi", "texinfo", NULL };

char *Yacc_suffixes [] =
  { "y", "ym", "yy", "yxx", "y++", NULL }; /* .ym is Objective yacc file */

/*
 * Table of languages.
 *
 * It is ok for a given function to be listed under more than one
 * name.  I just didn't.
 */

language lang_names [] =
{
  { "ada",     Ada_funcs,           Ada_suffixes,         NULL              },
  { "asm",     Asm_labels,          Asm_suffixes,         NULL              },
  { "c",       default_C_entries,   default_C_suffixes,   NULL              },
  { "c++",     Cplusplus_entries,   Cplusplus_suffixes,   NULL              },
  { "c*",      Cstar_entries,       Cstar_suffixes,       NULL              },
  { "cobol",   Cobol_paragraphs,    Cobol_suffixes,       NULL              },
  { "erlang",  Erlang_functions,    Erlang_suffixes,      NULL              },
  { "fortran", Fortran_functions,   Fortran_suffixes,     NULL              },
  { "java",    Cjava_entries,       Cjava_suffixes,       NULL              },
  { "lisp",    Lisp_functions,      Lisp_suffixes,        NULL              },
  { "pascal",  Pascal_functions,    Pascal_suffixes,      NULL              },
  { "perl",    Perl_functions,      Perl_suffixes,        Perl_interpreters },
  { "postscript", Postscript_functions, Postscript_suffixes, NULL           },
  { "proc",    plain_C_entries,     plain_C_suffixes,     NULL              },
  { "prolog",  Prolog_functions,    Prolog_suffixes,      NULL              },
  { "python",  Python_functions,    Python_suffixes,      NULL              },
  { "scheme",  Scheme_functions,    Scheme_suffixes,      NULL              },
  { "tex",     TeX_functions,       TeX_suffixes,         NULL              },
  { "texinfo", Texinfo_functions,   Texinfo_suffixes,     NULL              },
  { "yacc",    Yacc_entries,        Yacc_suffixes,        NULL              },
  { "auto", NULL },             /* default guessing scheme */
  { "none", just_read_file },   /* regexp matching only */
  { NULL, NULL }                /* end of list */
};


	case 'S':		/* for backward compatibility */
	  noindentypedefs = TRUE;
	  break;
	case 'l':
	  {
	    language *lang = get_language_from_name (optarg);
	    if (lang != NULL)
	      {
		argbuffer[current_arg].lang = lang;
		argbuffer[current_arg].arg_type = at_language;
		++current_arg;

/*
 * Return a language given the name.
 */
static language *
get_language_from_name (name)
     char *name;
{
  language *lang;

  if (name == NULL)

/*
 * Return a language given the file name.
 */
static language *
get_language_from_suffix (file)
     char *file;
{
  language *lang;
  char **ext, *suffix;

  suffix = etags_strrchr (file, '.');
  if (suffix == NULL)
    return NULL;
  suffix += 1;
  for (lang = lang_names; lang->name != NULL; lang++)

      lang->function (inf);
      return;
    }

  /* Try to guess the language given the file name. */
  lang = get_language_from_suffix (file);
  if (lang != NULL && lang->function != NULL)
    {
      curlang = lang;
      lang->function (inf);
      return;

     which is unlikely. */
  rewind (inf);

  /* Try Fortran. */
  old_last_node = last_node;
  curlang = get_language_from_name ("fortran");
  Fortran_functions (inf);

  /* No Fortran entries found.  Try C. */
  if (old_last_node == last_node)
    {
      /* We do not tag if rewind fails.
	 Only the file name will be recorded in the tags file. */
      rewind (inf);
      curlang = get_language_from_name (cplusplus ? "c++" : "c");
      default_C_entries (inf);
    }
  return;
}


	pfnote (savenstr (bp, ep-bp), TRUE,
		lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
    }
}

/* Added by Mosur Mohan, 4/22/88 */
/* Pascal parsing                */

/*
 *  Locates tags for procedures & functions.  Doesn't do any type- or

/*
 * TeX/LaTeX scanning loop.
 */
static void
TeX_functions (inf)
     FILE *inf;
{
  char *cp, *lasthit;
  register int i;

  return -1;
}

/* Texinfo support.  Dave Love, Mar. 2000.  */
static void
Texinfo_functions (inf)
     FILE * inf;
{
  char *cp, *start;
  LOOP_ON_INPUT_LINES (inf, lb, cp)
    {
      if ((*cp++ == '@' && *cp++ == 'n' && *cp++ == 'o' && *cp++ == 'd'
	      && *cp++ == 'e' && iswhite (*cp++)))
	{
	    while (iswhite (*cp))
	          cp++;
	      start = cp;
	        while (*cp != '\0' && *cp != ',')
		      cp++;
		  pfnote (savenstr (start, cp - start), TRUE,
			    lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
		  }
    }
}

/*
 * Prolog support (rewritten) by Anders Lindgren, Mar. 96

	    {
	      error ("unterminated language name in regex: %s", regex_arg);
	      return;
	    }
	*cp = '\0';
	lang = get_language_from_name (lang_name);
	if (lang == NULL)
	  return;
	add_regex (cp + 1, ignore_case, lang);
      }
      break;

canonicalize_filename (fn)
     register char *fn;
{
#ifdef DOS_NT
  /* Canonicalize drive letter case.  */
  if (fn[0] && fn[1] == ':' && ISLOWER (fn[0]))
    fn[0] = UPCASE (fn[0]);
  /* Convert backslashes to slashes.  */
  for (; *fn != '\0'; fn++)
    if (*fn == '\\')
      *fn = '/';
#else

35297:e268b7b500f0

 *	Regexp tags by Tom Tromey.
 *
 *	Francesco Potorti` ([e-mail]) is the current maintainer.
 */

char pot_etags_version[] = "@(#) pot revision number is 13.47";

#define	TRUE	1
#define	FALSE	0

#ifndef DEBUG

#define streq(s,t)	((DEBUG && (s) == NULL && (t) == NULL	\
			  && (abort (), 1)) || !strcmp (s, t))
#define strneq(s,t,n)	((DEBUG && (s) == NULL && (t) == NULL	\
			  && (abort (), 1)) || !strncmp (s, t, n))

#define CHARS 256		/* 2^sizeof(char) */
#define CHAR(x)		((unsigned int)(x) & (CHARS - 1))
#define	iswhite(c)	(_wht[CHAR(c)]) /* c is white */
#define notinname(c)	(_nin[CHAR(c)]) /* c is not in a name */
#define	begtoken(c)	(_btk[CHAR(c)]) /* c can start token */
#define	intoken(c)	(_itk[CHAR(c)]) /* c can be in token */
#define	endtoken(c)	(_etk[CHAR(c)]) /* c ends tokens */

#define ISALNUM(c)	isalnum (CHAR(c))
#define ISALPHA(c)	isalpha (CHAR(c))
#define ISDIGIT(c)	isdigit (CHAR(c))
#define ISLOWER(c)	islower (CHAR(c))

#define lowcase(c)	tolower (CHAR(c))
#define upcase(c)	toupper (CHAR(c))


/*
 *	xnew, xrnew -- allocate, reallocate storage
 *

typedef struct
{
  char *name;
  Lang_function *function;
  char **filenames;
  char **suffixes;
  char **interpreters;
} language;

typedef struct node_st

static void Cstar_entries P_((FILE *));
static void Erlang_functions P_((FILE *));
static void Fortran_functions P_((FILE *));
static void Yacc_entries P_((FILE *));
static void Lisp_functions P_((FILE *));
static void Makefile_targets P_((FILE *));
static void Pascal_functions P_((FILE *));
static void Perl_functions P_((FILE *));
static void Postscript_functions P_((FILE *));
static void Prolog_functions P_((FILE *));
static void Python_functions P_((FILE *));
static void Scheme_functions P_((FILE *));
static void TeX_commands P_((FILE *));
static void Texinfo_nodes P_((FILE *));
static void just_read_file P_((FILE *));

static void print_language_names P_((void));
static void print_version P_((void));
static void print_help P_((void));
int main P_((int, char **));
static int number_len P_((long));

static compressor *get_compressor_from_suffix P_((char *, char **));
static language *get_language_from_langname P_((char *));
static language *get_language_from_interpreter P_((char *));
static language *get_language_from_filename P_((char *));
static int total_size_of_entries P_((node *));
static long readline P_((linebuffer *, FILE *));
static long readline_internal P_((linebuffer *, FILE *));
static void get_tag P_((char *));

   given.  That is why default_C_entries is called here. */
char *default_C_suffixes [] =
  { "c", "h", NULL };

char *Cplusplus_suffixes [] =
  { "C", "c++", "cc", "cpp", "cxx", "H", "h++", "hh", "hpp", "hxx",
    "M",			/* Objective C++ */
    "pdb",			/* Postscript with C syntax */
    NULL };

char *Cjava_suffixes [] =

char *Fortran_suffixes [] =
  { "F", "f", "f90", "for", NULL };

char *Lisp_suffixes [] =
  { "cl", "clisp", "el", "l", "lisp", "LSP", "lsp", "ml", NULL };

char *Makefile_filenames [] =
  { "Makefile", "makefile", "GNUMakefile", "Makefile.in", "Makefile.am", NULL};

char *Pascal_suffixes [] =
  { "p", "pas", NULL };

char *Perl_suffixes [] =
  { "pl", "pm", NULL };
char *Perl_interpreters [] =
  { "perl", "@PERL@", NULL };

char *plain_C_suffixes [] =
  { "lm",			/* Objective lex file */
    "m",			/* Objective C file */
    "pc",			/* Pro*C file */
     NULL };

char *Postscript_suffixes [] =
  { "ps", "psw", NULL };	/* .psw is for PSWrap */

char *Python_suffixes [] =
  { "py", NULL };

/* Can't do the `SCM' or `scm' prefix with a version number. */
char *Scheme_suffixes [] =
  { "oak", "sch", "scheme", "SCM", "scm", "SM", "sm", "ss", "t", NULL };

char *TeX_suffixes [] =
  { "bib", "clo", "cls", "ltx", "sty", "TeX", "tex", NULL };

char *Texinfo_suffixes [] =
  { "texi", "texinfo", "txi", NULL };

char *Yacc_suffixes [] =
  { "y", "y++", "ym", "yxx", "yy", NULL }; /* .ym is Objective yacc file */

/*
 * Table of languages.
 *
 * It is ok for a given function to be listed under more than one
 * name.  I just didn't.
 */

language lang_names [] =
{
  { "ada",     	  Ada_funcs,           	NULL, Ada_suffixes,        	NULL },
  { "asm",     	  Asm_labels,          	NULL, Asm_suffixes,        	NULL },
  { "c",       	  default_C_entries,   	NULL, default_C_suffixes,  	NULL },
  { "c++",     	  Cplusplus_entries,   	NULL, Cplusplus_suffixes,  	NULL },
  { "c*",      	  Cstar_entries,       	NULL, Cstar_suffixes,      	NULL },
  { "cobol",   	  Cobol_paragraphs,    	NULL, Cobol_suffixes,      	NULL },
  { "erlang",  	  Erlang_functions,    	NULL, Erlang_suffixes,     	NULL },
  { "fortran", 	  Fortran_functions,   	NULL, Fortran_suffixes,    	NULL },
  { "java",    	  Cjava_entries,       	NULL, Cjava_suffixes,      	NULL },
  { "lisp",    	  Lisp_functions,      	NULL, Lisp_suffixes,       	NULL },
  { "makefile",   Makefile_targets,     Makefile_filenames, NULL,     	NULL },
  { "pascal",  	  Pascal_functions,    	NULL, Pascal_suffixes,     	NULL },
  { "perl",    	  Perl_functions,       NULL, Perl_suffixes, Perl_interpreters },
  { "postscript", Postscript_functions, NULL, Postscript_suffixes, 	NULL },
  { "proc",    	  plain_C_entries,     	NULL, plain_C_suffixes,    	NULL },
  { "prolog",  	  Prolog_functions,    	NULL, Prolog_suffixes,     	NULL },
  { "python",  	  Python_functions,    	NULL, Python_suffixes,     	NULL },
  { "scheme",  	  Scheme_functions,    	NULL, Scheme_suffixes,     	NULL },
  { "tex",     	  TeX_commands,        	NULL, TeX_suffixes,        	NULL },
  { "texinfo", 	  Texinfo_nodes,       	NULL, Texinfo_suffixes,    	NULL },
  { "yacc",    	  Yacc_entries,        	NULL, Yacc_suffixes,       	NULL },
  { "auto", NULL },             /* default guessing scheme */
  { "none", just_read_file },   /* regexp matching only */
  { NULL, NULL }                /* end of list */
};


	case 'S':		/* for backward compatibility */
	  noindentypedefs = TRUE;
	  break;
	case 'l':
	  {
	    language *lang = get_language_from_langname (optarg);
	    if (lang != NULL)
	      {
		argbuffer[current_arg].lang = lang;
		argbuffer[current_arg].arg_type = at_language;
		++current_arg;

/*
 * Return a language given the name.
 */
static language *
get_language_from_langname (name)
     char *name;
{
  language *lang;

  if (name == NULL)

/*
 * Return a language given the file name.
 */
static language *
get_language_from_filename (file)
     char *file;
{
  language *lang;
  char **name, **ext, *suffix;

  /* Try whole file name first. */
  for (lang = lang_names; lang->name != NULL; lang++)
    if (lang->filenames != NULL)
      for (name = lang->filenames; *name != NULL; name++)
	if (streq (*name, file))
	  return lang;

  /* If not found, try suffix after last dot. */
  suffix = etags_strrchr (file, '.');
  if (suffix == NULL)
    return NULL;
  suffix += 1;
  for (lang = lang_names; lang->name != NULL; lang++)

      lang->function (inf);
      return;
    }

  /* Try to guess the language given the file name. */
  lang = get_language_from_filename (file);
  if (lang != NULL && lang->function != NULL)
    {
      curlang = lang;
      lang->function (inf);
      return;

     which is unlikely. */
  rewind (inf);

  /* Try Fortran. */
  old_last_node = last_node;
  curlang = get_language_from_langname ("fortran");
  Fortran_functions (inf);

  /* No Fortran entries found.  Try C. */
  if (old_last_node == last_node)
    {
      /* We do not tag if rewind fails.
	 Only the file name will be recorded in the tags file. */
      rewind (inf);
      curlang = get_language_from_langname (cplusplus ? "c++" : "c");
      default_C_entries (inf);
    }
  return;
}


	pfnote (savenstr (bp, ep-bp), TRUE,
		lb.buffer, ep - lb.buffer + 1, lineno, linecharno);
    }
}

/*
 * Makefile support
 */
static void
Makefile_targets (inf)
     FILE *inf;
{
  register char *bp;

  LOOP_ON_INPUT_LINES (inf, lb, bp)
    {
      if (*bp == '\t' || *bp == '#')
	continue;
      while (*bp != '\0' && *bp != '=' && *bp != ':')
	bp++;
      if (*bp == ':')
	pfnote (savenstr (lb.buffer, bp - lb.buffer), TRUE,
		lb.buffer, bp - lb.buffer + 1, lineno, linecharno);
    }
}

/* Added by Mosur Mohan, 4/22/88 */
/* Pascal parsing                */

/*
 *  Locates tags for procedures & functions.  Doesn't do any type- or

/*
 * TeX/LaTeX scanning loop.
 */
static void
TeX_commands (inf)
     FILE *inf;
{
  char *cp, *lasthit;
  register int i;

  return -1;
}

/* Texinfo support.  Dave Love, Mar. 2000.  */
static void
Texinfo_nodes (inf)
     FILE * inf;
{
  char *cp, *start;
  LOOP_ON_INPUT_LINES (inf, lb, cp)
    {
      if ((*cp++ == '@'
	   && *cp++ == 'n'
	   && *cp++ == 'o'
	   && *cp++ == 'd'
	   && *cp++ == 'e' && iswhite (*cp++)))
	{
	  start = cp = skip_spaces(cp);
	  while (*cp != '\0' && *cp != ',')
	    cp++;
	  pfnote (savenstr (start, cp - start), TRUE,
		  lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
	}
    }
}

/*
 * Prolog support (rewritten) by Anders Lindgren, Mar. 96

	    {
	      error ("unterminated language name in regex: %s", regex_arg);
	      return;
	    }
	*cp = '\0';
	lang = get_language_from_langname (lang_name);
	if (lang == NULL)
	  return;
	add_regex (cp + 1, ignore_case, lang);
      }
      break;

canonicalize_filename (fn)
     register char *fn;
{
#ifdef DOS_NT
  /* Canonicalize drive letter case.  */
  if (fn[0] != '\0' && fn[1] == ':' && ISLOWER (fn[0]))
    fn[0] = upcase (fn[0]);
  /* Convert backslashes to slashes.  */
  for (; *fn != '\0'; fn++)
    if (*fn == '\\')
      *fn = '/';
#else