/* Tags file maker to go with GNU Emacs
- Copyright (C) 1984, 1987, 1988, 1989 Free Software Foundation, Inc. and Ken Arnold
+ Copyright (C) 1984,87,88,89,93,94 Free Software Foundation, Inc. and Ken Arnold
+This file is not considered part of GNU Emacs.
-This file is part of GNU Emacs.
-
-GNU Emacs is free software; you can redistribute it and/or modify
+This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
-the Free Software Foundation; either version 1, or (at your option)
-any later version.
+the Free Software Foundation; either version 2 of the License, or
+(at your option) any later version.
-GNU Emacs is distributed in the hope that it will be useful,
+This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
-along with GNU Emacs; see the file COPYING. If not, write to
-the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA. */
+along with this program; if not, write to the Free Software
+Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
/*
* Authors:
* Ctags originally by Ken Arnold.
- * FORTRAN added by Jim Kleckner.
+ * Fortran added by Jim Kleckner.
* Ed Pelegri-Llopart added C typedefs.
* Gnu Emacs TAGS format and modifications by RMS?
* Sam Kendall added C++.
+ * Francesco Potorti` reorganised C and C++ based on work by Joe Wells.
+#ifdef ETAGS_REGEXPS
+ * Regexp tags by Tom Tromey.
+#endif
+ *
+ * Francesco Potorti` (pot@cnuce.cnr.it) is the current maintainer.
*/
+char pot_etags_version[] = "@(#) pot revision number is 11.14";
+
+#ifdef MSDOS
+#include <fcntl.h>
+#include <sys/param.h>
+#endif /* MSDOS */
+
+#ifdef WINDOWSNT
+#include <stdlib.h>
+#include <fcntl.h>
+#include <string.h>
+#define MAXPATHLEN _MAX_PATH
+#endif
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+/* On some systems, Emacs defines static as nothing for the sake
+ of unexec. We don't want that here since we don't use unexec. */
+#undef static
+#endif
+
#include <stdio.h>
#include <ctype.h>
+#include <errno.h>
+#ifndef errno
+extern int errno;
+#endif
#include <sys/types.h>
#include <sys/stat.h>
-#ifdef __GNUC__
-#define alloca __builtin_alloca
-#else
-#ifdef sparc
-#include <alloca.h>
-#else
-extern char *alloca ();
-#endif
+#if !defined (S_ISREG) && defined (S_IFREG)
+# define S_ISREG(m) (((m) & S_IFMT) == S_IFREG)
#endif
-extern char *malloc (), *realloc ();
-extern char *getenv ();
-extern char *index (), *rindex ();
-extern char *strcpy (), *strncpy ();
-extern int strcmp ();
+#include <getopt.h>
-#ifdef hpux
-#define notdef
-#endif
+#ifdef ETAGS_REGEXPS
+#include <regex.h>
+#endif /* ETAGS_REGEXPS */
-/* Define the symbol ETAGS to make the program "etags",
- which makes emacs-style tag tables by default.
- Define CTAGS to make the program "ctags" compatible with the usual one.
- Define neither one to get behavior that depends
- on the name with which the program is invoked
- (but we don't normally compile it that way). */
-
-#if !defined(ETAGS) && !defined(CTAGS)
-/* If neither is defined, program can be run as either. */
-#define ETAGS
-#define CTAGS
-#endif
+extern char *getenv ();
-/* On VMS, CTAGS is not useful, so always do ETAGS. */
-#ifdef VMS
-#ifndef ETAGS
-#define ETAGS
-#endif
+/* Define CTAGS to make the program "ctags" compatible with the usual one.
+ Let it undefined to make the program "etags", which makes emacs-style
+ tag tables and tags typedefs, #defines and struct/union/enum by default. */
+#ifdef CTAGS
+# undef CTAGS
+# define CTAGS TRUE
+#else
+# define CTAGS FALSE
#endif
/* Exit codes for success and failure. */
#ifdef VMS
-#define GOOD (1)
-#define BAD (0)
+#define GOOD 1
+#define BAD 0
#else
-#define GOOD (0)
-#define BAD (1)
+#define GOOD 0
+#define BAD 1
#endif
/*
*
* long GET_CHARNO (pos)
* returns absolute char number.
- * long GET_COOKIE (pos)
- * returns ftell () cookie.
* void SET_FILEPOS (pos, fp, charno)
* FILE *fp; long charno;
* sets `pos' from the current file
*
* Implementation notes: the `+ 0' is to enforce rvalue-ness.
*/
-#ifdef VMS
-typedef struct
-{
- long cookie;
- long charno;
-} FILEPOS;
-#define GET_CHARNO(pos) ((pos).charno + 0)
-#define GET_COOKIE(pos) ((pos).cookie + 0)
-#define SET_FILEPOS(pos, fp, cno) \
- ((void) ((pos).cookie = ftell (fp), (pos).charno = (cno)))
-#else
#ifndef DEBUG
- /* UNIX real implementation */
+#define DEBUG FALSE
+
+ /* real implementation */
typedef long FILEPOS;
-#define GET_CHARNO(pos) ((pos) + 0)
-#define GET_COOKIE(pos) GET_CHARNO (pos)
-#define SET_FILEPOS(pos, fp, cno) ((void) ((pos) = (cno)))
+#define GET_CHARNO(pos) ((pos) + 0)
+#define SET_FILEPOS(pos,fp,cno) ((void) ((pos) = (cno)))
+
#else
- /* UNIX debugging implementation */
+#undef DEBUG
+#define DEBUG TRUE
+
+ /* debugging implementation */
typedef struct
{
long charno;
} FILEPOS;
#define GET_CHARNO(pos) ((pos).charno + 0)
-#define GET_COOKIE(pos) GET_CHARNO (pos)
-#define SET_FILEPOS(pos, fp, cno) \
+#define SET_FILEPOS(pos,fp,cno) \
((void) ((pos).charno = (cno), \
(cno) != ftell (fp) ? (error ("SET_FILEPOS inconsistency"), 0) \
: 0))
#endif
-#endif
-#define streq(s, t) (strcmp (s, t) == 0)
-#define strneq(s, t, n) (strncmp (s, t, n) == 0)
-#define reg register
-#define logical char
+#define streq(s,t) (strcmp (s, t) == 0)
+#define strneq(s,t,n) (strncmp (s, t, n) == 0)
#define TRUE 1
#define FALSE 0
#define begtoken(arg) (_btk[arg]) /* T if char can start token */
#define intoken(arg) (_itk[arg]) /* T if char can be in token */
#define endtoken(arg) (_etk[arg]) /* T if char ends tokens */
-#define isgood(arg) (_gd[arg]) /* T if char can be after ')' */
#define max(I1,I2) ((I1) > (I2) ? (I1) : (I2))
+typedef int logical;
+
struct nd_st
{ /* sorting structure */
char *name; /* function or type name */
char *file; /* file name */
logical is_func; /* use pattern or line no */
- logical rewritten; /* list name separately */
+ logical named; /* list name separately */
logical been_warned; /* set if noticed dup */
int lno; /* line number tag is on */
long cno; /* character number line starts on */
struct nd_st *left, *right; /* left and right sons */
};
-long ftell ();
typedef struct nd_st NODE;
-logical gotone, /* found a func already on line */
- /* boolean "func" (see init) */
- header_file, /* TRUE if .h file, FALSE o.w. */
- _wht[0177], _etk[0177], _itk[0177], _btk[0177], _gd[0177];
+/* boolean "functions" (see init) */
+logical _wht[0177], _etk[0177], _itk[0177], _btk[0177];
+char *cwd; /* current working directory */
+char *tagfiledir; /* directory of tagfile */
char *concat ();
-char *savenstr ();
-char *savestr ();
-char *xmalloc ();
-char *xrealloc ();
-int L_isdef ();
-int PF_funcs ();
+char *savenstr (), *savestr ();
+char *etags_strchr (), *etags_strrchr ();
+char *etags_getcwd ();
+char *relative_filename (), *absolute_filename (), *absolute_dirname ();
+char *xmalloc (), *xrealloc ();
+
+typedef void Lang_function ();
+#if FALSE /* many compilers barf on this */
+Lang_function Asm_labels;
+Lang_function default_C_entries;
+Lang_function C_entries;
+Lang_function Cplusplus_entries;
+Lang_function Cstar_entries;
+Lang_function Fortran_functions;
+Lang_function Yacc_entries;
+Lang_function Lisp_functions;
+Lang_function Pascal_functions;
+Lang_function Prolog_functions;
+Lang_function Scheme_functions;
+Lang_function TeX_functions;
+Lang_function just_read_file;
+#else /* so let's write it this way */
+void Asm_labels ();
+void default_C_entries ();
+void C_entries ();
+void Cplusplus_entries ();
+void Cstar_entries ();
+void Fortran_functions ();
+void Yacc_entries ();
+void Lisp_functions ();
+void Pascal_functions ();
+void Prolog_functions ();
+void Scheme_functions ();
+void TeX_functions ();
+void just_read_file ();
+#endif
+
+logical get_language ();
int total_size_of_entries ();
-logical consider_token ();
-logical tail ();
long readline ();
-void Asm_funcs ();
-void C_entries ();
-void L_funcs ();
-void L_getit ();
-void PAS_funcs ();
-void Scheme_funcs ();
-void TEX_funcs ();
+long readline_internal ();
+#ifdef ETAGS_REGEXPS
+void add_regex ();
+#endif
void add_node ();
void error ();
-void fatal ();
-void find_entries ();
+void fatal (), pfatal ();
+logical find_entries ();
void free_tree ();
void getit ();
-void getline ();
void init ();
void initbuffer ();
-void initbuffer ();
void pfnote ();
void process_file ();
void put_entries ();
* SYNOPSIS
* Type *xnew (int n, Type);
*/
-#define xnew(n, Type) ((Type *) xmalloc ((n) * sizeof (Type)))
-\f
-
+#define xnew(n,Type) ((Type *) xmalloc ((n) * sizeof (Type)))
/*
- * Symbol table stuff.
- *
- * Should probably be implemented with hash table; linked list for now.
+ * Symbol table types.
*/
-
enum sym_type
{
st_none, st_C_struct, st_C_enum, st_C_define, st_C_typedef, st_C_typespec
};
-
-struct stab_entry
-{
- char *sym;
- int symlen;
- enum sym_type type;
- struct stab_entry *next;
-};
-
-typedef struct stab_entry Stab_entry;
-typedef Stab_entry *Stab;
-
-/*
- * NAME
- * Stab, Stab_entry, stab_create, stab_search, stab_find -- symbol table
- *
- * SYNOPSIS
- * Types: Stab, Stab_entry, enum sym_type
- *
- * Stab * stab_create ()
- *
- * Stab_entry * stab_find (stab, sym)
- * Stab *stab;
- * char *sym;
- *
- * Stab_entry * stab_search (stab, sym)
- * Stab *stab;
- * char *sym;
- *
- * DESCRIPTION
- * stab_create creates a Stab, a symbol table object, and returns a
- * pointer to it. stab_find finds a symbol in a Stab; it returns a
- * pointer to the Stab_entry if found, otherwise NULL. stab_search
- * is like stab_find, except that it creates a new Stab_entry,
- * initialized with type = st_none, if one did not exist already
- * (it never returns NULL).
- *
- * A Stab_entry is a structure that contains at least the following
- * members:
- *
- * char *name; // must not be modified
- * enum sym_type type; // should be set
- *
- * The type field is initially set to st_none; it should be set to
- * something else by the caller of stab_search. Other possible values
- * of an enum sym_type can be added.
- */
-
-Stab *
-stab_create ()
-{
- Stab *sp;
- sp = xnew (1, Stab);
- *sp = NULL; /* a Stab starts out as a null Stab_entry* */
- return sp;
-}
-
-Stab_entry *
-stab_find (stab, sym, symlen)
- Stab *stab;
- register char *sym;
- register int symlen;
-{
- register Stab_entry *se;
- for (se = *stab; se != NULL; se = se->next)
- {
- if (se->symlen == symlen && strneq (se->sym, sym, symlen))
- return se;
- }
-
- return NULL;
-}
-
-Stab_entry *
-stab_search (stab, sym, symlen)
- register Stab *stab;
- char *sym;
- int symlen;
-{
- register Stab_entry *se;
- se = stab_find (stab, sym, symlen);
-
- if (se == NULL)
- {
- /* make a new one */
- se = xnew (1, Stab_entry);
- se->sym = savenstr (sym, symlen);
- se->symlen = symlen;
- se->type = st_none;
- se->next = *stab;
- *stab = se;
- }
-
- return se;
-}
-
-/*
- * NAME
- * stab_type -- type of a symbol table entry
- *
- * SYNOPSIS
- * enum sym_type stab_type (Stab_entry *se);
- *
- * WARNING
- * May evaluate its argument more than once.
- */
-
-#define stab_type(se) ((se)==NULL ? st_none : (se)->type)
\f
{
char *p;
int len;
- FILEPOS linestart;
LINENO lineno;
- logical rewritten;
+ logical named;
} TOKEN;
-\f
-
- /* typedefs are recognized using a simple finite automaton.
- * tydef is its state variable.
- */
-typedef enum
-{
- none, begin, middle, end
-} TYST;
-
-TYST tydef = none;
-
-
- /* struct tags for C++ are recognized using another simple
- * finite automaton. `structdef' is its state variable.
- * This machinery is only invoked for C++; otherwise structdef
- * should remain snone. However, this machinery can easily be
- * adapted to find structure tags in normal C code.
- */
-typedef enum
-{
- snone, /* nothing seen yet */
- skeyseen, /* struct-like keyword seen */
- stagseen, /* struct-like tag seen */
- scolonseen, /* colon seen after struct-like tag */
- sinbody /* in a class body: recognize member func defs */
-} STRUCTST;
-STRUCTST structdef = snone;
-/*
- * When structdef is stagseen, scolonseen, or sinbody, structtag is the
- * struct tag, and structkey is the preceding struct-like keyword.
- */
-char structtag[512];
-Stab_entry *structkey;
-
-/*
- * Yet another little state machine to deal with preprocessor lines.
- */
-typedef enum
-{
- dnone, /* nothing seen */
- dsharpseen, /* '#' seen as first char on line */
- ddefineseen, /* '#' and 'define' seen */
- dignorerest /* ignore rest of line */
-} DEFINEST;
-DEFINEST definedef;
-
-/*
- * LEVEL_OK_FOR_FUNCDEF allows C++ function definition within class body.
- * Currently tydef and structdef stuff (typedefs and struct definitions) are
- * only noticed when level==0, but that may change.
- *
- * Note that this macro may only be evaluated inside C_entries(). It is
- * for self-documentation only.
- */
-#define LEVEL_OK_FOR_FUNCDEF() \
- (level==0 || c_ext && level==1 && structdef==sinbody)
-/* C extensions. Currently all listed extensions are C++ dialects, so
- * `c_ext' is used as an abbreviation for `c_ext&C_PLPL'. If a non-C++
- * dialect is added, this must change.
+/* C extensions.
*/
-#define C_PLPL 0x1 /* C++ */
-#define C_STAR 0x3 /* C* */
+#define C_PLPL 0x00001 /* C++ */
+#define C_STAR 0x00003 /* C* */
+#define YACC 0x10000 /* yacc file */
char searchar = '/'; /* use /.../ searches */
LINENO lineno; /* line number of current line */
long charno; /* current character number */
-FILEPOS linepos; /* start of line (C only) */
-FILEPOS prev_linepos; /* start of previous line (C only) */
long linecharno; /* charno of start of line; not used by C, but
* by every other language.
*/
char *curfile, /* current input file name */
- *outfile, /* output file */
- *white = " \f\t\n", /* white chars */
- *endtk = " \t\n\"'#()[]{}=-+%*/&|^~!<>;,.:?", /* token ending chars */
- *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$", /* token starting chars */
- *intk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789", /* valid in-token chars */
- *notgd = ",;"; /* non-valid after-function chars */
-
-int file_num; /* current file number */
+ *tagfile, /* output file */
+ *white = " \f\t\n\013", /* white chars */
+ *endtk = " \t\n\013\"'#()[]{}=-+%*/&|^~!<>;,.:?", /* token ending chars */
+ /* token starting chars */
+ *begtk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$~",
+ /* valid in-token chars */
+ *intk = "ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz$0123456789";
+
int append_to_tagfile; /* -a: append to tags */
-int emacs_tags_format; /* emacs style output (no -e option any more) */
/* The following three default to 1 for etags, but to 0 for ctags. */
int typedefs; /* -t: create tags for typedefs */
int typedefs_and_cplusplus; /* -T: create tags for typedefs, level */
/* 0 struct/enum/union decls, and C++ */
- /* member functions */
+ /* member functions. */
int constantypedefs; /* -d: create tags for C #define and enum */
- /* constants. Default under etags. Enum */
- /* constants not implemented. */
+ /* constants. Enum consts not implemented. */
/* -D: opposite of -d. Default under ctags. */
int update; /* -u: update tags */
int vgrind_style; /* -v: create vgrind style index output */
/* Name this program was invoked with. */
char *progname;
-FILE *inf, /* ioptr for current input file */
- *outf; /* ioptr for tags file */
-
-NODE *head; /* the head of the binary tree of tags */
-
-int permit_duplicates = 1; /* Nonzero means allow duplicate tags. */
+struct option longopts[] =
+{
+ { "append", no_argument, NULL, 'a' },
+ { "backward-search", no_argument, NULL, 'B' },
+ { "c++", no_argument, NULL, 'C' },
+ { "cxref", no_argument, NULL, 'x' },
+ { "defines", no_argument, NULL, 'd' },
+ { "help", no_argument, NULL, 'h' },
+ { "help", no_argument, NULL, 'H' },
+ { "ignore-indentation", no_argument, NULL, 'S' },
+ { "include", required_argument, NULL, 'i' },
+ { "language", required_argument, NULL, 'l' },
+ { "no-defines", no_argument, NULL, 'D' },
+ { "no-regex", no_argument, NULL, 'R' },
+ { "no-warn", no_argument, NULL, 'w' },
+ { "output", required_argument, NULL, 'o' },
+ { "regex", required_argument, NULL, 'r' },
+ { "typedefs", no_argument, NULL, 't' },
+ { "typedefs-and-c++", no_argument, NULL, 'T' },
+ { "update", no_argument, NULL, 'u' },
+ { "version", no_argument, NULL, 'V' },
+ { "vgrind", no_argument, NULL, 'v' },
+ { 0 }
+};
-/* A `struct linebuffer' is a structure which holds a line of text.
- `readline' reads a line from a stream into a linebuffer
- and works regardless of the length of the line. */
+FILE *tagf; /* ioptr for tags file */
+NODE *head; /* the head of the binary tree of tags */
+logical permit_duplicates = TRUE; /* allow duplicate tags */
+/*
+ * A `struct linebuffer' is a structure which holds a line of text.
+ * `readline' reads a line from a stream into a linebuffer and works
+ * regardless of the length of the line.
+ */
struct linebuffer
{
long size;
};
struct linebuffer lb; /* the current line */
-struct linebuffer lb1; /* sometimes, a previous line in which a token lies */
struct linebuffer filename_lb; /* used to read in filenames */
+struct
+{
+ FILEPOS linepos;
+ struct linebuffer lb; /* used by C_entries instead of lb */
+} lbs[2];
+
+#ifdef ETAGS_REGEXPS
+/* Structure defining a regular expression. Elements are
+ the compiled pattern, and the name string. */
+struct pattern
+{
+ struct re_pattern_buffer *pattern;
+ struct re_registers regs;
+ char *name_pattern;
+ logical error_signaled;
+};
+
+/* Number of regexps found. */
+int num_patterns = 0;
+
+/* Array of all regexps. */
+struct pattern *patterns = NULL;
+#endif /* ETAGS_REGEXPS */
+
+/* Language stuff. */
+struct lang_entry
+{
+ char *extension;
+ Lang_function *function;
+};
+
+/* Table of language names and corresponding functions. */
+/* It is ok for a given function to be listed under more than one
+ name. I just didn't. */
+/* "auto" language reverts to default behavior. */
+struct lang_entry lang_names[] =
+{
+ { "asm", Asm_labels },
+ { "c", default_C_entries },
+ { "c++", Cplusplus_entries },
+ { "c*", Cstar_entries },
+ { "fortran", Fortran_functions },
+ { "lisp", Lisp_functions },
+ { "none", just_read_file },
+ { "pascal", Pascal_functions },
+ { "scheme" , Scheme_functions },
+ { "tex", TeX_functions },
+ { "auto", NULL },
+ { NULL, NULL }
+};
+
+/* Table of file extensions and corresponding language functions. */
+struct lang_entry lang_extensions[] =
+{
+ /* Assume that ".s" or ".a" is assembly code. -wolfgang.
+ Or even ".sa". */
+ { "a", Asm_labels }, /* Unix assembler */
+ { "asm", Asm_labels }, /* Microcontroller assembly */
+ { "def", Asm_labels }, /* BSO/Tasking definition includes */
+ { "inc", Asm_labels }, /* Microcontroller include files */
+ { "ins", Asm_labels }, /* Microcontroller include files */
+ { "s", Asm_labels },
+ { "sa", Asm_labels }, /* Unix assembler */
+ { "src", Asm_labels }, /* BSO/Tasking C compiler output */
+
+ /* .aux, .bbl, .clo, .cls, .dtx or .tex implies LaTeX source code. */
+ { "aux", TeX_functions },
+ { "bbl", TeX_functions },
+ { "clo", TeX_functions },
+ { "cls", TeX_functions },
+ { "dtx", TeX_functions },
+ { "sty", TeX_functions },
+ { "tex", TeX_functions },
+
+ /* .l or .el or .lisp (or .cl or .clisp or ...) implies lisp source code */
+ { "cl", Lisp_functions },
+ { "clisp", Lisp_functions },
+ { "el", Lisp_functions },
+ { "l", Lisp_functions },
+ { "lisp", Lisp_functions },
+ { "lsp", Lisp_functions },
+
+ /* .scm or .sm or .scheme implies scheme source code */
+ { "SCM", Scheme_functions },
+ { "SM", Scheme_functions },
+ { "oak", Scheme_functions },
+ { "sch", Scheme_functions },
+ { "scheme", Scheme_functions },
+ { "scm", Scheme_functions },
+ { "sm", Scheme_functions },
+ { "t", Scheme_functions },
+ /* FIXME Can't do the `SCM' or `scm' prefix with a version number */
+
+ /* Note that ".c" and ".h" can be considered C++, if the --c++
+ flag was given. That is why default_C_entries is called here. */
+ { "c", default_C_entries },
+ { "h", default_C_entries },
+
+ /* .C or .H or .cpp or .cxx or .hxx or .hh or .cc or .cpp: a C++ file */
+ { "C", Cplusplus_entries },
+ { "H", Cplusplus_entries },
+ { "cc", Cplusplus_entries },
+ { "cpp", Cplusplus_entries },
+ { "cxx", Cplusplus_entries },
+ { "hh", Cplusplus_entries },
+ { "hxx", Cplusplus_entries },
+
+ /* .y: a yacc file */
+ { "y", Yacc_entries },
+
+ /* .cs or .hs: a C* file */
+ { "cs", Cstar_entries },
+ { "hs", Cstar_entries },
+
+ /* .f and .for are FORTRAN. */
+ { "F", Fortran_functions },
+ { "f", Fortran_functions },
+ { "for", Fortran_functions },
+
+ /* .pl implies prolog source code */
+ { "pl", Prolog_functions },
+
+ /* .p or .pas: a Pascal file */
+ { "p", Pascal_functions },
+ { "pas", Pascal_functions },
+
+ { NULL, NULL }
+};
+
+/* Non-NULL if language fixed. */
+Lang_function *lang_func = NULL;
+
+\f
+void
+print_language_names ()
+{
+ struct lang_entry *name, *ext;
+
+ puts ("\nThese are the currently supported languages, along with the\n\
+default extensions for files:");
+ for (name = lang_names; name->extension; ++name)
+ {
+ printf ("\t%s\t", name->extension);
+ for (ext = lang_extensions; ext->extension; ++ext)
+ if (name->function == ext->function)
+ printf (" .%s", ext->extension);
+ puts ("");
+ }
+ puts ("Where `auto' means use default language for files based on filename\n\
+extension, and `none' means only do regexp processing on files.\n\
+If no language is specified and no extension is found for some file,\n\
+Fortran is tried first; if no tags are found, C is tried next.");
+}
+
+void
+print_version ()
+{
+#ifdef VERSION
+ printf ("%s for Emacs version %s.\n", (CTAGS) ? "CTAGS" : "ETAGS", VERSION);
+#else
+ printf ("%s for Emacs version 19.\n", (CTAGS) ? "CTAGS" : "ETAGS");
+#endif
+
+ exit (GOOD);
+}
+
+void
+print_help ()
+{
+ printf ("These are the options accepted by %s. You may use unambiguous\n\
+abbreviations for the long option names. A - as file name means read\n\
+names from stdin.\n\n", progname);
+
+ puts ("-a, --append\n\
+ Append tag entries to existing tags file.");
+
+ if (CTAGS)
+ puts ("-B, --backward-search\n\
+ Write the search commands for the tag entries using '?', the\n\
+ backward-search command instead of '/', the forward-search command.");
+
+ puts ("-C, --c++\n\
+ Treat files whose extension defaults to C language as C++ files.");
+
+ if (CTAGS)
+ puts ("-d, --defines\n\
+ Create tag entries for constant C #defines, too.");
+ else
+ puts ("-D, --no-defines\n\
+ Don't create tag entries for constant C #defines. This makes\n\
+ the tags file smaller.");
+
+ if (!CTAGS)
+ {
+ puts ("-i FILE, --include=FILE\n\
+ Include a note in tag file indicating that, when searching for\n\
+ a tag, one should also consult the tags file FILE after\n\
+ checking the current file.");
+ puts ("-l LANG, --language=LANG\n\
+ Force the following files to be considered as written in the\n\
+ named language up to the next --language=LANG option.");
+ }
+
+#ifdef ETAGS_REGEXPS
+ puts ("-r /REGEXP/, --regex=/REGEXP/\n\
+ Make a tag for each line matching pattern REGEXP in the\n\
+ following files. REGEXP is anchored (as if preceded by ^).\n\
+ The form /REGEXP/NAME/ creates a named tag. For example Tcl\n\
+ named tags can be created with:\n\
+ --regex=/proc[ \\t]+\\([^ \\t]+\\)/\\1/.");
+ puts ("-R, --no-regex\n\
+ Don't create tags from regexps for the following files.");
+#endif /* ETAGS_REGEXPS */
+ puts ("-o FILE, --output=FILE\n\
+ Write the tags to FILE.");
+ puts ("-S, --ignore-indentation\n\
+ Don't rely on indentation quite as much as normal. Currently,\n\
+ this means not to assume that a closing brace in the first\n\
+ column is the final brace of a function or structure\n\
+ definition in C and C++.");
+
+ if (CTAGS)
+ {
+ puts ("-t, --typedefs\n\
+ Generate tag entries for C typedefs.");
+ puts ("-T, --typedefs-and-c++\n\
+ Generate tag entries for C typedefs, C struct/enum/union tags,\n\
+ and C++ member functions.");
+ puts ("-u, --update\n\
+ Update the tag entries for the given files, leaving tag\n\
+ entries for other files in place. Currently, this is\n\
+ implemented by deleting the existing entries for the given\n\
+ files and then rewriting the new entries at the end of the\n\
+ tags file. It is often faster to simply rebuild the entire\n\
+ tag file than to use this.");
+ puts ("-v, --vgrind\n\
+ Generates an index of items intended for human consumption,\n\
+ similar to the output of vgrind. The index is sorted, and\n\
+ gives the page number of each item.");
+ puts ("-w, --no-warn\n\
+ Suppress warning messages about entries defined in multiple\n\
+ files.");
+ puts ("-x, --cxref\n\
+ Like --vgrind, but in the style of cxref, rather than vgrind.\n\
+ The output uses line numbers instead of page numbers, but\n\
+ beyond that the differences are cosmetic; try both to see\n\
+ which you like.");
+ }
+
+ puts ("-V, --version\n\
+ Print the version of the program.\n\
+-h, --help\n\
+ Print this help message.");
+
+ print_language_names ();
+
+ exit (GOOD);
+}
+
\f
+enum argument_type
+{
+ at_language,
+ at_regexp,
+ at_filename
+};
+
+/* This structure helps us allow mixing of --lang and filenames. */
+typedef struct
+{
+ enum argument_type arg_type;
+ char *what;
+ Lang_function *function;
+} ARGUMENT;
+
+#ifdef VMS /* VMS specific functions */
+
+#define EOS '\0'
+
+/* This is a BUG! ANY arbitrary limit is a BUG!
+ Won't someone please fix this? */
+#define MAX_FILE_SPEC_LEN 255
+typedef struct {
+ short curlen;
+ char body[MAX_FILE_SPEC_LEN + 1];
+} vspec;
+
+/*
+ v1.05 nmm 26-Jun-86 fn_exp - expand specification of list of file names
+ returning in each successive call the next filename matching the input
+ spec. The function expects that each in_spec passed
+ to it will be processed to completion; in particular, up to and
+ including the call following that in which the last matching name
+ is returned, the function ignores the value of in_spec, and will
+ only start processing a new spec with the following call.
+ If an error occurs, on return out_spec contains the value
+ of in_spec when the error occurred.
+
+ With each successive filename returned in out_spec, the
+ function's return value is one. When there are no more matching
+ names the function returns zero. If on the first call no file
+ matches in_spec, or there is any other error, -1 is returned.
+*/
+
+#include <rmsdef.h>
+#include <descrip.h>
+#define OUTSIZE MAX_FILE_SPEC_LEN
+short
+fn_exp (out, in)
+ vspec *out;
+ char *in;
+{
+ static long context = 0;
+ static struct dsc$descriptor_s o;
+ static struct dsc$descriptor_s i;
+ static logical pass1 = TRUE;
+ long status;
+ short retval;
+
+ if (pass1)
+ {
+ pass1 = FALSE;
+ o.dsc$a_pointer = (char *) out;
+ o.dsc$w_length = (short)OUTSIZE;
+ i.dsc$a_pointer = in;
+ i.dsc$w_length = (short)strlen(in);
+ i.dsc$b_dtype = DSC$K_DTYPE_T;
+ i.dsc$b_class = DSC$K_CLASS_S;
+ o.dsc$b_dtype = DSC$K_DTYPE_VT;
+ o.dsc$b_class = DSC$K_CLASS_VS;
+ }
+ if ((status = lib$find_file(&i, &o, &context, 0, 0)) == RMS$_NORMAL)
+ {
+ out->body[out->curlen] = EOS;
+ return 1;
+ }
+ else if (status == RMS$_NMF)
+ retval = 0;
+ else
+ {
+ strcpy(out->body, in);
+ retval = -1;
+ }
+ lib$find_file_end(&context);
+ pass1 = TRUE;
+ return retval;
+}
+
+/*
+ v1.01 nmm 19-Aug-85 gfnames - return in successive calls the
+ name of each file specified by the provided arg expanding wildcards.
+*/
+char *
+gfnames (arg, p_error)
+ char *arg;
+ logical *p_error;
+{
+ static vspec filename = {MAX_FILE_SPEC_LEN, "\0"};
+
+ switch (fn_exp (&filename, arg))
+ {
+ case 1:
+ *p_error = FALSE;
+ return filename.body;
+ case 0:
+ *p_error = FALSE;
+ return NULL;
+ default:
+ *p_error = TRUE;
+ return filename.body;
+ }
+}
+
+#ifndef OLD /* Newer versions of VMS do provide `system'. */
+system (cmd)
+ char *cmd;
+{
+ fprintf (stderr, "system() function not implemented under VMS\n");
+}
+#endif
+#define VERSION_DELIM ';'
+char *massage_name (s)
+ char *s;
+{
+ char *start = s;
+
+ for ( ; *s; s++)
+ if (*s == VERSION_DELIM)
+ {
+ *s = EOS;
+ break;
+ }
+ else
+ *s = tolower(*s);
+ return start;
+}
+#endif /* VMS */
+
+\f
void
main (argc, argv)
int argc;
{
char cmd[100];
int i;
- int outfflag = 0;
unsigned int nincluded_files = 0;
- char **included_files = (char **) alloca (argc * sizeof (char *));
+ char **included_files = xnew (argc, char *);
char *this_file;
+ ARGUMENT *argbuffer;
+ int current_arg = 0, file_count = 0;
#ifdef VMS
- char got_err;
-
- extern char *gfnames ();
- extern char *massage_name ();
+ logical got_err;
#endif
+
+#ifdef DOS_NT
+ _fmode = O_BINARY; /* all of files are treated as binary files */
+#endif /* DOS_NT */
progname = argv[0];
-#ifndef CTAGS
- emacs_tags_format = 1;
-#else
- emacs_tags_format = 0;
-#endif
+ /* Allocate enough no matter what happens. Overkill, but each one
+ is small. */
+ argbuffer = xnew (argc, ARGUMENT);
+
+#ifdef ETAGS_REGEXPS
+ /* Set syntax for regular expression routines. */
+ re_set_syntax (RE_SYNTAX_EMACS);
+#endif /* ETAGS_REGEXPS */
/*
* If etags, always find typedefs and structure tags. Why not?
* Also default is to find macro constants.
*/
- if (emacs_tags_format)
+ if (!CTAGS)
typedefs = typedefs_and_cplusplus = constantypedefs = 1;
- for (; argc > 1 && argv[1][0] == '-' && argv[1][1] != '\0'; argc--, argv++)
+ while (1)
{
- for (i = 1; argv[1][i]; i++)
+ int opt = getopt_long (argc, argv,
+ "-aCdDf:l:o:r:RStTi:BuvxwVhH", longopts, 0);
+
+ if (opt == EOF)
+ break;
+
+ switch (opt)
{
- switch (argv[1][i])
- {
- /* Common options. */
- case 'a':
- append_to_tagfile++;
- break;
- case 'C':
- cplusplus = 1;
- break;
- case 'd':
- constantypedefs = 1;
- break;
- case 'D':
- constantypedefs = 0;
- break;
- case 'o':
- if (outfflag)
- {
- fprintf (stderr,
- "%s: -o flag may only be given once\n", progname);
- goto usage;
- }
- outfflag++, argc--;
- argv++;
- if (argc <= 1 || argv[1][0] == '\0')
- {
- fprintf (stderr,
- "%s: -o flag must be followed by a filename\n",
- progname);
- goto usage;
- }
- outfile = argv[1];
- goto next_arg;
- case 'S':
- noindentypedefs++;
- break;
- case 't':
- typedefs++;
- break;
- case 'T':
- typedefs++;
- typedefs_and_cplusplus++;
- break;
+ case 0:
+ /* If getopt returns 0, then it has already processed a
+ long-named option. We should do nothing. */
+ break;
- /* Etags options */
- case 'i':
- if (!emacs_tags_format)
- goto usage;
- --argc;
- ++argv;
- if (argc <= 1 || argv[1][0] == '\0')
- {
- fprintf (stderr,
- "%s: -i flag must be followed by a filename\n",
- progname);
- goto usage;
- }
- included_files[nincluded_files++] = argv[1];
- goto next_arg;
-
- /* Ctags options. */
- case 'B':
- searchar = '?';
- if (emacs_tags_format)
- goto usage;
- break;
- case 'F':
- searchar = '/';
- if (emacs_tags_format)
- goto usage;
- break;
- case 'u':
- update++;
- if (emacs_tags_format)
- goto usage;
- break;
- case 'v':
- vgrind_style++;
- /*FALLTHRU*/
- case 'x':
- cxref_style++;
- if (emacs_tags_format)
- goto usage;
- break;
- case 'w':
- no_warnings++;
- if (emacs_tags_format)
- goto usage;
- break;
+ case 1:
+ /* This means that a filename has been seen. Record it. */
+ argbuffer[current_arg].arg_type = at_filename;
+ argbuffer[current_arg].what = optarg;
+ ++current_arg;
+ ++file_count;
+ break;
- default:
+ /* Common options. */
+ case 'a':
+ append_to_tagfile++;
+ break;
+ case 'C':
+ cplusplus = 1;
+ break;
+ case 'd':
+ constantypedefs = 1;
+ break;
+ case 'D':
+ constantypedefs = 0;
+ break;
+ case 'f': /* for compatibility with old makefiles */
+ case 'o':
+ if (tagfile)
+ {
+ fprintf(stderr,
+ "%s: -%c flag may only be given once.\n", progname, opt);
goto usage;
}
- }
- next_arg:;
- }
-
- if (argc <= 1)
- {
- usage:
- fprintf (stderr, "Usage:\n");
-#ifndef CTAGS
- fprintf (stderr, "\tetags [-aDiS] [-o tagsfile] file ...\n");
-#else
- fprintf (stderr, "\tctags [-aBdeFTStuwvx] [-o tagsfile] file ...\n");
-#endif
- exit (BAD);
- }
-
- if (outfile == 0)
- {
- outfile = emacs_tags_format ? "TAGS" : "tags";
- }
-
- init (); /* set up boolean "functions" */
+ tagfile = optarg;
+ break;
+ case 'l':
+ if (!get_language (optarg, &argbuffer[current_arg].function))
+ {
+ fprintf (stderr, "%s: language \"%s\" not recognized.\n",
+ progname, optarg);
+ goto usage;
+ }
+ argbuffer[current_arg].arg_type = at_language;
+ ++current_arg;
+ break;
+#ifdef ETAGS_REGEXPS
+ case 'r':
+ argbuffer[current_arg].arg_type = at_regexp;
+ argbuffer[current_arg].what = optarg;
+ ++current_arg;
+ break;
+ case 'R':
+ argbuffer[current_arg].arg_type = at_regexp;
+ argbuffer[current_arg].what = NULL;
+ ++current_arg;
+ break;
+#endif /* ETAGS_REGEXPS */
+ case 'S':
+ noindentypedefs++;
+ break;
+ case 'V':
+ print_version ();
+ break;
+ case 'h':
+ case 'H':
+ print_help ();
+ break;
+ case 't':
+ typedefs++;
+ break;
+ case 'T':
+ typedefs++;
+ typedefs_and_cplusplus++;
+ break;
+#if (!CTAGS)
+ /* Etags options */
+ case 'i':
+ included_files[nincluded_files++] = optarg;
+ break;
+#else /* CTAGS */
+ /* Ctags options. */
+ case 'B':
+ searchar = '?';
+ break;
+ case 'u':
+ update++;
+ break;
+ case 'v':
+ vgrind_style++;
+ /*FALLTHRU*/
+ case 'x':
+ cxref_style++;
+ break;
+ case 'w':
+ no_warnings++;
+ break;
+#endif /* CTAGS */
+ default:
+ fprintf (stderr,
+ "%s: -%c flag not recognised.\n", progname, opt);
+ goto usage;
+ }
+ }
+
+ for (; optind < argc; ++optind)
+ {
+ argbuffer[current_arg].arg_type = at_filename;
+ argbuffer[current_arg].what = argv[optind];
+ ++current_arg;
+ ++file_count;
+ }
+
+ if (nincluded_files == 0 && file_count == 0)
+ {
+ fprintf (stderr, "%s: No input files specified.\n", progname);
+
+ usage:
+ fprintf (stderr, "%s: Try `%s --help' for a complete list of options.\n",
+ progname, progname);
+ exit (BAD);
+ }
+
+ if (tagfile == NULL)
+ {
+ tagfile = CTAGS ? "tags" : "TAGS";
+ }
+ cwd = etags_getcwd (); /* the current working directory */
+ strcat (cwd, "/");
+ if (streq (tagfile, "-"))
+ {
+ tagfiledir = cwd;
+ }
+ else
+ {
+ tagfiledir = absolute_dirname (tagfile, cwd);
+ }
+
+ init (); /* set up boolean "functions" */
initbuffer (&lb);
- initbuffer (&lb1);
+ initbuffer (&lbs[0].lb);
+ initbuffer (&lbs[1].lb);
initbuffer (&filename_lb);
- /*
- * loop through files finding functions
- */
- if (emacs_tags_format)
+
+ if (!CTAGS)
{
- if (streq (outfile, "-"))
- outf = stdout;
+ if (streq (tagfile, "-"))
+ tagf = stdout;
else
- outf = fopen (outfile, append_to_tagfile ? "a" : "w");
- if (!outf)
- {
- perror (outfile);
- exit (1);
- }
+ tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
+ if (tagf == NULL)
+ pfatal (tagfile);
}
- file_num = 1;
-#ifdef VMS
- for (argc--, argv++;
- (this_file = gfnames (&argc, &argv, &got_err)) != NULL; file_num++)
+ /*
+ * Loop through files finding functions.
+ */
+ for (i = 0; i < current_arg; ++i)
{
- if (got_err)
- {
- error ("Can't find file %s\n", this_file);
- argc--, argv++;
- }
- else
+ switch (argbuffer[i].arg_type)
{
- this_file = massage_name (this_file);
-#if 0
- }
- } /* solely to balance out the ifdef'd parens above */
+ case at_language:
+ lang_func = argbuffer[i].function;
+ break;
+#ifdef ETAGS_REGEXPS
+ case at_regexp:
+ add_regex (argbuffer[i].what);
+ break;
#endif
+ case at_filename:
+#ifdef VMS
+ while ((this_file = gfnames (argbuffer[i].what, &got_err)) != NULL)
+ {
+ if (got_err)
+ {
+ error ("Can't find file %s\n", this_file);
+ argc--, argv++;
+ }
+ else
+ {
+ this_file = massage_name (this_file);
+ }
#else
- for (; file_num < argc; file_num++)
- {
- this_file = argv[file_num];
- if (1)
- {
+ this_file = argbuffer[i].what;
#endif
- /* Input file named "-" means read file names from stdin
- and use them. */
- if (streq (this_file, "-"))
- {
- while (!feof (stdin))
+ /* Input file named "-" means read file names from stdin
+ and use them. */
+ if (streq (this_file, "-"))
{
- (void) readline (&filename_lb, stdin);
- if (strlen (filename_lb.buffer) > 0)
- process_file (filename_lb.buffer);
+ while (!feof (stdin))
+ {
+ /* Use readline_internal so that regexp matching */
+ /* is not done on filenames. */
+ (void) readline_internal (&filename_lb, stdin);
+ if (strlen (filename_lb.buffer) > 0)
+ process_file (filename_lb.buffer);
+ }
}
+ else
+ process_file (this_file);
+#ifdef VMS
}
- else
- process_file (this_file);
+#endif
+ break;
}
}
- if (emacs_tags_format)
+ if (!CTAGS)
{
while (nincluded_files-- > 0)
- fprintf (outf, "\f\n%s,include\n", *included_files++);
+ fprintf (tagf, "\f\n%s,include\n", *included_files++);
- (void) fclose (outf);
- exit (0);
+ (void) fclose (tagf);
+ exit (GOOD);
}
if (cxref_style)
put_entries (head);
exit (GOOD);
}
- if (update) /* update cannot be set under VMS */
+ if (update)
{
- for (i = 1; i < argc; i++)
+ for (i = 0; i < current_arg; ++i)
{
+ if (argbuffer[i].arg_type == at_language)
+ continue;
sprintf (cmd,
"mv %s OTAGS;fgrep -v '\t%s\t' OTAGS >%s;rm OTAGS",
- outfile, argv[i], outfile);
+ tagfile, argbuffer[i].what, tagfile);
(void) system (cmd);
}
append_to_tagfile++;
}
- outf = fopen (outfile, append_to_tagfile ? "a" : "w");
- if (outf == NULL)
+ tagf = fopen (tagfile, append_to_tagfile ? "a" : "w");
+ if (tagf == NULL)
{
- perror (outfile);
+ perror (tagfile);
exit (GOOD);
}
put_entries (head);
- (void) fclose (outf);
+ (void) fclose (tagf);
if (update)
{
- sprintf (cmd, "sort %s -o %s", outfile, outfile);
+ sprintf (cmd, "sort %s -o %s", tagfile, tagfile);
(void) system (cmd);
}
exit (GOOD);
}
+/*
+ * Set the language, given the name.
+ */
+logical
+get_language (language, func)
+ char *language;
+ Lang_function **func;
+{
+ struct lang_entry *lang;
+
+ for (lang = lang_names; lang->extension; ++lang)
+ {
+ if (streq (language, lang->extension))
+ {
+ *func = lang->function;
+ return TRUE;
+ }
+ }
+
+ return FALSE;
+}
+
+
/*
* This routine is called on each file argument.
*/
{
struct stat stat_buf;
- stat (file, &stat_buf);
- if (!(stat_buf.st_mode & S_IFREG) || !(stat_buf.st_mode & S_IFLNK))
+ if (stat (file, &stat_buf) == 0 && !S_ISREG (stat_buf.st_mode))
{
fprintf (stderr, "Skipping %s: it is not a regular file.\n", file);
return;
}
-
- if (streq (file, outfile) && !streq (outfile, "-"))
+ if (streq (file, tagfile) && !streq (tagfile, "-"))
{
fprintf (stderr, "Skipping inclusion of %s in self.\n", file);
return;
}
- if (emacs_tags_format)
+ if (!find_entries (file))
{
- char *cp = rindex (file, '/');
- if (cp)
- ++cp;
- else
- cp = file;
+ return;
}
- find_entries (file);
- if (emacs_tags_format)
+ if (!CTAGS)
{
- fprintf (outf, "\f\n%s,%d\n",
- file, total_size_of_entries (head));
+ char *filename;
+
+ if (file[0] == '/')
+ {
+ /* file is an absolute filename. Canonicalise it. */
+ filename = absolute_filename (file, cwd);
+ }
+ else
+ {
+ /* file is a filename relative to cwd. Make it relative
+ to the directory of the tags file. */
+ filename = relative_filename (file, tagfiledir);
+ }
+ fprintf (tagf, "\f\n%s,%d\n", filename, total_size_of_entries (head));
put_entries (head);
free_tree (head);
head = NULL;
}
/*
- * This routine sets up the boolean psuedo-functions which work
- * by seting boolean flags dependent upon the corresponding character
+ * This routine sets up the boolean pseudo-functions which work
+ * by setting boolean flags dependent upon the corresponding character
* Every char which is NOT in that string is not a white char. Therefore,
* all of the array "_wht" is set to FALSE, and then the elements
* subscripted by the chars in "white" are set to TRUE. Thus "_wht"
void
init ()
{
- reg char *sp;
- reg int i;
+ register char *sp;
+ register int i;
for (i = 0; i < 0177; i++)
- {
- _wht[i] = _etk[i] = _itk[i] = _btk[i] = FALSE;
- _gd[i] = TRUE;
- }
+ _wht[i] = _etk[i] = _itk[i] = _btk[i] = FALSE;
for (sp = white; *sp; sp++)
_wht[*sp] = TRUE;
for (sp = endtk; *sp; sp++)
_itk[*sp] = TRUE;
for (sp = begtk; *sp; sp++)
_btk[*sp] = TRUE;
- for (sp = notgd; *sp; sp++)
- _gd[*sp] = FALSE;
_wht[0] = _wht['\n'];
_etk[0] = _etk['\n'];
_btk[0] = _btk['\n'];
_itk[0] = _itk['\n'];
- _gd[0] = _gd['\n'];
}
/*
* This routine opens the specified file and calls the function
* which finds the function and type definitions.
*/
-void
+logical
find_entries (file)
char *file;
{
+ FILE *inf;
char *cp;
- void prolog_funcs ();
+ struct lang_entry *lang;
+ NODE *old_last_node;
+ extern NODE *last_node;
inf = fopen (file, "r");
if (inf == NULL)
{
perror (file);
- return;
+ return FALSE;
}
curfile = savestr (file);
- cp = rindex (file, '.');
-
- header_file = (cp && (streq (cp + 1, "h")));
+ cp = etags_strrchr (file, '.');
- /* .tex, .aux or .bbl implies LaTeX source code */
- if (cp && (streq (cp + 1, "tex") || streq (cp + 1, "aux")
- || streq (cp + 1, "bbl")))
- {
- TEX_funcs (inf);
- goto close_and_return;
- }
- /* .l or .el or .lisp (or .cl or .clisp or ...) implies lisp source code */
- if (cp && (streq (cp + 1, "l")
- || streq (cp + 1, "el")
- || streq (cp + 1, "lsp")
- || streq (cp + 1, "lisp")
- || streq (cp + 1, "cl")
- || streq (cp + 1, "clisp")))
- {
- L_funcs (inf);
- goto close_and_return;
- }
- /* .scm or .sm or .scheme or ... implies scheme source code */
- if (cp && (streq (cp + 1, "sm")
- || streq (cp + 1, "scm")
- || streq (cp + 1, "scheme")
- || streq (cp + 1, "t")
- || streq (cp + 1, "sch")
- || streq (cp + 1, "SM")
- || streq (cp + 1, "SCM")
- /* The `SCM' or `scm' prefix with a version number */
- || (cp[-1] == 'm' && cp[-2] == 'c' && cp[-3] == 's'
- && string_numeric_p (cp + 1))
- || (cp[-1] == 'M' && cp[-2] == 'C' && cp[-3] == 'S'
- && string_numeric_p (cp + 1))))
- {
- Scheme_funcs (inf);
- fclose (inf);
- return;
- }
- /* Assume that ".s" or ".a" is assembly code. -wolfgang. */
- if (cp && (cp[1] == 's' || cp[1] == 'a') && cp[2] == '\0')
+ /* If user specified a language, use it. */
+ if (lang_func != NULL)
{
- Asm_funcs (inf);
+ lang_func (inf);
fclose (inf);
- return;
- }
- /* .C or .H or .cxx or .hxx or .cc: a C++ file */
- if (cp && (streq (cp + 1, "C")
- || streq (cp + 1, "H")
- || streq (cp + 1, "cxx")
- || streq (cp + 1, "hxx")
- || streq (cp + 1, "cc")))
- {
- C_entries (C_PLPL); /* C++ */
- goto close_and_return;
- }
- /* .cs or .hs: a C* file */
- if (cp && (cp[1] == 'c' || cp[1] == 'h') && cp[2] == 's' && cp[3] == '\0')
- {
- C_entries (C_STAR);
- goto close_and_return;
+ return TRUE;
}
- /* .pl implies prolog source code */
- if (cp && !strcmp (cp + 1, "pl"))
- {
- prolog_funcs (inf);
- goto close_and_return;
- }
- /* .p or .pas: a Pascal file */
- if (cp && (streq (cp + 1, "p")
- || streq (cp + 1, "pas")))
- {
- PAS_funcs (inf);
- goto close_and_return;
- }
- /* if not a .c or .h or .y file, try fortran */
- else if (cp && ((cp[1] != 'c' && cp[1] != 'h' && cp[1] != 'y')
- || (cp[1] != 0 && cp[2] != 0)))
+
+ if (cp)
{
- if (PF_funcs (inf) != 0)
- goto close_and_return;
- rewind (inf); /* no fortran tags found, try C */
+ ++cp;
+ for (lang = lang_extensions; lang->extension; ++lang)
+ {
+ if (streq (cp, lang->extension))
+ {
+ lang->function (inf);
+ fclose (inf);
+ return TRUE;
+ }
+ }
}
- C_entries (cplusplus ? C_PLPL : 0);
-close_and_return:
- (void) fclose (inf);
-}
-
-/* Nonzero if string STR is composed of digits. */
+ /* Try Fortran. */
+ old_last_node = last_node;
+ Fortran_functions (inf);
-int
-string_numeric_p (str)
- char *str;
-{
- while (*str)
- {
- if (*str < '0' || *str > '9')
- return 0;
- }
- return 1;
+ /* No Fortran entries found. Try C. */
+ if (old_last_node == last_node)
+ default_C_entries (inf);
+ fclose (inf);
+ return TRUE;
}
\f
/* Record a tag. */
/* Should take a TOKEN* instead!! */
-
void
-pfnote (name, is_func, rewritten, linestart, linelen, lno, cno)
+pfnote (name, is_func, named, linestart, linelen, lno, cno)
char *name; /* tag name */
logical is_func; /* function or type name? */
- logical rewritten; /* tag different from text of definition? */
+ logical named; /* tag different from text of definition? */
char *linestart;
int linelen;
int lno;
char tem[51];
char c;
- np = (NODE *) malloc (sizeof (NODE));
+ np = xnew (1, NODE);
if (np == NULL)
{
- if (!emacs_tags_format)
+ if (CTAGS)
{
/* It's okay to output early in etags -- it only disrupts the
* character count of the tag entries, which is no longer used
* by tags.el anyway.
*/
- error ("too many entries to sort");
+ error ("too many entries to sort", 0);
}
put_entries (head);
free_tree (head);
np = xnew (1, NODE);
}
/* If ctags mode, change name "main" to M<thisfilename>. */
- if (!emacs_tags_format && !cxref_style && streq (name, "main"))
+ if (CTAGS && !cxref_style && streq (name, "main"))
{
- fp = rindex (curfile, '/');
+ fp = etags_strrchr (curfile, '/');
name = concat ("M", fp == 0 ? curfile : fp + 1, "");
- fp = rindex (name, '.');
+ fp = etags_strrchr (name, '.');
if (fp && fp[1] != '\0' && fp[2] == '\0')
*fp = 0;
- rewritten = TRUE;
+ named = TRUE;
}
np->name = savestr (name);
np->file = curfile;
np->is_func = is_func;
- np->rewritten = rewritten;
+ np->named = named;
np->lno = lno;
- /* UNCOMMENT THE +1 HERE: */
- np->cno = cno /* + 1 */ ; /* our char numbers are 0-base; emacs's are 1-base */
+ /* Our char numbers are 0-base, because of C language tradition?
+ ctags compatibility? old versions compatibility? I don't know.
+ Anyway, since emacs's are 1-base we espect etags.el to take care
+ of the difference. If we wanted to have 1-based numbers, we would
+ uncomment the +1 below. */
+ np->cno = cno /* + 1 */ ;
np->left = np->right = 0;
- if (emacs_tags_format)
+ if (!CTAGS)
{
c = linestart[linelen];
linestart[linelen] = 0;
linestart = tem;
}
np->pat = savestr (linestart);
- if (emacs_tags_format)
+ if (!CTAGS)
{
linestart[linelen] = c;
}
* add_node is the only function allowed to add nodes, so it can
* maintain state.
*/
+NODE *last_node = NULL;
void
add_node (node, cur_node_p)
NODE *node, **cur_node_p;
{
register int dif;
register NODE *cur_node = *cur_node_p;
- static NODE *last_node = NULL;/* careful */
if (cur_node == NULL)
{
return;
}
- if (emacs_tags_format)
+ if (!CTAGS)
{
/* Etags Mode */
- if (!last_node)
- fatal ("internal error in add_node");
+ if (last_node == NULL)
+ fatal ("internal error in add_node", 0);
last_node->right = node;
last_node = node;
}
/* Maybe refuse to add duplicate nodes. */
if (!permit_duplicates)
{
- if (!strcmp (node->name, cur_node->name)
- && !strcmp (node->file, cur_node->file))
+ if (streq (node->name, cur_node->name)
+ && streq (node->file, cur_node->file))
return;
}
\f
void
put_entries (node)
- reg NODE *node;
+ register NODE *node;
{
- reg char *sp;
+ register char *sp;
if (node == NULL)
return;
/* Output this entry */
- if (emacs_tags_format)
+ if (!CTAGS)
{
- if (node->rewritten)
+ if (node->named)
{
- fprintf (outf, "%s\177%s\001%d,%d\n",
- node->name, node->pat, node->lno, node->cno);
+ fprintf (tagf, "%s\177%s\001%d,%d\n",
+ node->pat, node->name,
+ node->lno, node->cno);
}
else
{
- fprintf (outf, "%s\177%d,%d\n",
- node->pat, node->lno, node->cno);
+ fprintf (tagf, "%s\177%d,%d\n",
+ node->pat,
+ node->lno, node->cno);
}
}
else if (!cxref_style)
{
- fprintf (outf, "%s\t%s\t",
+ fprintf (tagf, "%s\t%s\t",
node->name, node->file);
if (node->is_func)
{ /* a function */
- putc (searchar, outf);
- putc ('^', outf);
+ putc (searchar, tagf);
+ putc ('^', tagf);
for (sp = node->pat; *sp; sp++)
{
if (*sp == '\\' || *sp == searchar)
- putc ('\\', outf);
- putc (*sp, outf);
+ putc ('\\', tagf);
+ putc (*sp, tagf);
}
- putc (searchar, outf);
+ putc (searchar, tagf);
}
else
{ /* a typedef; text pattern inadequate */
- fprintf (outf, "%d", node->lno);
+ fprintf (tagf, "%d", node->lno);
}
- putc ('\n', outf);
+ putc ('\n', tagf);
}
else if (vgrind_style)
fprintf (stdout, "%s %s %d\n",
node->name, node->file, (node->lno + 63) / 64);
else
- fprintf (stdout, "%-16s%4d %-16s %s\n",
+ fprintf (stdout, "%-16s %3d %-16s %s\n",
node->name, node->lno, node->file, node->pat);
/* Output subentries that follow this one */
/*
* Return total number of characters that put_entries will output for
- * the nodes in the subtree of the specified node. Works only if emacs_tags_format
- * is set, but called only in that case. This count is irrelevant with
- * the new tags.el, but is still supplied for backward compatibility.
+ * the nodes in the subtree of the specified node. Works only if
+ * we are not ctags, but called only in that case. This count
+ * is irrelevant with the new tags.el, but is still supplied for
+ * backward compatibility.
*/
int
total_size_of_entries (node)
- reg NODE *node;
+ register NODE *node;
{
- reg int total;
+ register int total;
if (node == NULL)
return 0;
/* Count this entry */
total += strlen (node->pat) + 1;
total += number_len ((long) node->lno) + 1 + number_len (node->cno) + 1;
- if (node->rewritten)
+ if (node->named)
total += 1 + strlen (node->name); /* \001name */
}
* The C symbol tables.
*/
-Stab *C_stab, *C_PLPL_stab, *C_STAR_stab;
-
+/* Feed stuff between (but not including) %[ and %] lines to:
+ gperf -c -k1,3 -o -p -r -t
+%[
+struct C_stab_entry { char *name; int c_ext; enum sym_type type; }
+%%
+class, C_PLPL, st_C_struct
+domain, C_STAR, st_C_struct
+union, 0, st_C_struct
+struct, 0, st_C_struct
+enum, 0, st_C_enum
+typedef, 0, st_C_typedef
+define, 0, st_C_define
+long, 0, st_C_typespec
+short, 0, st_C_typespec
+int, 0, st_C_typespec
+char, 0, st_C_typespec
+float, 0, st_C_typespec
+double, 0, st_C_typespec
+signed, 0, st_C_typespec
+unsigned, 0, st_C_typespec
+auto, 0, st_C_typespec
+void, 0, st_C_typespec
+extern, 0, st_C_typespec
+static, 0, st_C_typespec
+const, 0, st_C_typespec
+volatile, 0, st_C_typespec
+%]
+and replace lines between %< and %> with its output. */
+/*%<*/
+/* C code produced by gperf version 1.8.1 (K&R C version) */
+/* Command-line: gperf -c -k1,3 -o -p -r -t */
+
+
+struct C_stab_entry { char *name; int c_ext; enum sym_type type; };
+
+#define MIN_WORD_LENGTH 3
+#define MAX_WORD_LENGTH 8
+#define MIN_HASH_VALUE 10
+#define MAX_HASH_VALUE 62
/*
- * SYNOPSIS
- * Stab *get_C_stab (int c_ext);
- */
-#define get_C_stab(c_ext) ((c_ext&C_STAR) ? C_STAR_stab : \
- c_ext ? C_PLPL_stab : \
- C_stab)
+ 21 keywords
+ 53 is the maximum key range
+*/
+
+static int
+hash (str, len)
+ register char *str;
+ register int len;
+{
+ static unsigned char hash_table[] =
+ {
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62, 62, 62,
+ 62, 62, 62, 62, 62, 62, 62, 2, 62, 7,
+ 6, 9, 15, 30, 62, 24, 62, 62, 1, 24,
+ 7, 27, 13, 62, 19, 26, 18, 27, 1, 62,
+ 62, 62, 62, 62, 62, 62, 62, 62,
+ };
+ return len + hash_table[str[2]] + hash_table[str[0]];
+}
-void
-add_keyword (stab, sym, type)
- Stab *stab;
- char *sym;
- enum sym_type type;
+struct C_stab_entry *
+in_word_set (str, len)
+ register char *str;
+ register int len;
{
- stab_search (stab, sym, strlen (sym))->type = type;
+
+ static struct C_stab_entry wordlist[] =
+ {
+ {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
+ {"",},
+ {"volatile", 0, st_C_typespec},
+ {"",},
+ {"long", 0, st_C_typespec},
+ {"char", 0, st_C_typespec},
+ {"class", C_PLPL, st_C_struct},
+ {"",}, {"",}, {"",}, {"",},
+ {"const", 0, st_C_typespec},
+ {"",}, {"",}, {"",}, {"",},
+ {"auto", 0, st_C_typespec},
+ {"",}, {"",},
+ {"define", 0, st_C_define},
+ {"",},
+ {"void", 0, st_C_typespec},
+ {"",}, {"",}, {"",},
+ {"extern", 0, st_C_typespec},
+ {"static", 0, st_C_typespec},
+ {"",},
+ {"domain", C_STAR, st_C_struct},
+ {"",},
+ {"typedef", 0, st_C_typedef},
+ {"double", 0, st_C_typespec},
+ {"enum", 0, st_C_enum},
+ {"",}, {"",}, {"",}, {"",},
+ {"int", 0, st_C_typespec},
+ {"",},
+ {"float", 0, st_C_typespec},
+ {"",}, {"",}, {"",},
+ {"struct", 0, st_C_struct},
+ {"",}, {"",}, {"",}, {"",},
+ {"union", 0, st_C_struct},
+ {"",},
+ {"short", 0, st_C_typespec},
+ {"",}, {"",},
+ {"unsigned", 0, st_C_typespec},
+ {"signed", 0, st_C_typespec},
+ };
+
+ if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)
+ {
+ register int key = hash (str, len);
+
+ if (key <= MAX_HASH_VALUE && key >= MIN_HASH_VALUE)
+ {
+ register char *s = wordlist[key].name;
+
+ if (*s == *str && strneq (str + 1, s + 1, len - 1))
+ return &wordlist[key];
+ }
+ }
+ return 0;
}
+/*%>*/
-Stab *
-C_create_stab (c_ext)
+enum sym_type
+C_symtype(str, len, c_ext)
+ char *str;
+ int len;
int c_ext;
{
- Stab *stab;
-
- stab = stab_create ();
-
- /* C, C++ and C* */
- if (c_ext & C_PLPL)
- add_keyword (stab, "class", st_C_struct);
- if (c_ext & C_STAR)
- add_keyword (stab, "domain", st_C_struct);
- add_keyword (stab, "union", st_C_struct);
- add_keyword (stab, "struct", st_C_struct);
- add_keyword (stab, "enum", st_C_enum);
- add_keyword (stab, "typedef", st_C_typedef);
- add_keyword (stab, "define", st_C_define);
- add_keyword (stab, "long", st_C_typespec);
- add_keyword (stab, "short", st_C_typespec);
- add_keyword (stab, "int", st_C_typespec);
- add_keyword (stab, "char", st_C_typespec);
- add_keyword (stab, "float", st_C_typespec);
- add_keyword (stab, "double", st_C_typespec);
- add_keyword (stab, "signed", st_C_typespec);
- add_keyword (stab, "unsigned", st_C_typespec);
- add_keyword (stab, "const", st_C_typespec);
- add_keyword (stab, "volatile", st_C_typespec);
-
- return stab;
+ register struct C_stab_entry *se = in_word_set(str, len);
+
+ if (se == NULL || (se->c_ext && !(c_ext & se->c_ext)))
+ return st_none;
+ return se->type;
}
+\f
+ /*
+ * C functions are recognized using a simple finite automaton.
+ * funcdef is its state variable.
+ */
+typedef enum
+{
+ fnone, /* nothing seen */
+ ftagseen, /* function-like tag seen */
+ fstartlist, /* just after open parenthesis */
+ finlist, /* in parameter list */
+ flistseen, /* after parameter list */
+ fignore /* before open brace */
+} FUNCST;
+FUNCST funcdef;
+
+
+ /*
+ * typedefs are recognized using a simple finite automaton.
+ * typeddef is its state variable.
+ */
+typedef enum
+{
+ tnone, /* nothing seen */
+ ttypedseen, /* typedef keyword seen */
+ tinbody, /* inside typedef body */
+ tend, /* just before typedef tag */
+ tignore /* junk after typedef tag */
+} TYPEDST;
+TYPEDST typdef;
+
+
+ /*
+ * struct-like structures (enum, struct and union) are recognized
+ * using another simple finite automaton. `structdef' is its state
+ * variable.
+ */
+typedef enum
+{
+ snone, /* nothing seen yet */
+ skeyseen, /* struct-like keyword seen */
+ stagseen, /* struct-like tag seen */
+ scolonseen, /* colon seen after struct-like tag */
+ sinbody /* in struct body: recognize member func defs*/
+} STRUCTST;
+STRUCTST structdef;
-void
-C_create_stabs ()
+/*
+ * When structdef is stagseen, scolonseen, or sinbody, structtag is the
+ * struct tag, and structtype is the type of the preceding struct-like
+ * keyword.
+ */
+char structtag[BUFSIZ];
+enum sym_type structtype;
+
+/*
+ * Yet another little state machine to deal with preprocessor lines.
+ */
+typedef enum
+{
+ dnone, /* nothing seen */
+ dsharpseen, /* '#' seen as first char on line */
+ ddefineseen, /* '#' and 'define' seen */
+ dignorerest /* ignore rest of line */
+} DEFINEST;
+DEFINEST definedef;
+
+/*
+ * Set this to TRUE, and the next token considered is called a function.
+ * Used only for GNU emacs's function-defining macros.
+ */
+logical next_token_is_func;
+
+/*
+ * TRUE in the rules part of a yacc file, FALSE outside (parse as C).
+ */
+logical yacc_rules;
+
+/*
+ * consider_token ()
+ * checks to see if the current token is at the start of a
+ * function, or corresponds to a typedef, or is a struct/union/enum
+ * tag.
+ *
+ * *IS_FUNC gets TRUE iff the token is a function or macro with args.
+ * C_EXT is which language we are looking at.
+ *
+ * In the future we will need some way to adjust where the end of
+ * the token is; for instance, implementing the C++ keyword
+ * `operator' properly will adjust the end of the token to be after
+ * whatever follows `operator'.
+ *
+ * Globals
+ * funcdef IN OUT
+ * structdef IN OUT
+ * definedef IN OUT
+ * typdef IN OUT
+ * next_token_is_func IN OUT
+ */
+
+logical
+consider_token (c, tokp, c_ext, cblev, is_func)
+ register char c; /* IN: first char after the token */
+ register TOKEN *tokp; /* IN: token pointer */
+ int c_ext; /* IN: C extensions mask */
+ int cblev; /* IN: curly brace level */
+ logical *is_func; /* OUT: function found */
{
- C_stab = C_create_stab (0);
- C_PLPL_stab = C_create_stab (C_PLPL);
- C_STAR_stab = C_create_stab (C_STAR | C_PLPL);
+ enum sym_type toktype = C_symtype(tokp->p, tokp->len, c_ext);
+
+ /*
+ * Advance the definedef state machine.
+ */
+ switch (definedef)
+ {
+ case dnone:
+ /* We're not on a preprocessor line. */
+ break;
+ case dsharpseen:
+ if (toktype == st_C_define)
+ {
+ definedef = ddefineseen;
+ }
+ else
+ {
+ definedef = dignorerest;
+ }
+ return FALSE;
+ case ddefineseen:
+ /*
+ * Make a tag for any macro, unless it is a constant
+ * and constantypedefs is FALSE.
+ */
+ definedef = dignorerest;
+ *is_func = (c == '(');
+ if (!*is_func && !constantypedefs)
+ return FALSE;
+ else
+ return TRUE;
+ case dignorerest:
+ return FALSE;
+ default:
+ error ("internal error: definedef value.", 0);
+ }
+
+ /*
+ * Now typedefs
+ */
+ switch (typdef)
+ {
+ case tnone:
+ if (toktype == st_C_typedef)
+ {
+ if (typedefs)
+ typdef = ttypedseen;
+ funcdef = fnone;
+ return FALSE;
+ }
+ break;
+ case ttypedseen:
+ switch (toktype)
+ {
+ case st_none:
+ case st_C_typespec:
+ typdef = tend;
+ break;
+ case st_C_struct:
+ case st_C_enum:
+ break;
+ }
+ /* Do not return here, so the structdef stuff has a chance. */
+ break;
+ case tend:
+ switch (toktype)
+ {
+ case st_C_typespec:
+ case st_C_struct:
+ case st_C_enum:
+ return FALSE;
+ }
+ return TRUE;
+ }
+
+ /*
+ * This structdef business is currently only invoked when cblev==0.
+ * It should be recursively invoked whatever the curly brace level,
+ * and a stack of states kept, to allow for definitions of structs
+ * within structs.
+ *
+ * This structdef business is NOT invoked when we are ctags and the
+ * file is plain C. This is because a struct tag may have the same
+ * name as another tag, and this loses with ctags.
+ *
+ * This if statement deals with the typdef state machine as
+ * follows: if typdef==ttypedseen and token is struct/union/class/enum,
+ * return FALSE. All the other code here is for the structdef
+ * state machine.
+ */
+ switch (toktype)
+ {
+ case st_C_struct:
+ case st_C_enum:
+ if (typdef == ttypedseen
+ || (typedefs_and_cplusplus && cblev == 0 && structdef == snone))
+ {
+ structdef = skeyseen;
+ structtype = toktype;
+ }
+ return FALSE;
+ }
+ if (structdef == skeyseen)
+ {
+ if (structtype == st_C_struct)
+ {
+ strncpy (structtag, tokp->p, tokp->len);
+ structtag[tokp->len] = '\0'; /* for struct/union/class */
+ }
+ else
+ {
+ structtag[0] = '\0'; /* for enum (why is it treated differently?) */
+ }
+ structdef = stagseen;
+ return TRUE;
+ }
+
+ /* Avoid entering funcdef stuff if typdef is going on. */
+ if (typdef != tnone)
+ {
+ definedef = dnone;
+ return FALSE;
+ }
+
+ /* Detect GNU macros. */
+ if (definedef == dnone)
+ if (strneq (tokp->p, "DEFUN", 5) /* Used in emacs */
+#if FALSE
+ These are defined inside C functions, so currently they
+ are not met anyway.
+ || strneq (tokp->p, "EXFUN", 5) /* Used in glibc */
+ || strneq (tokp->p, "DEFVAR_", 7) /* Used in emacs */
+#endif
+ || strneq (tokp->p, "SYSCALL", 7) /* Used in glibc (mach) */
+ || strneq (tokp->p, "ENTRY", 5) /* Used in glibc */
+ || strneq (tokp->p, "PSEUDO", 6)) /* Used in glibc */
+
+ {
+ next_token_is_func = TRUE;
+ return FALSE;
+ }
+ if (next_token_is_func)
+ {
+ next_token_is_func = FALSE;
+ funcdef = fignore;
+ *is_func = TRUE;
+ return TRUE;
+ }
+
+ /* A function? */
+ switch (toktype)
+ {
+ case st_C_typespec:
+ if (funcdef != finlist && funcdef != fignore)
+ funcdef = fnone; /* should be useless */
+ return FALSE;
+ default:
+ if (funcdef == fnone)
+ {
+ funcdef = ftagseen;
+ *is_func = TRUE;
+ return TRUE;
+ }
+ }
+
+ return FALSE;
}
-\f
+
/*
* C_entries ()
- * This routine finds functions and typedefs in C syntax and adds them
+ * This routine finds functions, typedefs, #define's and
+ * struct/union/enum definitions in C syntax and adds them
* to the list.
*/
+#define curlb (lbs[curndx].lb)
+#define othlb (lbs[1-curndx].lb)
+#define newlb (lbs[newndx].lb)
+#define curlinepos (lbs[curndx].linepos)
+#define othlinepos (lbs[1-curndx].linepos)
+#define newlinepos (lbs[newndx].linepos)
+
+/* Save and restore token state. This is used when preprocessor defines
+ are handled, to avoid disturbing active function/typedef/struct states. */
+#define TOKEN_SAVED_P (savetok.lineno > 0)
+#define SAVE_TOKEN (savetok = tok, savetok.p = (char *) tokoff, \
+ savetok.len = toklen, strcpy(savenameb, nameb))
+#define RESTORE_TOKEN (tok = savetok, tokoff = (int) tok.p, \
+ toklen = tok.len, strcpy(nameb, savenameb), \
+ savetok.lineno = 0)
+
#define CNL_SAVE_DEFINEDEF \
-{ \
- prev_linepos = linepos; \
- SET_FILEPOS (linepos, inf, charno); \
+do { \
+ SET_FILEPOS (curlinepos, inf, charno); \
lineno++; \
- charno += readline (&lb, inf); \
- lp = lb.buffer; \
-}
+ charno += readline (&curlb, inf); \
+ lp = curlb.buffer; \
+ quotednl = FALSE; \
+ newndx = curndx; \
+} while (0)
#define CNL \
-{ \
+do { \
CNL_SAVE_DEFINEDEF; \
+ if (TOKEN_SAVED_P) \
+ RESTORE_TOKEN; \
definedef = dnone; \
-}
+} while (0)
+
+#define MAKE_TAG_FROM_NEW_LB(isfun) pfnote (nameb, isfun, tok.named, \
+ newlb.buffer, tokoff + toklen + 1, tok.lineno, GET_CHARNO (newlinepos))
+#define MAKE_TAG_FROM_OTH_LB(isfun) pfnote (nameb, isfun, tok.named, \
+ othlb.buffer, tokoff + toklen + 1, tok.lineno, GET_CHARNO (othlinepos))
void
-C_entries (c_ext)
- int c_ext; /* extension of C? */
+C_entries (c_ext, inf)
+ int c_ext; /* extension of C */
+ FILE *inf; /* input file */
{
- register int c; /* latest char read; '\0' for end of line */
- register int tokoff; /* offset in line of beginning of latest token */
- register int toklen; /* length of latest token */
+ register char c; /* latest char read; '\0' for end of line */
register char *lp; /* pointer one beyond the character `c' */
- logical incomm, inquote, inchar, midtoken;
- int level; /* current curly brace level */
- char tokb[BUFSIZ];
-
+ int curndx, newndx; /* indices for current and new lb */
+ TOKEN tok; /* latest token read for funcdef & structdef */
+ char nameb[BUFSIZ]; /* latest token name for funcdef & structdef */
+ register int tokoff; /* offset in line of start of latest token */
+ register int toklen; /* length of latest token */
+ int cblev; /* current curly brace level */
+ int parlev; /* current parenthesis level */
+ logical incomm, inquote, inchar, quotednl, midtoken;
+ logical cplpl;
+ TOKEN savetok; /* saved token during preprocessor handling */
+ char savenameb[BUFSIZ]; /* ouch! */
+
+ savetok.lineno = 0;
+ curndx = newndx = 0;
lineno = 0;
charno = 0;
- lp = lb.buffer;
+ lp = curlb.buffer;
*lp = 0;
- definedef = dnone;
- gotone = midtoken = inquote = inchar = incomm = FALSE;
- level = 0;
-
- C_create_stabs ();
+ definedef = dnone; funcdef = fnone; typdef = tnone; structdef = snone;
+ next_token_is_func = yacc_rules = FALSE;
+ midtoken = inquote = inchar = incomm = quotednl = FALSE;
+ cblev = 0;
+ parlev = 0;
+ cplpl = c_ext & C_PLPL;
while (!feof (inf))
{
c = *lp++;
if (c == '\\')
{
- if (*lp == 0)
- continue;
+ /* If we're at the end of the line, the next character is a
+ '\0'; don't skip it, because it's the thing that tells us
+ to read the next line. */
+ if (*lp == '\0')
+ {
+ quotednl = TRUE;
+ continue;
+ }
lp++;
c = ' ';
}
else if (incomm)
{
- if (c == '*' && *lp == '/')
+ switch (c)
{
- c = *lp++;
- incomm = FALSE;
+ case '*':
+ if (*lp == '/')
+ {
+ c = *lp++;
+ incomm = FALSE;
+ }
+ break;
+ case '\0':
+ /* Newlines inside comments do not end macro definitions in
+ traditional cpp. */
+ CNL_SAVE_DEFINEDEF;
+ break;
}
+ continue;
}
else if (inquote)
{
- /*
- * Too dumb to know about \" not being magic, but
- * they usually occur in pairs anyway.
- */
- if (c == '"')
- inquote = FALSE;
+ switch (c)
+ {
+ case '"':
+ inquote = FALSE;
+ break;
+ case '\0':
+ /* Newlines inside strings do not end macro definitions
+ in traditional cpp, even though compilers don't
+ usually accept them. */
+ CNL_SAVE_DEFINEDEF;
+ break;
+ }
continue;
}
else if (inchar)
{
- if (c == '\'')
- inchar = FALSE;
+ switch (c)
+ {
+ case '\0':
+ /* Hmmm, something went wrong. */
+ CNL;
+ /* FALLTHRU */
+ case '\'':
+ inchar = FALSE;
+ break;
+ }
continue;
}
- else
+ else
switch (c)
{
case '"':
inquote = TRUE;
+ if (funcdef != finlist && funcdef != fignore)
+ funcdef = fnone;
continue;
case '\'':
inchar = TRUE;
+ if (funcdef != finlist && funcdef != fignore)
+ funcdef = fnone;
continue;
case '/':
if (*lp == '*')
{
lp++;
incomm = TRUE;
+ continue;
}
- else if (c_ext && *lp == '/')
- {
- c = 0; /* C++ comment: skip rest of line */
- }
- continue;
- case '#':
- if (lp == lb.buffer + 1 && definedef == dnone)
- definedef = dsharpseen;
- continue;
-
- /*
- * The next two are to help the strucdef state machine.
- * They break when they are finished, so they don't interfere
- * with anything else that is going on.
- */
- case ':':
- if (structdef == stagseen)
- structdef = scolonseen;
- break;
- /* Not a struct definition when semicolon seen in non-sinbody context. */
- case ';':
- if (structdef != snone && structdef != sinbody)
- {
- structdef = snone;
- (void) strcpy (structtag, "<error 1>");
- }
- break;
-
- case '{':
- if (tydef == begin)
- {
- tydef = middle;
- }
- switch (structdef)
+ else if (cplpl && *lp == '/')
{
- case skeyseen: /* unnamed struct */
- structtag[0] = '\0';
- /* FALLTHRU */
- case stagseen:
- case scolonseen: /* named struct */
- structdef = sinbody;
+ c = 0;
break;
}
- level++;
- continue;
- case '}':
- if (!noindentypedefs && lp == lb.buffer + 1)
- level = 0; /* reset level if first column */
- else if (level > 0)
- level--;
- if (level == 0 && tydef == middle)
+ else
+ break;
+ case '%':
+ if ((c_ext & YACC) && *lp == '%')
{
- tydef = end;
- }
- if (level == 0)
+ /* entering or exiting rules section in yacc file */
+ lp++;
+ definedef = dnone; funcdef = fnone;
+ typdef = tnone; structdef = snone;
+ next_token_is_func = FALSE;
+ midtoken = inquote = inchar = incomm = quotednl = FALSE;
+ cblev = 0;
+ yacc_rules = !yacc_rules;
+ continue;
+ }
+ else
+ break;
+ case '#':
+ if (definedef == dnone)
{
- structdef = snone;
- (void) strcpy (structtag, "<error 2>");
- }
+ char *cp;
+ logical cpptoken = TRUE;
+
+ /* Look back on this line. If all blanks, or nonblanks
+ followed by an end of comment, this is a preprocessor
+ token. */
+ for (cp = newlb.buffer; cp < lp-1; cp++)
+ if (!iswhite (*cp))
+ {
+ if (*cp == '*' && *(cp+1) == '/')
+ {
+ cp++;
+ cpptoken = TRUE;
+ }
+ else
+ cpptoken = FALSE;
+ }
+ if (cpptoken)
+ definedef = dsharpseen;
+ } /* if (definedef == dnone) */
+
continue;
- }
- if (LEVEL_OK_FOR_FUNCDEF () && !inquote && !incomm && gotone == FALSE)
+ } /* switch (c) */
+
+
+ /* Consider token only if some complicated conditions are satisfied. */
+ if ((definedef != dnone
+ || (cblev == 0 && structdef != scolonseen)
+ || (cblev == 1 && cplpl && structdef == sinbody))
+ && typdef != tignore
+ && definedef != dignorerest
+ && funcdef != finlist)
{
if (midtoken)
{
if (endtoken (c))
{
- if (c_ext && c == ':' && *lp == ':' && intoken (*(lp + 1)))
+ if (cplpl && c == ':' && *lp == ':' && begtoken(*(lp + 1)))
{
/*
- * This handles :: in the middle, but not at beginning
- * of an identifier.
+ * This handles :: in the middle, but not at the
+ * beginning of an identifier.
*/
lp += 2;
toklen += 3;
}
else
{
- /* The following is no longer true,
- now that we advance to the next line
- at the end of processing the character. */
- /*
- * We've just finished lexing an identifier.
- * Note that if `c' is '\0', `lb' is the NEXT
- * line, `lp' points to the beginning of it, and
- * old pointers into `lb.buffer' may no longer be
- * valid, since `lb.buffer' may have been
- * reallocated. In this case (which corresponds
- * to an identifier followed immediately by a
- * newline), we re-read the line into lb1.
- *
- * This would be faster if the previous line's
- * buffer were always saved.
- */
- logical is_func;
- char *tok_linebuf;
- TOKEN tok;
- logical bingo, tok_at_end_of_line;
- char *lp_tmp; /* addressable */
-
-#if 0
- if (c == '\0')
- {
- getline (GET_COOKIE (prev_linepos));
- tok_linebuf = lb1.buffer;
- tok_at_end_of_line = TRUE;
- tok.linestart = prev_linepos;
- tok.lineno = lineno - 1;
- }
- else
-#endif
- {
- tok_linebuf = lb.buffer;
- tok_at_end_of_line = FALSE;
- tok.linestart = linepos;
- tok.lineno = lineno;
- }
- tok.p = tok_linebuf + tokoff;
+ logical is_func = FALSE;
+
+ tok.lineno = lineno;
+ tok.p = newlb.buffer + tokoff;
tok.len = toklen;
- tok.rewritten = FALSE;
- lp_tmp = lp;
- bingo = consider_token (c, &lp_tmp, &tok,
- &is_func, c_ext, level);
- lp = lp_tmp;
- if (bingo)
+ tok.named = FALSE;
+ if (yacc_rules
+ || consider_token (c, &tok, c_ext, cblev, &is_func))
{
- if (GET_CHARNO (tok.linestart) != GET_CHARNO (linepos)
- && !tok_at_end_of_line)
+ if (structdef == sinbody
+ && definedef == dnone
+ && is_func)
+ /* function defined in C++ class body */
{
- /*
- * Resynchronize tok.p to point into the right
- * linebuffer.
- */
- getline (GET_COOKIE (tok.linestart));
- if (!tok.rewritten)
- tok.p = lb1.buffer + (tok.p - tok_linebuf);
- tok_linebuf = lb1.buffer;
- }
- if (structdef == sinbody && definedef == dnone && is_func)
- { /* function defined in C++ class body */
- sprintf (tokb, "%s::%.*s",
- structtag[0] == '\0' ? "_anonymous_"
- : structtag,
+ sprintf (nameb, "%s::%.*s",
+ ((structtag[0] == '\0')
+ ? "_anonymous_" : structtag),
tok.len, tok.p);
- tok.rewritten = TRUE;
+ tok.named = TRUE;
}
else
{
- sprintf (tokb, "%.*s", tok.len, tok.p);
+ sprintf (nameb, "%.*s", tok.len, tok.p);
+ }
+
+ if (structdef == stagseen
+ || typdef == tend
+ || (is_func
+ && definedef == dignorerest)) /* macro */
+ tok.named = TRUE;
+
+ if (definedef == dnone
+ && (funcdef == ftagseen
+ || structdef == stagseen
+ || typdef == tend))
+ {
+ if (newndx == curndx)
+ curndx = 1 - curndx; /* switch line buffers */
}
- pfnote (tokb, is_func, tok.rewritten, tok_linebuf,
- tokoff + toklen + (tok_at_end_of_line ? 0 : 1),
- tok.lineno, GET_CHARNO (tok.linestart));
- gotone = is_func; /* function */
+ else
+ MAKE_TAG_FROM_NEW_LB (is_func);
}
midtoken = FALSE;
}
- }
+ } /* if (endtoken (c)) */
else if (intoken (c))
- toklen++;
- }
+ {
+ toklen++;
+ continue;
+ }
+ } /* if (midtoken) */
else if (begtoken (c))
{
- tokoff = lp - 1 - lb.buffer;
- toklen = 1;
- midtoken = TRUE;
+ switch (definedef)
+ {
+ case dnone:
+ switch (funcdef)
+ {
+ case fstartlist:
+ funcdef = finlist;
+ continue;
+ case flistseen:
+ MAKE_TAG_FROM_OTH_LB (TRUE);
+ funcdef = fignore;
+ break;
+ case ftagseen:
+ funcdef = fnone;
+ break;
+ }
+ if (structdef == stagseen)
+ structdef = snone;
+ break;
+ case dsharpseen:
+ /* Take a quick peek ahead for a define directive,
+ so we can avoid saving the token when not absolutely
+ necessary. [This is a speed hack.] */
+ if (c == 'd' && strneq (lp, "efine", 5)
+ && iswhite (*(lp + 5)))
+ {
+ SAVE_TOKEN;
+ definedef = ddefineseen;
+ lp += 6;
+ }
+ else
+ definedef = dignorerest;
+ continue;
+ }
+ if (!yacc_rules || lp == newlb.buffer + 1)
+ {
+ tokoff = lp - 1 - newlb.buffer;
+ toklen = 1;
+ midtoken = TRUE;
+ }
+ continue;
}
- }
- /* Detect end of line, after having handled the last token on the line. */
- if (c == 0)
- {
- CNL;
- gotone = FALSE;
- }
- if (c == ';' && tydef == end) /* clean with typedefs */
- tydef = none;
- }
-}
-
-/*
- * consider_token ()
- * checks to see if the current token is at the start of a
- * function, or corresponds to a typedef. It updates the input
- * line pointer *LPP so that the '(' will be in it when it returns.
- *
- * *IS_FUNC gets TRUE iff the token is a function.
- * C_EXT is which language we are looking at.
- *
- * In the future we will need some way to adjust where the end of
- * the token is; for instance, implementing the C++ keyword
- * `operator' properly will adjust the end of the token to be after
- * whatever follows `operator'.
- *
- * Globals
- * structdef IN OUT
- * definedef IN OUT
- * tydef IN OUT
- */
-
-logical
-consider_token (c, lpp, tokp, is_func, c_ext, level)
- reg char c; /* IN: first char after the token */
- char **lpp; /* IN OUT: *lpp points to 2nd char after the token */
- reg TOKEN *tokp; /* IN */
- logical *is_func; /* OUT */
- int c_ext; /* IN */
- int level; /* IN */
-{
- reg char *lp = *lpp;
- /*
- * next_token_is_func
- * set this to TRUE, and the next token considered is called a function.
- */
- static logical next_token_is_func;
- logical firsttok; /* TRUE if have seen first token in ()'s */
- Stab_entry *tokse = stab_find (get_C_stab (c_ext), tokp->p, tokp->len);
- enum sym_type toktype = stab_type (tokse);
-
- *is_func = TRUE; /* a function */
+ } /* if must look at token */
- /*
- * Advance the definedef state machine. We set `gotone' for good measure;
- * it's redundant.
- */
- switch (definedef)
- {
- case dnone:
- /* We're not on a preprocessor line. */
- break;
- case dsharpseen:
- if (toktype == st_C_define)
- {
- definedef = ddefineseen;
- gotone = FALSE;
- }
- else
- {
- definedef = dignorerest;
- gotone = TRUE;
- }
- goto badone;
- case ddefineseen:
- /*
- * Make a tag for any macro.
- * This will flub up if there is a newline immediately following
- * the macro name.
- */
- *is_func = (c == '(');
- definedef = dignorerest;
- gotone = TRUE;
- if (!*is_func && !constantypedefs)
- goto badone;
- goto goodone;
- case dignorerest:
- goto badone;
- default:
- error ("internal error: definedef value");
- }
- /*
- * Skip whitespace and comments after the token. This loop should
- * also skip C++ comments.
- */
- while (1)
- {
- /* At whitespace => skip it. */
- if (iswhite (c))
- {
- c = *lp++;
- }
- /* At a comment => skip to end of comment. */
- else if (c == '/' && *lp == '*')
+ /* Detect end of line, colon, comma, semicolon and various braces
+ after having handled a token.*/
+ switch (c)
{
- /* If we find a comment, skip it. */
- while (!(c == '*' && *lp == '/'))
+ case ':':
+ if (definedef != dnone)
+ break;
+ if (structdef == stagseen)
+ structdef = scolonseen;
+ else
+ switch (funcdef)
+ {
+ case ftagseen:
+ if (yacc_rules)
+ {
+ MAKE_TAG_FROM_OTH_LB (FALSE);
+ funcdef = fignore;
+ }
+ break;
+ case fstartlist:
+ funcdef = fnone;
+ break;
+ }
+ break;
+ case ';':
+ if (definedef != dnone)
+ break;
+ if (cblev == 0)
+ switch (typdef)
+ {
+ case tend:
+ MAKE_TAG_FROM_OTH_LB (FALSE);
+ /* FALLTHRU */
+ default:
+ typdef = tnone;
+ }
+ if (funcdef != fignore)
+ funcdef = fnone;
+ if (structdef == stagseen)
+ structdef = snone;
+ break;
+ case ',':
+ if (definedef != dnone)
+ break;
+ if (funcdef != finlist && funcdef != fignore)
+ funcdef = fnone;
+ if (structdef == stagseen)
+ structdef = snone;
+ break;
+ case '[':
+ if (definedef != dnone)
+ break;
+ if (cblev == 0 && typdef == tend)
+ {
+ typdef = tignore;
+ MAKE_TAG_FROM_OTH_LB (FALSE);
+ break;
+ }
+ if (funcdef != finlist && funcdef != fignore)
+ funcdef = fnone;
+ if (structdef == stagseen)
+ structdef = snone;
+ break;
+ case '(':
+ if (definedef != dnone)
+ break;
+ switch (funcdef)
+ {
+ case fnone:
+ switch (typdef)
+ {
+ case ttypedseen:
+ case tend:
+ /* Make sure that the next char is not a '*'.
+ This handles constructs like:
+ typedef void OperatorFun (int fun); */
+ if (*lp != '*')
+ {
+ typdef = tignore;
+ MAKE_TAG_FROM_OTH_LB (FALSE);
+ }
+ break;
+ } /* switch (typdef) */
+ break;
+ case ftagseen:
+ funcdef = fstartlist;
+ break;
+ case flistseen:
+ funcdef = finlist;
+ break;
+ }
+ parlev++;
+ break;
+ case ')':
+ if (definedef != dnone)
+ break;
+ if (--parlev == 0)
{
- c = *lp++;
- if (c == 0)
+ switch (funcdef)
{
- lp--;
+ case fstartlist:
+ case finlist:
+ funcdef = flistseen;
break;
}
+ if (cblev == 0 && typdef == tend)
+ {
+ typdef = tignore;
+ MAKE_TAG_FROM_OTH_LB (FALSE);
+ }
+ }
+ else if (parlev < 0) /* can happen due to ill-conceived #if's. */
+ parlev = 0;
+ break;
+ case '{':
+ if (definedef != dnone)
+ break;
+ if (typdef == ttypedseen)
+ typdef = tinbody;
+ switch (structdef)
+ {
+ case skeyseen: /* unnamed struct */
+ structtag[0] = '\0';
+ structdef = sinbody;
+ break;
+ case stagseen:
+ case scolonseen: /* named struct */
+ structdef = sinbody;
+ MAKE_TAG_FROM_OTH_LB (FALSE);
+ break;
}
- if (c == '*' && *lp == '/')
+ switch (funcdef)
{
- lp++; /* lp now points past the '/' */
- c = *lp++; /* c is now the --whatever-- after the '/' */
+ case flistseen:
+ MAKE_TAG_FROM_OTH_LB (TRUE);
+ /* FALLTHRU */
+ case fignore:
+ funcdef = fnone;
+ break;
+ case fnone:
+ /* Neutralize `extern "C" {' grot and look inside structs. */
+ if (cblev == 0 && structdef == snone && typdef == tnone)
+ cblev = -1;
}
- }
- else
- break;
-
- /* If we arrived at eof or eol, decide which one it is.
- If it's eol, advance to the next line. */
-
- if (c == 0)
- {
- lp--;
+ cblev++;
break;
- }
- }
-
- /*
- * If you have custom token types, or when configuration files can
- * define custom token types, this switch will be larger.
- */
- switch (toktype)
- {
- case st_C_typedef:
- if (typedefs)
- {
- tydef = begin;
- goto badone;
- }
- break;
- case st_C_typespec:
- if (tydef == begin || tydef == end)
- {
- tydef = end;
- goto badone;
- }
- break;
- }
-
- /*
- * This structdef business is currently only invoked when level==0.
- * It should be recursively invoked whatever the level, and a stack of
- * states kept, to allow for definitions of structs within structs.
- *
- * This structdef business is NOT invoked when we are ctags and the
- * file is plain C. This is because a struct tag may have the same
- * name as another tag, and this loses with ctags.
- *
- * This if statement deals with the tydef state machine as follows: if
- * tydef==begin and token is struct/union/class/enum, goto badone.
- * All the other code here is for the structdef state machine.
- */
- switch (toktype)
- {
- case st_C_struct:
- case st_C_enum:
- if (tydef == begin || (typedefs_and_cplusplus && level == 0 && structdef == snone))
- {
- structdef = skeyseen;
- structkey = tokse;
- }
- goto badone;
- }
-
- if (structdef == skeyseen)
- {
- /* If next char is '{' or (for C++) ':', found a structure tag. */
- if (c == '{' || (c_ext && c == ':'))
- {
- /*
- * We should do this slightly differently for straight C:
- * instead of defining `tag', as we now do, we should define
- * `struct tag'. (Do this only if the find-tag defaulting is
- * done on a sophisticated per-mode basis, so that if the user
- * says meta-. anywhere in `struct foo', the default comes out
- * `struct foo', not `struct' or `foo'.) This will require
- * remembering which keyword (struct/union/class/enum) we saw, as a
- * Stab_entry* -- this will also make it possible to merge the
- * skeyseen and senumseen states, if we want.
- */
- if (stab_type (structkey) == st_C_struct)
+ case '*':
+ if (definedef != dnone)
+ break;
+ if (funcdef == fstartlist)
+ funcdef = fnone; /* avoid tagging `foo' in `foo (*bar()) ()' */
+ break;
+ case '}':
+ if (definedef != dnone)
+ break;
+ if (!noindentypedefs && lp == newlb.buffer + 1)
{
- (void) strncpy (structtag, tokp->p, tokp->len);
- structtag[tokp->len] = '\0'; /* for struct/union/class */
- structdef = stagseen;
+ cblev = 0; /* reset curly brace level if first column */
+ parlev = 0; /* also reset paren level, just in case... */
}
- else
+ else if (cblev > 0)
+ cblev--;
+ if (cblev == 0)
{
- structtag[0] = '\0'; /* for enum */
+ if (typdef == tinbody)
+ typdef = tend;
+ structdef = snone;
+ strcpy (structtag, "<error 2>");
}
- *is_func = FALSE; /* not a function */
- goto goodone;
- }
- else
- {
- /* Not a definition: reset structdef */
- structdef = snone;
- (void) strcpy (structtag, "<error 3>");
- }
- /* Now what? And how does/should this stuff interact with tydef?? */
- /* Also maybe reset lp to *lpp for benefit of the function finding code. */
- }
- if (tydef == begin)
- {
- tydef = end;
- goto badone;
- }
- if (tydef == end)
- {
- *is_func = 0;
- goto goodone;
- }
- /* Detect GNUmacs's function-defining macros. */
- if (definedef == dnone && strneq (tokp->p, "DEF", 3))
- {
- next_token_is_func = TRUE;
- goto badone;
- }
- if (next_token_is_func)
- {
- next_token_is_func = FALSE;
- goto goodone;
- }
- if (c != '(')
- goto badone;
- firsttok = FALSE;
- while ((c = *lp++) != ')')
- {
- if (c == 0)
- {
- lp--;
break;
- }
- /*
- * This line used to confuse ctags:
- * int (*oldhup)();
- * This fixes it. A nonwhite char before the first
- * token, other than a / (in case of a comment in there)
- * makes this not a declaration.
- */
- if (begtoken (c) || c == '/')
- firsttok++;
- else if (!iswhite (c) && !firsttok)
- goto badone;
- }
- while (iswhite (c = *lp++))
- {
- if (c == 0)
- {
- lp--;
+ case '=':
+ case '#': case '+': case '-': case '~': case '&': case '%': case '/':
+ case '|': case '^': case '!': case '<': case '>': case '.': case '?':
+ if (definedef != dnone)
+ break;
+ /* These surely cannot follow a function tag. */
+ if (funcdef != finlist && funcdef != fignore)
+ funcdef = fnone;
break;
- }
- }
- if (!isgood (c))
- goto badone;
+ case '\0':
+ /* If a macro spans multiple lines don't reset its state. */
+ if (quotednl)
+ CNL_SAVE_DEFINEDEF;
+ else
+ CNL;
+ break;
+ } /* switch (c) */
-goodone:
- *lpp = lp - 1;
- return TRUE;
+ } /* while not eof */
+}
-badone:
- *lpp = lp - 1;
- return FALSE;
+/*
+ * Process either a C++ file or a C file depending on the setting
+ * of a global flag.
+ */
+void
+default_C_entries (inf)
+ FILE *inf;
+{
+ C_entries (cplusplus ? C_PLPL : 0, inf);
+}
+
+/* Always do C++. */
+void
+Cplusplus_entries (inf)
+ FILE *inf;
+{
+ C_entries (C_PLPL, inf);
}
+/* Always do C*. */
void
-getline (atcookie)
- long atcookie;
+Cstar_entries (inf)
+ FILE *inf;
{
- long saveftell = ftell (inf);
+ C_entries (C_STAR, inf);
+}
- (void) fseek (inf, atcookie, 0);
- (void) readline (&lb1, inf);
- (void) fseek (inf, saveftell, 0);
+/* Always do Yacc. */
+void
+Yacc_entries (inf)
+ FILE *inf;
+{
+ C_entries (YACC, inf);
}
\f
/* Fortran parsing */
char *dbp;
int pfcnt;
-int
-PF_funcs (fi)
- FILE *fi;
+logical
+tail (cp)
+ char *cp;
+{
+ register int len = 0;
+
+ while (*cp && (*cp | ' ') == (dbp[len] | ' '))
+ cp++, len++;
+ if (*cp == 0)
+ {
+ dbp += len;
+ return TRUE;
+ }
+ return FALSE;
+}
+
+void
+takeprec ()
+{
+ while (isspace (*dbp))
+ dbp++;
+ if (*dbp != '*')
+ return;
+ dbp++;
+ while (isspace (*dbp))
+ dbp++;
+ if (tail ("(*)"))
+ return;
+ if (!isdigit (*dbp))
+ {
+ --dbp; /* force failure */
+ return;
+ }
+ do
+ dbp++;
+ while (isdigit (*dbp));
+}
+
+void
+getit (inf)
+ FILE *inf;
+{
+ register char *cp;
+ char c;
+ char nambuf[BUFSIZ];
+
+ while (isspace (*dbp))
+ dbp++;
+ if (*dbp == '\0')
+ {
+ lineno++;
+ linecharno = charno;
+ charno += readline (&lb, inf);
+ dbp = lb.buffer;
+ if (dbp[5] != '&')
+ return;
+ dbp += 6;
+ while (isspace (*dbp))
+ dbp++;
+ }
+ if (!isalpha (*dbp)
+ && *dbp != '_'
+ && *dbp != '$')
+ return;
+ for (cp = dbp + 1;
+ (*cp
+ && (isalpha (*cp) || isdigit (*cp) || (*cp == '_') || (*cp == '$')));
+ cp++)
+ continue;
+ c = *cp;
+ *cp = '\0';
+ strcpy (nambuf, dbp);
+ *cp = c;
+ pfnote (nambuf, TRUE, FALSE, lb.buffer,
+ cp - lb.buffer + 1, lineno, linecharno);
+ pfcnt++;
+}
+
+void
+Fortran_functions (inf)
+ FILE *inf;
{
lineno = 0;
charno = 0;
pfcnt = 0;
- while (!feof (fi))
+ while (!feof (inf))
{
lineno++;
linecharno = charno;
- charno += readline (&lb, fi);
+ charno += readline (&lb, inf);
dbp = lb.buffer;
if (*dbp == '%')
dbp++; /* Ratfor escape to fortran */
{
case 'f':
if (tail ("function"))
- getit ();
+ getit (inf);
continue;
case 's':
if (tail ("subroutine"))
- getit ();
+ getit (inf);
+ continue;
+ case 'e':
+ if (tail ("entry"))
+ getit (inf);
continue;
case 'p':
if (tail ("program"))
{
- getit ();
+ getit (inf);
continue;
}
if (tail ("procedure"))
- getit ();
+ getit (inf);
continue;
}
}
- return (pfcnt);
-}
-
-logical
-tail (cp)
- char *cp;
-{
- register int len = 0;
-
- while (*cp && (*cp & ~' ') == ((*(dbp + len)) & ~' '))
- cp++, len++;
- if (*cp == 0)
- {
- dbp += len;
- return (1);
- }
- return (0);
-}
-
-void
-takeprec ()
-{
- while (isspace (*dbp))
- dbp++;
- if (*dbp != '*')
- return;
- dbp++;
- while (isspace (*dbp))
- dbp++;
- if (!isdigit (*dbp))
- {
- --dbp; /* force failure */
- return;
- }
- do
- dbp++;
- while (isdigit (*dbp));
}
-
+\f
+/*
+ * Bob Weiner, Motorola Inc., 4/3/94
+ * Unix and microcontroller assembly tag handling
+ * look for '^[a-zA-Z_.$][a-zA_Z0-9_.$]*[: ^I^J]'
+ */
void
-getit ()
+Asm_labels (inf)
+ FILE *inf;
{
+ char nambuf[BUFSIZ];
register char *cp;
char c;
- char nambuf[BUFSIZ];
-
- while (isspace (*dbp))
- dbp++;
- if (*dbp == 0 || (!isalpha (*dbp)) && (*dbp != '_') && (*dbp != '$'))
- return;
- for (cp = dbp + 1; *cp && (isalpha (*cp) || isdigit (*cp)
- || (*cp == '_') || (*cp == '$')); cp++)
- continue;
- c = cp[0];
- cp[0] = 0;
- (void) strcpy (nambuf, dbp);
- cp[0] = c;
- pfnote (nambuf, TRUE, FALSE, lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
- pfcnt++;
-}
-
-/* Handle a file of assembler code. */
-
-void
-Asm_funcs (fi)
- FILE *fi;
-{
- int i;
- register char c;
lineno = 0;
charno = 0;
pfcnt = 0;
- while (!feof (fi))
+ while (!feof (inf))
{
lineno++;
linecharno = charno;
- charno += readline (&lb, fi);
- dbp = lb.buffer;
-
- for (i = 0; ((c = dbp[i]) && !isspace (c)) && (c != ':'); i++)
- ;
-
- if ((i > 0) && (c == ':'))
- getit ();
+ charno += readline (&lb, inf);
+ cp = lb.buffer;
+
+ /* If first char is alphabetic or one of [_.$], test for colon
+ following identifier. */
+ if (isalpha (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
+ {
+ /* Read past label. */
+ cp++;
+ while (isalnum (*cp) || *cp == '_' || *cp == '.' || *cp == '$')
+ cp++;
+ if (*cp == ':' || isspace (*cp))
+ {
+ /* Found end of label, so copy it and add it to the table. */
+ c = *cp;
+ *cp = '\0';
+ strcpy (nambuf, lb.buffer);
+ *cp = c;
+ pfnote (nambuf, TRUE, FALSE, lb.buffer,
+ cp - lb.buffer + 1, lineno, linecharno);
+ pfcnt++;
+ }
+ }
}
}
\f
dbp = lb.buffer; \
}
-/* Locates tags for procedures & functions.
- * Doesn't do any type- or var-definitions.
- * It does look for the keyword "extern" or "forward"
- * immediately following the procedure statement;
- * if found, the tag is skipped.
+/*
+ * Locates tags for procedures & functions. Doesn't do any type- or
+ * var-definitions. It does look for the keyword "extern" or
+ * "forward" immediately following the procedure statement; if found,
+ * the tag is skipped.
*/
-
void
-PAS_funcs (fi)
- FILE *fi;
+Pascal_functions (inf)
+ FILE *inf;
{
struct linebuffer tline; /* mostly copied from C_entries */
long save_lcno;
char nambuf[BUFSIZ];
logical /* each of these flags is TRUE iff: */
- incomm1, /* point is inside {..} comment */
- incomm2, /* point is inside (*..*) comment */
+ incomment, /* point is inside a comment */
inquote, /* point is inside '..' string */
get_tagname, /* point is after PROCEDURE/FUNCTION */
/* keyword, so next item = potential tag */
*dbp = 0;
initbuffer (&tline);
- incomm1 = incomm2 = inquote = FALSE;
+ incomment = inquote = FALSE;
found_tag = FALSE; /* have a proc name; check if extern */
get_tagname = FALSE; /* have found "procedure" keyword */
inparms = FALSE; /* found '(' after "proc" */
verify_tag = FALSE; /* check if "extern" is ahead */
/* long main loop to get next char */
- while (!feof (fi))
+ while (!feof (inf))
{
c = *dbp++;
if (c == 0) /* if end of line */
/* to the beginning of the name of */
/* the procedure or function */
}
- if (incomm1) /* within { - } comments */
- {
- if (c == '}')
- incomm1 = FALSE;
- continue;
- }
- else if (incomm2) /* within (* - *) comments */
+ if (incomment)
{
- if (c == '*')
+ if (c == '}') /* within { - } comments */
+ incomment = FALSE;
+ else if (c == '*' && dbp[1] == ')') /* within (* - *) comments */
{
- while ((c = *dbp++) == '*')
- continue;
- if (c == 0)
- GET_NEW_LINE;
- if (c == ')')
- incomm2 = FALSE;
+ dbp++;
+ incomment = FALSE;
}
continue;
}
inquote = FALSE;
continue;
}
- else
+ else
switch (c)
{
case '\'':
inquote = TRUE; /* found first quote */
continue;
case '{': /* found open-{-comment */
- incomm1 = TRUE;
+ incomment = TRUE;
continue;
case '(':
if (*dbp == '*') /* found open-(*-comment */
{
- incomm2 = TRUE;
+ incomment = TRUE;
dbp++;
}
else if (found_tag) /* found '(' after tag, i.e., parm-list */
verify_tag = FALSE;
}
}
- if ((found_tag) && (verify_tag)) /* not external proc, so make tag */
+ if ((found_tag) && (verify_tag)) /* not external proc, so make tag */
{
found_tag = FALSE;
verify_tag = FALSE;
/* and proceed to check for "extern" */
}
- if ((!incomm1) && (!incomm2) && (!inquote) &&
- (!found_tag) && (!get_tagname))
+ if (!incomment && !inquote && !found_tag && !get_tagname)
{
/* check for proc/fn keywords */
switch (c | ' ')
continue;
}
}
- } /* while not e-o-f */
+ } /* while not eof */
+}
+\f
+/*
+ * lisp tag functions
+ * just look for (def or (DEF
+ */
+
+int
+L_isdef (tokp)
+ register char *tokp;
+{
+ return ((tokp[1] == 'd' || tokp[1] == 'D')
+ && (tokp[2] == 'e' || tokp[2] == 'E')
+ && (tokp[3] == 'f' || tokp[3] == 'F'));
+}
+
+int
+L_isquote (tokp)
+ register char *tokp;
+{
+ return ((*(++tokp) == 'q' || *tokp == 'Q')
+ && (*(++tokp) == 'u' || *tokp == 'U')
+ && (*(++tokp) == 'o' || *tokp == 'O')
+ && (*(++tokp) == 't' || *tokp == 'T')
+ && (*(++tokp) == 'e' || *tokp == 'E')
+ && isspace(*(++tokp)));
+}
+
+void
+L_getit ()
+{
+ register char *cp;
+ char c;
+ char nambuf[BUFSIZ];
+
+ if (*dbp == '\'') /* Skip prefix quote */
+ dbp++;
+ else if (*dbp == '(' && L_isquote (dbp)) /* Skip "(quote " */
+ {
+ dbp += 7;
+ while (isspace(*dbp))
+ dbp++;
+ }
+ for (cp = dbp /*+1*/; *cp && *cp != '(' && *cp != ' ' && *cp != ')'; cp++)
+ continue;
+ if (cp == dbp)
+ return;
+
+ c = cp[0];
+ cp[0] = 0;
+ strcpy (nambuf, dbp);
+ cp[0] = c;
+ pfnote (nambuf, TRUE, FALSE, lb.buffer,
+ cp - lb.buffer + 1, lineno, linecharno);
+ pfcnt++;
}
-\f
-/*
- * lisp tag functions
- * just look for (def or (DEF
- */
void
-L_funcs (fi)
- FILE *fi;
+Lisp_functions (inf)
+ FILE *inf;
{
lineno = 0;
charno = 0;
pfcnt = 0;
- while (!feof (fi))
+ while (!feof (inf))
{
lineno++;
linecharno = charno;
- charno += readline (&lb, fi);
+ charno += readline (&lb, inf);
dbp = lb.buffer;
if (dbp[0] == '(')
{
else
{
/* Check for (foo::defmumble name-defined ... */
- while (*dbp && *dbp != ':' && !isspace (*dbp)
- && *dbp != '(' && *dbp != ')')
+ do
dbp++;
+ while (*dbp && !isspace (*dbp)
+ && *dbp != ':' && *dbp != '(' && *dbp != ')');
if (*dbp == ':')
{
- while (*dbp == ':')
+ do
dbp++;
+ while (*dbp == ':');
- if (L_isdef (dbp))
+ if (L_isdef (dbp - 1))
{
while (!isspace (*dbp))
dbp++;
}
}
}
-
-int
-L_isdef (dbp)
- char *dbp;
-{
- return ((dbp[1] == 'D' || dbp[1] == 'd') &&
- (dbp[2] == 'E' || dbp[2] == 'e') &&
- (dbp[3] == 'F' || dbp[3] == 'f'));
-}
-
-void
-L_getit ()
-{
- register char *cp;
- char c;
- char nambuf[BUFSIZ];
-
- if (*dbp == 0)
- return;
- for (cp = dbp + 1; *cp && *cp != '(' && *cp != ' '; cp++)
- continue;
- c = cp[0];
- cp[0] = 0;
- (void) strcpy (nambuf, dbp);
- cp[0] = c;
- pfnote (nambuf, TRUE, FALSE, lb.buffer, cp - lb.buffer + 1, lineno, linecharno);
- pfcnt++;
-}
\f
/*
* Scheme tag functions
* look for (set! xyzzy
*/
-static void get_scheme ();
+void get_scheme ();
void
-Scheme_funcs (fi)
- FILE *fi;
+Scheme_functions (inf)
+ FILE *inf;
{
lineno = 0;
charno = 0;
pfcnt = 0;
- while (!feof (fi))
+ while (!feof (inf))
{
lineno++;
linecharno = charno;
- charno += readline (&lb, fi);
+ charno += readline (&lb, inf);
dbp = lb.buffer;
if (dbp[0] == '(' &&
(dbp[1] == 'D' || dbp[1] == 'd') &&
}
}
-static void
+void
get_scheme ()
{
register char *cp;
/* TEX_toktab is a table of TeX control sequences that define tags.
Each TEX_tabent records one such control sequence.
CONVERT THIS TO USE THE Stab TYPE!! */
-
struct TEX_tabent
{
char *name;
/* Default set of control sequences to put into TEX_toktab.
The value of environment var TEXTAGS is prepended to this. */
-static char *TEX_defenv =
-":chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem:typeout";
+char *TEX_defenv = "\
+:chapter:section:subsection:subsubsection:eqno:label:ref:cite:bibitem:typeout";
void TEX_mode ();
struct TEX_tabent *TEX_decode_env ();
void TEX_getit ();
int TEX_Token ();
-static char TEX_esc = '\\';
-static char TEX_opgrp = '{';
-static char TEX_clgrp = '}';
+char TEX_esc = '\\';
+char TEX_opgrp = '{';
+char TEX_clgrp = '}';
/*
* TeX/LaTeX scanning loop.
*/
-
void
-TEX_funcs (fi)
- FILE *fi;
+TeX_functions (inf)
+ FILE *inf;
{
char *lasthit;
pfcnt = 0;
/* Select either \ or ! as escape character. */
- TEX_mode (fi);
+ TEX_mode (inf);
/* Initialize token table once from environment. */
if (!TEX_toktab)
TEX_toktab = TEX_decode_env ("TEXTAGS", TEX_defenv);
- while (!feof (fi))
- {
+ while (!feof (inf))
+ { /* Scan each line in file */
lineno++;
linecharno = charno;
- charno += readline (&lb, fi);
+ charno += readline (&lb, inf);
dbp = lb.buffer;
lasthit = dbp;
+ while (dbp = etags_strchr (dbp, TEX_esc)) /* Look at each esc in line */
+ {
+ register int i;
- while (!feof (fi))
- { /* Scan each line in file */
- lineno++;
- linecharno = charno;
- charno += readline (&lb, fi);
- dbp = lb.buffer;
+ if (!*(++dbp))
+ break;
+ linecharno += dbp - lasthit;
lasthit = dbp;
- while (dbp = index (dbp, TEX_esc)) /* Look at each escape in line */
+ i = TEX_Token (lasthit);
+ if (0 <= i)
{
- register int i;
-
- if (!*(++dbp))
- break;
- linecharno += dbp - lasthit;
- lasthit = dbp;
- i = TEX_Token (lasthit);
- if (0 <= i)
- {
- TEX_getit (lasthit, TEX_toktab[i].len);
- break; /* We only save a line once */
- }
+ TEX_getit (lasthit, TEX_toktab[i].len);
+ break; /* We only save a line once */
}
}
}
#define TEX_SESC '!'
#define TEX_cmt '%'
-/* Figure out whether TeX's escapechar is '\\' or '!' and set grouping */
-/* chars accordingly. */
-
+/* Figure out whether TeX's escapechar is '\\' or '!' and set grouping
+ chars accordingly. */
void
-TEX_mode (f)
- FILE *f;
+TEX_mode (inf)
+ FILE *inf;
{
int c;
- while ((c = getc (f)) != EOF)
+ while ((c = getc (inf)) != EOF)
{
/* Skip to next line if we hit the TeX comment char. */
if (c == TEX_cmt)
while (c != '\n')
- c = getc (f);
+ c = getc (inf);
else if (c == TEX_LESC || c == TEX_SESC )
break;
}
TEX_opgrp = '<';
TEX_clgrp = '>';
}
- rewind (f);
+ rewind (inf);
}
-/* Read environment and prepend it to the default string. */
-/* Build token table. */
-
+/* Read environment and prepend it to the default string.
+ Build token table. */
struct TEX_tabent *
TEX_decode_env (evarname, defenv)
char *evarname;
char *defenv;
{
register char *env, *p;
- extern char *savenstr (), *index ();
struct TEX_tabent *tab;
int size, i;
/* Allocate a token table */
for (size = 1, p = env; p;)
- if ((p = index (p, ':')) && *(++p))
+ if ((p = etags_strchr (p, ':')) && *(++p))
size++;
- tab = xnew (size, struct TEX_tabent);
+ /* Add 1 to leave room for null terminator. */
+ tab = xnew (size + 1, struct TEX_tabent);
/* Unpack environment string into token table. Be careful about */
/* zero-length strings (leading ':', "::" and trailing ':') */
for (i = 0; *env;)
{
- p = index (env, ':');
+ p = etags_strchr (env, ':');
if (!p) /* End of environment string. */
p = env + strlen (env);
if (p - env > 0)
/* Record a tag defined by a TeX command of length LEN and starting at NAME.
The name being defined actually starts at (NAME + LEN + 1).
But we seem to include the TeX command in the tag name. */
-
void
TEX_getit (name, len)
char *name;
/* Let tag name extend to next group close (or end of line) */
while (*p && *p != TEX_clgrp)
p++;
- (void) strncpy (nambuf, name, p - name);
+ strncpy (nambuf, name, p - name);
nambuf[p - name] = 0;
pfnote (nambuf, TRUE, FALSE, lb.buffer, strlen (lb.buffer), lineno, linecharno);
}
/* If the text at CP matches one of the tag-defining TeX command names,
- return the index of that command in TEX_toktab.
- Otherwise return -1. */
-
-/* Keep the capital `T' in `Token' for dumb truncating compilers
+ return the pointer to the first occurrence of that command in TEX_toktab.
+ Otherwise return -1.
+ Keep the capital `T' in `Token' for dumb truncating compilers
(this distinguishes it from `TEX_toktab' */
int
TEX_Token (cp)
int i;
for (i = 0; TEX_toktab[i].len > 0; i++)
- if (strncmp (TEX_toktab[i].name, cp, TEX_toktab[i].len) == 0)
+ if (strneq (TEX_toktab[i].name, cp, TEX_toktab[i].len))
return i;
return -1;
}
\f
/* Support for Prolog. */
-/* whole head (not only functor, but also arguments)
+/* Whole head (not only functor, but also arguments)
is gotten in compound term. */
-
void
-prolog_getit (s, lineno, linecharno)
+prolog_getit (s)
char *s;
- int lineno;
- long linecharno;
{
char nambuf[BUFSIZ], *save_s, tmpc;
int insquote, npar;
*s = '\0';
strcpy (nambuf, save_s);
*s = tmpc;
- pfnote (nambuf, TRUE, save_s, strlen (nambuf), lineno, linecharno);
+ pfnote (nambuf, TRUE, FALSE, save_s, strlen (nambuf), lineno, linecharno);
}
/* It is assumed that prolog predicate starts from column 0. */
-
void
-prolog_funcs (fi)
- FILE *fi;
+Prolog_functions (inf)
+ FILE *inf;
{
void skip_comment (), prolog_getit ();
lineno = linecharno = charno = 0;
- while (!feof (fi))
+ while (!feof (inf))
{
lineno++;
linecharno += charno;
- charno = readline (&lb, fi) + 1; /* 1 for newline. */
+ charno = readline (&lb, inf) + 1; /* 1 for newline. */
dbp = lb.buffer;
if (isspace (dbp[0])) /* not predicate header. */
continue;
else if (dbp[0] == '%') /* comment. */
continue;
else if (dbp[0] == '/' && dbp[1] == '*') /* comment. */
- skip_comment (&lb, fi, &lineno, &linecharno);
+ skip_comment (&lb, inf, &lineno, &linecharno);
else /* found. */
- prolog_getit (dbp, lineno, linecharno);
+ prolog_getit (dbp);
}
}
void
-skip_comment (plb, fi, plineno, plinecharno)
+skip_comment (plb, inf, plineno, plinecharno)
struct linebuffer *plb;
- FILE *fi;
+ FILE *inf;
int *plineno; /* result */
long *plinecharno; /* result */
{
- while (!substr ("*/", plb->buffer))
+ char *cp;
+
+ do
{
+ for (cp = plb->buffer; *cp != '\0'; cp++)
+ if (cp[0] == '*' && cp[1] == '/')
+ return;
(*plineno)++;
- *plinecharno += readline (plb, fi) + 1;
- } /* 1 for newline. */
+ *plinecharno += readline (plb, inf) + 1; /* 1 for newline. */
+ }
+ while (!feof(inf));
}
+\f
+#ifdef ETAGS_REGEXPS
+/* Take a string like "/blah/" and turn it into "blah", making sure
+ that the first and last characters are the same, and handling
+ quoted separator characters. Actually, stops on the occurence of
+ an unquoted separator. Also turns "\t" into a Tab character.
+ Returns pointer to terminating separator. Works in place. Null
+ terminates name string. */
+char *
+scan_separators (name)
+ char *name;
+{
+ char sep = name[0];
+ char *copyto = name;
+ logical quoted = FALSE;
-/* Return TRUE if 'sub' exists somewhere in 's'. */
+ for (++name; *name != '\0'; ++name)
+ {
+ if (quoted)
+ {
+ if (*name == 't')
+ *copyto++ = '\t';
+ else if (*name == sep)
+ *copyto++ = sep;
+ else
+ {
+ /* Something else is quoted, so preserve the quote. */
+ *copyto++ = '\\';
+ *copyto++ = *name;
+ }
+ quoted = FALSE;
+ }
+ else if (*name == '\\')
+ quoted = TRUE;
+ else if (*name == sep)
+ break;
+ else
+ *copyto++ = *name;
+ }
-int
-substr (sub, s)
- char *sub;
- char *s;
-{
- while (*s && (s = index (s, *sub)))
- if (prestr (sub, s))
- return (TRUE);
- else
- s++;
- return (FALSE);
+ /* Terminate copied string. */
+ *copyto = '\0';
+ return name;
}
-/* Return TRUE if 'pre' is prefix of string 's'. */
-
-int
-prestr (pre, s)
- char *pre;
- char *s;
+/* Turn a name, which is an ed-style (but Emacs syntax) regular
+ expression, into a real regular expression by compiling it. */
+void
+add_regex (regexp_pattern)
+ char *regexp_pattern;
{
- if (*pre == '\0')
- return (TRUE);
- else if (*pre == *s)
- return (prestr (pre + 1, s + 1));
+ char *name;
+ const char *err;
+ struct re_pattern_buffer *patbuf;
+
+ if (regexp_pattern == NULL)
+ {
+ /* Remove existing regexps. */
+ num_patterns = 0;
+ patterns = NULL;
+ return;
+ }
+
+ if (regexp_pattern[0] == '\0')
+ {
+ error ("missing regexp", 0);
+ return;
+ }
+ if (regexp_pattern[strlen(regexp_pattern)-1] != regexp_pattern[0])
+ {
+ error ("%s: unterminated regexp", regexp_pattern);
+ return;
+ }
+ name = scan_separators (regexp_pattern);
+ if (regexp_pattern[0] == '\0')
+ {
+ error ("null regexp", 0);
+ return;
+ }
+ (void) scan_separators (name);
+
+ patbuf = xnew (1, struct re_pattern_buffer);
+ patbuf->translate = NULL;
+ patbuf->fastmap = NULL;
+ patbuf->buffer = NULL;
+ patbuf->allocated = 0;
+
+ err = re_compile_pattern (regexp_pattern, strlen (regexp_pattern), patbuf);
+ if (err != NULL)
+ {
+ error ("%s while compiling pattern", err);
+ return;
+ }
+
+ num_patterns += 1;
+ if (num_patterns == 1)
+ patterns = xnew (1, struct pattern);
else
- return (FALSE);
+ patterns = ((struct pattern *)
+ xrealloc (patterns,
+ (num_patterns * sizeof (struct pattern))));
+ patterns[num_patterns - 1].pattern = patbuf;
+ patterns[num_patterns - 1].name_pattern = savestr (name);
+ patterns[num_patterns - 1].error_signaled = FALSE;
+}
+
+/*
+ * Do the subtitutions indicated by the regular expression and
+ * arguments.
+ */
+char *
+substitute (in, out, regs)
+ char *in, *out;
+ struct re_registers *regs;
+{
+ char *result = NULL, *t;
+ int size = 0;
+
+ /* Pass 1: figure out how much size to allocate. */
+ for (t = out; *t; ++t)
+ {
+ if (*t == '\\')
+ {
+ ++t;
+ if (!*t)
+ {
+ fprintf (stderr, "%s: pattern subtitution ends prematurely\n",
+ progname);
+ return NULL;
+ }
+ if (isdigit (*t))
+ {
+ int dig = *t - '0';
+ size += regs->end[dig] - regs->start[dig];
+ }
+ }
+ }
+
+ /* Allocate space and do the substitutions. */
+ result = xnew (size + 1, char);
+ size = 0;
+ for (; *out; ++out)
+ {
+ if (*out == '\\')
+ {
+ ++out;
+ if (isdigit (*out))
+ {
+ /* Using "dig2" satisfies my debugger. Bleah. */
+ int dig2 = *out - '0';
+ strncpy (result + size, in + regs->start[dig2],
+ regs->end[dig2] - regs->start[dig2]);
+ size += regs->end[dig2] - regs->start[dig2];
+ }
+ else
+ {
+ switch (*out)
+ {
+ case '\t':
+ result[size++] = '\t';
+ break;
+ case '\\':
+ *out = '\\';
+ break;
+ default:
+ result[size++] = *out;
+ break;
+ }
+ }
+ }
+ else
+ result[size++] = *out;
+ }
+ result[size] = '\0';
+
+ return result;
}
\f
+#endif /* ETAGS_REGEXPS */
/* Initialize a linebuffer for use */
-
void
initbuffer (linebuffer)
struct linebuffer *linebuffer;
* which is the length of the line including the newline, if any.
*/
long
-readline (linebuffer, stream)
+readline_internal (linebuffer, stream)
struct linebuffer *linebuffer;
register FILE *stream;
{
char *buffer = linebuffer->buffer;
register char *p = linebuffer->buffer;
register char *pend;
- int newline; /* 1 if ended with newline, 0 if ended with EOF */
+ int chars_deleted;
- pend = p + linebuffer->size; /* Separate to avoind 386/IX compiler bug. */
+ pend = p + linebuffer->size; /* Separate to avoid 386/IX compiler bug. */
while (1)
{
pend = buffer + linebuffer->size;
linebuffer->buffer = buffer;
}
- if (c < 0 || c == '\n')
+ if (c == EOF)
+ {
+ chars_deleted = 0;
+ break;
+ }
+ if (c == '\n')
{
- *p = 0;
- newline = (c == '\n' ? 1 : 0);
+ if (p[-1] == '\r' && p > buffer)
+ {
+ *--p = '\0';
+ chars_deleted = 2;
+ }
+ else
+ {
+ *p = '\0';
+ chars_deleted = 1;
+ }
break;
}
*p++ = c;
}
- return p - buffer + newline;
+ return p - buffer + chars_deleted;
+}
+
+/*
+ * Like readline_internal, above, but try to match the input
+ * line against any existing regular expressions.
+ */
+long
+readline (linebuffer, stream)
+ struct linebuffer *linebuffer;
+ FILE *stream;
+{
+ /* Read new line. */
+ int i;
+ long result = readline_internal (linebuffer, stream);
+
+#ifdef ETAGS_REGEXPS
+ /* Match against all listed patterns. */
+ for (i = 0; i < num_patterns; ++i)
+ {
+ int match = re_match (patterns[i].pattern, linebuffer->buffer,
+ (int)result, 0, &patterns[i].regs);
+ switch (match)
+ {
+ case -2:
+ /* Some error. */
+ if (!patterns[i].error_signaled)
+ {
+ error ("error while matching pattern %d", i);
+ patterns[i].error_signaled = TRUE;
+ }
+ break;
+ case -1:
+ /* No match. */
+ break;
+ default:
+ /* Match occurred. Construct a tag. */
+ if (patterns[i].name_pattern[0] != '\0')
+ {
+ /* Make a named tag. */
+ char *name = substitute (linebuffer->buffer,
+ patterns[i].name_pattern,
+ &patterns[i].regs);
+ if (name != NULL)
+ pfnote (name, TRUE, TRUE, linebuffer->buffer,
+ match, lineno, linecharno);
+ }
+ else
+ {
+ /* Make an unnamed tag. */
+ pfnote (NULL, TRUE, FALSE, linebuffer->buffer,
+ match, lineno, linecharno);
+ }
+ break;
+ }
+ }
+#endif /* ETAGS_REGEXPS */
+
+ return result;
+}
+
+/*
+ * Read a file, but do no processing. This is used to do regexp
+ * matching on files that have no language defined.
+ */
+void
+just_read_file (inf)
+ FILE *inf;
+{
+ while (!feof (inf))
+ {
+ ++lineno;
+ linecharno = charno;
+ charno += readline (&lb, inf) + 1;
+ }
}
+
\f
char *
savestr (cp)
register char *dp;
dp = xnew (len + 1, char);
- (void) strncpy (dp, cp, len);
+ strncpy (dp, cp, len);
dp[len] = '\0';
return dp;
}
-#ifdef notdef
/*
* Return the ptr in sp at which the character c last
* appears; NULL if not found
*
- * Identical to v7 rindex, included for portability.
+ * Identical to System V strrchr, included for portability.
*/
-
char *
-rindex (sp, c)
+etags_strrchr (sp, c)
register char *sp, c;
{
register char *r;
if (*sp == c)
r = sp;
} while (*sp++);
- return (r);
+ return r;
}
+
/*
* Return the ptr in sp at which the character c first
* appears; NULL if not found
*
- * Identical to v7 index, included for portability.
+ * Identical to System V strchr, included for portability.
*/
-
char *
-index (sp, c)
+etags_strchr (sp, c)
register char *sp, c;
{
do
{
if (*sp == c)
- return (sp);
- } while (*sp++);
- return (NULL);
+ return sp;
+ } while (*sp++);
+ return NULL;
}
-#endif /* notdef */
-
/* Print error message and exit. */
-
-/* VARARGS1 */
void
fatal (s1, s2)
char *s1, *s2;
{
error (s1, s2);
- exit (1);
+ exit (BAD);
}
-/* Print error message. `s1' is printf control string, `s2' is arg for it. */
+void
+pfatal (s1)
+ char *s1;
+{
+ perror (s1);
+ exit (BAD);
+}
-/* VARARGS1 */
+/* Print error message. `s1' is printf control string, `s2' is arg for it. */
void
error (s1, s2)
char *s1, *s2;
fprintf (stderr, "\n");
}
-/* Return a newly-allocated string whose contents concatenate those of s1, s2, s3. */
-
+/* Return a newly-allocated string whose contents
+ concatenate those of s1, s2, s3. */
char *
concat (s1, s2, s3)
char *s1, *s2, *s3;
int len1 = strlen (s1), len2 = strlen (s2), len3 = strlen (s3);
char *result = xnew (len1 + len2 + len3 + 1, char);
- (void) strcpy (result, s1);
- (void) strcpy (result + len1, s2);
- (void) strcpy (result + len1 + len2, s3);
- *(result + len1 + len2 + len3) = 0;
+ strcpy (result, s1);
+ strcpy (result + len1, s2);
+ strcpy (result + len1 + len2, s3);
+ result[len1 + len2 + len3] = '\0';
return result;
}
+\f
+/* Does the same work as the system V getcwd, but does not need to
+ guess buffer size in advance. */
+char *
+etags_getcwd ()
+#ifdef DOS_NT
+{
+ char *p, path[MAXPATHLEN + 1]; /* Fixed size is safe on MSDOS. */
-/* Like malloc but get fatal error if memory is exhausted. */
+ getwd (path);
+ p = path;
+ while (*p)
+ if (*p == '\\')
+ *p++ = '/';
+ else
+ *p++ = tolower (*p);
+
+ return strdup (path);
+}
+#elif HAVE_GETCWD /* not DOS_NT */
+{
+ int bufsize = 200;
+ char *path = xnew (bufsize, char);
+
+ while (getcwd (path, bufsize) == NULL)
+ {
+ if (errno != ERANGE)
+ pfatal ("pwd");
+ bufsize *= 2;
+ path = xnew (bufsize, char);
+ }
+
+ return path;
+}
+#else /* not DOS_NT and not HAVE_GETCWD */
+{
+ struct linebuffer path;
+ FILE *pipe;
+
+ initbuffer (&path);
+ pipe = (FILE *) popen ("pwd 2>/dev/null", "r");
+ if (pipe == NULL || readline_internal (&path, pipe) == 0)
+ pfatal ("pwd");
+ pclose (pipe);
+
+ return path.buffer;
+}
+#endif /* not DOS_NT and not HAVE_GETCWD */
+
+/* Return a newly allocated string containing the filename
+ of FILE relative to the absolute directory DIR (which
+ should end with a slash). */
+char *
+relative_filename (file, dir)
+ char *file, *dir;
+{
+ char *fp, *dp, *res;
+
+ /* Find the common root of file and dir. */
+ fp = absolute_filename (file, cwd);
+ dp = dir;
+ while (*fp++ == *dp++)
+ continue;
+ do
+ {
+ fp--;
+ dp--;
+ }
+ while (*fp != '/');
+
+ /* Build a sequence of "../" strings for the resulting relative filename. */
+ for (dp = etags_strchr (dp + 1, '/'), res = "";
+ dp != NULL;
+ dp = etags_strchr (dp + 1, '/'))
+ {
+ res = concat (res, "../", "");
+ }
+ /* Add the filename relative to the common root of file and dir. */
+ res = concat (res, fp + 1, "");
+
+ return res; /* temporary stub */
+}
+
+/* Return a newly allocated string containing the
+ absolute filename of FILE given CWD (which should
+ end with a slash). */
+char *
+absolute_filename (file, cwd)
+ char *file, *cwd;
+{
+ char *slashp, *cp, *res;
+
+ if (file[0] == '/')
+ res = concat (file, "", "");
+ else
+ res = concat (cwd, file, "");
+
+ /* Delete the "/dirname/.." and "/." substrings. */
+ slashp = etags_strchr (res, '/');
+ while (slashp != NULL && slashp[0] != '\0')
+ {
+ if (slashp[1] == '.')
+ {
+ if (slashp[2] == '.'
+ && (slashp[3] == '/' || slashp[3] == '\0'))
+ {
+ cp = slashp;
+ do
+ cp--;
+ while (cp >= res && *cp != '/');
+ if (*cp == '/')
+ {
+ strcpy (cp, slashp + 3);
+ }
+ else /* else (cp == res) */
+ {
+ if (slashp[3] != '\0')
+ strcpy (cp, slashp + 4);
+ else
+ return ".";
+ }
+ slashp = cp;
+ continue;
+ }
+ else if (slashp[2] == '/' || slashp[2] == '\0')
+ {
+ strcpy (slashp, slashp + 2);
+ continue;
+ }
+ }
+
+ slashp = etags_strchr (slashp + 1, '/');
+ }
+
+ return res;
+}
+
+/* Return a newly allocated string containing the absolute
+ filename of dir where FILE resides given CWD (which should
+ end with a slash). */
+char *
+absolute_dirname (file, cwd)
+ char *file, *cwd;
+{
+ char *slashp, *res;
+ char save;
+
+ slashp = etags_strrchr (file, '/');
+ if (slashp == NULL)
+ return cwd;
+ save = slashp[1];
+ slashp[1] = '\0';
+ res = absolute_filename (file, cwd);
+ slashp[1] = save;
+
+ return res;
+}
+
+/* Like malloc but get fatal error if memory is exhausted. */
char *
xmalloc (size)
- int size;
+ unsigned int size;
{
- char *result = malloc (size);
- if (!result)
+ char *result = (char *) malloc (size);
+ if (result == NULL)
fatal ("virtual memory exhausted", 0);
return result;
}
char *
xrealloc (ptr, size)
char *ptr;
- int size;
+ unsigned int size;
{
- char *result = realloc (ptr, size);
- if (!result)
+ char *result = (char *) realloc (ptr, size);
+ if (result == NULL)
fatal ("virtual memory exhausted");
return result;
}