RE: [21.5] make-docfile needs to deal with UNUSED in DEFUN

-----Original Message----- From: xemacs-patches-bounces(a)xemacs.org [mailto:xemacs-patches-bounces＠xemacs.org] On Behalf Of Stephen J. Turnbull Sent: Tuesday, January 25, 2005 3:44 AM To: xemacs-patches(a)xemacs.org Subject: [21.5] make-docfile needs to deal with UNUSED in DEFUN Not for 21.4. Yes, there are UNUSED in DEFUNs. Dunno why; GNU or backward compatibility, I guess. This results in a badly broken docstring: the "UNUSED" itself is interpreted as a formal argument, and the following close paren ends the argument list. This patch works for me so far, but the whole thing seems quite delicate: any mistakes in the parser here will cost you the docstrings of everything defined in C.... Tests and comments welcome. Eventually I'll commit if nobody complains. Index: lib-src/ChangeLog =================================================================== RCS file: /Users/steve/Software/Repositories/cvs.xemacs.org/XEmacs/xemac s/lib-src/ChangeLog,v retrieving revision 1.178.2.1 diff -u -r1.178.2.1 ChangeLog --- ChangeLog 2004/12/11 05:12:52 1.178.2.1 +++ ChangeLog 2005/01/22 04:53:31 ＠＠ -0,0 +1,5 ＠＠ +2004-12-19 Stephen J. Turnbull <stephen(a)xemacs.org&gt; + + * make-docfile.c (write_c_args): Handle UNUSED, USED_IF macros. + (scan_c_file): Handle nested parens in DEFUN argument lists. + Index: lib-src/make-docfile.c =================================================================== RCS file: /Users/steve/Software/Repositories/cvs.xemacs.org/XEmacs/xemac s/lib-src/make-docfile.c,v retrieving revision 1.15.2.1 diff -u -r1.15.2.1 make-docfile.c --- make-docfile.c 2004/12/11 05:12:54 1.15.2.1 +++ make-docfile.c 2005/01/23 11:07:27 ＠＠ -525,9 +525,16 ＠＠ char c = *p; int ident_start = 0; - /* XEmacs addition: add support for ANSI prototypes. Hop over - "Lisp_Object" string (the only C type allowed in DEFUNs) */ + /* XEmacs addition: add support for ANSI prototypes and the UNUSED + macros. Hop over them. "Lisp_Object" is the only C type allowed + in DEFUNs. For the UNUSED macros we need to eat parens, too. */ + static char uu [] = "UNUSED"; + static char ui [] = "USED_IF_"; static char lo[] = "Lisp_Object"; + + /* aren't these all vulnerable to buffer overrun? I guess that + means that the .c is busted, so we may as well just die ... */ + /* skip over "Lisp_Object" */ if ((C_IDENTIFIER_CHAR_P (c) != in_ident) && !in_ident && (strncmp (p, lo, sizeof (lo) - 1) == 0) && isspace ((unsigned char) p[sizeof (lo) - 1])) ＠＠ -538,6 +545,45 ＠＠ c = *p; } + /* skip over "UNUSED" invocation */ + if ((C_IDENTIFIER_CHAR_P (c) != in_ident) && !in_ident && + (strncmp (p, uu, sizeof (uu) - 1) == 0)) + { + char *here = p; + p += (sizeof (uu) - 1); + while (isspace ((unsigned char) (*p))) + p++; + if (*p == '(') + { + while (isspace ((unsigned char) (*++p))) + ; + c = *p; + } + else + p = here; + } + + /* skip over "USED_IF_*" invocation (only if USED failed) */ + else if ((C_IDENTIFIER_CHAR_P (c) != in_ident) && !in_ident && + (strncmp (p, ui, sizeof (ui) - 1) == 0)) + { + char *here = p; + p += (sizeof (ui) - 1); + /* There should be a law against parsing in C: + this allows a broken USED_IF call, skipping to next macro's + parens. *You* can fix that, I don't see how offhand. ;-) */ + while (*p && *p++ != '(') + ; + if (*p) + { + while (isspace ((unsigned char) (*p))) + p++; + c = *p; + } + else + p = here; + } + /* Notice when we start printing a new identifier. */ if (C_IDENTIFIER_CHAR_P (c) != in_ident) { ＠＠ -836,6 +882,7 ＠＠ if (defunflag && maxargs != -1) { char argbuf[1024], *p = argbuf; + int paren_level = 1; #if 0 /* For old DEFUN's only */ while (c != ')') { ＠＠ -858,8 +905,13 ＠＠ *p++ = c = getc (infile); if (c < 0) goto eof; + /* XEmacs change: handle macros with args (eg, UNUSED) */ + if (c == ')') + paren_level--; + if (c == '(') + paren_level++; } - while (c != ')'); + while (paren_level > 0); *p = '\0'; /* Output them. */ if (ellcc) -- Graduate School of Systems and Information Engineering University of Tsukuba http://turnbull.sk.tsukuba.ac.jp/ Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Economics of Information Communication and Computation Systems Experimental Economics, Microeconomic Theory, Game Theory