Weird File Names (Removing them)

[William E. Davidsen Jr]

In article <91 at raider.MFEE.TN.US> root at raider.MFEE.TN.US (Bob Reineri) writes:
| I run a public access site, and have some users that are fairly new to the
| Uni* world (myself included). It takes my users a while to get used to the 
| environment, and often they press many wrong keystrokes before getting the
| hang of things.

Will you settle for a utility that turns all of the crap names into
legible names? Thought you would. Also does control characters, high bit
set, embedded blanks, etc. Converts uppercase to lowercase because a lot
of DOS archivers and stuff generate it.

No fancy shar stuff, trim it by hand:
________________________________________________________________

/*  X X U  --  20-to-Unix filename converter  */

/*
 Change DEC-20 or VAX/VMS style filenames into normal Unix names.
 Handy for use after ftp MGETs, when you find your directory full of
 files with names like LIB:<KERMIT>CKUFIO.C.2 or FRED::[ETHEL]A.B;37
 when all you really wanted was ckufio.c and a.b.

 Usage: xxu file(s)

 Action: Renames argument files as follows:
   strips Unix path name from front (up to rightmost '/') if present
   strips DEC device:, node:: names from front (up to rightmost ':') if present
   strips DEC-20 <directory> or VMS [directory] name if present
   strips DEC-20 version number from end (everything after 2nd dot) if present
   strips VMS generation number from end (everything after ';') if present
   lowercases any uppercase letters
   honors DEC-20 CTRL-V quote for special characters
   discards unquoted unprintable characters
   if result is null, file is renamed to xxfile-n, where n is a number.
   if result would write over an existing file, file also renamed to xxfile-n.

 Recommended procedure: make a new directory, cd to it, then FTP files
 from DEC-20 or VMS system, then do "xxu *".

 Author:  F. da Cruz, CUCCA, July 85
*/

#include <stdio.h>
#if	SYSV | M_XENIX
#include <sys/types.h>
#endif
#include <ctype.h>
#include <sys/file.h>			/* For access() */
/* <<<<<<<< define NO_RENAME on cc line if missing >>>>>>>> */

char name[500];				/* File name buffer */
char *pp, *cp, *xp;			/* Character pointers */
char delim;				/* Directory Delimiter */
int dc = 0, n = 0;			/* Counters */
int quote = 0, indir = 0; done = 0;	/* Flags */

main(argc,argv) int argc; char **argv; {

    if (argc < 2) {			/* Give message if no args */
	fprintf(stderr,"Usage: xxu file(s)\n");
	exit(1);
    }
    n = 0;				/* Unfixable filename counter */
    while (--argc > 0) {		/* For all files on command line... */
	argv++;
	xp = *argv;			/* Copy pointer for simplicity */
	printf("%s ",*argv);		/* Echo name of this file */

	pp = name;			/* Point to translation buffer */
	*name = '\0';			/* Initialize buffer */
	dc = 0;				/* Filename dot counter */
	done = 0;			/* Flag for early completion */

	for (cp = xp; (*cp != '\0') && !done; cp++) { /* Loop thru chars... */

	    if (quote) {		/* If this char quoted, */
		*pp++ = *cp;		/*  include it literally. */
		quote = 0;
	    }
	    else if (indir) {		/* If in directory name, */
		if (*cp == delim) indir = 0; /* look for end delimiter. */
	    }
	    else switch (*cp) {
		case '<':		/* Discard DEC-20 directory name */
		    indir = 1;
		    delim = '>';
		    break;
		case '[':		/* Discard VMS directory name */
		    indir = 1;
		    delim = ']';
		    break;
		case '/':		/* Discard Unix path name */
		case ':':   	    	/*  or DEC dev: or node:: name */
		    pp = name; 
		    break;
		case '.':		/* DEC -20 generation number */
	    	    if (++dc == 1)	/* Keep first dot */
		    	*pp++ = *cp;
		    else		/* Discard everything starting */
		    	done = 1;	/* with second dot. */
		    break;
		case ';':		/* VMS generation or DEC-20 attrib */
		    done = 1;		/* Discard everything starting with */
		    break;		/* semicolon */
	    	case '\026':		/* Control-V quote for special chars */
		    quote = 1;		/* Set flag for next time. */
		    break;
		default:
		    if (isupper(*cp))  	/* Uppercase letter to lowercase */
	    	    	*pp++ = tolower(*cp);
		    else if (*cp == ' ')/* change blanks to underscore */
		        *pp++ = '_';
		    else if (isprint(*cp)) /* Other printable, just keep */
		    	*pp++ = *cp;
	    }
	}
	*pp = '\0';			/* Done with name, terminate it */
	if (strcmp(name,xp) == 0) {	/* If no renaming necessary, */
	    printf("(ok)\n");		/*  just give message. */
	    continue;
        }
	while (*name == '\0' || access(name,0) == 0) { /* Find unique name */
	    sprintf(name,"xxfile-%d",n++);
	}
	printf("=> %s ",name);		/* Tell what new name will be */
	if (rename(xp,name) == 0)	/* Try to rename it */
	    printf("(ok)\n");		/* Say what happened */
	else
	    perror("failed");
    }
    exit(0);				/* Done. */
}

/*****************************************************************
 |  rename - for systems lacking the rename system call
 |----------------------------------------------------------------
 |  Arguments:
 |   1) string - current filename
 |   2) string - new filename
 ****************************************************************/

#if	NO_RENAME
rename(oldname, newname)
    char *oldname, *newname;
{
    char cmdline[133];
    
    sprintf(cmdline, "mv \"%s\" %s", oldname, newname);
    return system(cmdline);
}
#endif
________________________________________________________________

That's all folks!
-- 
	bill davidsen		(wedu at ge-crd.arpa)
  {uunet | philabs}!steinmetz!crdos1!davidsen
"Stupidity, like virtue, is its own reward" -me