daemon@caip.UUCP (04/23/86)
From: Chris Metcalf <princeton!spock!yale!metcalf@topaz.rutgers.edu> Jeff Dickson at jpl-milvax posted this some days ago: > Back in November '85, someone posted a message to the effect > that they had run their Lattice 'C' include files through some kind > of text filter and freed up a sizable amount of disk space for other > things.[:-)] > > Well to make a long story short, that person never did make > the "fabulous filter" available to us netters [:-(]. So when I pur- > chased my Lattice 'C' compiler, I decided to write my own text filter. Let me apologize for that. I posted my program, stripc.c, to USENET, and it never seems to have made it to the info-amiga list. I sent copies to a few people who hadn't received it, and sent a note to Eric Lavitsky, who said he'd repost it for me. Here (hopefully for the last time :-) is stripc.c. It has a simpler user interface than pack, since it works either as a simple filter or else strips the filenames specified on the command line. I didn't include any of the fancy wildcarding stuff. However, the code itself should produce more accurately (and neatly) packed C code. "pack", as posted, will take strings like "many spaces" and pack them, which tends to distort the code a bit. However, it can be a bit clumsy to use, since each file has to be individually specified, and it uses the physical disk rather than the RAM disk as it works. Use whichever you like. Me, I've switched to Aztec's symbol-table dump files... just imagine, a 30K file on your RAM disk that includes all of <intuition/intuition.h> and its friends, and loads into the compiler in just over a second when you use the +i flag! (No, I'm not affiliated with Manx, I just think it's a great product.) Chris Metcalf (metcalf@yale.ARPA) ...!ihnp4!hsi!yale!metcalf (metcalf@yalecs.BITNET) ...!decvax!yale!metcalf ----------------------------- cut here --------------------------- /* * stripc.c * * This program removes all comments from the selected file (either * a command-line argument or standard input). It also removes much of * the excess whitespace while attempting to preserve the basic * arrangement of the program text. * * SYNOPSIS * stripc [ filename ... ] * * Chris Metcalf * Jan 2, 1986 * reposted Apr 21, 1986 */ #include <stdio.h> #include <ctype.h> #define bool char #define TRUE 1 #define FALSE 0 char nextchar(); /* returns next non-comment character */ void strip_blanks(); #define WS 0x80 /* ORed into whitespace characters */ main (argc, argv) char **argv; { char tmpfile[32]; /* temp name of stripped file */ char *name; /* name of this program (stripc) */ FILE *in, *out; /* file descriptors for parameter files */ /* read filename arguments -- syntax: stripc [filename filename ...] */ name = argv[0]; if (argc == 1) strip_blanks(stdin, stdout); else while (*++argv) { if ((in = fopen(*argv, "r")) == NULL) { fprintf(stderr, "%s: couldn't open file %s\n", name, *argv); exit(1); } if ((out = fopen(strcat(strcpy(tmpfile, *argv), ".t"), "w")) == NULL) { fprintf(stderr, "%s: couldn't open temp file %s\n", name, tmpfile); exit(1); } strip_blanks(in, out); fclose(in); fclose(out); if (unlink(*argv)) { fprintf(stderr, "%s: couldn't remove %s\n", name, *argv); exit(1); } if (rename(tmpfile, *argv)) { fprintf(stderr, "%s: couldn't rename %s as %s\n",name,*argv,tmpfile); exit(1); } } } /* * read characters through nextchar(), eliminating white space */ void strip_blanks(in, out) FILE *in, *out; { char c; /* current character */ char lastws = 0; /* the white space character being buffered */ bool ws; /* set if current character if white space */ while ((c = nextchar(in)) != EOF) { ws = c & WS; c &= ~WS; if (lastws) { if (!ws) { putc(lastws, out); putc(c, out); lastws = 0; } else if (c == '\n' || c == '\r' || c == '\f') lastws = '\n'; else if (c == '\t' && lastws != '\n') lastws = '\t'; else if (c == ' ' && lastws != '\n' && lastws != '\t') lastws = ' '; } else if (ws) lastws = (c=='\r'||c=='\f')?'\n':c; else putc(c, out); } if (lastws) putc(lastws, out); } char nextchar (in) FILE *in; { char c; /* current character being read */ bool modes; /* a temp variable for testing all modes */ static char next; /* the character waiting to be read */ static bool init = TRUE, /* is the first time through the loop */ comments = FALSE, /* whether we're in a block of comments */ oldcomments = FALSE, /* whether we WERE in a block of comments */ quotes = FALSE, /* whether we're in double quotes */ single = FALSE, /* in single quotes */ preproc = FALSE, /* in a # line (preprocessor command) */ demagic = FALSE, /* was the previous character a backslash */ newline = TRUE; /* was the previous character a newline */ /* establish lookahead the first time the routine is called */ if (init) { next = getc(in); init = FALSE; } /* * The main do statement loops until a "real" character is found, * or the end of file is encountered. The initial state of preproc, * quotes, single, and comments are saved so that transitions from * state to state during the "switch" can be eliminated in the output. * * The switch statement handles the work and the strange cases. * If a # is found and the previous character was a newline, and * we're not in comments mode, preprocessor mode is set TRUE. * Preprocessor mode is ended by a newline (unless it's escaped). * Comments mode is set if a / is followed by a *, and none of * the other modes (preproc, quotes, single) are set. Comments mode * is ended by * followed by / unless it's in a preprocessor line. * Single quotes mode is toggled on and off by ', unless we're in * double quotes or a comment mode, or the ' is escaped. Double * quotes function similarly. Note that single and double quotes * mode can be erroneously set during preprocessor lines, but this * is unimportant, since the code for newline already turns off quotes * mode at the end of a line (to guard against mismatched quotes). */ do { if ( (c = next) == EOF) { init = TRUE; return (EOF); } next = getc(in); modes = preproc || quotes || single; oldcomments = comments; switch (c) { case '#': if (newline && !comments) preproc = TRUE; break; case '\n': if (!demagic) preproc = single = quotes = FALSE; break; case '/': if (!quotes && !single && next == '*') comments = TRUE; break; case '*': if (comments && next == '/') comments = FALSE; break; case '\'': if (!comments && !demagic && !quotes) single = !single; break; case '"': if (!comments && !demagic && !single) quotes = !quotes; break; } demagic = ( !demagic && (c == '\\') ); newline = ( !demagic && (c == '\n') ); if ( comments != oldcomments ) next = getc(in); /* we don't want to have to read it again */ } while (comments || oldcomments); if ((!isprint(c) || c == ' ') && !quotes && !single) c |= WS; return c; }