/* rtfwc - read rtf input, write word count (actually, char, word and paragraph counts). This installs callbacks for the ascii and control token classes. The control class is necessary so that special characters such as \par, \tab, \sect, etc. can be counted. Counts paragraphs instead of lines, since the concept of "line" is relatively meaningless. It's problematic how to count text in headers and footers, and what to do about tables. 04 Feb 91 Paul DuBois dubois@primate.wisc.edu 04 Feb 91 V1.0. Created. 27 Feb 91 V1.01. Updated for distribution 1.05. 16 Mar 91 V1.02. Updated for distribution 1.06. Multiple files allowed. */ # include # include "rtf.h" static long chars, tchars = 0; static long words, twords = 0; static long paras, tparas = 0; static long wchars = 0; /* chars in current word */ static void Count (); static void Text (); static void Control (); int main (argc, argv) int argc; char **argv; { int i; --argc; ++argv; if (argc == 0) { Count (); printf ("\n"); } else { for (i = 0; i < argc; i++) { if (freopen (argv[i], "r", stdin) == NULL) { fprintf (stderr, "Can't open \"%s\"\n", argv[i]); exit (1); } Count (); printf ("\t%s\n", argv[i]); } if (argc > 1) /* multiple files, print totals */ printf ("%ld chars\t%ld words\t%ld paragraphs\ttotal\n", tchars, twords, tparas); } exit (0); } static void Count () { chars = words = paras = wchars = 0; RTFInit (); /* install counting hooks and process the input stream */ RTFSetClassCallback (rtfText, Text); RTFSetClassCallback (rtfControl, Control); RTFRead (); printf ("%ld chars\t%ld words\t%ld paragraphs", chars, words, paras); tchars += chars; twords += words; tparas += paras; } static void Text () { ++chars; if (rtfMajor != ' ') ++wchars; else if (wchars > 0) ++words; } static void Control () { if (rtfMajor != rtfSpecialChar) return; switch (rtfMinor) { case rtfPage: case rtfSect: case rtfLine: case rtfPar: ++paras; ++chars; if (wchars > 0) { ++words; wchars = 0; } break; case rtfNoBrkSpace: case rtfTab: ++chars; if (wchars > 0) { ++words; wchars = 0; } break; case rtfNoBrkHyphen: ++chars; ++wchars; break; } }