ms2html.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

		case 'c':
			r = getrune();
			if (r == '\n') {
				sol = ignore_nl = 1;
				if (indirective)
					break;
				}
			else
				ungetrune();
			return getnext();

		/* escape backslash */
		case 'e':
			return "\\";

		/* font change */
		case 'f':
			r = getrune();
			switch(r){
			case '(':
				str[0] = getrune();
				str[1] = getrune();
				str[2] = 0;
				token[0] = 0;
				if(strcmp("BI", str) == 0)
					return changefont(&bifont);
				else if(strcmp("CW", str) == 0)
					return changefont(&cwfont);
				else
					return changefont(nil);
			case '3':
			case 'B':
				return changefont(&bfont);
			case '2':
			case 'I':
				return changefont(&ifont);
			case '4':
				return changefont(&bifont);
			case '5':
				return changefont(&cwfont);
			case 'P':
				return changebackfont();
			case 'R':
			default:
				return changefont(nil);
			}

		/* number register */
		case 'n':
			r = getrune();
			if (r == '(') /*)*/ {
				r = getrune();
				if (r < 0)
					return nil;
				str[0] = r;
				r = getrune();
				if (r < 0)
					return nil;
				str[1] = r;
				str[2] = 0;
				}
			else {
				str[0] = r;
				str[1] = 0;
				}
			pushstr(getnr(str));
			return getnext();

		/* font size */
		case 's':
			r = getrune();
			switch(r){
			case '0':
				return changesize(0);
			case '-':
				r = getrune();
				if (!isdigit(r))
					return getnext();
				return changesize(-(r - '0'));
			case '+':
				r = getrune();
				if (!isdigit(r))
					return getnext();
				return changesize(r - '0');
			}
			return getnext();
		/* vertical movement */
		case 'v':
			r = getrune();
			if(r != '\''){
				ungetrune();
				return getnext();
			}
			r = getrune();
			if(r != '-')
				vert--;
			else
				vert++;
			while(r != '\'' && r != '\n')
				r = getrune();
			if(r != '\'')
				ungetrune();
			
			if(vert > 0)
				return "^";
			return getnext();
			

		/* horizontal line */
		case 'l':
			r = getrune();
			if(r != '\''){
				ungetrune();
				return "<HR>";
			}
			while(getrune() != '\'')
				;
			return "<HR>";

		/* character height and slant */
		case 'S':
		case 'H':
			r = getrune();
			if(r != '\''){
				ungetrune();
				return "<HR>";
			}
			while(getrune() != '\'')
				;
			return getnext();

		/* digit-width space */
		case '0':
			return " ";

		/*for .if, .ie, .el */
		case '{':
			return "\\{"; /*}*/
		case '}':
			return "";
		/* up and down */
		case 'u':
			if (isdown) {
				isdown = 0;
				return "</sub>";
			}
			isup = 1;
			return "<sup>";
		case 'd':
			if (isup) {
				isup = 0;
				return "</sup>";
			}
			isdown = 1;
			return "<sub>";
		}
		break;
	case '&':
		if(msp >= 0 || strsp >= 0)
			return "&";
		return "&amp;";
	case '<':
		if(msp >= 0 || strsp >= 0)
			return "<";
		return "&#60;";
	case '>':
		if(msp >= 0 || strsp >= 0)
			return ">";
		return "&#62;";
	}
	if (r < Runeself) {
		token[0] = r;
		token[1] = 0;
		}
	else {
		R = r;
		token[runetochar(token,&R)] = 0;
	}
	return token;
}

/* if arg0 is set, read up to (and expand) to the next whitespace, else to the end of line */
char*
copyline(char *p, char *e, int arg0)
{
	int c;
	Rune r;
	char *p1;

	while((c = getrune()) == ' ' || c == '\t')
		;
	for(indirective = 1; p < e; c = getrune()) {
		if (c < 0)
			goto done;
		switch(c) {
		case '\\':
			break;
		case '\n':
			if (arg0)
				ungetrune();
			goto done;
		case ' ':
		case '\t':
			if (arg0)
				goto done;
		default:
			r = c;
			p += runetochar(p,&r);
			continue;
		}
		ungetrune();
		p1 = getnext();
		if (p1 == nil)
			goto done;
		if (*p1 == '\n') {
			if (arg0)
				ungetrune();
			break;
		}
		while((*p = *p1++) && p < e)
			p++;
	}
done:
	indirective = 0;
	*p++ = 0;
	return p;
}

char*
copyarg(char *p, char *e, int *nullarg)
{
	int c, quoted, last;
	Rune r;

	*nullarg = 0;
	quoted = 0;
	do{
		c = getrune();
	} while(c == ' ' || c == '\t');

	if(c == '"'){
		quoted = 1;
		*nullarg = 1;
		c = getrune();
	}

	if(c == '\n')
		goto done;

	last = 0;
	for(; p < e; c = getrune()) {
		if (c < 0)
			break;
		switch(c) {
		case '\n':
			ungetrune();
			goto done;
		case '\\':
			r = c;
			p += runetochar(p,&r);
			if(last == '\\')
				r = 0;
			break;
		case ' ':
		case '\t':
			if(!quoted && last != '\\')
				goto done;
			r = c;
			p += runetochar(p,&r);
			break;
		case '"':
			if(quoted && last != '\\')
				goto done;
			r = c;
			p += runetochar(p,&r);
			break;
		default:
			r = c;
			p += runetochar(p,&r);
			break;
		}
		last = r;
	}
done:
	*p++ = 0;
	return p;

}

int
parseargs(char *p, char *e, char **argv)
{
	int argc;
	char *np;
	int nullarg;

	indirective = 1;
	*p++ = 0;
	for(argc = 1; argc < Narg; argc++){
		np = copyarg(p, e, &nullarg);
		if(nullarg==0 && np == p+1)
			break;
		argv[argc] = p;
		p = np;
	}
	argv[argc] = nil;
	indirective = 0;


	return argc;
}

void
dodirective(void)
{
	char *p, *e;
	Goobie *g;
	Goobieif *gif;
	char line[Nline], *line1;
	int i, argc;
	char *argv[Narg];
	Mstack *m;

	/* read line, translate special bytes */
	e = line + sizeof(line) - UTFmax - 1;
	line1 = copyline(line, e, 1);
	if (!line[0])
		return;
	argv[0] = line;

	/* first look through user defined macros */
	p = getmacro(argv[0]);
	if(p != nil){
		if(msp == Maxmstack-1){
			fprint(2, "ms2html: macro stack overflow\n");
			return;
		}
		argc = parseargs(line1, e, argv);
		m = &mstack[++msp];
		m->ptr = p;
		memset(m->argv, 0, sizeof(m->argv));
		for(i = 0; i < argc; i++)
			m->argv[i] = strdup(argv[i]);
		return;
	}

	/* check for .if or .ie */
	for(gif = gtabif; gif->name; gif++)
		if(strcmp(gif->name, argv[0]) == 0){
			(*gif->f)(line1, e);
			return;
		}

	argc = parseargs(line1, e, argv);

	/* try standard ms macros */
	for(g = gtab; g->name; g++)
		if(strcmp(g->name, argv[0]) == 0){
			(*g->f)(argc, argv);
			return;
		}

	if(debug)
		fprint(2, "stdin %d(%s:%d): unknown directive %s\n",
			ssp->lno, ssp->filename, ssp->rlno, line);
}

void
printarg(char *a)
{
	char *e, *p;
	
	e = a + strlen(a);
	pushstr(a);
	while(strsp >= 0 && strstack[strsp] >= a && strstack[strsp] < e){
		p = getnext();
		if(p == nil)
			return;
		Bprint(&bout, "%s", p);
	}
}

void
printargs(int argc, char **argv)
{
	argc--;
	argv++;
	while(--argc > 0){
		printarg(*argv++);
		Bprint(&bout, " ");
	}
	if(argc == 0)
		printarg(*argv);
}

void
dohangingdt(void)
{
	switch(hangingdt){
	case 3:
		hangingdt--;
		break;
	case 2:
		Bprint(&bout, "<dd>");
		hangingdt = 0;
		break;
	}

}

void
dohangingau(void)
{
	if(hangingau == 0)
		return;
	Bprint(&bout, "</I></DL>\n");
	hangingau = 0;
}

void
dohanginghead(void)
{
	if(hanginghead == 0)
		return;
	Bprint(&bout, "</H%d>\n", hanginghead);
	hanginghead = 0;
}

/*
 *  convert a man page to html and output
 */
void
doconvert(void)
{
	char c, *p;
	Tm *t;

	pushsrc(nil);

	sol = 1;
	Bprint(&bout, "<html>\n");
	Bflush(&bout);
	for(;;){
		p = getnext();
		if(p == nil)
			break;
		c = *p;
		if(c == '.' && sol){
			dodirective();
			dohangingdt();
			ssp->lno++;
			ssp->rlno++;
			sol = 1;
		} else if(c == '\n'){
			if (ignore_nl)
				ignore_nl = 0;
			else {
				if(hangingau)
					Bprint(&bout, "<br>\n");
				else
					Bprint(&bout, "%s", p);
				dohangingdt();
				}
			ssp->lno++;
			ssp->rlno++;
			sol = 1;
		} else{
			Bprint(&bout, "%s", p);
			ignore_nl = sol = 0;
		}
	}
	dohanginghead();
	dohangingdt();
	closel();
	if(fsp >= 0 && fstack[fsp])
		Bprint(&bout, "%s", fstack[fsp]->end);
	Bprint(&bout, "<br>&#32;<br>\n");
	Bprint(&bout, "<A href=http://www.lucent.com/copyright.html>\n");
	t = localtime(time(nil));
	Bprint(&bout, "Copyright</A> &#169; %d Lucent Technologies Inc.  All rights reserved.\n",
			t->year+1900);
	Bprint(&bout, "</body></html>\n");
}

static void
usage(void)
{
	sysfatal("usage: ms2html [-q] [-b basename] [-d '$$'] [-t title]\n");
}

void
main(int argc, char **argv)
{
	quiet = 1;
	ARGBEGIN {
	case 't':
		title = EARGF(usage());
		break;
	case 'b':
		basename = EARGF(usage());
		break;
	case 'q':
		quiet = 0;
		break;
	case 'd':
		delim = EARGF(usage());
		break;
	case '?':
	default:
		usage();
	} ARGEND;

	Binit(&bout, 1, OWRITE);

	ds("R", "&#174;");

	doconvert();
	exits(nil);
}

void
g_notyet(int, char **argv)
{
	fprint(2, "ms2html: .%s not yet supported\n", argv[0]);
}

void
g_ignore(int, char **argv)
{
	if(quiet)
		return;
	fprint(2, "ms2html: line %d: ignoring .%s\n", ssp->lno, argv[0]);
}

void
g_PP(int, char**)
{
	dohanginghead();
	closel();
	closefont();
	Bprint(&bout, "<P>\n");
	paragraph = 1;
}

void
g_LP(int, char**)
{
	dohanginghead();
	closel();
	closefont();
	Bprint(&bout, "<br>&#32;<br>\n");
}

/* close a list */
void
closel(void)
{
	g_P2(1, nil);
	dohangingau();
	if(paragraph){
		Bprint(&bout, "</P>\n");
		paragraph = 0;
	}
	switch(list){
	case Lordered:
		Bprint(&bout, "</ol>\n");
		break;
	case Lunordered:
		Bprint(&bout, "</ul>\n");
		break;
	case Lother:
	case Ldef:
		Bprint(&bout, "</dl>\n");
		break;
	}
	list = 0;
	
}


void
g_IP(int argc, char **argv)
{
	switch(list){
	default:
		closel();
		if(argc > 1){
			if(strcmp(argv[1], "1") == 0){
				list = Lordered;
				listnum = 1;
				Bprint(&bout, "<OL>\n");
			} else if(strcmp(argv[1], "\\(bu") == 0){
				list = Lunordered;
				Bprint(&bout, "<UL>\n");
			} else {
				list = Lother;
				Bprint(&bout, "<DL COMPACT>\n");
			}
		} else {
			list = Lother;
			Bprint(&bout, "<DL>\n");
		}
		break;
	case Lother:
	case Lordered:
	case Lunordered:
		break;
	}

	switch(list){
	case Lother:
		Bprint(&bout, "<DT>");
		if(argc > 1)
			printarg(argv[1]);
		else
			Bprint(&bout, "<DT>&#32;");
		Bprint(&bout, "<DD>\n");
		break;
	case Lordered:
	case Lunordered:
		Bprint(&bout, "<LI>\n");
		break;
	}
}

/*
 *  .5i is one <DL><DT><DD>
 */
void
g_in(int argc, char **argv)
{
	float	f;
	int	delta, x;
	char	*p;

	f = indent/0.5;
	delta = f;
	if(argc <= 1){
		indent = 0.0;
	} else {
		f = strtod(argv[1], &p);
		switch(*p){
		case 'i':
			break;
		case 'c':
			f = f / 2.54;
			break;
		case 'P':
			f = f / 6;
			break;
		default:
		case 'u':
		case 'm':
			f = f * (12 / 72);
			break;
		case 'n':
			f = f * (6 / 72);
			break;
		case 'p':
			f = f / 72.0;
			break;
		}
		switch(argv[1][0]){
		case '+':
		case '-':
			indent += f;
			break;
		default:
			indent = f;
			break;
		}
	}
	if(indent < 0.0)
		indent = 0.0;
	f = (indent/0.5);
	x = f;
	delta = x - delta;
	while(delta < 0){
		Bprint(&bout, "</DL>\n");
		delta++;
	}
	while(delta > 0){
		Bprint(&bout, "<DL><DT><DD>\n");
		delta--;
	}
}

void
g_HP(int, char**)
{
	switch(list){
	default:
		closel();
		list = Ldef;
		hangingdt = 1;
		Bprint(&bout, "<DL><DT>\n");
		break;
	case Ldef:
		if(hangingdt)
			Bprint(&bout, "<DD>");
		Bprint(&bout, "<DT>");
		hangingdt = 1;
		break;
	}
}

void
g_SH(int, char**)
{
	dohanginghead();
	dohangingcenter();
	closel();
	closefont();
	Bprint(&bout, "<H%d>", HH);
	hanginghead = HH;
}

void
g_NH(int argc, char **argv)
{
	int i, level;

	closel();
	closefont();

	dohangingcenter();
	if(argc == 1)
		level = 0;
	else {
		level = atoi(argv[1])-1;
		if(level < 0 || level >= Maxnh)
			level = Maxnh - 1;
	}
	nh[level]++;

	Bprint(&bout, "<H%d>", HH);
	hanginghead = HH;

	Bprint(&bout, "%d", nh[0]);
	for(i = 1; i <= level; i++)
		Bprint(&bout, ".%d", nh[i]);
	Bprint(&bout, " ");

	for(i = level+1; i < Maxnh; i++)
		nh[i] = 0;
}

void
g_TL(int, char**)
{
	char *p, *np;
	char name[128];

	closefont();

	if(!titleseen){
		if(!title){
			/* get base part of filename */
			p = strrchr(ssp->filename, '/');
			if(p == nil)
				p = ssp->filename;
			else
				p++;
			strncpy(name, p, sizeof(name));
			name[sizeof(name)-1] = 0;
		
			/* dump any extensions */
			np = strchr(name, '.');
			if(np)
				*np = 0;
			title = p;
		}
		Bprint(&bout, "<title>\n");
		Bprint(&bout, "%s\n", title);
		Bprint(&bout, "</title>\n");
		Bprint(&bout, "<body BGCOLOR=\"#FFFFFF\" TEXT=\"#000000\" LINK=\"#0000FF\" VLINK=\"#330088\" ALINK=\"#FF0044\">\n");
		titleseen = 1;
	}

	Bprint(&bout, "<center>");
	hangingcenter = 1;
	Bprint(&bout, "<H%d>", 1);
	hanginghead = 1;
}

void
dohangingcenter(void)
{
	if(hangingcenter){
		Bprint(&bout, "</center>");
		hangingcenter = 1;
	}
}

void
g_AU(int, char**)
{
	closel();
	dohanginghead();
	Bprint(&bout, "<DL><DD><I>");
	hangingau = 1;
}

void
pushfont(Font *f)
{
	if(fsp == Maxfsp)
		return;
	if(fsp >= 0 && fstack[fsp])
		Bprint(&bout, "%s", fstack[fsp]->end);
	if(f != nil)
		Bprint(&bout, "%s", f->start);
	fstack[++fsp] = f;
}

void
popfont(void)
{
	if(fsp >= 0){
		if(fstack[fsp])
			Bprint(&bout, "%s", fstack[fsp]->end);
		fsp--;
	}
}

/*
 *  for 3 args print arg3 \fxarg1\fP arg2
 *  for 2 args print arg1 \fxarg2\fP
 *  for 1 args print \fxarg1\fP
 */
void
font(Font *f, int argc, char **argv)
{
	if(argc == 1){
		pushfont(nil);
		return;
	}
	if(argc > 3)
		printarg(argv[3]);
	pushfont(f);
	printarg(argv[1]);
	popfont();
	if(argc > 2)
		printarg(argv[2]);
	Bprint(&bout, "\n");
}

void
closefont(void)
{
	if(fsp >= 0 && fstack[fsp])
		Bprint(&bout, "%s", fstack[fsp]->end);
	fsp = -1;
}

void
g_B(int argc, char **argv)
{
	font(&bfont, argc, argv);
}

void
g_R(int argc, char **argv)
{
	font(nil, argc, argv);
}

void
g_BI(int argc, char **argv)
{
	font(&bifont, argc, argv);
}

void
g_CW(int argc, char **argv)
{
	font(&cwfont, argc, argv);
}

char*