parse.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

 */
static void
mimeifrange(Hlex *h, char *)
{
	int c, d, et;

	et = 0;
	c = getc(h);
	while(c == ' ' || c == '\t')
		c = getc(h);
	if(c == '"')
		et = 1;
	else if(c == 'W'){
		d = getc(h);
		if(d == '/')
			et = 1;
		ungetc(h);
	}
	ungetc(h);
	if(et){
		h->c->head.ifrangeetag = mimeetag(h, h->c->head.ifrangeetag);
	}else{
		lexhead(h);
		h->c->head.ifrangedate = hdate2sec(h->wordval);
	}
}

static void
mimerange(Hlex *h, char *)
{
	h->c->head.range = mimeranges(h, h->c->head.range);
}

/*
 * note: netscape and ie through versions 4.7 and 4
 * support only basic authorization, so that is all that is supported here
 *
 * "Authorization" ":" "Basic" base64-user-pass
 * where base64-user-pass is the base64 encoding of
 * username ":" password
 */
static void
mimeauthorization(Hlex *h, char *)
{
	char *up, *p;
	int n;

	if(lex(h) != Word || cistrcmp(h->wordval, "basic") != 0)
		return;

	n = lexbase64(h);
	if(!n)
		return;

	/*
	 * wipe out source for password, so it won't be logged.
	 * it is replaced by a single =,
	 * which is valid base64, but not ok for an auth reponse.
	 * therefore future parses of the header field will not overwrite
	 * authuser and authpass.
	 */
	memmove(h->c->hpos - (n - 1), h->c->hpos, h->c->hstop - h->c->hpos);
	h->c->hstop -= n - 1;
	*h->c->hstop = '\0';
	h->c->hpos -= n - 1;
	h->c->hpos[-1] = '=';

	up = halloc(h->c, n + 1);
	n = dec64((uchar*)up, n, h->wordval, n);
	up[n] = '\0';
	p = strchr(up, ':');
	if(p != nil){
		*p++ = '\0';
		h->c->head.authuser = hstrdup(h->c, up);
		h->c->head.authpass = hstrdup(h->c, p);
	}
}

static void
mimeagent(Hlex *h, char *)
{
	lexhead(h);
	h->c->head.client = hstrdup(h->c, h->wordval);
}

static void
mimefrom(Hlex *h, char *)
{
	lexhead(h);
}

static void
mimehost(Hlex *h, char *)
{
	char *hd;

	lexhead(h);
	for(hd = h->wordval; *hd == ' ' || *hd == '\t'; hd++)
		;
	h->c->head.host = hlower(hstrdup(h->c, hd));
}

/*
 * if present, implies that a message body follows the headers
 * "content-length" ":" digits
 */
static void
mimecontlen(Hlex *h, char *)
{
	char *e;
	ulong v;

	if(lex(h) != Word)
		return;
	e = h->wordval;
	v = digtoul(e, &e);
	if(v == ~0UL || *e != '\0')
		return;
	h->c->head.contlen = v;
}

/*
 * mimexpect	: "expect" ":" expects
 * expects	: | expects "," expect
 * expect	: "100-continue" | token | token "=" token expectparams | token "=" qstring expectparams
 * expectparams	: ";" token | ";" token "=" token | token "=" qstring
 * for now, we merely parse "100-continue" or anything else.
 */
static void
mimeexpect(Hlex *h, char *)
{
	if(lex(h) != Word || cistrcmp(h->wordval, "100-continue") != 0 || lex(h) != '\n')
		h->c->head.expectother = 1;
	h->c->head.expectcont = 1;
}

static void
mimetransenc(Hlex *h, char *)
{
	h->c->head.transenc = mimehfields(h);
}

static void
mimefresh(Hlex *h, char *)
{
	char *s;

	lexhead(h);
	for(s = h->wordval; *s && (*s==' ' || *s=='\t'); s++)
		;
	if(strncmp(s, "pathstat/", 9) == 0)
		h->c->head.fresh_thresh = atoi(s+9);
	else if(strncmp(s, "have/", 5) == 0)
		h->c->head.fresh_have = atoi(s+5);
}

static void
mimeignore(Hlex *h, char *)
{
	lexhead(h);
}

static void
parsejump(Hlex *h, char *k)
{
	int l, r, m;

	l = 1;
	r = nelem(mimehead) - 1;
	while(l <= r){
		m = (r + l) >> 1;
		if(cistrcmp(mimehead[m].name, k) <= 0)
			l = m + 1;
		else
			r = m - 1;
	}
	m = l - 1;
	if(cistrcmp(mimehead[m].name, k) == 0 && !mimehead[m].ignore){
		mimehead[m].seen = 1;
		(*mimehead[m].parse)(h, k);
	}else
		mimeignore(h, k);
}

static int
lex(Hlex *h)
{
	return h->tok = lex1(h, 0);
}

static int
lexbase64(Hlex *h)
{
	int c, n;

	n = 0;
	lex1(h, 1);

	while((c = getc(h)) >= 0){
		if(!(c >= 'A' && c <= 'Z'
		|| c >= 'a' && c <= 'z'
		|| c >= '0' && c <= '9'
		|| c == '+' || c == '/')){
			ungetc(h);
			break;
		}

		if(n < HMaxWord-1)
			h->wordval[n++] = c;
	}
	h->wordval[n] = '\0';
	return n;
}

/*
 * rfc 822/rfc 1521 lexical analyzer
 */
static int
lex1(Hlex *h, int skipwhite)
{
	int level, c;

	if(h->eol)
		return '\n';

top:
	c = getc(h);
	switch(c){
	case '(':
		level = 1;
		while((c = getc(h)) >= 0){
			if(c == '\\'){
				c = getc(h);
				if(c < 0)
					return '\n';
				continue;
			}
			if(c == '(')
				level++;
			else if(c == ')' && --level == 0)
				break;
			else if(c == '\n'){
				c = getc(h);
				if(c < 0)
					return '\n';
				if(c == ')' && --level == 0)
					break;
				if(c != ' ' && c != '\t'){
					ungetc(h);
					return '\n';
				}
			}
		}
		goto top;

	case ' ': case '\t':
		goto top;

	case '\r':
		c = getc(h);
		if(c != '\n'){
			ungetc(h);
			goto top;
		}

	case '\n':
		if(h->tok == '\n'){
			h->eol = 1;
			h->eoh = 1;
			return '\n';
		}
		c = getc(h);
		if(c < 0){
			h->eol = 1;
			return '\n';
		}
		if(c != ' ' && c != '\t'){
			ungetc(h);
			h->eol = 1;
			return '\n';
		}
		goto top;

	case ')':
	case '<': case '>':
	case '[': case ']':
	case '@': case '/':
	case ',': case ';': case ':': case '?': case '=':
		if(skipwhite){
			ungetc(h);
			return c;
		}
		return c;

	case '"':
		if(skipwhite){
			ungetc(h);
			return c;
		}
		word(h, "\"");
		getc(h);		/* skip the closing quote */
		return QString;

	default:
		ungetc(h);
		if(skipwhite)
			return c;
		word(h, "\"(){}<>@,;:/[]?=\r\n \t");
		if(h->wordval[0] == '\0'){
			h->c->head.closeit = 1;
			hfail(h->c, HSyntax);
			longjmp(h->jmp, -1);
		}
		return Word;
	}
	/* not reached */
}

/*
 * return the rest of an rfc 822, including \n
 * do not map to lower case
 */
static void
lexhead(Hlex *h)
{
	int c, n;

	n = 0;
	while((c = getc(h)) >= 0){
		if(c == '\r')
			c = wordcr(h);
		else if(c == '\n')
			c = wordnl(h);
		if(c == '\n')
			break;
		if(c == '\\'){
			c = getc(h);
			if(c < 0)
				break;
		}

		if(n < HMaxWord-1)
			h->wordval[n++] = c;
	}
	h->tok = '\n';
	h->eol = 1;
	h->wordval[n] = '\0';
}

static void
word(Hlex *h, char *stop)
{
	int c, n;

	n = 0;
	while((c = getc(h)) >= 0){
		if(c == '\r')
			c = wordcr(h);
		else if(c == '\n')
			c = wordnl(h);
		if(c == '\\'){
			c = getc(h);
			if(c < 0)
				break;
		}else if(c < 32 || strchr(stop, c) != nil){
			ungetc(h);
			break;
		}

		if(n < HMaxWord-1)
			h->wordval[n++] = c;
	}
	h->wordval[n] = '\0';
}

static int
wordcr(Hlex *h)
{
	int c;

	c = getc(h);
	if(c == '\n')
		return wordnl(h);
	ungetc(h);
	return ' ';
}

static int
wordnl(Hlex *h)
{
	int c;

	c = getc(h);
	if(c == ' ' || c == '\t')
		return c;
	ungetc(h);

	return '\n';
}

static int
getc(Hlex *h)
{
	if(h->eoh)
		return -1;
	if(h->c->hpos < h->c->hstop)
		return *h->c->hpos++;
	h->eoh = 1;
	h->eol = 1;
	return -1;
}

static void
ungetc(Hlex *h)
{
	if(h->eoh)
		return;
	h->c->hpos--;
}

static ulong
digtoul(char *s, char **e)
{
	ulong v;
	int c, ovfl;

	v = 0;
	ovfl = 0;
	for(;;){
		c = *s;
		if(c < '0' || c > '9')
			break;
		s++;
		c -= '0';
		if(v > UlongMax/10 || v == UlongMax/10 && c >= UlongMax%10)
			ovfl = 1;
		v = v * 10 + c;
	}

	if(e)
		*e = s;
	if(ovfl)
		return UlongMax;
	return v;
}

int
http11(HConnect *c)
{
	return c->req.vermaj > 1 || c->req.vermaj == 1 && c->req.vermin > 0;
}

char*
hmkmimeboundary(HConnect *c)
{
	char buf[32];
	int i;

	srand((time(0)<<16)|getpid());
	strcpy(buf, "upas-");
	for(i = 5; i < sizeof(buf)-1; i++)
		buf[i] = 'a' + nrand(26);
	buf[i] = 0;
	return hstrdup(c, buf);
}

HSPairs*
hmkspairs(HConnect *c, char *s, char *t, HSPairs *next)
{
	HSPairs *sp;

	sp = halloc(c, sizeof *sp);
	sp->s = s;
	sp->t = t;
	sp->next = next;
	return sp;
}

HSPairs*
hrevspairs(HSPairs *sp)
{
	HSPairs *last, *next;

	last = nil;
	for(; sp != nil; sp = next){
		next = sp->next;
		sp->next = last;
		last = sp;
	}
	return last;
}

HFields*
hmkhfields(HConnect *c, char *s, HSPairs *p, HFields *next)
{
	HFields *hf;

	hf = halloc(c, sizeof *hf);
	hf->s = s;
	hf->params = p;
	hf->next = next;
	return hf;
}

HFields*
hrevhfields(HFields *hf)
{
	HFields *last, *next;

	last = nil;
	for(; hf != nil; hf = next){
		next = hf->next;
		hf->next = last;
		last = hf;
	}
	return last;
}

HContent*
hmkcontent(HConnect *c, char *generic, char *specific, HContent *next)
{
	HContent *ct;

	ct = halloc(c, sizeof(HContent));
	ct->generic = generic;
	ct->specific = specific;
	ct->next = next;
	ct->q = 1;
	ct->mxb = 0;
	return ct;
}