msg.c

来自「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」· C语言 代码 · 共 1,781 行 · 第 1/3 页

{
	int m, mm, nn;

	nn = 0;
	for(; n > 0; n -= m){
		m = 18 * 3;
		if(m > n)
			m = n;
		mm = enc64(out, lim - nn, in, m);
		in += m;
		out += mm;
		*out++ = '\r';
		*out++ = '\n';
		nn += mm + 2;
	}
	return nn;
}

static void
body64(int in, int out)
{
	uchar buf[3*18*54];
	char obuf[3*18*54*2];
	int m, n;

	for(;;){
		n = read(in, buf, sizeof(buf));
		if(n < 0)
			return;
		if(n == 0)
			break;
		m = enc64x18(obuf, sizeof(obuf), buf, n);
		if(write(out, obuf, m) < 0)
			return;
	}
}

/*
 * strip all non-printable characters from a file
 */
static void
bodystrip(int in, int out)
{
	uchar buf[3*18*54];
	int m, n, i, c;

	for(;;){
		n = read(in, buf, sizeof(buf));
		if(n < 0)
			return;
		if(n == 0)
			break;
		m = 0;
		for(i = 0; i < n; i++){
			c = buf[i];
			if(c > 0x1f && c < 0x7f		/* normal characters */
			|| c >= 0x9 && c <= 0xd)	/* \t, \n, vertical tab, form feed, \r */
				buf[m++] = c;
		}

		if(m && write(out, buf, m) < 0)
			return;
	}
}

/*
 * read in the message body to count \n without a preceding \r
 */
static int
msgBodySize(Msg *m)
{
	Dir *d;
	char buf[BufSize + 2], *s, *se;
	vlong length;
	ulong size, lines, bad;
	int n, fd, c;

	if(m->lines != ~0UL)
		return 1;
	fd = msgFile(m, "rawbody");
	if(fd < 0)
		return 0;
	d = dirfstat(fd);
	if(d == nil){
		close(fd);
		return 0;
	}
	length = d->length;
	free(d);

	size = 0;
	lines = 0;
	bad = 0;
	buf[0] = ' ';
	for(;;){
		n = read(fd, &buf[1], BufSize);
		if(n <= 0)
			break;
		size += n;
		se = &buf[n + 1];
		for(s = &buf[1]; s < se; s++){
			c = *s;
			if(c == '\0'){
				close(fd);
				return msgBogus(m, BogusBody);
			}
			if(c != '\n')
				continue;
			if(s[-1] != '\r')
				bad++;
			lines++;
		}
		buf[0] = buf[n];
	}
	if(size != length)
		bye("bad length reading rawbody");
	size += bad;
	m->size = size;
	m->lines = lines;
	close(fd);
	return 1;
}

/*
 * retrieve information from the unixheader file
 */
static int
msgUnix(Msg *m, int top)
{
	Tm tm;
	char *s, *ss;

	if(m->unixDate != nil)
		return 1;

	if(!top){
bogus:
		m->unixDate = estrdup("");
		m->unixFrom = unixFrom(nil);
		return 1;
	}

	if(msgReadFile(m, "unixheader", &ss) < 0)
		return 0;
	s = ss;
	s = strchr(s, ' ');
	if(s == nil){
		free(ss);
		goto bogus;
	}
	s++;
	m->unixFrom = unixFrom(s);
	s = (char*)headStr;
	if(date2tm(&tm, s) == nil)
		s = m->info[IUnixDate];
	if(s == nil){
		free(ss);
		goto bogus;
	}
	m->unixDate = estrdup(s);
	free(ss);
	return 1;
}

/*
 * parse the address in the unix header
 * last line of defence, so must return something
 */
static MAddr *
unixFrom(char *s)
{
	MAddr *a;
	char *e, *t;

	if(s == nil)
		return nil;
	headStr = (uchar*)s;
	t = emalloc(strlen(s) + 2);
	e = headAddrSpec(t, nil);
	if(e == nil)
		a = nil;
	else{
		if(*e != '\0')
			*e++ = '\0';
		else
			e = site;
		a = MKZ(MAddr);

		a->box = estrdup(t);
		a->host = estrdup(e);
	}
	free(t);
	return a;
}

/*
 * read in the entire header,
 * and parse out any existing mime headers
 */
static int
msgHeader(Msg *m, Header *h, char *file)
{
	char *s, *ss, *t, *te;
	ulong lines, n, nn;
	long ns;
	int dated, c;

	if(h->buf != nil)
		return 1;

	ns = msgReadFile(m, file, &ss);
	if(ns < 0)
		return 0;
	s = ss;
	n = ns;

	/*
	 * count lines ending with \n and \r\n
	 * add an extra line at the end, since upas/fs headers
	 * don't have a terminating \r\n
	 */
	lines = 1;
	te = s + ns;
	for(t = s; t < te; t++){
		c = *t;
		if(c == '\0')
			return msgBogus(m, BogusHeader);
		if(c != '\n')
			continue;
		if(t == s || t[-1] != '\r')
			n++;
		lines++;
	}
	if(t > s && t[-1] != '\n'){
		if(t[-1] != '\r')
			n++;
		n++;
	}
	n += 2;
	h->buf = emalloc(n + 1);
	h->size = n;
	h->lines = lines;

	/*
	 * make sure all headers end in \r\n
	 */
	nn = 0;
	for(t = s; t < te; t++){
		c = *t;
		if(c == '\n'){
			if(!nn || h->buf[nn - 1] != '\r')
				h->buf[nn++] = '\r';
			lines++;
		}
		h->buf[nn++] = c;
	}
	if(nn && h->buf[nn-1] != '\n'){
		if(h->buf[nn-1] != '\r')
			h->buf[nn++] = '\r';
		h->buf[nn++] = '\n';
	}
	h->buf[nn++] = '\r';
	h->buf[nn++] = '\n';
	h->buf[nn] = '\0';
	if(nn != n)
		bye("misconverted header %d %d", nn, n);
	free(s);

	/*
	 * and parse some mime headers
	 */
	headStr = (uchar*)h->buf;
	dated = 0;
	while(s = headAtom(headFieldStop)){
		if(cistrcmp(s, "content-type") == 0)
			mimeType(h);
		else if(cistrcmp(s, "content-transfer-encoding") == 0)
			mimeEncoding(h);
		else if(cistrcmp(s, "content-id") == 0)
			mimeId(h);
		else if(cistrcmp(s, "content-description") == 0)
			mimeDescription(h);
		else if(cistrcmp(s, "content-disposition") == 0)
			mimeDisposition(h);
		else if(cistrcmp(s, "content-md5") == 0)
			mimeMd5(h);
		else if(cistrcmp(s, "content-language") == 0)
			mimeLanguage(h);
		else if(h == &m->head && cistrcmp(s, "from") == 0)
			m->from = headMAddr(m->from);
		else if(h == &m->head && cistrcmp(s, "to") == 0)
			m->to = headMAddr(m->to);
		else if(h == &m->head && cistrcmp(s, "reply-to") == 0)
			m->replyTo = headMAddr(m->replyTo);
		else if(h == &m->head && cistrcmp(s, "sender") == 0)
			m->sender = headMAddr(m->sender);
		else if(h == &m->head && cistrcmp(s, "cc") == 0)
			m->cc = headMAddr(m->cc);
		else if(h == &m->head && cistrcmp(s, "bcc") == 0)
			m->bcc = headMAddr(m->bcc);
		else if(h == &m->head && cistrcmp(s, "date") == 0)
			dated = 1;
		headSkip();
		free(s);
	}

	if(h == &m->head){
		if(m->from == nil){
			m->from = m->unixFrom;
			if(m->from != nil){
				s = maddrStr(m->from);
				msgAddHead(m, "From", s);
				free(s);
			}
		}
		if(m->sender == nil)
			m->sender = m->from;
		if(m->replyTo == nil)
			m->replyTo = m->from;

		if(infoIsNil(m->info[IDate]))
			m->info[IDate] = m->unixDate;
		if(!dated && m->from != nil)
			msgAddDate(m);
	}
	return 1;
}

/*
 * prepend head: body to the cached header
 */
static void
msgAddHead(Msg *m, char *head, char *body)
{
	char *s;
	long size, n;

	n = strlen(head) + strlen(body) + 4;
	size = m->head.size + n;
	s = emalloc(size + 1);
	snprint(s, size + 1, "%s: %s\r\n%s", head, body, m->head.buf);
	free(m->head.buf);
	m->head.buf = s;
	m->head.size = size;
	m->head.lines++;
}

static void
msgAddDate(Msg *m)
{
	Tm tm;
	char buf[64];

	/* don't bother if we don't have a date */
	if(infoIsNil(m->info[IDate]))
		return;

	date2tm(&tm, m->info[IDate]);
	rfc822date(buf, sizeof(buf), &tm);
	msgAddHead(m, "Date", buf);
}

static MimeHdr*
mkMimeHdr(char *s, char *t, MimeHdr *next)
{
	MimeHdr *mh;

	mh = MK(MimeHdr);
	mh->s = s;
	mh->t = t;
	mh->next = next;
	return mh;
}

static void
freeMimeHdr(MimeHdr *mh)
{
	MimeHdr *last;

	while(mh != nil){
		last = mh;
		mh = mh->next;
		free(last->s);
		free(last->t);
		free(last);
	}
}

static void
cleanupHeader(Header *h)
{
	freeMimeHdr(h->type);
	freeMimeHdr(h->id);
	freeMimeHdr(h->description);
	freeMimeHdr(h->encoding);
	freeMimeHdr(h->md5);
	freeMimeHdr(h->disposition);
	freeMimeHdr(h->language);
}

/*
 * parser for rfc822 & mime header fields
 */

/*
 * type		: 'content-type' ':' token '/' token params
 */
static void
mimeType(Header *h)
{
	char *s, *t;

	if(headChar(1) != ':')
		return;
	s = headAtom(mimeTokenStop);
	if(s == nil || headChar(1) != '/'){
		free(s);
		return;
	}
	t = headAtom(mimeTokenStop);
	if(t == nil){
		free(s);
		return;
	}
	h->type = mkMimeHdr(s, t, mimeParams());
}

/*
 * params	:
 *		| params ';' token '=' token
 * 		| params ';' token '=' quoted-str
 */
static MimeHdr*
mimeParams(void)
{
	MimeHdr head, *last;
	char *s, *t;

	head.next = nil;
	last = &head;
	for(;;){
		if(headChar(1) != ';')
			break;
		s = headAtom(mimeTokenStop);
		if(s == nil || headChar(1) != '='){
			free(s);
			break;
		}
		if(headChar(0) == '"'){
			t = headQuoted('"', '"');
			stripQuotes(t);
		}else
			t = headAtom(mimeTokenStop);
		if(t == nil){
			free(s);
			break;
		}
		last->next = mkMimeHdr(s, t, nil);
		last = last->next;
	}
	return head.next;
}

/*
 * encoding	: 'content-transfer-encoding' ':' token
 */
static void
mimeEncoding(Header *h)
{
	char *s;

	if(headChar(1) != ':')
		return;
	s = headAtom(mimeTokenStop);
	if(s == nil)
		return;
	h->encoding = mkMimeHdr(s, nil, nil);
}

/*
 * mailaddr	: ':' addresses
 */
static MAddr*
headMAddr(MAddr *old)
{
	MAddr *a;

	if(headChar(1) != ':')
		return old;

	if(headChar(0) == '\n')
		return old;

	a = headAddresses();
	if(a == nil)
		return old;

	freeMAddr(old);
	return a;
}

/*
 * addresses	: address | addresses ',' address
 */
static MAddr*
headAddresses(void)
{
	MAddr *addr, *tail, *a;

	addr = headAddress();
	if(addr == nil)
		return nil;
	tail = addr;
	while(headChar(0) == ','){
		headChar(1);
		a = headAddress();
		if(a == nil){
			freeMAddr(addr);
			return nil;
		}
		tail->next = a;
		tail = a;
	}
	return addr;
}

/*
 * address	: mailbox | group
 * group	: phrase ':' mboxes ';' | phrase ':' ';'
 * mailbox	: addr-spec
 *		| optphrase '<' addr-spec '>'
 *		| optphrase '<' route ':' addr-spec '>'
 * optphrase	: | phrase
 * route	: '@' domain
 *		| route ',' '@' domain
 * personal names are the phrase before '<',
 * or a comment before or after a simple addr-spec
 */
static MAddr*
headAddress(void)
{
	MAddr *addr;
	uchar *hs;
	char *s, *e, *w, *personal;
	int c;

	s = emalloc(strlen((char*)headStr) + 2);
	e = s;
	personal = headSkipWhite(1);
	c = headChar(0);
	if(c == '<')
		w = nil;
	else{
		w = headWord();
		c = headChar(0);
	}
	if(c == '.' || c == '@' || c == ',' || c == '\n' || c == '\0'){
		lastWhite = headStr;
		e = headAddrSpec(s, w);
		if(personal == nil){
			hs = headStr;
			headStr = lastWhite;
			personal = headSkipWhite(1);
			headStr = hs;
		}
	}else{
		if(c != '<' || w != nil){
			free(personal);
			if(!headPhrase(e, w)){
				free(s);
				return nil;
			}

			/*
			 * ignore addresses with groups,
			 * so the only thing left if <
			 */
			c = headChar(1);
			if(c != '<'){
				free(s);
				return nil;
			}
			personal = estrdup(s);
		}else
			headChar(1);

		/*
		 * after this point, we need to free personal before returning.
		 * set e to nil to everything afterwards fails.
		 *
		 * ignore routes, they are useless, and heavily discouraged in rfc1123.
		 * imap4 reports them up to, but not including, the terminating :
		 */
		e = s;