msg.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

#include <u.h>
#include <libc.h>
#include <bio.h>
#include <libsec.h>
#include <auth.h>
#include <fcall.h>
#include "imap4d.h"

static void	body64(int in, int out);
static void	bodystrip(int in, int out);
static void	cleanupHeader(Header *h);
static char	*domBang(char *s);
static void	freeMAddr(MAddr *a);
static void	freeMimeHdr(MimeHdr *mh);
static char	*headAddrSpec(char *e, char *w);
static MAddr	*headAddresses(void);
static MAddr	*headAddress(void);
static char	*headAtom(char *disallowed);
static int	headChar(int eat);
static char	*headDomain(char *e);
static MAddr	*headMAddr(MAddr *old);
static char	*headPhrase(char *e, char *w);
static char	*headQuoted(int start, int stop);
static char	*headSkipWhite(int);
static void	headSkip(void);
static char	*headSubDomain(void);
static char	*headText(void);
static void	headToEnd(void);
static char	*headWord(void);
static void	mimeDescription(Header *h);
static void	mimeDisposition(Header *h);
static void	mimeEncoding(Header *h);
static void	mimeId(Header *h);
static void	mimeLanguage(Header *h);
static void	mimeMd5(Header *h);
static MimeHdr	*mimeParams(void);
static void	mimeType(Header *h);
static MimeHdr	*mkMimeHdr(char *s, char *t, MimeHdr *next);
static void	msgAddDate(Msg *m);
static void	msgAddHead(Msg *m, char *head, char *body);
static int	msgBodySize(Msg *m);
static int	msgHeader(Msg *m, Header *h, char *file);
static long	msgReadFile(Msg *m, char *file, char **ss);
static int	msgUnix(Msg *m, int top);
static void	stripQuotes(char *q);
static MAddr	*unixFrom(char *s);


static char bogusBody[] = 
	"This message contains null characters, so it cannot be displayed correctly.\r\n"
	"Most likely you were sent a bogus message or a binary file.\r\n"
	"\r\n"
	"Each of the following attachments has a different version of the message.\r\n"
	"The first is inlined with all non-printable characters stripped.\r\n"
	"The second contains the message as it was stored in your mailbox.\r\n"
	"The third has the initial header stripped.\r\n";

static char bogusMimeText[] =
	"Content-Disposition: inline\r\n"
	"Content-Type: text/plain; charset=\"US-ASCII\"\r\n"
	"Content-Transfer-Encoding: 7bit\r\n";

static char bogusMimeBinary[] =
	"Content-Disposition: attachment\r\n"
	"Content-Type: application/octet-stream\r\n"
	"Content-Transfer-Encoding: base64\r\n";

/*
 * stop list for header fields
 */
static char	*headFieldStop = ":";
static char	*mimeTokenStop = "()<>@,;:\\\"/[]?=";
static char	*headAtomStop = "()<>@,;:\\\".[]";
static uchar	*headStr;
static uchar	*lastWhite;

long
selectFields(char *dst, long n, char *hdr, SList *fields, int matches)
{
	SList *f;
	uchar *start;
	char *s;
	long m, nf;

	headStr = (uchar*)hdr;
	m = 0;
	for(;;){
		start = headStr;
		s = headAtom(headFieldStop);
		if(s == nil)
			break;
		headSkip();
		for(f = fields; f != nil; f = f->next){
			if(cistrcmp(s, f->s) == !matches){
				nf = headStr - start;
				if(m + nf > n)
					return 0;
				memmove(&dst[m], start, nf);
				m += nf;
			}
		}
		free(s);
	}
	if(m + 3 > n)
		return 0;
	dst[m++] = '\r';
	dst[m++] = '\n';
	dst[m] = '\0';
	return m;
}

void
freeMsg(Msg *m)
{
	Msg *k, *last;

	free(m->iBuf);
	freeMAddr(m->to);
	if(m->replyTo != m->from)
		freeMAddr(m->replyTo);
	if(m->sender != m->from)
		freeMAddr(m->sender);
	if(m->from != m->unixFrom)
		freeMAddr(m->from);
	freeMAddr(m->unixFrom);
	freeMAddr(m->cc);
	freeMAddr(m->bcc);
	free(m->unixDate);
	cleanupHeader(&m->head);
	cleanupHeader(&m->mime);
	for(k = m->kids; k != nil; ){
		last = k;
		k = k->next;
		freeMsg(last);
	}
	free(m);
}

ulong
msgSize(Msg *m)
{
	return m->head.size + m->size;
}

int
infoIsNil(char *s)
{
	return s == nil || s[0] == '\0';
}

char*
maddrStr(MAddr *a)
{
	char *host, *addr;
	int n;

	host = a->host;
	if(host == nil)
		host = "";
	n = strlen(a->box) + strlen(host) + 2;
	if(a->personal != nil)
		n += strlen(a->personal) + 3;
	addr = emalloc(n);
	if(a->personal != nil)
		snprint(addr, n, "%s <%s@%s>", a->personal, a->box, host);
	else
		snprint(addr, n, "%s@%s", a->box, host);
	return addr;
}

/*
 * return actual name of f in m's fs directory
 * this is special cased when opening m/rawbody, m/mimeheader, or m/rawheader,
 * if the message was corrupted.  in that case,
 * a temporary file is made to hold the base64 encoding of m/raw.
 */
int
msgFile(Msg *m, char *f)
{
	Msg *parent, *p;
	Dir d;
	Tm tm;
	char buf[64], nbuf[2];
	uchar dbuf[64];
	int i, n, fd, fd1, fd2;

	if(!m->bogus
	|| strcmp(f, "") != 0 && strcmp(f, "rawbody") != 0
	&& strcmp(f, "rawheader") != 0 && strcmp(f, "mimeheader") != 0
	&& strcmp(f, "info") != 0 && strcmp(f, "unixheader") != 0){
		if(strlen(f) > MsgNameLen)
			bye("internal error: msgFile name too long");
		strcpy(m->efs, f);
		return cdOpen(m->fsDir, m->fs, OREAD);
	}

	/*
	 * walk up the stupid runt message parts for non-multipart messages
	 */
	parent = m->parent;
	if(parent != nil && parent->parent != nil){
		m = parent;
		parent = m->parent;
	}
	p = m;
	if(parent != nil)
		p = parent;

	if(strcmp(f, "info") == 0 || strcmp(f, "unixheader") == 0){
		strcpy(p->efs, f);
		return cdOpen(p->fsDir, p->fs, OREAD);
	}

	fd = imapTmp();
	if(fd < 0)
		return -1;

	/*
	 * craft the message parts for bogus messages
	 */
	if(strcmp(f, "") == 0){
		/*
		 * make a fake directory for each kid
		 * all we care about is the name
		 */
		if(parent == nil){
			nulldir(&d);
			d.mode = DMDIR|0600;
			d.qid.type = QTDIR;
			d.name = nbuf;
			nbuf[1] = '\0';
			for(i = '1'; i <= '4'; i++){
				nbuf[0] = i;
				n = convD2M(&d, dbuf, sizeof(dbuf));
				if(n <= BIT16SZ)
					fprint(2, "bad convD2M %d\n", n);
				write(fd, dbuf, n);
			}
		}
	}else if(strcmp(f, "mimeheader") == 0){
		if(parent != nil){
			switch(m->id){
			case 1:
			case 2:
				fprint(fd, "%s", bogusMimeText);
				break;
			case 3:
			case 4:
				fprint(fd, "%s", bogusMimeBinary);
				break;
			}
		}
	}else if(strcmp(f, "rawheader") == 0){
		if(parent == nil){
			date2tm(&tm, m->unixDate);
			rfc822date(buf, sizeof(buf), &tm);
			fprint(fd,
				"Date: %s\r\n"
				"From: imap4 daemon <%s@%s>\r\n"
				"To: <%s@%s>\r\n"
				"Subject: This message was illegal or corrupted\r\n"
				"MIME-Version: 1.0\r\n"
				"Content-Type: multipart/mixed;\r\n\tboundary=\"upas-%s\"\r\n",
					buf, username, site, username, site, m->info[IDigest]);
		}
	}else if(strcmp(f, "rawbody") == 0){
		fd1 = msgFile(p, "raw");
		strcpy(p->efs, "rawbody");
		fd2 = cdOpen(p->fsDir, p->fs, OREAD);
		if(fd1 < 0 || fd2 < 0){
			close(fd);
			close(fd1);
			close(fd2);
			return -1;
		}
		if(parent == nil){
			fprint(fd,
				"This is a multi-part message in MIME format.\r\n"
				"--upas-%s\r\n"
				"%s"
				"\r\n"
				"%s"
				"\r\n",
					m->info[IDigest], bogusMimeText, bogusBody);

			fprint(fd,
				"--upas-%s\r\n"
				"%s"
				"\r\n",
					m->info[IDigest], bogusMimeText);
			bodystrip(fd1, fd);

			fprint(fd,
				"--upas-%s\r\n"
				"%s"
				"\r\n",
					m->info[IDigest], bogusMimeBinary);
			seek(fd1, 0, 0);
			body64(fd1, fd);

			fprint(fd,
				"--upas-%s\r\n"
				"%s"
				"\r\n",
					m->info[IDigest], bogusMimeBinary);
			body64(fd2, fd);

			fprint(fd, "--upas-%s--\r\n", m->info[IDigest]);
		}else{
			switch(m->id){
			case 1:
				fprint(fd, "%s", bogusBody);
				break;
			case 2:
				bodystrip(fd1, fd);
				break;
			case 3:
				body64(fd1, fd);
				break;
			case 4:
				body64(fd2, fd);
				break;
			}
		}
		close(fd1);
		close(fd2);
	}
	seek(fd, 0, 0);
	return fd;
}

int
msgIsMulti(Header *h)
{
	return h->type != nil && cistrcmp("multipart", h->type->s) == 0;
}

int
msgIsRfc822(Header *h)
{
	return h->type != nil && cistrcmp("message", h->type->s) == 0 && cistrcmp("rfc822", h->type->t) == 0;
}

/*
 * check if a message has been deleted by someone else
 */
void
msgDead(Msg *m)
{
	if(m->expunged)
		return;
	*m->efs = '\0';
	if(!cdExists(m->fsDir, m->fs))
		m->expunged = 1;
}

/*
 * make sure the message has valid associated info
 * used for ISubject, IDigest, IInReplyTo, IMessageId.
 */
int
msgInfo(Msg *m)
{
	char *s;
	int i;

	if(m->info[0] != nil)
		return 1;

	i = msgReadFile(m, "info", &m->iBuf);
	if(i < 0)
		return 0;

	s = m->iBuf;
	for(i = 0; i < IMax; i++){
		m->info[i] = s;
		s = strchr(s, '\n');
		if(s == nil)
			break;
		*s++ = '\0';
	}
	for(; i < IMax; i++)
		m->info[i] = nil;

	for(i = 0; i < IMax; i++)
		if(infoIsNil(m->info[i]))
			m->info[i] = nil;

	return 1;
}

/*
 * make sure the message has valid mime structure
 * and sub-messages
 */
int
msgStruct(Msg *m, int top)
{
	Msg *k, head, *last;
	Dir *d;
	char *s;
	ulong max, id;
	int i, nd, fd, ns;

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

	if(m->expunged
	|| !msgInfo(m)
	|| !msgUnix(m, top)
	|| !msgBodySize(m)
	|| !msgHeader(m, &m->mime, "mimeheader")
	|| (top || msgIsRfc822(&m->mime) || msgIsMulti(&m->mime)) && !msgHeader(m, &m->head, "rawheader")){
		if(top && m->bogus && !(m->bogus & BogusTried)){
			m->bogus |= BogusTried;
			return msgStruct(m, top);
		}
		msgDead(m);
		return 0;
	}

	/*
	 * if a message has no kids, it has a kid which is just the body of the real message
	 */
	if(!msgIsMulti(&m->head) && !msgIsMulti(&m->mime) && !msgIsRfc822(&m->head) && !msgIsRfc822(&m->mime)){
		k = MKZ(Msg);
		k->id = 1;
		k->fsDir = m->fsDir;
		k->bogus = m->bogus;
		k->parent = m->parent;
		ns = m->efs - m->fs;
		k->fs = emalloc(ns + (MsgNameLen + 1));
		memmove(k->fs, m->fs, ns);
		k->efs = k->fs + ns;
		*k->efs = '\0';
		k->size = m->size;
		m->kids = k;
		return 1;
	}

	/*
	 * read in all child messages messages
	 */
	fd = msgFile(m, "");
	if(fd < 0){
		msgDead(m);
		return 0;
	}

	max = 0;
	head.next = nil;
	last = &head;
	while((nd = dirread(fd, &d)) > 0){
		for(i = 0; i < nd; i++){
			s = d[i].name;
			id = strtol(s, &s, 10);
			if(id <= max || *s != '\0'
			|| (d[i].mode & DMDIR) != DMDIR)
				continue;

			max = id;

			k = MKZ(Msg);
			k->id = id;
			k->fsDir = m->fsDir;
			k->bogus = m->bogus;
			k->parent = m;
			ns = strlen(m->fs);
			k->fs = emalloc(ns + 2 * (MsgNameLen + 1));
			k->efs = seprint(k->fs, k->fs + ns + (MsgNameLen + 1), "%s%lud/", m->fs, id);
			k->prev = last;
			k->size = ~0UL;
			k->lines = ~0UL;
			last->next = k;
			last = k;
		}
	}
	close(fd);
	m->kids = head.next;

	/*
	 * if kids fail, just whack them
	 */
	top = top && (msgIsRfc822(&m->head) || msgIsMulti(&m->head));
	for(k = m->kids; k != nil; k = k->next){
		if(!msgStruct(k, top)){
			for(k = m->kids; k != nil; ){
				last = k;
				k = k->next;
				freeMsg(last);
			}
			m->kids = nil;
			break;
		}
	}
	return 1;
}

static long
msgReadFile(Msg *m, char *file, char **ss)
{
	Dir *d;
	char *s, buf[BufSize];
	vlong length;
	long n, nn;
	int fd;

	fd = msgFile(m, file);
	if(fd < 0){
		msgDead(m);
		return -1;
	}

	n = read(fd, buf, BufSize);
	if(n < BufSize){
		close(fd);
		if(n < 0){
			*ss = nil;
			return -1;
		}
		s = emalloc(n + 1);
		memmove(s, buf, n);
		s[n] = '\0';
		*ss = s;
		return n;
	}

	d = dirfstat(fd);
	if(d == nil){
		close(fd);
		return -1;
	}
	length = d->length;
	free(d);
	nn = length;
	s = emalloc(nn + 1);
	memmove(s, buf, n);
	if(nn > n)
		nn = readn(fd, s+n, nn-n) + n;
	close(fd);
	if(nn != length){
		free(s);
		return -1;
	}
	s[nn] = '\0';
	*ss = s;
	return nn;
}

static void
freeMAddr(MAddr *a)
{
	MAddr *p;

	while(a != nil){
		p = a;
		a = a->next;
		free(p->personal);
		free(p->box);
		free(p->host);
		free(p);
	}
}

/*
 * the message is corrupted or illegal.
 * reset message fields.  msgStruct will reparse the message,
 * relying on msgFile to make up corrected body parts.
 */
static int
msgBogus(Msg *m, int flags)
{
	if(!(m->bogus & flags))
		m->bogus |= flags;
	m->lines = ~0;
	free(m->head.buf);
	free(m->mime.buf);
	memset(&m->head, 0, sizeof(Header));
	memset(&m->mime, 0, sizeof(Header));
	return 0;
}

/*
 *  stolen from upas/marshal; base64 encodes from one fd to another.
 *
 *  the size of buf is very important to enc64.  Anything other than
 *  a multiple of 3 will cause enc64 to output a termination sequence.
 *  To ensure that a full buf corresponds to a multiple of complete lines,
 *  we make buf a multiple of 3*18 since that's how many enc64 sticks on
 *  a single line.  This avoids short lines in the output which is pleasing
 *  but not necessary.
 */
static int
enc64x18(char *out, int lim, uchar *in, int n)