mbox.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 "common.h"
#include <ctype.h>
#include <plumb.h>
#include <libsec.h>
#include "dat.h"

typedef struct Header Header;

struct Header {
	char *type;
	void (*f)(Message*, Header*, char*);
	int len;
};

/* headers */
static	void	ctype(Message*, Header*, char*);
static	void	cencoding(Message*, Header*, char*);
static	void	cdisposition(Message*, Header*, char*);
static	void	date822(Message*, Header*, char*);
static	void	from822(Message*, Header*, char*);
static	void	to822(Message*, Header*, char*);
static	void	sender822(Message*, Header*, char*);
static	void	replyto822(Message*, Header*, char*);
static	void	subject822(Message*, Header*, char*);
static	void	inreplyto822(Message*, Header*, char*);
static	void	cc822(Message*, Header*, char*);
static	void	bcc822(Message*, Header*, char*);
static	void	messageid822(Message*, Header*, char*);
static	void	mimeversion(Message*, Header*, char*);
static	void	nullsqueeze(Message*);
enum
{
	Mhead=	11,	/* offset of first mime header */
};

Header head[] =
{
	{ "date:", date822, },
	{ "from:", from822, },
	{ "to:", to822, },
	{ "sender:", sender822, },
	{ "reply-to:", replyto822, },
	{ "subject:", subject822, },
	{ "cc:", cc822, },
	{ "bcc:", bcc822, },
	{ "in-reply-to:", inreplyto822, },
	{ "mime-version:", mimeversion, },
	{ "message-id:", messageid822, },

[Mhead]	{ "content-type:", ctype, },
	{ "content-transfer-encoding:", cencoding, },
	{ "content-disposition:", cdisposition, },
	{ 0, },
};

static	void	fatal(char *fmt, ...);
static	void	initquoted(void);
static	void	startheader(Message*);
static	void	startbody(Message*);
static	char*	skipwhite(char*);
static	char*	skiptosemi(char*);
static	char*	getstring(char*, String*, int);
static	void	setfilename(Message*, char*);
static	char*	lowercase(char*);
static	int	is8bit(Message*);
static	int	headerline(char**, String*);
static	void	initheaders(void);
static void	parseattachments(Message*, Mailbox*);

int		debug;

char *Enotme = "path not served by this file server";

enum
{
	Chunksize = 1024,
};

Mailboxinit *boxinit[] = {
	imap4mbox,
	pop3mbox,
	plan9mbox,
};

char*
syncmbox(Mailbox *mb, int doplumb)
{
	return (*mb->sync)(mb, doplumb);
}

/* create a new mailbox */
char*
newmbox(char *path, char *name, int std)
{
	Mailbox *mb, **l;
	char *p, *rv;
	int i;

	initheaders();

	mb = emalloc(sizeof(*mb));
	strncpy(mb->path, path, sizeof(mb->path)-1);
	if(name == nil){
		p = strrchr(path, '/');
		if(p == nil)
			p = path;
		else
			p++;
		if(*p == 0){
			free(mb);
			return "bad mbox name";
		}
		strncpy(mb->name, p, sizeof(mb->name)-1);
	} else {
		strncpy(mb->name, name, sizeof(mb->name)-1);
	}

	rv = nil;
	// check for a mailbox type
	for(i=0; i<nelem(boxinit); i++)
		if((rv = (*boxinit[i])(mb, path)) != Enotme)
			break;
	if(i == nelem(boxinit)){
		free(mb);
		return "bad path";
	}

	// on error, give up
	if(rv){
		free(mb);
		return rv;
	}

	// make sure name isn't taken
	qlock(&mbllock);
	for(l = &mbl; *l != nil; l = &(*l)->next){
		if(strcmp((*l)->name, mb->name) == 0){
			if(strcmp(path, (*l)->path) == 0)
				rv = nil;
			else
				rv = "mbox name in use";
			if(mb->close)
				(*mb->close)(mb);
			free(mb);
			qunlock(&mbllock);
			return rv;
		}
	}

	// always try locking
	mb->dolock = 1;

	mb->refs = 1;
	mb->next = nil;
	mb->id = newid();
	mb->root = newmessage(nil);
	mb->std = std;
	*l = mb;
	qunlock(&mbllock);

	qlock(mb);
	if(mb->ctl){
		henter(PATH(mb->id, Qmbox), "ctl",
			(Qid){PATH(mb->id, Qmboxctl), 0, QTFILE}, nil, mb);
	}
	rv = syncmbox(mb, 0);
	qunlock(mb);

	return rv;
}

// close the named mailbox
void
freembox(char *name)
{
	Mailbox **l, *mb;

	qlock(&mbllock);
	for(l=&mbl; *l != nil; l=&(*l)->next){
		if(strcmp(name, (*l)->name) == 0){
			mb = *l;
			*l = mb->next;
			mboxdecref(mb);
			break;
		}
	}
	hfree(PATH(0, Qtop), name);
	qunlock(&mbllock);
}

static void
initheaders(void)
{
	Header *h;
	static int already;

	if(already)
		return;
	already = 1;

	for(h = head; h->type != nil; h++)
		h->len = strlen(h->type);
}

/*
 *  parse a Unix style header
 */
void
parseunix(Message *m)
{
	char *p;
	String *h;

	h = s_new();
	for(p = m->start + 5; *p && *p != '\r' && *p != '\n'; p++)
		s_putc(h, *p);
	s_terminate(h);
	s_restart(h);

	m->unixfrom = s_parse(h, s_reset(m->unixfrom));
	m->unixdate = s_append(s_reset(m->unixdate), h->ptr);

	s_free(h);
}

/*
 *  parse a message
 */
void
parseheaders(Message *m, int justmime, Mailbox *mb, int addfrom)
{
	String *hl;
	Header *h;
	char *p, *q;
	int i;

	if(m->whole == m->whole->whole){
		henter(PATH(mb->id, Qmbox), m->name,
			(Qid){PATH(m->id, Qdir), 0, QTDIR}, m, mb);
	} else {
		henter(PATH(m->whole->id, Qdir), m->name,
			(Qid){PATH(m->id, Qdir), 0, QTDIR}, m, mb);
	}
	for(i = 0; i < Qmax; i++)
		henter(PATH(m->id, Qdir), dirtab[i],
			(Qid){PATH(m->id, i), 0, QTFILE}, m, mb);

	// parse mime headers
	p = m->header;
	hl = s_new();
	while(headerline(&p, hl)){
		if(justmime)
			h = &head[Mhead];
		else
			h = head;
		for(; h->type; h++){
			if(cistrncmp(s_to_c(hl), h->type, h->len) == 0){
				(*h->f)(m, h, s_to_c(hl));
				break;
			}
		}
		s_reset(hl);
	}
	s_free(hl);

	// the blank line isn't really part of the body or header
	if(justmime){
		m->mhend = p;
		m->hend = m->header;
	} else {
		m->hend = p;
	}
	if(*p == '\n')
		p++;
	m->rbody = m->body = p;

	// if type is text, get any nulls out of the body.  This is
	// for the two seans and imap clients that get confused.
	if(strncmp(s_to_c(m->type), "text/", 5) == 0)
		nullsqueeze(m);

	//
	// cobble together Unix-style from line
	// for local mailbox messages, we end up recreating the
	// original header.
	// for pop3 messages, the best we can do is 
	// use the From: information and the RFC822 date.
	//
	if(m->unixdate == nil || strcmp(s_to_c(m->unixdate), "???") == 0
	|| strcmp(s_to_c(m->unixdate), "Thu Jan 1 00:00:00 GMT 1970") == 0){
		if(m->unixdate){
			s_free(m->unixdate);
			m->unixdate = nil;
		}
		// look for the date in the first Received: line.
		// it's likely to be the right time zone (it's
	 	// the local system) and in a convenient format.
		if(cistrncmp(m->header, "received:", 9)==0){
			if((q = strchr(m->header, ';')) != nil){
				p = q;
				while((p = strchr(p, '\n')) != nil){
					if(p[1] != ' ' && p[1] != '\t' && p[1] != '\n')
						break;
					p++;
				}
				if(p){
					*p = '\0';
					m->unixdate = date822tounix(q+1);
					*p = '\n';
				}
			}
		}

		// fall back on the rfc822 date	
		if(m->unixdate==nil && m->date822)
			m->unixdate = date822tounix(s_to_c(m->date822));
	}

	if(m->unixheader != nil)
		s_free(m->unixheader);

	// only fake header for top-level messages for pop3 and imap4
	// clients (those protocols don't include the unix header).
	// adding the unix header all the time screws up mime-attached
	// rfc822 messages.
	if(!addfrom && !m->unixfrom){
		m->unixheader = nil;
		return;
	}

	m->unixheader = s_copy("From ");
	if(m->unixfrom && strcmp(s_to_c(m->unixfrom), "???") != 0)
		s_append(m->unixheader, s_to_c(m->unixfrom));
	else if(m->from822)
		s_append(m->unixheader, s_to_c(m->from822));
	else
		s_append(m->unixheader, "???");

	s_append(m->unixheader, " ");
	if(m->unixdate)
		s_append(m->unixheader, s_to_c(m->unixdate));
	else
		s_append(m->unixheader, "Thu Jan  1 00:00:00 GMT 1970");

	s_append(m->unixheader, "\n");
}

String*
promote(String **sp)
{
	String *s;

	if(*sp != nil)
		s = s_clone(*sp);
	else
		s = nil;
	return s;
}

void
parsebody(Message *m, Mailbox *mb)
{
	Message *nm;

	// recurse
	if(strncmp(s_to_c(m->type), "multipart/", 10) == 0){
		parseattachments(m, mb);
	} else if(strcmp(s_to_c(m->type), "message/rfc822") == 0){
		decode(m);
		parseattachments(m, mb);
		nm = m->part;

		// promote headers
		if(m->replyto822 == nil && m->from822 == nil && m->sender822 == nil){
			m->from822 = promote(&nm->from822);
			m->to822 = promote(&nm->to822);
			m->date822 = promote(&nm->date822);
			m->sender822 = promote(&nm->sender822);
			m->replyto822 = promote(&nm->replyto822);
			m->subject822 = promote(&nm->subject822);
			m->unixdate = promote(&nm->unixdate);
		}
	}
}

void
parse(Message *m, int justmime, Mailbox *mb, int addfrom)
{
	parseheaders(m, justmime, mb, addfrom);
	parsebody(m, mb);
}

static void
parseattachments(Message *m, Mailbox *mb)
{
	Message *nm, **l;
	char *p, *x;

	// if there's a boundary, recurse...
	if(m->boundary != nil){
		p = m->body;
		nm = nil;
		l = &m->part;
		for(;;){
			x = strstr(p, s_to_c(m->boundary));

			/* no boundary, we're done */
			if(x == nil){
				if(nm != nil)
					nm->rbend = nm->bend = nm->end = m->bend;
				break;
			}

			/* boundary must be at the start of a line */
			if(x != m->body && *(x-1) != '\n'){
				p = x+1;
				continue;
			}

			if(nm != nil)
				nm->rbend = nm->bend = nm->end = x;
			x += strlen(s_to_c(m->boundary));

			/* is this the last part? ignore anything after it */
			if(strncmp(x, "--", 2) == 0)
				break;

			p = strchr(x, '\n');
			if(p == nil)
				break;
			nm = newmessage(m);
			nm->start = nm->header = nm->body = nm->rbody = ++p;
			nm->mheader = nm->header;
			*l = nm;
			l = &nm->next;
		}
		for(nm = m->part; nm != nil; nm = nm->next)
			parse(nm, 1, mb, 0);
		return;
	}

	// if we've got an rfc822 message, recurse...
	if(strcmp(s_to_c(m->type), "message/rfc822") == 0){
		nm = newmessage(m);
		m->part = nm;
		nm->start = nm->header = nm->body = nm->rbody = m->body;
		nm->end = nm->bend = nm->rbend = m->bend;
		parse(nm, 0, mb, 0);
	}
}

/*
 *  pick up a header line
 */
static int
headerline(char **pp, String *hl)
{
	char *p, *x;

	s_reset(hl);
	p = *pp;
	x = strpbrk(p, ":\n");
	if(x == nil || *x == '\n')
		return 0;
	for(;;){
		x = strchr(p, '\n');
		if(x == nil)
			x = p + strlen(p);
		s_nappend(hl, p, x-p);
		p = x;
		if(*p != '\n' || *++p != ' ' && *p != '\t')
			break;
		while(*p == ' ' || *p == '\t')
			p++;
		s_putc(hl, ' ');
	}
	*pp = p;
	return 1;
}

static String*
addr822(char *p)
{
	String *s, *list;
	int incomment, addrdone, inanticomment, quoted;
	int n;
	int c;

	list = s_new();
	s = s_new();
	quoted = incomment = addrdone = inanticomment = 0;
	n = 0;
	for(; *p; p++){
		c = *p;

		// whitespace is ignored
		if(!quoted && isspace(c) || c == '\r')
			continue;

		// strings are always treated as atoms
		if(!quoted && c == '"'){
			if(!addrdone && !incomment)
				s_putc(s, c);
			for(p++; *p; p++){
				if(!addrdone && !incomment)
					s_putc(s, *p);
				if(!quoted && *p == '"')
					break;
				if(*p == '\\')
					quoted = 1;
				else
					quoted = 0;
			}
			if(*p == 0)
				break;
			quoted = 0;
			continue;
		}

		// ignore everything in an expicit comment
		if(!quoted && c == '('){
			incomment = 1;
			continue;
		}
		if(incomment){
			if(!quoted && c == ')')
				incomment = 0;
			quoted = 0;
			continue;
		}

		// anticomments makes everything outside of them comments
		if(!quoted && c == '<' && !inanticomment){
			inanticomment = 1;
			s = s_reset(s);
			continue;
		}
		if(!quoted && c == '>' && inanticomment){
			addrdone = 1;
			inanticomment = 0;
			continue;
		}

		// commas separate addresses
		if(!quoted && c == ',' && !inanticomment){
			s_terminate(s);
			addrdone = 0;
			if(n++ != 0)
				s_append(list, " ");
			s_append(list, s_to_c(s));
			s = s_reset(s);
			continue;
		}

		// what's left is part of the address
		s_putc(s, c);

		// quoted characters are recognized only as characters
		if(c == '\\')
			quoted = 1;
		else
			quoted = 0;

	}

	if(*s_to_c(s) != 0){
		s_terminate(s);
		if(n++ != 0)
			s_append(list, " ");
		s_append(list, s_to_c(s));
	}
	s_free(s);

	if(n == 0){
		s_free(list);
		return nil;
	}
	return list;
}

static void
to822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->to822);
	m->to822 = addr822(p);
}

static void
cc822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->cc822);
	m->cc822 = addr822(p);
}

static void
bcc822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->bcc822);
	m->bcc822 = addr822(p);
}

static void
from822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->from822);
	m->from822 = addr822(p);
}

static void
sender822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->sender822);
	m->sender822 = addr822(p);
}

static void
replyto822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->replyto822);
	m->replyto822 = addr822(p);
}

static void
mimeversion(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	s_free(m->mimeversion);
	m->mimeversion = addr822(p);
}

static void
killtrailingwhite(char *p)
{
	char *e;

	e = p + strlen(p) - 1;
	while(e > p && isspace(*e))
		*e-- = 0;
}

static void
date822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	p = skipwhite(p);
	s_free(m->date822);
	m->date822 = s_copy(p);
	p = s_to_c(m->date822);
	killtrailingwhite(p);
}

static void
subject822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	p = skipwhite(p);
	s_free(m->subject822);
	m->subject822 = s_copy(p);
	p = s_to_c(m->subject822);
	killtrailingwhite(p);
}

static void
inreplyto822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	p = skipwhite(p);
	s_free(m->inreplyto822);
	m->inreplyto822 = s_copy(p);
	p = s_to_c(m->inreplyto822);
	killtrailingwhite(p);
}

static void
messageid822(Message *m, Header *h, char *p)
{
	p += strlen(h->type);
	p = skipwhite(p);
	s_free(m->messageid822);
	m->messageid822 = s_copy(p);
	p = s_to_c(m->messageid822);
	killtrailingwhite(p);
}

static int
isattribute(char **pp, char *attr)
{
	char *p;
	int n;

	n = strlen(attr);
	p = *pp;
	if(cistrncmp(p, attr, n) != 0)
		return 0;
	p += n;
	while(*p == ' ')
		p++;
	if(*p++ != '=')
		return 0;
	while(*p == ' ')
		p++;
	*pp = p;
	return 1;
}

static void
ctype(Message *m, Header *h, char *p)
{
	String *s;

	p += h->len;
	p = skipwhite(p);

	p = getstring(p, m->type, 1);
	
	while(*p){
		if(isattribute(&p, "boundary")){
			s = s_new();
			p = getstring(p, s, 0);
			m->boundary = s_reset(m->boundary);
			s_append(m->boundary, "--");
			s_append(m->boundary, s_to_c(s));
			s_free(s);
		} else if(cistrncmp(p, "multipart", 9) == 0){
			/*
			 *  the first unbounded part of a multipart message,
			 *  the preamble, is not displayed or saved
			 */
		} else if(isattribute(&p, "name")){
			if(m->filename == nil)
				setfilename(m, p);
		} else if(isattribute(&p, "charset")){
			p = getstring(p, s_reset(m->charset), 0);
		}
		
		p = skiptosemi(p);