vf.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

	if(np && np->boundary){
		cp = Brdline(&in, '\n');
		Bwrite(&out, cp, Blinelen(&in));
	}

fprint(2, "a %p\n", np);
	return np;
}

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;
}

/*
 *  parse content type header 
 */
static void
ctype(Part *p, Hdef *h, char *cp)
{
	String *s;

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

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

/*
 *  parse content encoding header 
 */
static void
cencoding(Part *m, Hdef *h, char *p)
{
	p += h->len;
	p = skipwhite(p);
	if(cistrncmp(p, "base64", 6) == 0)
		m->encoding = Ebase64;
	else if(cistrncmp(p, "quoted-printable", 16) == 0)
		m->encoding = Equoted;
}

/*
 *  parse content disposition header 
 */
static void
cdisposition(Part *p, Hdef *h, char *cp)
{
	cp += h->len;
	cp = skipwhite(cp);
	while(*cp){
		if(cistrncmp(cp, "inline", 6) == 0){
			p->disposition = Dinline;
		} else if(cistrncmp(cp, "attachment", 10) == 0){
			p->disposition = Dfile;
		} else if(cistrncmp(cp, "filename=", 9) == 0){
			cp += 9;
			setfilename(p, cp);
		}
		cp = skiptosemi(cp);
	}

}

static void
setfilename(Part *p, char *name)
{
	if(p->filename == nil)
		p->filename = s_new();
	getstring(name, s_reset(p->filename), 0);
	p->filename = tokenconvert(p->filename);
	p->badfile = badfile(s_to_c(p->filename));
}

static char*
skipwhite(char *p)
{
	while(isspace(*p))
		p++;
	return p;
}

static char*
skiptosemi(char *p)
{
	while(*p && *p != ';')
		p++;
	while(*p == ';' || isspace(*p))
		p++;
	return p;
}

/*
 *  parse a possibly "'d string from a header.  A
 *  ';' terminates the string.
 */
static char*
getstring(char *p, String *s, int dolower)
{
	s = s_reset(s);
	p = skipwhite(p);
	if(*p == '"'){
		p++;
		for(;*p && *p != '"'; p++)
			if(dolower)
				s_putc(s, tolower(*p));
			else
				s_putc(s, *p);
		if(*p == '"')
			p++;
		s_terminate(s);

		return p;
	}

	for(; *p && !isspace(*p) && *p != ';'; p++)
		if(dolower)
			s_putc(s, tolower(*p));
		else
			s_putc(s, *p);
	s_terminate(s);

	return p;
}

static void
init_hdefs(void)
{
	Hdef *hd;
	static int already;

	if(already)
		return;
	already = 1;

	for(hd = hdefs; hd->type != nil; hd++)
		hd->len = strlen(hd->type);
}

/*
 *  create a new boundary
 */
static String*
mkboundary(void)
{
	char buf[32];
	int i;
	static int already;

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

/*
 *  skip blank lines till header
 */
static void
passnotheader(void)
{
	char *cp;
	int i, n;

	while((cp = Brdline(&in, '\n')) != nil){
		n = Blinelen(&in);
		for(i = 0; i < n-1; i++)
			if(cp[i] != ' ' && cp[i] != '\t' && cp[i] != '\r'){
				Bseek(&in, -n, 1);
				return;
			}
		Bwrite(&out, cp, n);
	}
}

/*
 *  pass unix header lines
 */
static void
passunixheader(void)
{
	char *p;
	int n;

	while((p = Brdline(&in, '\n')) != nil){
		n = Blinelen(&in);
		if(strncmp(p, "From ", 5) != 0){
			Bseek(&in, -n, 1);
			break;
		}
		Bwrite(&out, p, n);
	}
}

/*
 *  Read mime types
 */
static void
readmtypes(void)
{
	Biobuf *b;
	char *p;
	char *f[6];
	Mtype *m;
	Mtype **l;

	b = Bopen("/sys/lib/mimetype", OREAD);
	if(b == nil)
		return;

	l = &mtypes;
	while((p = Brdline(b, '\n')) != nil){
		if(*p == '#')
			continue;
		p[Blinelen(b)-1] = 0;
		if(tokenize(p, f, nelem(f)) < 5)
			continue;
		m = mallocz(sizeof *m, 1);
		if(m == nil)
			goto err;
		m->ext = strdup(f[0]);
		if(m->ext == 0)
			goto err;
		m->gtype = strdup(f[1]);
		if(m->gtype == 0)
			goto err;
		m->stype = strdup(f[2]);
		if(m->stype == 0)
			goto err;
		m->class = *f[4];
		*l = m;
		l = &(m->next);
	}
	Bterm(b);
	return;
err:
	if(m == nil)
		return;
	free(m->ext);
	free(m->gtype);
	free(m->stype);
	free(m);
	Bterm(b);
}

/*
 *  if the class is 'm' or 'y', accept it
 *  if the class is 'p' check a previous extension
 *  otherwise, filename is bad
 */
static int
badfile(char *name)
{
	char *p;
	Mtype *m;
	int rv;

	p = strrchr(name, '.');
	if(p == nil)
		return 0;

	for(m = mtypes; m != nil; m = m->next)
		if(cistrcmp(p, m->ext) == 0){
			switch(m->class){
			case 'm':
			case 'y':
				return 0;
			case 'p':
				*p = 0;
				rv = badfile(name);
				*p = '.';
				return rv;
			case 'r':
				return 2;
			}
		}
	return 1;
}

/*
 *  if the class is 'm' or 'y' or 'p', accept it
 *  otherwise, filename is bad
 */
static int
badtype(char *type)
{
	Mtype *m;
	char *s, *fix;
	int rv = 1;

	fix = s = strchr(type, '/');
	if(s != nil)
		*s++ = 0;
	else
		s = "-";

	for(m = mtypes; m != nil; m = m->next){
		if(cistrcmp(type, m->gtype) != 0)
			continue;
		if(cistrcmp(s, m->stype) != 0)
			continue;
		switch(m->class){
		case 'y':
		case 'p':
		case 'm':
			rv = 0;
			break;
		}
		break;
	}

	if(fix != nil)
		*fix = '/';
	return rv;
}

/* rfc2047 non-ascii */
typedef struct Charset Charset;
struct Charset {
	char *name;
	int len;
	int convert;
} charsets[] =
{
	{ "us-ascii",		8,	1, },
	{ "utf-8",		5,	0, },
	{ "iso-8859-1",		10,	1, },
};

/*
 *  convert to UTF if need be
 */
static String*
tokenconvert(String *t)
{
	String *s;
	char decoded[1024];
	char utfbuf[2*1024];
	int i, len;
	char *e;
	char *token;

	token = s_to_c(t);
	len = s_len(t);

	if(token[0] != '=' || token[1] != '?' ||
	   token[len-2] != '?' || token[len-1] != '=')
		goto err;
	e = token+len-2;
	token += 2;

	// bail if we don't understand the character set
	for(i = 0; i < nelem(charsets); i++)
		if(cistrncmp(charsets[i].name, token, charsets[i].len) == 0)
		if(token[charsets[i].len] == '?'){
			token += charsets[i].len + 1;
			break;
		}
	if(i >= nelem(charsets))
		goto err;

	// bail if it doesn't fit 
	if(strlen(token) > sizeof(decoded)-1)
		goto err;

	// bail if we don't understand the encoding
	if(cistrncmp(token, "b?", 2) == 0){
		token += 2;
		len = dec64((uchar*)decoded, sizeof(decoded), token, e-token);
		decoded[len] = 0;
	} else if(cistrncmp(token, "q?", 2) == 0){
		token += 2;
		len = decquoted(decoded, token, e);
		if(len > 0 && decoded[len-1] == '\n')
			len--;
		decoded[len] = 0;
	} else
		goto err;

	s = nil;
	switch(charsets[i].convert){
	case 0:
		s = s_copy(decoded);
		break;
	case 1:
		s = s_new();
		latin1toutf(utfbuf, decoded, decoded+len);
		s_append(s, utfbuf);
		break;
	}

	return s;
err:
	return s_clone(t);
}

/*
 *  decode quoted 
 */
enum
{
	Self=	1,
	Hex=	2,
};
uchar	tableqp[256];

static void
initquoted(void)
{
	int c;

	memset(tableqp, 0, 256);
	for(c = ' '; c <= '<'; c++)
		tableqp[c] = Self;
	for(c = '>'; c <= '~'; c++)
		tableqp[c] = Self;
	tableqp['\t'] = Self;
	tableqp['='] = Hex;
}

static int
hex2int(int x)
{
	if(x >= '0' && x <= '9')
		return x - '0';
	if(x >= 'A' && x <= 'F')
		return (x - 'A') + 10;
	if(x >= 'a' && x <= 'f')
		return (x - 'a') + 10;
	return 0;
}

static char*
decquotedline(char *out, char *in, char *e)
{
	int c, soft;

	/* dump trailing white space */
	while(e >= in && (*e == ' ' || *e == '\t' || *e == '\r' || *e == '\n'))
		e--;

	/* trailing '=' means no newline */
	if(*e == '='){
		soft = 1;
		e--;
	} else
		soft = 0;

	while(in <= e){
		c = (*in++) & 0xff;
		switch(tableqp[c]){
		case Self:
			*out++ = c;
			break;
		case Hex:
			c = hex2int(*in++)<<4;
			c |= hex2int(*in++);
			*out++ = c;
			break;
		}
	}
	if(!soft)
		*out++ = '\n';
	*out = 0;

	return out;
}

static int
decquoted(char *out, char *in, char *e)
{
	char *p, *nl;

	if(tableqp[' '] == 0)
		initquoted();

	p = out;
	while((nl = strchr(in, '\n')) != nil && nl < e){
		p = decquotedline(p, in, nl);
		in = nl + 1;
	}
	if(in < e)
		p = decquotedline(p, in, e-1);

	// make sure we end with a new line
	if(*(p-1) != '\n'){
		*p++ = '\n';
		*p = 0;
	}

	return p - out;
}

/* translate latin1 directly since it fits neatly in utf */
static int
latin1toutf(char *out, char *in, char *e)
{
	Rune r;
	char *p;

	p = out;
	for(; in < e; in++){
		r = (*in) & 0xff;
		p += runetochar(p, &r);
	}
	*p = 0;
	return p - out;
}