url.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

	 * a NUL then someone is unescaping too many times.
	 */
	if(strstr(url, "%00")){
		werrstr("escaped NUL in URI");
		return -1;
	}

	m[0].sp = m[0].ep = nil;
	t = &retab[REsplit];
	if(!regx(t->prog, url, m, t->size)){
		werrstr("malformed URI: %q", url);
		return -1;
	}

	su->url.s = m[0].sp;
	su->url.e = m[0].ep;
	su->scheme.s = m[t->ind[0]].sp;
	su->scheme.e = m[t->ind[0]].ep;
	su->authority.s = m[t->ind[1]].sp;
	su->authority.e = m[t->ind[1]].ep;
	su->path.s = m[t->ind[2]].sp;
	su->path.e = m[t->ind[2]].ep;
	su->query.s = m[t->ind[3]].sp;
	su->query.e = m[t->ind[3]].ep;
	su->fragment.s = m[t->ind[4]].sp;
	su->fragment.e = m[t->ind[4]].ep;

	if(urldebug)
		fprint(2, "split url %s into %.*q %.*q %.*q %.*q %.*q %.*q\n",
			url,
			su->url.s ? utfnlen(su->url.s, su->url.e-su->url.s) : 10, su->url.s ? su->url.s : "",
			su->scheme.s ? utfnlen(su->scheme.s, su->scheme.e-su->scheme.s) : 10, su->scheme.s ? su->scheme.s : "",
			su->authority.s ? utfnlen(su->authority.s, su->authority.e-su->authority.s) : 10, su->authority.s ? su->authority.s : "",
			su->path.s ? utfnlen(su->path.s, su->path.e-su->path.s) : 10, su->path.s ? su->path.s : "",
			su->query.s ? utfnlen(su->query.s, su->query.e-su->query.s) : 10, su->query.s ? su->query.s : "",
			su->fragment.s ? utfnlen(su->fragment.s, su->fragment.e-su->fragment.s) : 10, su->fragment.s ? su->fragment.s : "");

	return 0;
}

static int
parse_scheme(SplitUrl *su, Url *u)
{
	if(su->scheme.s == nil){
		werrstr("missing scheme");
		return -1;
	}
	u->scheme = estredup(su->scheme.s, su->scheme.e);
	strlower(u->scheme);

	if(!ismatch(REscheme, u->scheme, "scheme"))
		return -1;

	u->ischeme = ischeme(u->scheme);
	if(urldebug)
		fprint(2, "parse_scheme %s => %d\n", u->scheme, u->ischeme);
	return 0;
}

static int
parse_unknown_part(SplitUrl *su, Url *u)
{
	char *s, *e;

	assert(u->ischeme == USunknown);
	assert(su->scheme.e[0] == ':');

	s = su->scheme.e+1;
	if(su->fragment.s){
		e = su->fragment.s-1;
		assert(*e == '#');
	}else
		e = s+strlen(s);

	u->schemedata = estredup(s, e);
	if(!ismatch(REunknowndata, u->schemedata, "unknown scheme data"))
		return -1;
	return 0;
}

static int
parse_userinfo(char *s, char *e, Url *u)
{
	Resub m[MaxResub];
	Retab *t;

	m[0].sp = s;
	m[0].ep = e;
	t = &retab[REuserinfo];
	if(!regx(t->prog, nil, m, t->size)){
		werrstr("malformed userinfo: %.*q", utfnlen(s, e-s), s);
		return -1;
	}
	if(m[t->ind[0]].sp)
		u->user = estredup(m[t->ind[0]].sp, m[t->ind[0]].ep);
	if(m[t->ind[1]].sp)
		u->user = estredup(m[t->ind[1]].sp, m[t->ind[1]].ep);
	return 0;
}

static int
parse_host(char *s, char *e, Url *u)
{
	Resub m[MaxResub];
	Retab *t;

	m[0].sp = s;
	m[0].ep = e;
	t = &retab[REhost];
	if(!regx(t->prog, nil, m, t->size)){
		werrstr("malformed host: %.*q", utfnlen(s, e-s), s);
		return -1;
	}

	assert(m[t->ind[0]].sp || m[t->ind[1]].sp);

	if(m[t->ind[0]].sp)	/* regular */
		u->host = estredup(m[t->ind[0]].sp, m[t->ind[0]].ep);
	else
		u->host = estredup(m[t->ind[1]].sp, m[t->ind[1]].ep);
	return 0;
}

static int
parse_authority(SplitUrl *su, Url *u)
{
	Resub m[MaxResub];
	Retab *t;

	if(su->authority.s == nil)
		return 0;

	u->authority = estredup(su->authority.s, su->authority.e);
	m[0].sp = m[0].ep = nil;
	t = &retab[REauthority];
	if(!regx(t->prog, u->authority, m, t->size)){
		werrstr("malformed authority: %q", u->authority);
		return -1;
	}

	if(m[t->ind[0]].sp)
		if(parse_userinfo(m[t->ind[0]].sp, m[t->ind[0]].ep, u) < 0)
			return -1;
	if(m[t->ind[1]].sp)
		if(parse_host(m[t->ind[1]].sp, m[t->ind[1]].ep, u) < 0)
			return -1;
	if(m[t->ind[2]].sp)
		u->port = estredup(m[t->ind[2]].sp, m[t->ind[2]].ep);

	return 0;
}

static int
parse_abspath(SplitUrl *su, Url *u)
{
	if(su->path.s == nil)
		return 0;
	u->path = estredup(su->path.s, su->path.e);
	if(!ismatch(REabspath, u->path, "absolute path"))
		return -1;
	return 0;
}

static int
parse_query(SplitUrl *su, Url *u)
{
	if(su->query.s == nil)
		return 0;
	u->query = estredup(su->query.s, su->query.e);
	if(!ismatch(REquery, u->query, "query"))
		return -1;
	return 0;
}

static int
parse_fragment(SplitUrl *su, Url *u)
{
	if(su->fragment.s == nil)
		return 0;
	u->fragment = estredup(su->fragment.s, su->fragment.e);
	if(!ismatch(REfragment, u->fragment, "fragment"))
		return -1;
	return 0;
}

static int
postparse_http(Url *u)
{
	u->open = httpopen;
	u->read = httpread;
	u->close = httpclose;

	if(u->authority==nil){
		werrstr("missing authority (hostname, port, etc.)");
		return -1;
	}
	if(u->host == nil){
		werrstr("missing host specification");
		return -1;
	}

	if(u->path == nil){
		u->http.page_spec = estrdup("/");
		return 0;
	}

	if(!ismatch(REhttppath, u->path, "http path"))
		return -1;
	if(u->query){
		u->http.page_spec = emalloc(strlen(u->path)+1+strlen(u->query)+1);
		strcpy(u->http.page_spec, u->path);
		strcat(u->http.page_spec, "?");
		strcat(u->http.page_spec, u->query);
	}else
		u->http.page_spec = estrdup(u->path);

	return 0;
}

static int
postparse_ftp(Url *u)
{
	Resub m[MaxResub];
	Retab *t;

	if(u->authority==nil){
		werrstr("missing authority (hostname, port, etc.)");
		return -1;
	}
	if(u->query){
		werrstr("unexpected \"?query\" in ftp path");
		return -1;
	}
	if(u->host == nil){
		werrstr("missing host specification");
		return -1;
	}

	if(u->path == nil){
		u->ftp.path_spec = estrdup("/");
		return 0;
	}

	m[0].sp = m[0].ep = nil;
	t = &retab[REftppath];
	if(!regx(t->prog, u->path, m, t->size)){
		werrstr("malformed ftp path: %q", u->path);
		return -1;
	}

	if(m[t->ind[0]].sp){
		u->ftp.path_spec = estredup(m[t->ind[0]].sp, m[t->ind[0]].ep);
		if(strchr(u->ftp.path_spec, ';')){
			werrstr("unexpected \";param\" in ftp path");
			return -1;
		}
	}else
		u->ftp.path_spec = estrdup("/");

	if(m[t->ind[1]].sp){
		u->ftp.type = estredup(m[t->ind[1]].sp, m[t->ind[1]].ep);
		strlower(u->ftp.type);
	}
	return 0;
}

static int
postparse_file(Url *u)
{
	if(u->user || u->passwd){
		werrstr("user information not valid with file scheme");
		return -1;
	}
	if(u->query){
		werrstr("unexpected \"?query\" in file path");
		return -1;
	}
	if(u->port){
		werrstr("port not valid with file scheme");
		return -1;
	}
	if(u->path == nil){
		werrstr("missing path in file scheme");
		return -1;
	}
	if(strchr(u->path, ';')){
		werrstr("unexpected \";param\" in file path");
		return -1;
	}

	if(!ismatch(REfilepath, u->path, "file path"))
		return -1;

	/* "localhost" is equivalent to no host spec, we'll chose the latter */
	if(u->host && cistrcmp(u->host, "localhost") == 0){
		free(u->host);
		u->host = nil;
	}
	return 0;
}

static int (*postparse[])(Url*) = {
	nil,
	postparse_http,
	postparse_http,
	postparse_ftp,
	postparse_file,
};

Url*
parseurl(char *url, Url *base)
{
	Url *u;
	SplitUrl su;

	if(urldebug)
		fprint(2, "parseurl %s with base %s\n", url, base ? base->url : "<none>");

	u = emalloc(sizeof(Url));
	u->url = estrdup(url);
	if(spliturl(u->url, &su) < 0){
	Fail:
		freeurl(u);
		return nil;
	}

	/* RFC2396 sec 3.1 says relative URIs are distinguished by absent scheme */ 
	if(su.scheme.s==nil){
		if(urldebug)
			fprint(2, "parseurl has nil scheme\n");
		if(resolve_relative(&su, base, u) < 0 || spliturl(u->url, &su) < 0)
			goto Fail;
		if(u->ischeme == UScurrent){
			/* 'u.url' refers to current document; set fragment and return */
			if(parse_fragment(&su, u) < 0)
				goto Fail;
			return u;
		}
	}

	if(parse_scheme(&su, u) < 0
	|| parse_fragment(&su, u) < 0)
		goto Fail;

	if(u->ischeme == USunknown){
		if(parse_unknown_part(&su, u) < 0)
			goto Fail;
		return u;
	}

	if(parse_query(&su, u) < 0
	|| parse_authority(&su, u) < 0
	|| parse_abspath(&su, u) < 0)
		goto Fail;

	if(u->ischeme < nelem(postparse) && postparse[u->ischeme])
		if((*postparse[u->ischeme])(u) < 0)
			goto Fail;

	setmalloctag(u, getcallerpc(&url));
	return u;
}

void
freeurl(Url *u)
{
	if(u == nil)
		return;
	free(u->url);
	free(u->scheme);
	free(u->schemedata);
	free(u->authority);
	free(u->user);
	free(u->passwd);
	free(u->host);
	free(u->port);
	free(u->path);
	free(u->query);
	free(u->fragment);
	switch(u->ischeme){
	case UShttp:
		free(u->http.page_spec);
		break;
	case USftp:
		free(u->ftp.path_spec);
		free(u->ftp.type);
		break;
	}
	free(u);
}

void
rewriteurl(Url *u)
{
	char *s;

	if(u->schemedata)
		s = estrmanydup(u->scheme, ":", u->schemedata, nil);
	else
		s = estrmanydup(u->scheme, "://", 
			u->user ? u->user : "",
			u->passwd ? ":" : "", u->passwd ? u->passwd : "",
			u->user ? "@" : "", u->host ? u->host : "", 
			u->port ? ":" : "", u->port ? u->port : "",
			u->path,
			u->query ? "?" : "", u->query ? u->query : "",
			u->fragment ? "#" : "", u->fragment ? u->fragment : "",
			nil);
	free(u->url);
	u->url = s;
}

int
seturlquery(Url *u, char *query)
{
	if(query == nil){
		free(u->query);
		u->query = nil;
		return 0;
	}

	if(!ismatch(REquery, query, "query"))
		return -1;

	free(u->query);
	u->query = estrdup(query);
	return 0;
}

static void
dupp(char **p)
{
	if(*p)
		*p = estrdup(*p);
}

Url*
copyurl(Url *u)
{
	Url *v;

	v = emalloc(sizeof(Url));
	*v = *u;
	dupp(&v->url);
	dupp(&v->scheme);
	dupp(&v->schemedata);
	dupp(&v->authority);
	dupp(&v->user);
	dupp(&v->passwd);
	dupp(&v->host);
	dupp(&v->port);
	dupp(&v->path);
	dupp(&v->query);
	dupp(&v->fragment);

	switch(v->ischeme){
	case UShttp:
		dupp(&v->http.page_spec);
		break;
	case USftp:
		dupp(&v->ftp.path_spec);
		dupp(&v->ftp.type);
		break;
	}
	return v;
}

static int
dhex(char c)
{
	if('0' <= c && c <= '9')
		return c-'0';
	if('a' <= c && c <= 'f')
		return c-'a'+10;
	if('A' <= c && c <= 'F')
		return c-'A'+10;
	return 0;
}

char*
escapeurl(char *s, int (*needesc)(int))
{
	int n;
	char *t, *u;
	Rune r;
	static char *hex = "0123456789abcdef";

	n = 0;
	for(t=s; *t; t++)
		if((*needesc)(*t))
			n++;

	u = emalloc(strlen(s)+2*n+1);
	t = u;
	for(; *s; s++){
		s += chartorune(&r, s);
		if(r >= 0xFF){
			werrstr("URLs cannot contain Runes > 0xFF");
			free(t);
			return nil;
		}
		if((*needesc)(r)){
			*u++ = '%';
			*u++ = hex[(r>>4)&0xF];
			*u++ = hex[r&0xF];
		}else
			*u++ = r;
	}
	*u = '\0';
	return t;
}

char*
unescapeurl(char *s)
{
	char *r, *w;
	Rune rune;

	s = estrdup(s);
	for(r=w=s; *r; r++){
		if(*r=='%'){
			r++;
			if(!isxdigit(r[0]) || !isxdigit(r[1])){
				werrstr("bad escape sequence '%.3s' in URL", r);
				return nil;
			}
			if(r[0]=='0' && r[2]=='0'){
				werrstr("escaped NUL in URL");
				return nil;
			}
			rune = (dhex(r[0])<<4)|dhex(r[1]);	/* latin1 */
			w += runetochar(w, &rune);
			r += 2;
		}else
			*w++ = *r;
	}
	*w = '\0';
	return s;
}