/*
* === FUNCTION ======================================================================
* Name: is_pchar
* Description: Test if a character is a pchar as desribed by RFC3986. A pchar
* is: unreserved / pct-encoded / sub-delims / ":" / "@", if character
* is a pchar return true otherwise return false.
* =====================================================================================
*/
static bool
is_pchar( char c)
{
bool result = false;
if( is_unreserved(c) || is_sub_delim(c) ){
result = true;
}
else if( c == ':' || c == '@'){
result = true;
}
return result;
}
static bool is_password(char c)
{
switch (c) {
case '&':
case '=':
case '+':
case '$':
case ',':
return true;
default:
return is_unreserved(c);
}
}
/************************************************************************
* Function : parse_uric
*
* Parameters :
* char *in ; string of characters
* int max ; maximum limit
* token *out ; token object where the string of characters is
* copied
*
* Description : Parses a string of uric characters starting at in[0]
* as defined in http://www.ietf.org/rfc/rfc2396.txt (RFC explaining
* URIs)
*
* Return : int ;
*
* Note :
************************************************************************/
int
parse_uric( char *in,
int max,
token * out )
{
int i = 0;
while( ( i < max )
&& ( ( is_unreserved( in[i] ) ) || ( is_reserved( in[i] ) )
|| ( ( i + 2 < max ) && ( is_escaped( &in[i] ) ) ) ) ) {
i++;
}
out->size = i;
out->buff = in;
return i;
}
std::string uri::encode(const std::experimental::string_view input) {
static const std::array<char,16> hex =
{{ '0','1','2','3','4','5','6','7','8','9','a','b','c','d','e','f' }};
std::string res;
res.reserve(input.size());
for (const auto chr : input)
if (is_unreserved(chr)) {
res += chr;
} else {
res += '%';
res += hex[ chr >> 4 ];
res += hex[ chr & 0xf ];
}
return res;
}
/*
* === FUNCTION ======================================================================
* Name: uri_norm_host
* Description: Normalize the host section of the URI if it is set. The host section
* that is normalized here is the reg-name, and not the IP address.
* The IP address should be inside the IP section of the structure.
*
* This section attempts to implement section 3.2.2 of RFC3986 with the
* exception that userinfo is not checked as HTTP and HTTPS do not
* specify the usage of it.
* =====================================================================================
*/
extern int
uri_norm_host( uriobj_t *uri)
{
int err = 0, i, len;
char *pct = (char *) malloc(4),
*host = URI_CP_PT(uri->uri_host);
if( host ) {
len = strlen(host);
for(i = 0;i < len; i ++ ) {
if( host[i] == '%'){
if( (i + 2) > len ) {
err = EILSEQ;
break;
}
pct[0] = host[i ++];
pct[1] = host[i ++];
pct[2] = host[i ++];
pct[3] = '\0';
err = norm_pct(&pct);
if( err ) {
break;
}
host[(i - 3)] = pct[0];
host[(i - 2)] = pct[1];
host[(i - 1)] = pct[2];
continue;
}
if(!(is_sub_delim(host[i]) || is_unreserved(host[i]))){
err = EILSEQ;
break;
}
if( isalpha(host[i]) && isupper(host[i]) ){
host[i] = tolower(host[i]);
}
}
if( ! err ) {
*(uri->uri_host) = NULL;
free(*uri->uri_host);
*(uri->uri_host) = host;
}
}
return err;
}
// No special destructor needed, fragment is just char *. A url does not need to have
// a fragment, so a NULL return value is not strictly an error.
static char *get_fragment(char **s,unsigned int *fragment_out_err) {
*fragment_out_err = NO_UPARSE_ERROR;
if ((NULL == s) || (0 == strlen(*s))) {
return NULL;
}
char *c = *s;
if (FRAGMENT_DELIM != c[0]) {
fprintf(stderr,"no %c as prefix\n",FRAGMENT_DELIM);
*fragment_out_err = UPARSE_ERROR;
return NULL;
}
c++;
size_t fragment_len = strlen(c);
if (fragment_len == 0) {
// no fragment
*fragment_out_err = NO_UPARSE_ERROR;
return NULL;
}
char *fragment = strdup(c);
if (NULL == fragment) {
fprintf(stderr,"cannot allocate fragment\n");
*fragment_out_err = UPARSE_ERROR;
return NULL;
}
char *test_fragment = fragment;
while (*test_fragment) {
if (!is_unreserved(*test_fragment)) {
fprintf(stderr,"'%c' is invalid\n",*test_fragment);
free(fragment);
*fragment_out_err = UPARSE_ERROR;
return NULL;
}
test_fragment++;
}
*fragment_out_err = NO_UPARSE_ERROR;
return fragment;
}
/*!
* \brief Parses a string of uric characters starting at in[0] as defined in
* http://www.ietf.org/rfc/rfc2396.txt (RFC explaining URIs).
*
* \return
*/
static size_t parse_uric(
/*! [in] String of characters. */
const char *in,
/*! [in] Maximum limit. */
size_t max,
/*! [out] Token object where the string of characters is copied. */
token *out)
{
size_t i = (size_t)0;
while (i < max &&
(is_unreserved(in[i]) ||
is_reserved(in[i]) ||
((i + (size_t)2 < max) && is_escaped(&in[i])))) {
i++;
}
out->size = i;
out->buff = in;
return i;
}
std::string uri::decode(const std::experimental::string_view input) {
std::string res;
res.reserve(input.size());
for (auto it = input.cbegin(), e = input.cend(); it not_eq e; ++it) {
if (*it == '%') {
if (++it >= e) return decode_error(std::move(res));
const uint8_t nibble1 = (*it);
if (++it >= e) return decode_error(std::move(res));
const uint8_t nibble2 = (*it);
res += static_cast<char>(from_hex(nibble1, nibble2));
} else {
if (is_reserved(*it) or is_unreserved(*it))
res += *it;
else
return decode_error(std::move(res));
}
}
return res;
}
static bool is_hvalue(char c)
{
return is_hnv_unreserved(c) || is_unreserved(c);
}
static bool is_paramchar(char c)
{
return is_param_unreserved(c) || is_unreserved(c);
}
// A url doesn't have to have a ?query arg list, so this can return NULL
// and not be an error.
static query_arg_list_t *get_query_arg_list(char **s,unsigned int *query_out_err) {
if ((NULL == s) || (0 == strlen(*s))) {
*query_out_err = NO_UPARSE_ERROR;
return NULL;
}
char *c = *s;
*query_out_err = UPARSE_ERROR; // default until end of function
if (QUERY_DELIM != c[0]) {
fprintf(stderr,"no %c as prefix\n",QUERY_DELIM);
return NULL;
}
c++;
if (NULL == c) {
fprintf(stderr,"after %c found, no text\n",QUERY_DELIM);
return NULL;
}
// at the end of the loop, t will point to the end of the query string, which
// may not be the end of the url; it may have a #fragment
char *t = c;
unsigned int query_delim_count = 0;
unsigned int query_pair_count = 0;
unsigned int query_str_len = 0;
while ((*t) && (FRAGMENT_DELIM != *t)) {
if (QUERY_PAIR_DELIM == *t) {
query_delim_count++;
}
t++;
query_str_len++;
}
query_pair_count = query_delim_count + 1;
char query_string[query_str_len+1];
bzero((void *) query_string,(query_str_len+1) * sizeof(char));
strlcpy(query_string,c,query_str_len+1);
// stringify the delims for strsep
char query_pair_delim_str[2];
snprintf(query_pair_delim_str,2,"%c",QUERY_PAIR_DELIM);
char query_key_value_delim_str[2];
snprintf(query_key_value_delim_str,2,"%c",QUERY_KEY_VAL_DELIM);
query_key_val_t **query_key_vals =
(query_key_val_t **) malloc(query_pair_count * sizeof(query_key_val_t *));
if (NULL == query_key_vals) {
fprintf(stderr,"could not allocate query_key_val\n");
return NULL;
}
bzero((void *) query_key_vals,query_pair_count * sizeof(query_key_val_t *));
unsigned int i = 0;
char *pair_tok;
char *free_query_string = query_string;
while ((pair_tok = strsep(&free_query_string,query_pair_delim_str)) != NULL) {
if (0 != strcmp("",pair_tok)) {
if (i >= query_pair_count) {
fprintf(stderr,"loop count %d >= previous pair count %d\n",
i,query_pair_count);
free_query_key_val_t_list(query_key_vals,query_pair_count);
return NULL;
}
char *sep_pair_tok = strdup(pair_tok);
if (NULL == sep_pair_tok) {
fprintf(stderr,"could not allocate sep_pair_tok\n");
free_query_key_val_t_list(query_key_vals,query_pair_count);
return NULL;
}
char *free_sep_pair_tok = sep_pair_tok;
// put the key and val in a struct, and set as item i in a list
// query_pair_count long (which will need to be allocated above)
query_key_val_t *query_key_val_tok =
(query_key_val_t *) malloc(sizeof(query_key_val_t));
if (NULL == query_key_val_tok) {
fprintf(stderr,"could not allocate query_key_val_tok\n");
free_query_key_val_t_list(query_key_vals,query_pair_count);
free(free_sep_pair_tok);
return NULL;
}
bzero((void *) query_key_val_tok,sizeof(query_key_val_t));
bool seen_key = false;
char *kv_tok = NULL;
char *key = NULL;
char *val = NULL;
// make sure sep_pair_tok contains an = to delimit the key and value
if (NULL == strstr(sep_pair_tok,query_key_value_delim_str)) {
fprintf(stderr,"could not find '%s' in query pair '%s'\n",
query_key_value_delim_str,sep_pair_tok);
free_query_key_val_t_list(query_key_vals,query_pair_count);
free_query_key_val_t(query_key_val_tok);
free(free_sep_pair_tok);
return NULL;
}
// break sep_pair_tok (a=b) into the key (a) and the value (b)
while ((kv_tok = strsep(&sep_pair_tok,query_key_value_delim_str)) != NULL) {
if (!seen_key) {
key = strdup(kv_tok);
seen_key = true;
} else {
val = strdup(kv_tok);
}
}
free(kv_tok);
if ((NULL == key) || (NULL == val)) {
fprintf(stderr,"either key or val from %s was null\n",free_sep_pair_tok);
free_query_key_val_t_list(query_key_vals,query_pair_count);
free_query_key_val_t(query_key_val_tok);
free(free_sep_pair_tok);
free(key);
free(val);
return NULL;
}
char *key_val_list[] = {key,val};
char *test_kv;
for (size_t j = 0;j < 2;j++) {
test_kv = key_val_list[i];
while (*test_kv) {
if (!is_unreserved(*test_kv)) {
fprintf(stderr,"'%c' is invalid\n",*test_kv);
free_query_key_val_t_list(query_key_vals,query_pair_count);
free(free_sep_pair_tok);
free_query_key_val_t(query_key_val_tok);
free(key);
free(val);
return NULL;
}
test_kv++;
}
}
query_key_val_tok->key = key;
query_key_val_tok->val = val;
query_key_vals[i] = query_key_val_tok;
free(free_sep_pair_tok);
i++;
}
}
free(pair_tok);
query_arg_list_t *query_arg_list =
(query_arg_list_t *) malloc(sizeof(query_arg_list_t));
if (NULL == query_arg_list) {
fprintf(stderr,"cannot allocate query_arg_list\n");
return NULL;
}
bzero((void *) query_arg_list,sizeof(query_arg_list_t));
query_arg_list->count = i;
query_arg_list->query_key_vals = query_key_vals;
*s = t; // pointer is now past query arg str, at # fragment delim if there
*query_out_err = NO_UPARSE_ERROR;
return query_arg_list;
}
static bool is_user(char c)
{
return is_unreserved(c) || is_user_unreserved(c);
}
// A url does not need to have a path, so this can return NULL without an error
// being thrown
static path_t *get_path(char **s, unsigned int *path_out_err) {
if ((NULL == s) || (0 == strlen(*s))) {
*path_out_err = NO_UPARSE_ERROR; // not an error: url can have no path
return NULL;
}
*path_out_err = UPARSE_ERROR; // default until end of function
size_t delim_count = 0;
size_t chars_until_query_delim = 0;
char *c = *s;
while (*c) {
if (PATH_DELIM == *c) {
delim_count++;
} else if (QUERY_DELIM == *c) {
break;
} else if (!is_unreserved(*c)) {
fprintf(stderr,"'%c' is invalid\n",*c);
return NULL;
}
chars_until_query_delim++;
c++;
}
char *just_path;
char path_delim_str[2] = {PATH_DELIM,'\0'}; // "/" (not '/')
char nonempty_path[chars_until_query_delim+1];
if (0 == delim_count) {
// in this case, the path was '', not even '/'. so we create
// a default empty path of '/' to represent what '' implies
just_path = (char *) &path_delim_str;
} else {
bzero((void *) nonempty_path,(chars_until_query_delim + 1) * sizeof(char));
strlcpy(nonempty_path,*s,chars_until_query_delim+1);
just_path = (char *) &nonempty_path;
}
path_t *path = (path_t *) malloc(sizeof(path_t));
if (NULL == path) {
fprintf(stderr,"cannot allocate path\n");
return NULL;
}
bzero((void *) path,sizeof(path_t));
path->path_str = strdup(just_path);
// set pointer to where the query delim may be found, now that *s
// has been copied into just_path
*s = c;
// case of single '/' path (like: http://foo.com/?key=val)
if (strlen(just_path) == 1) {
path->path_elts = (char **) malloc(1 * sizeof(char *));
path->path_elts[0] = (char *) calloc(1,sizeof(char));
path->count = 1;
*path_out_err = NO_UPARSE_ERROR;
return path;
}
path->path_elts = (char **) malloc(delim_count * sizeof(char *));
if (NULL == path->path_elts) {
fprintf(stderr,"cannot allocate path\n");
return NULL;
}
path->count = delim_count;
char *tok;
size_t i = 0;
while (((tok = strsep(&just_path,path_delim_str)) != NULL) && (i < delim_count)) {
if (0 != strcmp("",tok)) {
path->path_elts[i] = strdup(tok);
if (NULL == path->path_elts[i]) {
fprintf(stderr,"cannot dup %s\n",tok);
free_path_t(path);
free(tok);
return NULL;
}
i++;
}
}
*path_out_err = NO_UPARSE_ERROR;
return path;
}
// Get the host:port section of the url. Doesn't support ipv6, unicode hosts,
// username annotations etc.
// Every url must have a host (but the port is optional). If this returns NULL,
// it is an error.
static host_port_t *get_host_port(char **s, unsigned int *host_port_out_err) {
*host_port_out_err = UPARSE_ERROR; // default
if ((NULL == s) || (0 == strlen(*s))) {
fprintf(stderr,"arg pointer null\n");
return NULL;
}
char *c = *s;
size_t host_len = 0;
size_t port_len = 0;
char *host_start = *s;
char *port_start = NULL;
bool seen_host_delim = false;
while (*c) {
// normally we would see the PATH_DELIM to break this, but
// this can be '', so we must also look out for the
// QUERY_DELIM
if ((PATH_DELIM == c[0]) || (QUERY_DELIM == c[0])) {
break;
} else if (HOST_PORT_DELIM == c[0]) {
seen_host_delim = true;
c++;
port_start = c;
} else {
// *c is part of the host or port, depending on delimiters seen
if (!is_unreserved(*c)) {
fprintf(stderr,"'%c' is invalid\n",*c);
return NULL;
}
// If we have already seen the host/port delimiter (:), then
// count the chars in the length of the port part. Otherwise,
// count the chars in the length of the host part.
seen_host_delim ? port_len++ : host_len++;
c++;
}
}
if (0 == host_len) {
fprintf(stderr,"no host found\n");
return NULL;
}
if (seen_host_delim && (0 == port_len)) {
fprintf(stderr,"port delimiter seen but no port number string\n");
return NULL;
}
host_port_t *host_port = (host_port_t *) malloc(sizeof(host_port_t));
if (NULL == host_port) {
fprintf(stderr,"cannot allocate host_port\n");
return NULL;
}
bzero((void *) host_port,sizeof(host_port_t));
host_port->host = strndup(host_start,host_len);
if (NULL == host_port->host) {
fprintf(stderr,"cannot allocate host_port->host\n");
free_host_port_t(host_port);
return NULL;
}
host_port->port = 0;
if ((seen_host_delim) && (port_len > 0)) {
char *port_chars = strndup(port_start,port_len);
if (NULL == port_chars) {
fprintf(stderr,"cannot allocate chars for host_port->port\n");
free_host_port_t(host_port);
return NULL;
}
long port_long = (long) strtol(port_chars,NULL,10);
if (0 == port_long) {
fprintf(stderr,"zero port or cannot convert port_chars %s to long\n",port_chars);
free_host_port_t(host_port);
free(port_chars);
return NULL;
}
free(port_chars);
if (MAX_PORT < port_long) {
fprintf(stderr,"port %ld out of range\n",port_long);
free_host_port_t(host_port);
*host_port_out_err = UPARSE_ERROR;
return NULL;
}
host_port->port = (unsigned int) port_long;
}
*s = c; // set pointer past host/port
*host_port_out_err = NO_UPARSE_ERROR;
return host_port;
}
bool net::uri::uri::normalize(uri& other) const
{
if (!other._M_uri.allocate(_M_uri.length() + 1)) {
return false;
}
const char* d = _M_uri.data();
const char* end = d + _M_uri.length();
char* otherd = other._M_uri.data();
// Copy scheme.
size_t len = _M_scheme.length();
for (size_t i = 0; i < len; i++) {
*otherd++ = util::to_lower(*d++);
}
other._M_scheme.set(other._M_uri.data(), len);
// Skip colon.
d++;
*otherd++ = ':';
// If there is authority...
if (_M_hier_part.host.length() > 0) {
d += 2;
*otherd++ = '/';
*otherd++ = '/';
// If there is user information...
if ((len = _M_hier_part.userinfo.length()) > 0) {
// Save start of the userinfo.
char* userinfo = otherd;
size_t i = 0;
do {
if (*d == '%') {
uint8_t c = (util::hex2dec(d[1]) * 16) + util::hex2dec(d[2]);
if (is_unreserved(c)) {
*otherd++ = static_cast<char>(c);
} else {
*otherd++ = '%';
*otherd++ = util::to_upper(d[1]);
*otherd++ = util::to_upper(d[2]);
}
d += 3;
i += 2;
} else {
*otherd++ = *d++;
}
} while (++i < len);
other._M_hier_part.userinfo.set(userinfo, otherd - userinfo);
d++;
*otherd++ = '@';
}
// IP literal?
if (_M_hier_part.ip_literal) {
// Skip '['.
d++;
*otherd++ = '[';
}
// Save start of the host.
char* host = otherd;
// Copy host.
len = _M_hier_part.host.length();
size_t i = 0;
do {
if (*d == '%') {
uint8_t c = (util::hex2dec(d[1]) * 16) + util::hex2dec(d[2]);
if (is_unreserved(c)) {
*otherd++ = static_cast<char>(util::to_lower(c));
} else {
*otherd++ = '%';
*otherd++ = util::to_upper(d[1]);
*otherd++ = util::to_upper(d[2]);
}
d += 3;
i += 2;
} else {
*otherd++ = util::to_lower(*d++);
}
} while (++i < len);
other._M_hier_part.host.set(host, otherd - host);
// IP literal?
if (_M_hier_part.ip_literal) {
// Skip ']'.
d++;
*otherd++ = ']';
}
other._M_hier_part.ip_literal = _M_hier_part.ip_literal;
// If not a standard port...
if (_M_hier_part.port != 0) {
*otherd++ = ':';
const char* otherend = other._M_uri.data() + other._M_uri.capacity();
otherd += snprintf(otherd, otherend - otherd, "%u", _M_hier_part.port);
}
other._M_hier_part.port = _M_hier_part.port;
}
// Save start of the path.
char* path = otherd;
// If the path is empty...
if ((len = _M_hier_part.path.length()) == 0) {
*otherd++ = '/';
} else {
// Remove dot segments.
d = _M_hier_part.path.data();
size_t i = 0;
do {
if (*d == '%') {
uint8_t c = (util::hex2dec(d[1]) * 16) + util::hex2dec(d[2]);
if (is_unreserved(c)) {
*otherd++ = static_cast<char>(c);
} else {
*otherd++ = '%';
*otherd++ = util::to_upper(d[1]);
*otherd++ = util::to_upper(d[2]);
}
d += 3;
i += 2;
} else {
*otherd++ = *d++;
}
// A. If the input buffer begins with a prefix of "../" or "./",
// then remove that prefix from the input buffer; otherwise,
if (((path + 3 == otherd) &&
(path[0] == '.') &&
(path[1] == '.') &&
(path[2] == '/')) ||
((path + 2 == otherd) &&
(path[0] == '.') &&
(path[1] == '/'))) {
otherd = path;
// B. if the input buffer begins with a prefix of "/./" or "/.",
// where "." is a complete path segment, then replace that
// prefix with "/" in the input buffer; otherwise,
} else if ((path + 3 <= otherd) &&
(otherd[-3] == '/') &&
(otherd[-2] == '.') &&
(otherd[-1] == '/')) {
otherd -= 2;
} else if ((path + 2 <= otherd) &&
(d == end) &&
(otherd[-2] == '/') &&
(otherd[-1] == '.')) {
otherd--;
// C. if the input buffer begins with a prefix of "/../" or "/..",
// where ".." is a complete path segment, then replace that
// prefix with "/" in the input buffer and remove the last
// segment and its preceding "/" (if any) from the output
// buffer; otherwise,
} else if ((path + 4 <= otherd) &&
(otherd[-4] == '/') &&
(otherd[-3] == '.') &&
(otherd[-2] == '.') &&
(otherd[-1] == '/')) {
otherd -= 4;
while (otherd > path) {
if (*--otherd == '/') {
break;
}
}
*otherd++ = '/';
} else if ((path + 3 <= otherd) &&
(d == end) &&
(otherd[-3] == '/') &&
(otherd[-2] == '.') &&
(otherd[-1] == '.')) {
otherd -= 3;
while (otherd > path) {
if (*--otherd == '/') {
break;
}
}
*otherd++ = '/';
// D. if the input buffer consists only of "." or "..", then remove
// that from the input buffer; otherwise,
} else if (((path + 1 == otherd) &&
(d == end) &&
(otherd[-1] == '.')) ||
((path + 2 == otherd) &&
(d == end) &&
(otherd[-2] == '.') &&
(otherd[-1] == '.'))) {
otherd = path;
}
} while (++i < len);
}
if (path < otherd) {
other._M_hier_part.path.set(path, otherd - path);
}
// If there is query...
if ((len = _M_query.length()) > 0) {
*otherd++ = '?';
// Save start of the query.
char* query = otherd;
d = _M_query.data();
size_t i = 0;
do {
if (*d == '%') {
uint8_t c = (util::hex2dec(d[1]) * 16) + util::hex2dec(d[2]);
if (is_unreserved(c)) {
*otherd++ = static_cast<char>(c);
} else {
*otherd++ = '%';
*otherd++ = util::to_upper(d[1]);
*otherd++ = util::to_upper(d[2]);
}
d += 3;
i += 2;
} else {
*otherd++ = *d++;
}
} while (++i < len);
other._M_query.set(query, otherd - query);
}
// If there is fragment...
if ((len = _M_fragment.length()) > 0) {
*otherd++ = '#';
// Save start of the fragment.
char* fragment = otherd;
d = _M_fragment.data();
size_t i = 0;
do {
if (*d == '%') {
uint8_t c = (util::hex2dec(d[1]) * 16) + util::hex2dec(d[2]);
if (is_unreserved(c)) {
*otherd++ = static_cast<char>(c);
} else {
*otherd++ = '%';
*otherd++ = util::to_upper(d[1]);
*otherd++ = util::to_upper(d[2]);
}
d += 3;
i += 2;
} else {
*otherd++ = *d++;
}
} while (++i < len);
other._M_fragment.set(fragment, otherd - fragment);
}
other._M_uri.length(otherd - other._M_uri.data());
return true;
}
/**
Decodes URL-encoded data. Because encoded data is always bigger conversion is
done in-place.
@return Number of decoded bytes written to data. Negative integer if data is
not valid URL-encoded sequence.
*/
ssize_t urldecode(decoder_state *state, char *data, size_t size)
{
size_t inpos = 0, outpos = 0;
int d1, d2;
while (inpos < size)
{
char in = data[inpos++];
switch (in)
{
case '%':
switch (state->state)
{
case ST_SYM:
state->state = ST_PERCENT;
break;
default:
return -1;
}
break;
case '+':
switch (state->state)
{
case ST_SYM:
data[outpos++] = ' ';
break;
default:
return -1;
}
break;
default:
switch (state->state)
{
case ST_PERCENT_AND_SYM:
d1 = hexdigit(state->sym), d2 = hexdigit(in);
if (d1 >= 0 && d2 >= 0)
{
data[outpos++] = (d1 << 4) | d2;
}
else
{
return -1;
}
state->state = ST_SYM;
break;
case ST_PERCENT:
state->sym = in;
state->state = ST_PERCENT_AND_SYM;
break;
case ST_SYM:
if (is_unreserved(in))
{
data[outpos++] = in;
}
else
{
return -1;
}
break;
}
}
}
return outpos;
}
/** Built-in preprocessing callback
*
* Built-in preprocessing callback to break or not to break URLs according to
* some rules by Chicago Manual of Style 15th ed.
* If data is NULL, prohibit break.
* Otherwise, allow break by rule above.
*/
gcstring_t *linebreak_prep_URIBREAK(linebreak_t * lbobj, void *data,
unistr_t * str, unistr_t * text)
{
gcstring_t *gcstr;
size_t i;
unichar_t *ptr;
/* Pass I */
if (text != NULL) {
/*
* Search URL in str.
* Following code loosely refers RFC3986 but some practical
* assumptions are put:
*
* o Broken pct-encoded sequences (e.g. single "%") are allowed.
* o scheme names must end with alphanumeric, must be longer than
* or equal to two octets, and must not contain more than one
* non-alphanumeric ("+", "-" or ".").
* o URLs containing neither non-empty path, query part nor fragment
* (e.g. "about:") are omitted: they are treated as ordinal words.
*/
for (ptr = NULL, i = 0; i < str->len; ptr = NULL, i++) {
int has_double_slash, has_authority, has_empty_path,
has_no_query, has_no_fragment;
size_t alphadigit, nonalphadigit;
/* skip non-alpha. */
if (!is_alpha(str, i))
continue;
ptr = str->str + i;
/* "url:" - case insensitive */
if (startswith(str, i, "url:", 4, 0))
i += 4;
/* scheme */
if (is_alpha(str, i))
i++;
else
continue;
nonalphadigit = 0;
alphadigit = 1;
while (1) {
if (is_alpha(str, i) || is_digit(str, i))
alphadigit++;
else if (is(str, i, '+') || is(str, i, '-') || is(str, i, '.'))
nonalphadigit++;
else
break;
i++;
}
if (alphadigit < 2 || 1 < nonalphadigit ||
! (is_digit(str, i - 1) || is_alpha(str, i - 1)))
continue;
/* ":" */
if (is(str, i, ':'))
i++;
else
continue;
/* hier-part */
has_double_slash = 0;
has_authority = 0;
has_empty_path = 0;
has_no_query = 0;
has_no_fragment = 0;
if (startswith(str, i, "//", 2, 0)) {
/* "//" */
has_double_slash = 1;
i += 2;
/* authority - FIXME:syntax relaxed */
if (is(str, i, '[') || is(str, i, ':') || is(str, i, '@') ||
is_unreserved(str, i) || is_pct_encoded(str, i) ||
is_sub_delim(str, i)) {
has_authority = 1;
i++;
while (is(str, i, '[') || is(str, i, ']') ||
is(str, i, ':') || is(str, i, '@') ||
is_unreserved(str, i) || is_pct_encoded(str, i) ||
is_sub_delim(str, i))
i++;
}
}
/* path */
if (has_double_slash) {
if (has_authority)
goto path_abempty;
else
goto path_absolute;
} /* else goto path_rootless; */
/* path_rootless: */
if (is_pchar(str, i)) { /* FIXME:path-noscheme not concerned */
i++;
while (is_pchar(str, i))
i++;
goto path_abempty;
} else {
has_empty_path = 1;
goto path_empty;
}
path_absolute:
if (startswith(str, i, "//", 2, 0))
continue;
else if (is(str, i, '/')) {
i++;
if (is_pchar(str, i)) {
i++;
while (is_pchar(str, i))
i++;
}
goto path_abempty;
} else
continue;
path_abempty:
if (is(str, i, '/')) {
i++;
while (is(str, i, '/') || is_pchar(str, i))
i++;
} /* else goto path_empty; */
path_empty:
;
/* query */
if (is(str, i, '?')) {
i++;
while (is(str, i, '/') || is(str, i, '?') || is_pchar(str, i))
i++;
} else
has_no_query = 1;
/* fragment */
if (is(str, i, '#')) {
i++;
while (is(str, i, '/') || is(str, i, '?') || is_pchar(str, i))
i++;
} else
has_no_fragment = 1;
if (has_empty_path && has_no_query && has_no_fragment)
continue;
break;
}
if (ptr != NULL)
str->len = i - (ptr - str->str);
str->str = ptr;
return NULL;
}
/* Pass II */
if ((gcstr = gcstring_newcopy(str, lbobj)) == NULL) {
lbobj->errnum = errno ? errno : ENOMEM;
return NULL;
}
/* non-break URI. */
if (data == NULL) {
for (i = 1; i < gcstr->gclen; i++)
gcstr->gcstr[i].flag = LINEBREAK_FLAG_PROHIBIT_BEFORE;
return gcstr;
}
/* break URI. */
if (startswith((unistr_t *) gcstr, 0, "url:", 4, 0)) {
gcstr->gcstr[4].flag = LINEBREAK_FLAG_ALLOW_BEFORE;
i = 5;
} else
i = 1;
for (; i < gcstr->gclen; i++) {
unichar_t u, v;
u = gcstr->str[gcstr->gcstr[i - 1].idx];
v = gcstr->str[gcstr->gcstr[i].idx];
/*
* Some rules based on CMoS 15th ed.
* 17.11 1.1: [/] ÷ [^/]
* 17.11 2: [-] ×
* 6.17 2: [.] ×
* 17.11 1.2: ÷ [-~.,_?#%]
* 17.11 1.3: ÷ [=&]
* 17.11 1.3: [=&] ÷
* Default: ALL × ALL
*/
if (u == '/' && v != '/')
gcstr->gcstr[i].flag = LINEBREAK_FLAG_ALLOW_BEFORE;
else if (u == '-' || u == '.')
gcstr->gcstr[i].flag = LINEBREAK_FLAG_PROHIBIT_BEFORE;
else if (v == '-' || v == '~' || v == '.' || v == ',' ||
v == '_' || v == '?' || v == '#' || v == '%' ||
u == '=' || v == '=' || u == '&' || v == '&')
gcstr->gcstr[i].flag = LINEBREAK_FLAG_ALLOW_BEFORE;
else
gcstr->gcstr[i].flag = LINEBREAK_FLAG_PROHIBIT_BEFORE;
}
/* Won't break punctuations at end of matches. */
for (i = gcstr->gclen - 1; 1 <= i; i--) {
unichar_t u = gcstr->str[gcstr->gcstr[i].idx];
if (gcstr->gcstr[i].flag == LINEBREAK_FLAG_ALLOW_BEFORE &&
(u == '"' || u == '.' || u == ':' || u == ';' || u == ',' ||
u == '>'))
gcstr->gcstr[i].flag = LINEBREAK_FLAG_PROHIBIT_BEFORE;
else
break;
}
return gcstr;
}