tb_bool_t tb_async_transfer_init_istream_from_url(tb_async_transfer_ref_t transfer, tb_char_t const* url)
{
// check
tb_async_transfer_impl_t* impl = (tb_async_transfer_impl_t*)transfer;
tb_assert_and_check_return_val(impl && impl->aicp && url, tb_false);
// muse be closed
tb_assert_and_check_return_val(TB_STATE_CLOSED == tb_atomic_get(&impl->state), tb_false);
// check stream type
if (impl->istream)
{
// probe protocol
tb_size_t protocol = tb_url_protocol_probe(url);
tb_assert_static((tb_size_t)TB_URL_PROTOCOL_FILE == (tb_size_t)TB_STREAM_TYPE_FILE);
tb_assert_static((tb_size_t)TB_URL_PROTOCOL_HTTP == (tb_size_t)TB_STREAM_TYPE_HTTP);
tb_assert_static((tb_size_t)TB_URL_PROTOCOL_SOCK == (tb_size_t)TB_STREAM_TYPE_SOCK);
tb_assert_static((tb_size_t)TB_URL_PROTOCOL_DATA == (tb_size_t)TB_STREAM_TYPE_DATA);
// protocol => type
tb_size_t type = protocol;
if (!type || type > TB_STREAM_TYPE_DATA)
{
tb_trace_e("unknown stream for url: %s", url);
return tb_false;
}
// exit the previous stream first if be different stream type
if (tb_async_stream_type(impl->istream) != type)
{
if (impl->iowner) tb_async_stream_exit(impl->istream);
impl->istream = tb_null;
}
}
// using the previous stream?
if (impl->istream)
{
// ctrl stream
if (!tb_async_stream_ctrl(impl->istream, TB_STREAM_CTRL_SET_URL, url)) return tb_false;
}
else
{
// init stream
impl->istream = tb_async_stream_init_from_url(impl->aicp, url);
tb_assert_and_check_return_val(impl->istream, tb_false);
// init owner
impl->iowner = 1;
}
// ok
return tb_true;
}
static tb_size_t tb_demo_spider_parser_get_url(tb_xml_reader_ref_t reader, tb_url_ref_t url)
{
// check
tb_assert_and_check_return_val(reader && url, tb_false);
// done
tb_size_t ok = 0;
tb_size_t event = TB_XML_READER_EVENT_NONE;
while (!ok && (event = tb_xml_reader_next(reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
case TB_XML_READER_EVENT_ELEMENT_BEG:
{
// the element name
tb_char_t const* name = tb_xml_reader_element(reader);
tb_check_break(name);
// <a href="" />?
// <link href="" />
// <img src="" />?
// <script src="" />?
// <source src="" />?
// <frame src="" />?
if ( !tb_stricmp(name, "a")
|| !tb_stricmp(name, "link")
|| !tb_stricmp(name, "img")
|| !tb_stricmp(name, "frame")
|| !tb_stricmp(name, "source"))
{
// walk attributes
tb_xml_node_ref_t attr = (tb_xml_node_ref_t)tb_xml_reader_attributes(reader);
for (; attr; attr = attr->next)
{
// href or src?
if ( tb_string_size(&attr->data)
&& ( !tb_string_cstricmp(&attr->name, "href")
|| !tb_string_cstricmp(&attr->name, "src")))
{
// the url protocol
tb_size_t protocol = tb_url_protocol_probe(tb_string_cstr(&attr->data));
// http?
if(protocol == TB_URL_PROTOCOL_HTTP)
{
// save url
ok = tb_url_set(url, tb_string_cstr(&attr->data));
}
// file?
else if (protocol == TB_URL_PROTOCOL_FILE)
{
// save path
tb_url_path_set(url, tb_string_cstr(&attr->data));
// ok
ok = tb_true;
}
}
}
}
}
break;
default:
break;
}
}
// ok?
return ok;
}