/* //////////////////////////////////////////////////////////////////////////////////////
* implementation
*/
static tb_bool_t tb_demo_spider_parser_open_html(tb_stream_ref_t stream, tb_char_t const* url)
{
// check
tb_assert_and_check_return_val(stream && url, tb_false);
// done
tb_bool_t ok = tb_false;
do
{
// find the .suffix
tb_char_t const* p = tb_strrchr(url, '.');
if (p)
{
// not html?
tb_check_break ( tb_stricmp(p, ".css")
&& tb_stricmp(p, ".js")
&& tb_stricmp(p, ".png")
&& tb_stricmp(p, ".jpg")
&& tb_stricmp(p, ".gif")
&& tb_stricmp(p, ".ico")
&& tb_stricmp(p, ".bmp")
&& tb_stricmp(p, ".mp4")
&& tb_stricmp(p, ".mp3")
&& tb_stricmp(p, ".flv")
&& tb_stricmp(p, ".avi")
&& tb_stricmp(p, ".exe")
&& tb_stricmp(p, ".msi")
&& tb_stricmp(p, ".zip")
&& tb_stricmp(p, ".rar")
&& tb_stricmp(p, ".7z"));
}
// ctrl stream
if (!tb_stream_ctrl(stream, TB_STREAM_CTRL_SET_URL, url)) break;
// open stream
if (!tb_stream_open(stream)) break;
// the stream size
tb_hong_t size = tb_stream_size(stream);
tb_check_break(size);
// prefetch some data
tb_byte_t* data = tb_null;
tb_size_t need = tb_min((tb_size_t)size, 256);
if (!tb_stream_need(stream, &data, need)) break;
// is html?
if (tb_strnistr((tb_char_t const*)data, need, "<!DOCTYPE html>"))
{
ok = tb_true;
break;
}
// is html?
ok = tb_strnistr((tb_char_t const*)data, need, "<html")? tb_true : tb_false;
} while (0);
// failed?
if (!ok)
{
// clos stream
if (stream) tb_stream_clos(stream);
}
// ok?
return ok;
}
static tb_bool_t tb_demo_istream_open_func(tb_async_stream_ref_t stream, tb_size_t state, tb_cpointer_t priv)
{
// check
tb_demo_context_t* context = (tb_demo_context_t*)priv;
tb_assert_and_check_return_val(stream && context && context->option, tb_false);
// done
tb_bool_t ok = tb_false;
do
{
// check
if (state != TB_STATE_OK)
{
// print verbose info
if (context->verbose)
{
tb_char_t const* url = tb_null;
tb_async_stream_ctrl(stream, TB_STREAM_CTRL_GET_URL, &url);
tb_printf("open: %s: %s\n", url, tb_state_cstr(state));
}
break;
}
// trace
if (context->verbose) tb_printf("open: ok\n");
// init ostream
if (tb_option_find(context->option, "more0"))
{
// the path
tb_char_t const* path = tb_option_item_cstr(context->option, "more0");
// init
context->ostream = tb_async_stream_init_from_file(tb_async_stream_aicp((tb_async_stream_ref_t)stream), path, TB_FILE_MODE_RW | TB_FILE_MODE_CREAT | TB_FILE_MODE_BINARY | TB_FILE_MODE_TRUNC);
// print verbose info
if (context->verbose) tb_printf("save: %s: ..\n", path);
}
else
{
// the name
tb_char_t const* name = tb_strrchr(tb_option_item_cstr(context->option, "url"), '/');
if (!name) name = tb_strrchr(tb_option_item_cstr(context->option, "url"), '\\');
if (!name) name = "/async_stream.file";
// the path
tb_char_t path[TB_PATH_MAXN] = {0};
if (tb_directory_curt(path, TB_PATH_MAXN))
{
// append name
tb_strcat(path, name);
// init file
context->ostream = tb_async_stream_init_from_file(tb_async_stream_aicp((tb_async_stream_ref_t)stream), path, TB_FILE_MODE_RW | TB_FILE_MODE_CREAT | TB_FILE_MODE_BINARY | TB_FILE_MODE_TRUNC);
// print verbose info
if (context->verbose) tb_printf("save: %s: ..\n", path);
}
}
tb_assert_and_check_break(context->ostream);
// init transfer
context->transfer = tb_async_transfer_init(tb_null, tb_true);
tb_assert_and_check_break(context->transfer);
// init transfer stream
if (!tb_async_transfer_init_istream(context->transfer, stream)) break;
if (!tb_async_transfer_init_ostream(context->transfer, context->ostream)) break;
// the limit rate
if (tb_option_find(context->option, "limitrate"))
tb_async_transfer_limitrate(context->transfer, tb_option_item_uint32(context->option, "limitrate"));
// open and done transfer
if (!tb_async_transfer_open_done(context->transfer, 0, tb_demo_transfer_done_func, context)) break;
// ok
ok = tb_true;
} while (0);
// failed or closed? exit wait
if (state != TB_STATE_OK && context->event) tb_event_post(context->event);
// ok?
return ok;
}
/* //////////////////////////////////////////////////////////////////////////////////////
* main
*/
tb_int_t tb_demo_stream_main(tb_int_t argc, tb_char_t** argv)
{
// done
tb_option_ref_t option = tb_null;
tb_stream_ref_t istream = tb_null;
tb_stream_ref_t ostream = tb_null;
tb_stream_ref_t pstream = tb_null;
do
{
// init option
option = tb_option_init("stream", "the stream demo", g_options);
tb_assert_and_check_break(option);
// done option
if (tb_option_done(option, argc - 1, &argv[1]))
{
// debug & verbose
tb_bool_t debug = tb_option_find(option, "debug");
tb_bool_t verbose = tb_option_find(option, "no-verbose")? tb_false : tb_true;
// done url
if (tb_option_find(option, "url"))
{
// init istream
istream = tb_stream_init_from_url(tb_option_item_cstr(option, "url"));
tb_assert_and_check_break(istream);
// ctrl http
if (tb_stream_type(istream) == TB_STREAM_TYPE_HTTP)
{
// enable gzip?
if (tb_option_find(option, "gzip"))
{
// auto unzip
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_AUTO_UNZIP, 1)) break;
// need gzip
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_HEAD, "Accept-Encoding", "gzip,deflate")) break;
}
// enable debug?
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_HEAD_FUNC, debug? tb_demo_stream_head_func : tb_null)) break;
// custem header?
if (tb_option_find(option, "header"))
{
// init
tb_string_t key;
tb_string_t val;
tb_string_init(&key);
tb_string_init(&val);
// done
tb_bool_t k = tb_true;
tb_char_t const* p = tb_option_item_cstr(option, "header");
while (*p)
{
// is key?
if (k)
{
if (*p != ':' && !tb_isspace(*p)) tb_string_chrcat(&key, *p++);
else if (*p == ':')
{
// skip ':'
p++;
// skip space
while (*p && tb_isspace(*p)) p++;
// is val now
k = tb_false;
}
else p++;
}
// is val?
else
{
if (*p != ';') tb_string_chrcat(&val, *p++);
else
{
// skip ';'
p++;
// skip space
while (*p && tb_isspace(*p)) p++;
// set header
if (tb_string_size(&key) && tb_string_size(&val))
{
if (debug) tb_printf("header: %s: %s\n", tb_string_cstr(&key), tb_string_cstr(&val));
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_HEAD, tb_string_cstr(&key), tb_string_cstr(&val))) break;
}
// is key now
k = tb_true;
// clear key & val
tb_string_clear(&key);
tb_string_clear(&val);
}
}
}
// set header
if (tb_string_size(&key) && tb_string_size(&val))
{
if (debug) tb_printf("header: %s: %s\n", tb_string_cstr(&key), tb_string_cstr(&val));
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_HEAD, tb_string_cstr(&key), tb_string_cstr(&val))) break;
}
// exit
tb_string_exit(&key);
tb_string_exit(&val);
}
// keep alive?
if (tb_option_find(option, "keep-alive"))
{
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_HEAD, "Connection", "keep-alive")) break;
}
// post-data?
if (tb_option_find(option, "post-data"))
{
tb_char_t const* post_data = tb_option_item_cstr(option, "post-data");
tb_hize_t post_size = tb_strlen(post_data);
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_METHOD, TB_HTTP_METHOD_POST)) break;
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_POST_DATA, post_data, post_size)) break;
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_POST_FUNC, tb_demo_http_post_func)) break;
if (debug) tb_printf("post: %llu\n", post_size);
}
// post-file?
else if (tb_option_find(option, "post-file"))
{
tb_char_t const* url = tb_option_item_cstr(option, "post-file");
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_METHOD, TB_HTTP_METHOD_POST)) break;
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_POST_URL, url)) break;
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_POST_FUNC, tb_demo_http_post_func)) break;
if (debug) tb_printf("post: %s\n", url);
}
}
// set range
if (tb_option_find(option, "range"))
{
tb_char_t const* p = tb_option_item_cstr(option, "range");
if (p)
{
// the bof
tb_hize_t eof = 0;
tb_hize_t bof = tb_atoll(p);
while (*p && tb_isdigit(*p)) p++;
if (*p == '-')
{
p++;
eof = tb_atoll(p);
}
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_HTTP_SET_RANGE, bof, eof)) break;
}
}
// set timeout
if (tb_option_find(option, "timeout"))
{
tb_size_t timeout = tb_option_item_uint32(option, "timeout");
if (!tb_stream_ctrl(istream, TB_STREAM_CTRL_SET_TIMEOUT, timeout)) break;
}
// print verbose info
if (verbose) tb_printf("open: %s: ..\n", tb_option_item_cstr(option, "url"));
// open istream
if (!tb_stream_open(istream))
{
// print verbose info
if (verbose) tb_printf("open: %s\n", tb_state_cstr(tb_stream_state(istream)));
break;
}
// print verbose info
if (verbose) tb_printf("open: ok\n");
// init ostream
if (tb_option_find(option, "more0"))
{
// the path
tb_char_t const* path = tb_option_item_cstr(option, "more0");
// init
ostream = tb_stream_init_from_file(path, TB_FILE_MODE_RW | TB_FILE_MODE_CREAT | TB_FILE_MODE_BINARY | TB_FILE_MODE_TRUNC);
// print verbose info
if (verbose) tb_printf("save: %s\n", path);
}
else
{
// the name
tb_char_t const* name = tb_strrchr(tb_option_item_cstr(option, "url"), '/');
if (!name) name = tb_strrchr(tb_option_item_cstr(option, "url"), '\\');
if (!name) name = "/stream.file";
// the path
tb_char_t path[TB_PATH_MAXN] = {0};
if (tb_directory_current(path, TB_PATH_MAXN))
tb_strcat(path, name);
else break;
// init file
ostream = tb_stream_init_from_file(path, TB_FILE_MODE_RW | TB_FILE_MODE_CREAT | TB_FILE_MODE_BINARY | TB_FILE_MODE_TRUNC);
// print verbose info
if (verbose) tb_printf("save: %s\n", path);
}
tb_assert_and_check_break(ostream);
// the limit rate
tb_size_t limitrate = 0;
if (tb_option_find(option, "limitrate"))
limitrate = tb_option_item_uint32(option, "limitrate");
// save it
tb_hong_t save = 0;
tb_demo_context_t context = {0};
context.verbose = verbose;
if ((save = tb_transfer_done(istream, ostream, limitrate, tb_demo_stream_save_func, &context)) < 0) break;
}
else tb_option_help(option);
}
else tb_option_help(option);
} while (0);
// exit pstream
if (pstream) tb_stream_exit(pstream);
pstream = tb_null;
// exit istream
if (istream) tb_stream_exit(istream);
istream = tb_null;
// exit ostream
if (ostream) tb_stream_exit(ostream);
ostream = tb_null;
// exit option
if (option) tb_option_exit(option);
option = tb_null;
return 0;
}