tb_void_t tb_circle_queue_clear(tb_circle_queue_ref_t self)
{
// check
tb_circle_queue_t* queue = (tb_circle_queue_t*)self;
tb_assert_and_check_return(queue);
// clear it
while (!tb_circle_queue_null(self)) tb_circle_queue_pop(self);
queue->head = 0;
queue->tail = 0;
queue->size = 0;
}
static tb_void_t tb_demo_spider_parser_task_done(tb_thread_pool_worker_ref_t worker, tb_cpointer_t priv)
{
// check
tb_demo_spider_task_t* task = (tb_demo_spider_task_t*)priv;
tb_assert_and_check_return(worker && task && task->spider);
// init parser
tb_demo_spider_parser_t* parser = tb_demo_spider_parser_init(worker);
tb_assert_and_check_return(parser && parser->stream && parser->reader && parser->cache);
// open stream
if (tb_demo_spider_parser_open_html(parser->stream, task->ourl))
{
// open reader
if (tb_xml_reader_open(parser->reader, parser->stream, tb_false))
{
// trace
tb_trace_d("parser: open: %s", task->ourl);
// init url
tb_url_set(&parser->iurl, task->iurl);
// parse url
while ( TB_STATE_OK == tb_atomic_get(&task->spider->state)
&& tb_demo_spider_parser_get_url(parser->reader, &parser->iurl))
{
// trace
tb_trace_d("parser: done: %s", tb_url_get(&parser->iurl));
// done task
tb_bool_t full = tb_false;
if (!tb_demo_spider_task_done(task->spider, tb_url_get(&parser->iurl), &full))
{
// full?
tb_assert_and_check_break(full);
// cache url
if (!tb_circle_queue_full(parser->cache)) tb_circle_queue_put(parser->cache, tb_url_get(&parser->iurl));
// trace
tb_trace_d("parser: cache: save: %s, size: %lu", tb_url_get(&parser->iurl), tb_circle_queue_size(parser->cache));
}
}
// clos reader
tb_xml_reader_clos(parser->reader);
}
// clos stream
tb_stream_clos(parser->stream);
}
// done task from the cache
while (!tb_circle_queue_null(parser->cache))
{
// the url
tb_char_t const* url = (tb_char_t const*)tb_circle_queue_get(parser->cache);
tb_assert_and_check_break(url);
// done task
if (!tb_demo_spider_task_done(task->spider, url, tb_null)) break;
// trace
tb_trace_d("parser: cache: load: %s, size: %lu", url, tb_circle_queue_size(parser->cache));
// pop it
tb_circle_queue_pop(parser->cache);
}
}