void *parser(void *arg) {
struct input_args *in_args = (struct input_args *)arg;
while (1) {
struct page *page = (struct page *)unbounded_buffer_get(in_args->page_queue);
if (page == NULL)
break;
char *start = page->content;
while ((start = strstr(start, "link:")) != NULL) {
if (start > page->content && *(start - 1) != ' ' && *(start - 1) != '\n') {
start = start + 5;
continue;
}
char *end = start + 5;
while (*end != ' ' && *end != '\n' && *end != '\0')
end++;
if (*end == '\0') {
char *url = str_duplicate(start + 5);
in_args->edge(page->url, url);
bounded_buffer_put(in_args->url_queue, (void *)url);
break;
} else {
char tmp = *end;
*end = '\0';
char *url = str_duplicate(start + 5);
in_args->edge(page->url, url);
bounded_buffer_put(in_args->url_queue, (void *)url);
*end = tmp;
start = end + 1;
}
}
mem_free(page->url);
mem_free(page->content);
mem_free(page);
unbounded_buffer_done(in_args->page_queue);
mutex_lock(in_args->done_mutex);
cond_signal(in_args->done_cond);
mutex_unlock(in_args->done_mutex);
}
return NULL;
}
void*
fill_buffer(void* arg){
params* p;
buffer* buf;
char temp[BUFFER_LENGTH];
FILE* file;
p=(params*)arg;
buf=p->buf;
file = open_file(p->dictionary); /* Open file */
do{
if(sem_wait(p->empty)==-1)
perror("error on sem empty");
if(buf->nb_elem < buf->size){
if(get_next(file, temp)<0){
p->found = true;
printf("Password not found\n");
}
bounded_buffer_put(buf, temp);
}
sem_post(p->full);
}while(temp!=NULL && !p->found);
close_file(file);
pthread_exit(NULL);
}
int crawl(char *start_url, int download_workers, int parse_workers, int queue_size,
char *(*_fetch_fn)(char *url), void (*_edge_fn)(char *from, char *to)) {
int i;
bounded_buffer_t url_queue;
unbounded_buffer_t page_queue;
hashset_t url_set;
bounded_buffer_init(&url_queue, queue_size);
unbounded_buffer_init(&page_queue);
hashset_init(&url_set, HASHSET_BUCKETS);
bounded_buffer_put(&url_queue, (void *)str_duplicate(start_url));
mutex_t done_mutex;
cond_t done_cond;
mutex_init(&done_mutex);
cond_init(&done_cond);
struct input_args in_args;
in_args.url_queue = &url_queue;
in_args.page_queue = &page_queue;
in_args.url_set = &url_set;
in_args.fetch = _fetch_fn;
in_args.edge = _edge_fn;
in_args.done_mutex = &done_mutex;
in_args.done_cond = &done_cond;
thread_t downloaders[download_workers];
thread_t parsers[parse_workers];
for (i = 0; i < download_workers; i++)
thread_create(&downloaders[i], downloader, (void *)&in_args);
for (i = 0; i < parse_workers; i++)
thread_create(&parsers[i], parser, (void *)&in_args);
while (1) {
mutex_lock(&done_mutex);
mutex_lock(&url_queue.mutex);
mutex_lock(&url_queue.worker_mutex);
mutex_lock(&page_queue.mutex);
mutex_lock(&page_queue.worker_mutex);
if (url_queue.count == 0 && url_queue.workers == 0 &&
page_queue.count == 0 && page_queue.workers == 0) {
url_queue.done = 1;
page_queue.done = 1;
cond_broadcast(&url_queue.empty);
cond_broadcast(&url_queue.fill);
cond_broadcast(&page_queue.fill);
mutex_unlock(&url_queue.mutex);
mutex_unlock(&url_queue.worker_mutex);
mutex_unlock(&page_queue.mutex);
mutex_unlock(&page_queue.worker_mutex);
mutex_unlock(&done_mutex);
break;
} else {
mutex_unlock(&url_queue.mutex);
mutex_unlock(&url_queue.worker_mutex);
mutex_unlock(&page_queue.mutex);
mutex_unlock(&page_queue.worker_mutex);
cond_wait(&done_cond, &done_mutex);
mutex_unlock(&done_mutex);
}
}
for (i = 0; i < download_workers; i++)
thread_join(downloaders[i], NULL);
for (i = 0; i < parse_workers; i++)
thread_join(parsers[i], NULL);
bounded_buffer_destroy(&url_queue);
unbounded_buffer_destroy(&page_queue);
hashset_destroy(&url_set);
return 0;
}