void find_bcd_device() {
DIR *dirp;
struct dirent *dp;
dirp = opendir("/sys/bus/usb/devices/");
while ((dp = readdir(dirp)) != NULL) {
if (dp->d_type != DT_LNK) continue;
struct dirent *sub_dp;
DIR *sub_dirp;
char *dir_path = join_char(join_char("/sys/bus/usb/devices/", dp->d_name), "/");
sub_dirp = opendir(dir_path);
int pid_ok = 0;
int vid_ok = 0;
while ((sub_dp = readdir(sub_dirp)) != NULL) {
if (strcmp(sub_dp->d_name,"idProduct") == 0) {
int file = open(join_char(dir_path, sub_dp->d_name), O_RDONLY);
char spid[4];
read(file, spid, 4);
if (spid[0] == '0' && spid[1] == '0' && spid[2] == '0' && spid[3] == '8') {
pid_ok = 1;
}
close(file);
}
if (strcmp(sub_dp->d_name,"idVendor") == 0) {
int file = open(join_char(dir_path, sub_dp->d_name), O_RDONLY);
char svid[4];
read(file, svid, 4);
if (svid[0] == '2' && svid[1] == '0' && svid[2] == 'b' && svid[3] == '1') {
vid_ok = 1;
}
close(file);
}
}
if (pid_ok && vid_ok) {
int file = open(join_char(dir_path, "bcdDevice"), O_RDONLY);
char bcdD[5];
read(file, bcdD, 4);
bcdDevice = atoi(bcdD);
close(file);
break;
}
}
printf("bcdDevice = %d\n", bcdDevice);
closedir(dirp);
}
//有多少个词条就开多少个线程来模拟分叉树算法，选出词条最多的一个或几个作为分词结果
int _tree_cut_words(struct TREE_CUT_WORDS_DATA *parent_cut_words_data) {
english_word_cut((*parent_cut_words_data).cut_word_result, &((*parent_cut_words_data).remainder_words),
(*parent_cut_words_data).words_weight);
//取出第一个汉子，查询redis词条，for循环生产不同的父级分词结果，然后传给新建的子进程,进入递归
char *first_chinese_word;
first_chinese_word = (char *) malloc(chinese_char_byte);
memset(first_chinese_word, 0, chinese_char_byte);
if (!first_chinese_word) {
//todo.每一个可能的错误都要做处理或日志记录.not only malloc
printf("Not Enough Memory!/n");
}
strncpy(first_chinese_word, parent_cut_words_data->remainder_words, chinese_word_byte);
strcat(first_chinese_word, "\0");
char *get_word_item_cmd = join_char(redis_get_common, first_chinese_word);
//用完释放内存
free(first_chinese_word);
redisContext *redis_connect;
get_redis_connect(&redis_connect);
redisReply *redis_reply = redisCommand(redis_connect, get_word_item_cmd);
char *match_words_json_str = redis_reply->str;
free(get_word_item_cmd);
cJSON *match_words_json = cJSON_Parse(match_words_json_str);
//是否已经到最后一个词
bool NEED_not_match_word_end_cut;
//有多少词条是否包含在搜索词里
int8_t match_words_item_number = 0;
//判断是否查找到redis词条
if (redis_reply->type == REDIS_REPLY_STRING) {
//pullwords_result_number要先计算好有多少种结果。
for (int tree_arg_index = 0; tree_arg_index < cJSON_GetArraySize(match_words_json); tree_arg_index++) {
cJSON *subitem = cJSON_GetArrayItem(match_words_json, tree_arg_index);
//判断词条是否包含在搜索词里
if (strstr(parent_cut_words_data->remainder_words, subitem->valuestring) != NULL) {
match_words_item_number++;
}
};
} else {
NEED_not_match_word_end_cut = false;
}
if (match_words_item_number == 0) {
NEED_not_match_word_end_cut = true;
}
if (NEED_not_match_word_end_cut == false) {
int8_t add_result_number = 0;
//pullwords_result_number要做原子操作，递增，支持并发
add_result_number = match_words_item_number - 1;
__sync_fetch_and_add((*parent_cut_words_data).pullwords_result_number, add_result_number);
for (int tree_arg_index = 0; tree_arg_index < cJSON_GetArraySize(match_words_json); tree_arg_index++) {
cJSON *subitem = cJSON_GetArrayItem(match_words_json, tree_arg_index);
//判断词条是否包含在搜索词里,如果在，进入树分词算法
if (strstr(parent_cut_words_data->remainder_words, subitem->valuestring) != NULL) {
//复制一份分词结果,生出父级分词结果
struct TREE_CUT_WORDS_DATA *tree_cut_words_data = (struct TREE_CUT_WORDS_DATA *) malloc(
sizeof(struct TREE_CUT_WORDS_DATA));
memset(tree_cut_words_data, 0, sizeof(struct TREE_CUT_WORDS_DATA));
init_tree_cut_words_data(tree_cut_words_data);
strcpy(tree_cut_words_data->cut_word_result, parent_cut_words_data->cut_word_result);
strcpy(tree_cut_words_data->remainder_words, parent_cut_words_data->remainder_words);
*tree_cut_words_data->words_weight = *(parent_cut_words_data->words_weight);
//todo 把pullwords主线程的data封装成一个指针，减少指针创建,.
tree_cut_words_data->pullwords_result_queue = parent_cut_words_data->pullwords_result_queue;
tree_cut_words_data->pullwords_result_queue_write_lock = parent_cut_words_data->pullwords_result_queue_write_lock;
tree_cut_words_data->pullwords_cond = parent_cut_words_data->pullwords_cond;
(*tree_cut_words_data).pullwords_result_number = (*parent_cut_words_data).pullwords_result_number;
(*tree_cut_words_data).pullwords_result_finish_number = (*parent_cut_words_data).pullwords_result_finish_number;
chinese_word_cut(tree_cut_words_data->cut_word_result, &(tree_cut_words_data->remainder_words),
tree_cut_words_data->words_weight, subitem->valuestring);
//还有词没分完，进入递归分词
if (strlen(tree_cut_words_data->remainder_words) > 0) {
pthread_t pthread_id;
int pthread_create_result;
pthread_create_result = pthread_create(&pthread_id, NULL,
(void *) _tree_cut_words, tree_cut_words_data);
if (pthread_create_result != 0) {
//todo,deal the error,.
printf("Create pthread error!\n");
}
} else {
//创建线程最终分词结果，写入主线程队列
struct TREE_CUT_WORDS_RESULT *last_tree_cut_words_result = (struct TREE_CUT_WORDS_RESULT *) malloc(
sizeof(struct TREE_CUT_WORDS_RESULT));
memset(last_tree_cut_words_result, 0, sizeof(struct TREE_CUT_WORDS_RESULT));
strcpy(last_tree_cut_words_result->pullwords_result, tree_cut_words_data->cut_word_result);
last_tree_cut_words_result->words_weight = *(tree_cut_words_data->words_weight);
//分词结束，加锁，写入分词结果到主线程的分词队列
pthread_mutex_lock(tree_cut_words_data->pullwords_result_queue_write_lock);
EnQueue((*tree_cut_words_data).pullwords_result_queue, last_tree_cut_words_result);
*((*parent_cut_words_data).pullwords_result_finish_number) =
(*((*parent_cut_words_data).pullwords_result_finish_number)) + 1;
//加锁比较pullwords_result_finish_number跟pullwords_result_number，如果全部作业完成，通知主进程
if (*((*parent_cut_words_data).pullwords_result_finish_number) ==
*((*parent_cut_words_data).pullwords_result_number)) {
pthread_cond_signal((*parent_cut_words_data).pullwords_cond);
}
pthread_mutex_unlock(tree_cut_words_data->pullwords_result_queue_write_lock);
}
}
}
} else {
//首汉子没有匹配到字典词条 OR 没有查找到redis词条，选取剩下的字作为一个半词
not_match_word_end_cut(parent_cut_words_data->cut_word_result, &(parent_cut_words_data->remainder_words),
parent_cut_words_data->words_weight);
//创建线程最终分词结果，写入主线程队列
struct TREE_CUT_WORDS_RESULT *last_tree_cut_words_result = (struct TREE_CUT_WORDS_RESULT *) malloc(
sizeof(struct TREE_CUT_WORDS_RESULT));
memset(last_tree_cut_words_result, 0, sizeof(struct TREE_CUT_WORDS_RESULT));
strcpy(last_tree_cut_words_result->pullwords_result, parent_cut_words_data->cut_word_result);
last_tree_cut_words_result->words_weight = *(parent_cut_words_data->words_weight);
//分词结束，加锁，写入分词结果到主线程的分词队列
pthread_mutex_lock(parent_cut_words_data->pullwords_result_queue_write_lock);
EnQueue((*parent_cut_words_data).pullwords_result_queue, last_tree_cut_words_result);
*((*parent_cut_words_data).pullwords_result_finish_number) =
(*((*parent_cut_words_data).pullwords_result_finish_number)) + 1;
//加锁比较pullwords_result_finish_number跟pullwords_result_number，如果全部作业完成，通知主进程
if (*((*parent_cut_words_data).pullwords_result_finish_number) ==
*((*parent_cut_words_data).pullwords_result_number)) {
pthread_cond_signal((*parent_cut_words_data).pullwords_cond);
}
pthread_mutex_unlock(parent_cut_words_data->pullwords_result_queue_write_lock);
}
free(parent_cut_words_data->cut_word_result);
free(parent_cut_words_data->words_weight);
free(parent_cut_words_data->origin_remainder_words);
free(parent_cut_words_data);
return 0;
}
/*
* The function may product many thread to deal the participle.
* I use recursion
* */
int _tree_cut_words(struct TREE_CUT_WORDS_DATA *parent_cut_words_data)
{
not_chinese_word_cut((*parent_cut_words_data).cut_word_result, &((*parent_cut_words_data).remainder_words),
(*parent_cut_words_data).words_weight);
char *first_chinese_word;
first_chinese_word = (char *) malloc(chinese_char_byte);
memset(first_chinese_word, 0, chinese_char_byte);
if (!first_chinese_word) {
//todo.deal the error,not just printf.
printf("Not Enough Memory!/n");
}
strncpy(first_chinese_word, parent_cut_words_data->remainder_words, chinese_word_byte);
strcat(first_chinese_word, "\0");
char *get_word_item_cmd = join_char(redis_get_common, first_chinese_word);
//free Memory
free(first_chinese_word);
redisContext *redis_connect;
get_redis_connect(&redis_connect);
redisReply *redis_reply = redisCommand(redis_connect, get_word_item_cmd);
char *match_words_json_str = redis_reply->str;
free(get_word_item_cmd);
cJSON *match_words_json = cJSON_Parse(match_words_json_str);
//The variable is check the remain words is the end or not
bool NEED_not_match_word_end_cut;
/*
* count the number,how many words items that in dictionary is also a path in the remain words.
* */
int8_t match_words_item_number = 0;
//check that, if one or more word item in dictionary has this fhinese chinese word.
if (redis_reply->type == REDIS_REPLY_STRING) {
for (int tree_arg_index = 0; tree_arg_index < cJSON_GetArraySize(match_words_json); tree_arg_index++) {
cJSON *subitem = cJSON_GetArrayItem(match_words_json, tree_arg_index);
//check the valuestring is in the remain words or not.
if (strstr(parent_cut_words_data->remainder_words, subitem->valuestring) != NULL) {
match_words_item_number++;
}
};
} else {
NEED_not_match_word_end_cut = false;
}
if (match_words_item_number == 0) {
NEED_not_match_word_end_cut = true;
}
if (NEED_not_match_word_end_cut == false) {
int8_t add_result_number = 0;
add_result_number = match_words_item_number - 1;
__sync_fetch_and_add((*parent_cut_words_data).pullwords_result_number, add_result_number);
for (int tree_arg_index = 0; tree_arg_index < cJSON_GetArraySize(match_words_json); tree_arg_index++) {
cJSON *subitem = cJSON_GetArrayItem(match_words_json, tree_arg_index);
//check the valuestring is in the remain words or not.
if (strstr(parent_cut_words_data->remainder_words, subitem->valuestring) != NULL) {
/*
* copy the participle result.send it to the children thread.
* */
struct TREE_CUT_WORDS_DATA *tree_cut_words_data = (struct TREE_CUT_WORDS_DATA *) malloc(
sizeof(struct TREE_CUT_WORDS_DATA));
memset(tree_cut_words_data, 0, sizeof(struct TREE_CUT_WORDS_DATA));
init_tree_cut_words_data(tree_cut_words_data);
strcpy(tree_cut_words_data->cut_word_result, parent_cut_words_data->cut_word_result);
strcpy(tree_cut_words_data->remainder_words, parent_cut_words_data->remainder_words);
*tree_cut_words_data->words_weight = *(parent_cut_words_data->words_weight);
//todo,include the five point to one point.
tree_cut_words_data->pullwords_result_queue = parent_cut_words_data->pullwords_result_queue;
tree_cut_words_data->pullwords_result_queue_write_lock = parent_cut_words_data->pullwords_result_queue_write_lock;
tree_cut_words_data->pullwords_cond = parent_cut_words_data->pullwords_cond;
(*tree_cut_words_data).pullwords_result_number = (*parent_cut_words_data).pullwords_result_number;
(*tree_cut_words_data).pullwords_result_finish_number = (*parent_cut_words_data).pullwords_result_finish_number;
chinese_word_cut(tree_cut_words_data->cut_word_result, &(tree_cut_words_data->remainder_words),
tree_cut_words_data->words_weight, subitem->valuestring);
//not finish,create another thread to deal with it.
if (strlen(tree_cut_words_data->remainder_words) > 0) {
pthread_t pthread_id;
int pthread_create_result;
pthread_create_result = pthread_create(&pthread_id, NULL,
(void *) _tree_cut_words, tree_cut_words_data);
if (pthread_create_result != 0) {
//todo,deal the error,.
printf("Create pthread error!\n");
}
} else {
//participle is finished.
//create the last participle result;
struct TREE_CUT_WORDS_RESULT *last_tree_cut_words_result = (struct TREE_CUT_WORDS_RESULT *) malloc(
sizeof(struct TREE_CUT_WORDS_RESULT));
memset(last_tree_cut_words_result, 0, sizeof(struct TREE_CUT_WORDS_RESULT));
strcpy(last_tree_cut_words_result->pullwords_result, tree_cut_words_data->cut_word_result);
last_tree_cut_words_result->words_weight = *(tree_cut_words_data->words_weight);
//lock
pthread_mutex_lock(tree_cut_words_data->pullwords_result_queue_write_lock);
//write the result to the queue.
EnQueue((*tree_cut_words_data).pullwords_result_queue, last_tree_cut_words_result);
*((*parent_cut_words_data).pullwords_result_finish_number) =
(*((*parent_cut_words_data).pullwords_result_finish_number)) + 1;
//campare the need finish and had finish number. If equal.notify the first parent thread.
if (*((*parent_cut_words_data).pullwords_result_finish_number) ==
*((*parent_cut_words_data).pullwords_result_number)) {
pthread_cond_signal((*parent_cut_words_data).pullwords_cond);
}
pthread_mutex_unlock(tree_cut_words_data->pullwords_result_queue_write_lock);
}
}
}
} else {
/*
* use the first chinese word to find the word item in dictionary.If find nothing,It is end.finish
* If find some word items,but no one in the remain words,It is end.finish
* */
not_match_word_end_cut(parent_cut_words_data->cut_word_result, &(parent_cut_words_data->remainder_words),
parent_cut_words_data->words_weight);
//create the last participle result;
struct TREE_CUT_WORDS_RESULT *last_tree_cut_words_result = (struct TREE_CUT_WORDS_RESULT *) malloc(
sizeof(struct TREE_CUT_WORDS_RESULT));
memset(last_tree_cut_words_result, 0, sizeof(struct TREE_CUT_WORDS_RESULT));
strcpy(last_tree_cut_words_result->pullwords_result, parent_cut_words_data->cut_word_result);
last_tree_cut_words_result->words_weight = *(parent_cut_words_data->words_weight);
//lock
pthread_mutex_lock(parent_cut_words_data->pullwords_result_queue_write_lock);
//write the result to the queue.
EnQueue((*parent_cut_words_data).pullwords_result_queue, last_tree_cut_words_result);
*((*parent_cut_words_data).pullwords_result_finish_number) =
(*((*parent_cut_words_data).pullwords_result_finish_number)) + 1;
//campare the need finish and had finish number. If equal.notify the first parent thread.
if (*((*parent_cut_words_data).pullwords_result_finish_number) ==
*((*parent_cut_words_data).pullwords_result_number)) {
pthread_cond_signal((*parent_cut_words_data).pullwords_cond);
}
pthread_mutex_unlock(parent_cut_words_data->pullwords_result_queue_write_lock);
}
//free the data
free(parent_cut_words_data->cut_word_result);
free(parent_cut_words_data->words_weight);
free(parent_cut_words_data->origin_remainder_words);
free(parent_cut_words_data);
return 0;
}
