static int parse_advert_packet (const char *buf, char **port, char **fs2protocol, char **id)
{
char *fs2protocol_field, *port_or_auto_field, *id_field;
fs2protocol_field = get_next_field(&buf);
if (!fs2protocol_field) goto err_fs2proto;
*fs2protocol = fs2protocol_field;
/* autoindexnode packets are the clients saying they /can/ become indexnodes
* if needed. There is also a weight field - an indication of how powerful
* the machine is - that is used to vote on the client that becomes
* indexnode. All irrelevant to us. */
port_or_auto_field = get_next_field(&buf);
if (!port_or_auto_field) goto err_port_auto;
if (!strcmp(port_or_auto_field, "autoindexnode")) goto err_id;
*port = port_or_auto_field;
id_field = get_next_field(&buf);
if (!id_field) goto err_id;
*id = id_field;
return 0;
err_id:
free(port_or_auto_field);
err_port_auto:
free(fs2protocol_field);
err_fs2proto:
return 1;
}
//---------------------------------------------------------------------------
bool read_training_data_from_proben1_format(const char *filename, char list_separator, int num_classes, double **&training_data, double *&target, int &num_training_data, int &num_variables, char* error_message)
{
FILE* f = fopen(filename, "r");
if (!f) {
strcpy(error_message, "Cannot find file! Please specify the full path!");
return false;
}
char *buf = new char[10000];
char * start_buf = buf;
// count the number of training data and the number of variables
num_training_data = 0;
while (fgets(buf, 10000, f)) {
if (strlen(buf) > 1)
num_training_data++;
if (num_training_data == 1) {
num_variables = 0;
char tmp_str[10000];
int size;
int skip_size;
bool result = get_next_field(buf, list_separator, tmp_str, size, skip_size);
while (result) {
buf = buf + size + 1 + skip_size;
result = get_next_field(buf, list_separator, tmp_str, size, skip_size);
num_variables++;
}
}
buf = start_buf;
}
delete[] start_buf;
num_variables--;
rewind(f);
num_variables -= num_classes - 1;
allocate_training_data(training_data, target, num_training_data, num_variables);
int value;
for (int i = 0; i < num_training_data; i++) {
for (int j = 0; j < num_variables; j++)
fscanf(f, "%lf", &training_data[i][j]);
for (int j = 0; j < num_classes; j++) {
int num_read = fscanf(f, "%d", &value);
if (num_read != 1) {
strcpy(error_message, "Incorrect format!");
return false;
}
if (value == 1)
target[i] = j;
}
}
fclose(f);
return true;
}
/*
Add a new obj and the node id to the prefix tree.
Repeate obj and nid association will not be added twice
*/
int pft_add_obj(char *obj, uint32_t nid)
{
assert(strlen(obj) <= OBJ_NAME_LEN);
char buff[OBJ_NAME_LEN + 1];
int start_idx = 0;
pf_node_t **curr_lv_rt = &g_obj_pft; // current level root prefix
pf_node_t *curr_node = *curr_lv_rt;
pf_node_t *target = NULL;
while(0 == get_next_field(obj, start_idx, '/', buff, &start_idx))
{
// search the current level prefix
while(curr_node)
{
if(!strcmp(buff, curr_node->prefix))
break;
curr_node = curr_node->next_peer;
}
if(!curr_node) // not found
{
curr_node = new_node();
if(!curr_node)
return -1;
strcpy(curr_node->prefix, buff);
curr_node->next_peer = *curr_lv_rt;
*curr_lv_rt = curr_node;
}
target = curr_node;
curr_lv_rt = &curr_node->next_child;
curr_node = *curr_lv_rt;
}
assert(target != NULL);
// store the nid to the current prefix level node
if(target->darr == NULL)
{
target->darr = new_darr();
if(target->darr == NULL)
return -1;
}
// check repeating before add
if(-1 == get_elem_idx(target->darr, nid))
{
if(-1 == add_elem(target->darr, nid))
return -1;
}
return 0;
}
/*
Search the prefix tree for an obj, return the array of nodes that contains
the object, if there is.
The all_pre param will always be set with a not-NULL value. This array may
contain all all nodes that claimed an aggregate prefix.
The last_pre param may be set to point to a array containing nodes with
the longest aggragate prefix match.
For example, for a obj '/a/b/obj', if node 1, 2, 3 all claim to have the
obj and node 1 also claim to have /a/, node 2 claims to have /a/b/.
Then the darr return will indicate node 1, 2, 3,
The last_pre will indicate node 2
All pre will indicate node 1 and 2.
*/
darr_t *pft_get_obj_list(const char *obj, darr_t **last_pre, darr_t **all_pre)
{
char buff[OBJ_NAME_LEN + 1];
int start_idx = 0;
pf_node_t *curr_node = g_obj_pft;
pf_node_t *target = NULL;
static darr_t *al_pre = NULL;
if(al_pre == NULL)
{
al_pre = new_darr();
if(!al_pre)
return NULL;
}
*all_pre = NULL;
*last_pre = NULL;
al_pre->curr_elem = 0;
while(0 == get_next_field(obj, start_idx, '/', buff, &start_idx))
{
while(curr_node)
{
if(!strcmp(buff, curr_node->prefix))
break;
curr_node = curr_node->next_peer;
}
if(!curr_node) // not found
{
break; // longest prefix match up to now
}
target = curr_node;
curr_node = curr_node->next_child;
if(target->darr && target->darr->curr_elem != 0)
{
merge_darr(al_pre, target->darr);
*last_pre = target->darr;
}
}
if(!target)
return NULL;
if(al_pre->curr_elem > 0) //remove the match node's last appned id
{
al_pre->curr_elem -= target->darr->curr_elem;
}
*all_pre = al_pre;
return target->darr;
}
void read_stat(char *proc_id,override_type *entry,int k,int uid)
{
char *stat_filename=make_name(proc_id,1);
int done=0;
char *field;
FILE *fp;
int n; /* current field number */
char *proc_name;
unsigned utime;
int priority;
int nice;
if ((fp=fopen(stat_filename,"r"))==NULL)
fprintf(stderr,"Error: failed to open '%s'!\n",stat_filename);
else
{
for (n=0; !done && (field=get_next_field(fp)); n++)
switch (n)
{
case 1: /* proc name */
proc_name=alloca(strlen(field)+1);
strcpy(proc_name,strip_parens(field));
break;
case 13: /* utime */
utime=atol(field)/(float)USER_HZ;
break;
case 17: /* priority */
priority=atoi(field);
break;
case 18: /* nice */
nice=atoi(field);
done=1; /* end iteration after last interesting field */
break;
}
fclose(fp);
}
if (done) /* reached last field OK */
{
/* if exceeded defined kill threshold and not an exception... */
if (k>0 && (!entry || entry->elapsed!=-1) && utime>k)
kill_proc(atoi(proc_id),uid,k);
else
adjust_sched(atoi(proc_id),proc_name,utime,priority,nice,entry);
}
}
string get_field(char *data,
int length,
int num)
{
/*
* While num>1 and there are fields left, skip a field & decrement num:
*/
while (length && num>1) {
if (!*data)
num--;
length--;
data++;
}
/*
* If any more fields left, the first field is the one we want.
* Otherwise, there is no such field as num -- return "".
*/
if (length)
return(get_next_field(&data, &length));
else
return(string_Copy(""));
}
/*
Remove an object associated with a node
*/
void pft_rm_obj(char *obj, uint32_t nid)
{
char buff[OBJ_NAME_LEN + 1];
int start_idx = 0;
size_t idx;
pf_node_t *curr_node = g_obj_pft;
pf_node_t *target = NULL;
while(0 == get_next_field(obj, start_idx, '/', buff, &start_idx))
{
while(curr_node)
{
if(!strcmp(buff, curr_node->prefix))
break;
curr_node = curr_node->next_peer;
}
if(!curr_node) // not found
{
return ;
}
target = curr_node;
curr_node = curr_node->next_child;
}
if(!target)
return ;
// check existance before remove
if(-1 != (idx = get_elem_idx(target->darr, nid)))
{
rm_elem_by_idx(target->darr, idx);
}
}
//-----------------------------------------------------------------
bool t_mep_data::from_csv_double(const char *filename)
{
FILE *f = NULL;
#ifdef WIN32
int count_chars = MultiByteToWideChar(CP_UTF8, 0, filename, -1, NULL, 0);
wchar_t *w_filename = new wchar_t[count_chars];
MultiByteToWideChar(CP_UTF8, 0, filename, -1, w_filename, count_chars);
f = _wfopen(w_filename, L"r");
delete[] w_filename;
#else
f = fopen(filename, "r");
#endif
if (!f)
return false;
delete_data();
char *buf = new char[MAX_ROW_CHARS];
char * start_buf = buf;
num_data = 0;
data_type = MEP_DATA_DOUBLE;
while (my_fgets(buf, MAX_ROW_CHARS, f)) {
long len = strlen(buf);
if (len > 1)
num_data++;
if (num_data == 1) {
num_cols = 0;
char tmp_str[1000];
int size;
int skipped;
bool result = get_next_field(buf, list_separator, tmp_str, size, skipped);
while (result) {
num_cols++;
if (!buf[size])
break;
buf = buf + size + skipped;
result = get_next_field(buf, list_separator, tmp_str, size, skipped);
}
buf = start_buf;
}
}
// num_cols--;
rewind(f);
_data_double = new double*[num_data];
int count_mep_data = 0;
while (my_fgets(buf, MAX_ROW_CHARS, f)) {
long len = strlen(buf);
if (len > 1) {
int col = 0;
char tmp_str[MAX_ROW_CHARS];
int size;
int skipped;
_data_double[count_mep_data] = new double[num_cols];
bool result = get_next_field(buf, list_separator, tmp_str, size, skipped);
while (result) {
if (col < num_cols)
_data_double[count_mep_data][col] = atof(tmp_str);
else {
break;
}
buf = buf + size + skipped;
result = get_next_field(buf, list_separator, tmp_str, size, skipped);
col++;
}
count_mep_data++;
}
buf = start_buf;
}
fclose(f);
delete[] buf;
return true;
}
//-----------------------------------------------------------------
bool t_mep_data::from_csv_string(const char *filename)
{
FILE *f = NULL;
#ifdef WIN32
int count_chars = MultiByteToWideChar(CP_UTF8, 0, filename, -1, NULL, 0);
wchar_t *w_filename = new wchar_t[count_chars];
MultiByteToWideChar(CP_UTF8, 0, filename, -1, w_filename, count_chars);
f = _wfopen(w_filename, L"r");
delete[] w_filename;
#else
f = fopen(filename, "r");
#endif
if (!f)
return false;
delete_data();
char *buf = new char[MAX_ROW_CHARS];
char * start_buf = buf;
data_type = MEP_DATA_STRING;
num_data = 0;
while (my_fgets(buf, MAX_ROW_CHARS, f)) {
long len = strlen(buf);
if (len > 1)
num_data++;
if (num_data == 1) {
num_cols = 0;
int skipped;
char tmp_str[1000];
int size;
bool result = get_next_field(buf, list_separator, tmp_str, size, skipped);
while (result) {
num_cols++;
if (!buf[size + skipped])
break;
buf = buf + size + skipped;
result = get_next_field(buf, list_separator, tmp_str, size, skipped);
}
buf = start_buf;
}
}
// num_cols--;
rewind(f);
_data_string = new char**[num_data];
int count_mep_data = 0;
//has_missing_values = 0;
while (my_fgets(buf, MAX_ROW_CHARS, f)) {
long len = strlen(buf);
if (len > 1) {
int col = 0;
char tmp_str[1000];
int size;
_data_string[count_mep_data] = new char*[num_cols];
for (int c = 0; c < num_cols; c++)
_data_string[count_mep_data][col] = NULL;
int skipped = 0;
bool result = get_next_field(buf, list_separator, tmp_str, size, skipped);
while (result) {
if (col < num_cols) {
_data_string[count_mep_data][col] = new char[strlen(tmp_str) + 1];
strcpy(_data_string[count_mep_data][col], tmp_str);
}
else {
break;
}
buf = buf + size + skipped;
//if (buf - start_buf >= len)
//break;
result = get_next_field(buf, list_separator, tmp_str, size, skipped);
col++;
}
count_mep_data++;
}
buf = start_buf;
}
fclose(f);
//delete[] start_buf;
delete[] buf;
return true;
}
//-----------------------------------------------------------------
bool t_mep_data::from_PROBEN1_format(const char *filename, int num_classes)
{
FILE *f = NULL;
#ifdef WIN32
int count_chars = MultiByteToWideChar(CP_UTF8, 0, filename, -1, NULL, 0);
wchar_t *w_filename = new wchar_t[count_chars];
MultiByteToWideChar(CP_UTF8, 0, filename, -1, w_filename, count_chars);
f = _wfopen(w_filename, L"r");
delete[] w_filename;
#else
f = fopen(filename, "r");
#endif
if (!f)
return false;
delete_data();
char *buf = new char[MAX_ROW_CHARS];
char * start_buf = buf;
num_data = 0;
data_type = MEP_DATA_DOUBLE;
while (my_fgets(buf, MAX_ROW_CHARS, f)) {
long len = strlen(buf);
if (len > 1)
num_data++;
if (num_data == 1) {
num_cols = 0;
char tmp_str[1000];
int size;
int skipped;
bool result = get_next_field(buf, ' ', tmp_str, size, skipped);
while (result) {
num_cols++;
if (!buf[size])
break;
buf = buf + size + skipped;
result = get_next_field(buf, ' ', tmp_str, size, skipped);
}
buf = start_buf;
}
}
num_cols -= num_classes;
if (num_classes)
num_cols++;
rewind(f);
_data_double = new double*[num_data];
int out_class;
for (int r = 0; r < num_data; r++) {
_data_double[r] = new double[num_cols];
for (int c = 0; c < num_cols - 1; c++)
fscanf(f, "%lf", &_data_double[r][c]);
// now scan the outputs
if (num_classes) // classification problem
for (int c = 0; c < num_classes; c++) {
fscanf(f, "%d", &out_class);
if (out_class == 1)
_data_double[r][num_cols - 1] = c;
}
else// regression problem
fscanf(f, "%lf", &_data_double[r][num_cols - 1]);
}
fclose(f);
delete[] buf;
return true;
}
// ---------------------------------------------------------------------------
int t_mep_data::from_xml(pugi::xml_node parent)
{
clear_data();
pugi::xml_node node = parent.child("num_data");
if (node) {
const char *value_as_cstring = node.child_value();
num_data = atoi(value_as_cstring);
}
else
num_data = 0;
node = parent.child("num_variables");
if (node) {
const char *value_as_cstring = node.child_value();
int num_variables = atoi(value_as_cstring);
if (num_variables)
num_cols = num_variables + 1;
else
num_cols = 0;
}
else {
node = parent.child("num_cols");
if (node) {
const char *value_as_cstring = node.child_value();
num_cols = atoi(value_as_cstring);
}
else
num_cols = 0;
}
node = parent.child("num_outputs");
if (node) {
const char *value_as_cstring = node.child_value();
num_outputs = atoi(value_as_cstring);
}
else
num_outputs = 1;
node = parent.child("num_classes");
if (node) {
const char *value_as_cstring = node.child_value();
num_classes = atoi(value_as_cstring);
}
else
num_classes = 0;
node = parent.child("data_type");
if (node) {
const char *value_as_cstring = node.child_value();
data_type = atoi(value_as_cstring);
}
else
data_type = MEP_DATA_DOUBLE;// double by default
/*
node = parent.child("has_missing_values");
if (node) {
const char *value_as_cstring = node.child_value();
has_missing_values = atoi(value_as_cstring);
}
else
has_missing_values = 0;// double by default
*/
node = parent.child("list_separator");
if (node) {
const char *value_as_cstring = node.child_value();
list_separator = value_as_cstring[0];
}
else
list_separator = ' ';// blank space by default
if (data_type == MEP_DATA_DOUBLE) {
if (num_data) {
_data_double = new double*[num_data];
for (int r = 0; r < num_data; r++)
_data_double[r] = new double[num_cols];
}
}
else {
if (num_data) {
_data_string = new char**[num_data];
for (int r = 0; r < num_data; r++) {
_data_string[r] = new char*[num_cols];
for (int v = 0; v < num_cols; v++)
_data_string[r][v] = NULL;
}
}
}
pugi::xml_node node_data = parent.child("data");
if (!node_data)
return true;
int r = 0;
if (data_type == MEP_DATA_DOUBLE) {// double
for (pugi::xml_node row = node_data.child("row"); row; row = row.next_sibling("row"), r++) {
const char *value_as_cstring = row.child_value();
int num_jumped_chars = 0;
for (int c = 0; c < num_cols; c++) {
sscanf(value_as_cstring + num_jumped_chars, "%lf", &_data_double[r][c]);
long local_jump = strcspn(value_as_cstring + num_jumped_chars, " ");
num_jumped_chars += local_jump + 1;
}
}
}
else {// string
for (pugi::xml_node row = node_data.child("row"); row; row = row.next_sibling("row"), r++) {
const char *value_as_cstring = row.child_value();
char *buf = (char*)value_as_cstring;
char tmp_str[1000];
int size;
int c = 0;
int skipped;
bool result = get_next_field(buf, ' ', tmp_str, size, skipped);
while (result) {
if (c < num_cols) {
_data_string[r][c] = new char[strlen(tmp_str) + 1];
strcpy(_data_string[r][c], tmp_str);
}
buf = buf + size + 1 + skipped;
//if (buf - start_buf >= len)
// break;
result = get_next_field(buf, ' ', tmp_str, size, skipped);
c++;
}
}
}
_modified = false;
return true;
}
