static void scws_rule_json_set_attrs(rule_t rules, rule_item_t rule, cJSON *attrs)
{
char *attrname;
cJSON *json_attr;
json_attr = attrs->child;
// printf("Setting attr: %s\n", rule->name);
while (json_attr != NULL) {
attrname = json_attr->string;
if (!strcmp(attrname, "tf"))
rule->tf = (float)json_attr->valuedouble;
else if (!strcmp(attrname, "idf"))
rule->idf = (float)json_attr->valuedouble;
else if (!strcmp(attrname, "attr"))
strncpy(rule->attr, json_attr->valuestring, 2);
else if (!strcmp(attrname, "znum")) {
// znum: {min: a, max: b} | a
if (json_attr->type == cJSON_Object) {
if (cJSON_GetObjectItem(json_attr, "min"))
rule->zmin = (int)cJSON_GetObjectItem(json_attr, "min")->valueint;
if (cJSON_GetObjectItem(json_attr, "max"))
rule->zmax = (int)cJSON_GetObjectItem(json_attr, "max")->valueint;
}
else if (json_attr->type == cJSON_Number)
rule->zmin = (int)json_attr->valueint;
}
else if (!strcmp(attrname, "type")) {
if (!strncmp(json_attr->valuestring, "prefix", 6))
rule->flag |= SCWS_ZRULE_PREFIX;
else if(!strncmp(json_attr->valuestring, "suffix", 6))
rule->flag |= SCWS_ZRULE_SUFFIX;
}
else if (!strcmp(attrname, "include") || !strcmp(attrname, "exclude")) {
unsigned int *clude;
if (!strcmp(attrname, "include")) {
clude = &rule->inc;
rule->flag |= SCWS_ZRULE_INCLUDE;
}
else {
clude = &rule->exc;
rule->flag |= SCWS_ZRULE_EXCLUDE;
}
// loop thru xclude values
int k, j;
for (k = 0; k < cJSON_GetArraySize(json_attr); k++) {
char *cludename = cJSON_GetArrayItem(json_attr, k)->valuestring;
if ((j = _rule_index_get(rules, cludename)) >= 0)
*clude |= rules->items[j].bit;
}
}
json_attr = json_attr->next;
}
}
rule_t scws_rule_new(const char *fpath, unsigned char *mblen)
{
FILE *fp;
rule_t r;
rule_item_t cr;
int i, j, rbl, aflag;
rule_attr_t a,rtail;
unsigned char buf[512], *str, *ptr, *qtr;
/* loaded or open file failed */
if ((fp = fopen(fpath, "r")) == NULL)
return NULL;
/* alloc the memory */
r = (rule_t) malloc(sizeof(rule_st));
memset(r, 0, sizeof(rule_st));
/* quick scan to add the name to list */
i = j = rbl = aflag = 0;
while (fgets(buf, sizeof(buf)-1, fp))
{
if (buf[0] != '[' || !(ptr = strchr(buf, ']')))
continue;
str = buf + 1;
*ptr = '\0';
if (ptr == str || (ptr-str) > 15 || !strcasecmp(str, "attrs"))
continue;
if (_rule_index_get(r, str) >= 0)
continue;
strcpy(r->items[i].name, str);
r->items[i].tf = 5.0;
r->items[i].idf = 3.5;
strncpy(r->items[i].attr, "un", 2);
if (!strcasecmp(str, "special"))
r->items[i].bit = SCWS_RULE_SPECIAL;
else if (!strcasecmp(str, "nostats"))
r->items[i].bit = SCWS_RULE_NOSTATS;
else
{
r->items[i].bit = (1<<j);
j++;
}
if (++i >= SCWS_RULE_MAX)
break;
}
rewind(fp);
/* load the tree data */
if ((r->tree = xtree_new(0, 1)) == NULL)
{
free(r);
return NULL;
}
cr = NULL;
while (fgets(buf, sizeof(buf)-1, fp))
{
if (buf[0] == ';')
continue;
if (buf[0] == '[')
{
cr = NULL;
str = buf + 1;
aflag = 0;
if ((ptr = strchr(str, ']')) != NULL)
{
*ptr = '\0';
if (!strcasecmp(str, "attrs"))
{
aflag = 1;
}
else if ((i = _rule_index_get(r, str)) >= 0)
{
rbl = 1; /* default read by line = yes */
cr = &r->items[i];
}
}
continue;
}
/* attr flag open? */
if (aflag == 1)
{
/* parse the attr line */
str = buf;
while (*str == ' ' || *str == '\t') str++;
if ((ptr = strchr(str, '+')) == NULL) continue;
*ptr++ = '\0';
if ((qtr = strchr(ptr, '=')) == NULL) continue;
*qtr++ = '\0';
/* create new memory */
a = (rule_attr_t) malloc(sizeof(struct scws_rule_attr));
memset(a, 0, sizeof(struct scws_rule_attr));
/* get ratio */
while(*qtr == ' ' || *qtr == '\t') qtr++;
a->ratio = (short) atoi(qtr);
if (a->ratio < 1)
a->ratio = 1;
a->npath[0] = a->npath[1] = 0xff;
/* read attr1 & npath1? */
a->attr1[0] = *str++;
if (*str && *str != '(' && *str != ' ' && *str != '\t')
a->attr1[1] = *str++;
while (*str && *str != '(') str++;
if (*str == '(')
{
str++;
if ((qtr = strchr(str, ')')) != NULL)
{
*qtr = '\0';
a->npath[0] = (unsigned char) atoi(str);
if (a->npath[0] > 0)
a->npath[0]--;
else
a->npath[0] = 0xff;
}
}
/* read attr1 & npath2? */
str = ptr;
while (*str == ' ' || *str == '\t') str++;
a->attr2[0] = *str++;
if (*str && *str != '(' && *str != ' ' && *str != '\t')
a->attr2[1] = *str++;
while (*str && *str != '(') str++;
if (*str == '(')
{
str++;
if ((qtr = strchr(str, ')')) != NULL)
{
*qtr = '\0';
a->npath[1] = (unsigned char) atoi(str);
if (a->npath[1] > 0)
a->npath[1]--;
else
a->npath[1] = 0xff;
}
}
//printf("%c%c(%d)+%c%c(%d)=%d\n", a->attr1[0], a->attr1[1] ? a->attr1[1] : ' ', a->npath[0],
// a->attr2[0], a->attr2[1] ? a->attr2[1] : ' ', a->npath[1], a->ratio);
/* append to the chain list */
if (r->attr == NULL)
r->attr = rtail = a;
else
{
rtail->next = a;
rtail = a;
}
continue;
}
if (cr == NULL)
continue;
/* param set: line|znum|include|exclude|type|tf|idf|attr */
if (buf[0] == ':')
{
str = buf + 1;
if (!(ptr = strchr(str, '=')))
continue;
while (*str == ' ' || *str == '\t') str++;
qtr = ptr + 1;
while (ptr > str && (ptr[-1] == ' ' || ptr[-1] == '\t')) ptr--;
*ptr = '\0';
ptr = str;
str = qtr;
while (*str == ' ' || *str == '\t') str++;
if (!strcmp(ptr, "line"))
rbl = (*str == 'N' || *str == 'n') ? 0 : 1;
else if (!strcmp(ptr, "tf"))
cr->tf = (float) atof(str);
else if (!strcmp(ptr, "idf"))
cr->idf = (float) atof(str);
else if (!strcmp(ptr, "attr"))
strncpy(cr->attr, str, 2);
else if (!strcmp(ptr, "znum"))
{
if ((ptr = strchr(str, ',')) != NULL)
{
*ptr++ = '\0';
while (*ptr == ' ' || *ptr == '\t') ptr++;
cr->zmax = atoi(ptr);
cr->flag |= SCWS_ZRULE_RANGE;
}
cr->zmin = atoi(str);
}
else if (!strcmp(ptr, "type"))
{
if (!strncmp(str, "prefix", 6))
cr->flag |= SCWS_ZRULE_PREFIX;
else if (!strncmp(str, "suffix", 6))
cr->flag |= SCWS_ZRULE_SUFFIX;
}
else if (!strcmp(ptr, "include") || !strcmp(ptr, "exclude"))
{
unsigned int *clude;
if (!strcmp(ptr, "include"))
{
clude = &cr->inc;
cr->flag |= SCWS_ZRULE_INCLUDE;
}
else
{
clude = &cr->exc;
cr->flag |= SCWS_ZRULE_EXCLUDE;
}
while ((ptr = strchr(str, ',')) != NULL)
{
while (ptr > str && (ptr[-1] == '\t' || ptr[-1] == ' ')) ptr--;
*ptr = '\0';
if ((i = _rule_index_get(r, str)) >= 0)
*clude |= r->items[i].bit;
str = ptr + 1;
while (*str == ' ' || *str == '\t' || *str == ',') str++;
}
ptr = strlen(str) + str;
while (ptr > str && strchr(" \t\r\n", ptr[-1])) ptr--;
*ptr = '\0';
if (ptr > str && (i = _rule_index_get(r, str)))
*clude |= r->items[i].bit;
}
continue;
}
/* read the entries */
str = buf;
while (*str == ' ' || *str == '\t') str++;
ptr = str + strlen(str);
while (ptr > str && strchr(" \t\r\n", ptr[-1])) ptr--;
*ptr = '\0';
/* emptry line */
if (ptr == str)
continue;
if (rbl)
xtree_nput(r->tree, cr, sizeof(struct scws_rule_item), str, ptr - str);
else
{
while (str < ptr)
{
j = mblen[(*str)];
#ifdef DEBUG
/* try to check repeat */
if ((i = (int) xtree_nget(r->tree, str, j, NULL)) != 0)
fprintf(stderr, "Reapeat word on %s|%s: %.*s\n", cr->name, ((rule_item_t) i)->name, j, str);
#endif
xtree_nput(r->tree, cr, sizeof(struct scws_rule_item), str, j);
str += j;
}
}
}
fclose(fp);
/* optimize the tree */
xtree_optimize(r->tree);
return r;
}
