node_t * ROXML_API roxml_load_buf(char *buffer)
{
node_t *current_node = NULL;
if(buffer == NULL) { return NULL; }
current_node = roxml_create_node(0, buffer, ROXML_ELM_NODE | ROXML_BUFF);
current_node = roxml_append_node(NULL, current_node);
return roxml_load(current_node, NULL, buffer);
}
node_t * ROXML_API roxml_load_doc(char *filename)
{
node_t *current_node = NULL;
FILE* file = fopen(filename, "rb");
if(file == NULL) {
return NULL;
}
current_node = roxml_create_node(0, file, ROXML_ELM_NODE | ROXML_FILE);
current_node = roxml_append_node(NULL, current_node);
return roxml_load(current_node, file, NULL);
}
int _func_load_end_node(char * chunk, void * data)
{
#ifdef DEBUG_PARSING
fprintf(stderr, "calling func %s chunk %c\n",__func__,chunk[0]);
#endif /* DEBUG_PARSING */
int cur = 1;
roxml_load_ctx_t *context = (roxml_load_ctx_t*)data;
switch(context->state) {
case STATE_NODE_BEG:
roxml_process_begin_node(context, context->pos-1);
context->state = STATE_NODE_END;
break;
case STATE_NODE_NAME:
context->state = STATE_NODE_SINGLE;
break;
case STATE_NODE_ATTR:
if((context->mode != MODE_COMMENT_DQUOTE)&&(context->mode != MODE_COMMENT_QUOTE)) {
if(context->inside_node_state == STATE_INSIDE_VAL) {
node_t * to_be_closed = NULL;
if(context->content_quoted) {
context->content_quoted = 0;
to_be_closed = roxml_create_node(context->pos-1, context->src, ROXML_ATTR_NODE | context->type);
} else {
to_be_closed = roxml_create_node(context->pos, context->src, ROXML_ATTR_NODE | context->type);
}
roxml_close_node(context->candidat_val, to_be_closed);
}
context->inside_node_state = STATE_INSIDE_ARG_BEG;
context->state = STATE_NODE_SINGLE;
}
break;
}
context->pos += cur;
return cur;
}
node_t * ROXML_API roxml_load_fd(int fd)
{
FILE * file = NULL;
node_t *current_node = NULL;
if(fd < 0) {
return NULL;
}
file = fdopen(fd, "r");
if(file == NULL) {
return NULL;
}
current_node = roxml_create_node(0, file, ROXML_ELM_NODE | ROXML_FILE);
current_node = roxml_append_node(NULL, current_node);
return roxml_load(current_node, file, NULL);
}
int _func_load_colon(char * chunk, void * data)
{
int cur = 1;
roxml_load_ctx_t *context = (roxml_load_ctx_t*)data;
#ifdef DEBUG_PARSING
fprintf(stderr, "calling func %s chunk %c\n",__func__,chunk[0]);
#endif /* DEBUG_PARSING */
if(context->state == STATE_NODE_NAME) {
context->state = STATE_NODE_BEG;
context->candidat_node->ns = roxml_lookup_nsdef(context->namespaces, context->curr_name);
if(!context->candidat_node->ns) {
char *nsname = malloc(context->curr_name_len+1);
memcpy(nsname, context->curr_name, context->curr_name_len);
nsname[context->curr_name_len] = '\0';
context->candidat_node->ns = roxml_create_node(0, nsname, ROXML_NSDEF_NODE | ROXML_PENDING | ROXML_INVALID);
}
context->candidat_node->pos += context->curr_name_len+2;
context->ns = 1;
} else if(context->state == STATE_NODE_ATTR) {
if(context->inside_node_state == STATE_INSIDE_ARG) {
context->inside_node_state = STATE_INSIDE_ARG_BEG;
if((context->curr_name_len==5)&&(strncmp(context->curr_name, "xmlns", 5) == 0)) {
context->candidat_arg->type |= ROXML_NS_NODE;
context->nsdef = 1;
} else {
context->candidat_arg->ns = roxml_lookup_nsdef(context->namespaces, context->curr_name);
context->candidat_arg->pos += context->curr_name_len+2;
context->ns = 1;
}
}
}
context->pos += cur;
return cur;
}
int _func_load_close_node(char * chunk, void * data)
{
#ifdef DEBUG_PARSING
fprintf(stderr, "calling func %s chunk %c\n",__func__,chunk[0]);
#endif /* DEBUG_PARSING */
int cur = 1;
roxml_load_ctx_t *context = (roxml_load_ctx_t*)data;
switch(context->state) {
case STATE_NODE_NAME:
context->empty_text_node = 1;
context->current_node = roxml_parent_node(context->current_node, context->candidat_node, 0);
break;
case STATE_NODE_ATTR:
if((context->mode != MODE_COMMENT_DQUOTE)||(context->mode != MODE_COMMENT_QUOTE)) {
if(context->inside_node_state == STATE_INSIDE_VAL) {
node_t * to_be_closed = NULL;
if(context->content_quoted) {
context->content_quoted = 0;
to_be_closed = roxml_create_node(context->pos-1, context->src, ROXML_ATTR_NODE | context->type);
} else {
to_be_closed = roxml_create_node(context->pos, context->src, ROXML_ATTR_NODE | context->type);
}
roxml_close_node(context->candidat_val, to_be_closed);
}
context->current_node = roxml_parent_node(context->current_node, context->candidat_node, 0);
context->inside_node_state = STATE_INSIDE_ARG_BEG;
roxml_process_unaliased_ns(context);
} else {
context->pos++;
return 1;
}
break;
case STATE_NODE_SINGLE:
if(context->doctype) {
context->doctype--;
if(context->doctype > 0) {
context->pos++;
return 1;
}
context->candidat_node->end = context->pos;
}
context->empty_text_node = 1;
context->current_node = roxml_parent_node(context->current_node, context->candidat_node, 0);
if(context->current_node->prnt != NULL) { context->current_node = context->current_node->prnt; }
roxml_process_unaliased_ns(context);
break;
case STATE_NODE_END:
context->empty_text_node = 1;
roxml_close_node(context->current_node, context->candidat_node);
context->candidat_node = NULL;
if(context->current_node->prnt != NULL) { context->current_node = context->current_node->prnt; }
break;
case STATE_NODE_CDATA:
case STATE_NODE_CONTENT:
default:
context->pos++;
return 1;
break;
}
if(context->candidat_node && context->candidat_node->ns && ((context->candidat_node->ns->type & ROXML_INVALID) == ROXML_INVALID)) {
roxml_free_node(context->candidat_node->ns);
}
context->state = STATE_NODE_CONTENT;
context->previous_state = STATE_NODE_CONTENT;
context->candidat_txt = roxml_create_node(context->pos+1, context->src, ROXML_TXT_NODE | context->type);
#ifdef IGNORE_EMPTY_TEXT_NODES
while(chunk[cur] != '\0') {
if(chunk[cur] == '<') { break; }
else if(!ROXML_WHITE(chunk[cur])) { context->empty_text_node = 0; break; }
cur++;
}
#endif /* IGNORE_EMPTY_TEXT_NODES */
while((chunk[cur] != '<')&&(chunk[cur] != '\0')) { cur++; }
context->pos += cur;
return cur;
}
int _func_load_default(char * chunk, void * data)
{
node_t * to_be_closed;
int cur = 1;
roxml_load_ctx_t *context = (roxml_load_ctx_t*)data;
#ifdef DEBUG_PARSING
fprintf(stderr, "calling func %s chunk %c\n",__func__,chunk[0]);
#endif /* DEBUG_PARSING */
switch(context->state) {
case STATE_NODE_SINGLE:
context->state = context->previous_state;
break;
case STATE_NODE_BEG:
if(context->ns == 0) {
roxml_process_begin_node(context, context->pos-1);
}
context->ns = 0;
context->state = STATE_NODE_NAME;
context->curr_name = chunk;
while(!ROXML_WHITE(chunk[cur])&&(chunk[cur] != '>')&&(chunk[cur] != '/')&&(chunk[cur] != ':')&&(chunk[cur] != '\0')) { cur++; }
context->curr_name_len = cur;
break;
case STATE_NODE_ATTR:
if(context->inside_node_state == STATE_INSIDE_ARG_BEG) {
if(context->nsdef) {
if(context->namespaces == NULL) {
context->namespaces = context->candidat_arg;
context->last_ns = context->candidat_arg;
} else {
context->last_ns->next = context->candidat_arg;
context->last_ns = context->candidat_arg;
}
} else if(context->ns == 0) {
context->candidat_arg = roxml_create_node(context->pos-1, context->src, ROXML_ATTR_NODE | context->type);
context->candidat_arg = roxml_parent_node(context->candidat_node, context->candidat_arg, 0);
}
context->ns = 0;
context->inside_node_state = STATE_INSIDE_ARG;
context->curr_name = chunk;
while((chunk[cur] != '=')&&(chunk[cur] != '>')&&(chunk[cur] != ':')&&(chunk[cur] != '\0')) { cur++; }
context->curr_name_len = cur;
if(context->nsdef) {
roxml_ns_t * ns = calloc(1, sizeof(roxml_ns_t)+(1+context->curr_name_len));
ns->id = ROXML_NS_ID;
ns->alias = (char*)ns + sizeof(roxml_ns_t);
memcpy(ns->alias, context->curr_name, context->curr_name_len);
context->candidat_arg->priv = ns;
context->nsdef = 0;
if(context->candidat_node->ns) {
if((context->candidat_node->ns->type & ROXML_INVALID) == ROXML_INVALID) {
if(strcmp(context->candidat_arg->prnt->ns->src.buf, ns->alias) == 0) {
roxml_free_node(context->candidat_node->ns);
context->candidat_node->ns = context->candidat_arg;
}
}
}
}
} else if(context->inside_node_state == STATE_INSIDE_VAL_BEG) {
if(context->mode != MODE_COMMENT_NONE) {
context->content_quoted = 1;
context->candidat_val = roxml_create_node(context->pos+1, context->src, ROXML_TXT_NODE | context->type);
} else {
context->candidat_val = roxml_create_node(context->pos, context->src, ROXML_TXT_NODE | context->type);
}
context->candidat_val = roxml_parent_node(context->candidat_arg, context->candidat_val, 0);
context->inside_node_state = STATE_INSIDE_VAL;
} else if((context->inside_node_state == STATE_INSIDE_ARG)&&(chunk[0] == '=')) {
context->inside_node_state = STATE_INSIDE_VAL_BEG;
to_be_closed = roxml_create_node(context->pos, context->src, ROXML_ATTR_NODE | context->type);
roxml_close_node(context->candidat_arg, to_be_closed);
if((context->curr_name_len==5)&&(strncmp(context->curr_name, "xmlns", 5) == 0)) {
context->nsdef = 1;
}
}
break;
}
context->pos += cur;
return cur;
}
node_t * roxml_add_node(node_t * parent, int position, int type, char *name, char *value)
{
int name_l = 0;
int end_node = 0;
int content_l = 0;
int content_pos = 0;
int end_content = 0;
char * buffer = NULL;
node_t * new_node;
int allow_attrib_child;
if(parent) {
if(parent->type & ROXML_ATTR_NODE) {
if(((type & ROXML_TXT_NODE) == 0)||(parent->chld)) {
return NULL;
}
} else if((parent->type & ROXML_ELM_NODE) == 0) {
if(parent->prnt && (parent->prnt->type & ROXML_ELM_NODE)) {
parent = parent->prnt;
} else {
return NULL;
}
}
}
if(value) {
content_l = strlen(value);
}
if(name) {
name_l = strlen(name);
}
if(type & ROXML_ATTR_NODE) {
int xmlns_l = 0;
if(!name || !value) { return NULL; }
if(type & ROXML_NS_NODE) {
if(name_l > 0) {
xmlns_l = 6;
} else {
xmlns_l = 5;
}
buffer = (char*)malloc(sizeof(char)*(name_l+content_l+xmlns_l+4));
sprintf(buffer,"xmlns%s%s=\"%s\"", name_l?":":"", name, value);
} else {
buffer = (char*)malloc(sizeof(char)*(name_l+content_l+4));
sprintf(buffer,"%s=\"%s\"",name, value);
}
content_pos = name_l + 2 + xmlns_l;
end_node = name_l + 1 + xmlns_l;
end_content = name_l + content_l + 2 + xmlns_l;
} else if(type & ROXML_CMT_NODE) {
if(!value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(content_l+8));
sprintf(buffer,"<!--%s-->", value);
content_pos = 0;
end_node = content_l + 4;
end_content = content_l + 4;
} else if(type & ROXML_PI_NODE) {
if(!name) { return NULL; }
if(content_l) {
buffer = (char*)malloc(sizeof(char)*(name_l+content_l+8));
sprintf(buffer,"<?%s %s?>", name, value);
end_node = name_l + content_l + 3;
end_content = name_l + content_l + 5;
} else {
buffer = (char*)malloc(sizeof(char)*(name_l+7));
sprintf(buffer,"<?%s?>", name);
end_node = name_l + 2;
end_content = name_l + 4;
}
content_pos = 0;
} else if(type & ROXML_TXT_NODE) {
if(!value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(content_l+1));
sprintf(buffer,"%s", value);
content_pos = 0;
end_node = content_l + 1;
end_content = content_l + 1;
} else if(type & ROXML_ELM_NODE) {
if(!name) { return NULL; }
if(value) {
buffer = (char*)malloc(sizeof(char)*(name_l*2+content_l+6));
sprintf(buffer,"<%s>%s</%s>",name, value, name);
content_pos = name_l+2;
end_node = name_l + content_l + 2;
end_content = end_node;
} else {
buffer = (char*)malloc(sizeof(char)*(name_l+5));
sprintf(buffer,"<%s />",name);
}
} else {
return NULL;
}
new_node = roxml_create_node(0, buffer, type | ROXML_PENDING | ROXML_BUFF);
new_node->end = end_node;
if(type & ROXML_NS_NODE) {
roxml_ns_t * ns = calloc(1, sizeof(roxml_ns_t) + name_l + 1);
ns->id = ROXML_NS_ID;
ns->alias = (char*)ns + sizeof(roxml_ns_t);
strcpy(ns->alias, name);
new_node->priv = ns;
}
if(((type & ROXML_ELM_NODE) && content_l) || (type & ROXML_ATTR_NODE)) {
node_t *new_txt = roxml_create_node(content_pos, buffer, ROXML_TXT_NODE | ROXML_PENDING | ROXML_BUFF);
roxml_append_node(new_node, new_txt);
new_txt->end = end_content;
}
if(parent == NULL) {
xpath_tok_table_t * table = (xpath_tok_table_t*)calloc(1, sizeof(xpath_tok_table_t));
parent = roxml_create_node(0, NULL, ROXML_ELM_NODE | ROXML_PENDING | ROXML_BUFF);
parent->end = 1;
table->id = ROXML_REQTABLE_ID;
table->ids[ROXML_REQTABLE_ID] = 1;
pthread_mutex_init(&table->mut, NULL);
parent->priv = (void*)table;
roxml_append_node(parent, new_node);
} else {
roxml_parent_node(parent, new_node, position);
}
return new_node;
}
node_t * roxml_add_node(node_t * parent, int position, int type, char *name, char *value)
{
int name_l = 0;
int end_node = 0;
int content_l = 0;
int content_pos = 0;
int end_content = 0;
char * buffer = NULL;
node_t * new_node;
if(parent && !(parent->type & ROXML_ELM_NODE)) {
return NULL;
}
if(value) {
content_l = strlen(value);
}
if(name) {
name_l = strlen(name);
}
if(type & ROXML_ATTR_NODE) {
if(!name || !value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(name_l+content_l+4));
sprintf(buffer,"%s=\"%s\"",name, value);
content_pos = name_l+2;
end_node = name_l + 1;
end_content = name_l + content_l + 2;
} else if(type & ROXML_CMT_NODE) {
if(!value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(content_l+8));
sprintf(buffer,"<!--%s-->", value);
content_pos = 0;
end_node = content_l + 4;
end_content = content_l + 4;
} else if(type & ROXML_PI_NODE) {
if(!value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(content_l+7));
sprintf(buffer,"<? %s ?>", value);
content_pos = 0;
end_node = content_l + 4;
end_content = content_l + 4;
} else if(type & ROXML_TXT_NODE) {
if(!value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(content_l+1));
sprintf(buffer,"%s", value);
content_pos = 0;
end_node = content_l + 1;
end_content = content_l + 1;
} else if(type & ROXML_ELM_NODE) {
if(!name) { return NULL; }
if(content_l) {
if(!value) { return NULL; }
buffer = (char*)malloc(sizeof(char)*(name_l*2+content_l+6));
sprintf(buffer,"<%s>%s</%s>",name, value, name);
content_pos = name_l+2;
end_node = name_l + content_l + 2;
end_content = end_node;
} else {
buffer = (char*)malloc(sizeof(char)*(name_l+5));
sprintf(buffer,"<%s />",name);
}
}
new_node = roxml_create_node(0, buffer, type | ROXML_PENDING | ROXML_BUFF);
new_node->end = end_node;
if(value && name) {
node_t *new_txt = roxml_create_node(content_pos, buffer, ROXML_TXT_NODE | ROXML_PENDING | ROXML_BUFF);
roxml_parent_node(new_node, new_txt);
new_txt->end = end_content;
}
if(parent == NULL) {
parent = roxml_create_node(0, NULL, ROXML_ELM_NODE | ROXML_PENDING | ROXML_BUFF);
parent->end = 1;
roxml_parent_node(parent, new_node);
} else {
roxml_parent_node_at(parent, new_node, position);
}
return new_node;
}
