static const char *xml_parse_left_tag(ACL_XML *xml, const char *data)
{
int ch;
if (LEN(xml->curr_node->ltag) == 0) {
SKIP_SPACE(data);
}
while ((ch = *data) != 0) {
data++;
if (ch == '>') {
xml->curr_node->status = ACL_XML_S_LGT;
xml_parse_check_self_closed(xml);
if ((xml->curr_node->flag & ACL_XML_F_SELF_CL)
&& xml->curr_node->last_ch == '/')
{
acl_vstring_truncate(xml->curr_node->ltag,
LEN(xml->curr_node->ltag) - 1);
}
break;
} else if (IS_SPACE(ch)) {
xml->curr_node->status = ACL_XML_S_ATTR;
xml->curr_node->last_ch = ch;
break;
} else {
ACL_VSTRING_ADDCH(xml->curr_node->ltag, ch);
xml->curr_node->last_ch = ch;
}
}
ACL_VSTRING_TERMINATE(xml->curr_node->ltag);
return (data);
}
static const char *xml_parse_left_tag(ACL_XML2 *xml, const char *data)
{
int ch;
if (xml->curr_node->ltag == xml->addr)
SKIP_SPACE(data);
if (*data == 0)
return data;
if (xml->curr_node->ltag == xml->addr)
xml->curr_node->ltag = xml->ptr;
while ((ch = *data) != 0) {
if (ch == '>') {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
xml->curr_node->ltag_size = xml->ptr -
xml->curr_node->ltag;
*xml->ptr++ = 0;
xml_parse_check_self_closed(xml);
if ((xml->curr_node->flag & ACL_XML2_F_SELF_CL)
&& xml->curr_node->last_ch == '/')
{
size_t n = xml->curr_node->ltag_size;
if (n >= 2)
xml->curr_node->ltag[n - 2] = 0;
xml->curr_node->status = ACL_XML2_S_RGT;
} else
xml->curr_node->status = ACL_XML2_S_LGT;
break;
} else if (IS_SPACE(ch)) {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
xml->curr_node->ltag_size = xml->ptr -
xml->curr_node->ltag;
*xml->ptr++ = 0;
xml->curr_node->status = ACL_XML2_S_ATTR;
xml->curr_node->last_ch = ch;
break;
} else {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
*xml->ptr++ = ch;
xml->curr_node->last_ch = ch;
}
}
return data;
}
static const char *xml_parse_left_tag(ACL_XML2 *xml, const char *data)
{
int ch;
if (xml->curr_node->ltag == xml->dummy)
SKIP_SPACE(data);
if (*data == 0)
return data;
if (xml->curr_node->ltag == xml->dummy)
xml->curr_node->ltag = END(xml);
while ((ch = *data) != 0) {
if (ch == '>') {
if (NO_SPACE(xml))
return data;
xml->curr_node->ltag_size =
END(xml) - xml->curr_node->ltag;
ADD(xml, '\0');
data++;
xml_parse_check_self_closed(xml);
if ((xml->curr_node->flag & ACL_XML2_F_SELF_CL)
&& xml->curr_node->last_ch == '/')
{
if (xml->curr_node->ltag_size > 0) {
size_t n;
xml->curr_node->ltag_size--;
n = xml->curr_node->ltag_size;
xml->curr_node->ltag[n] = 0;
}
xml->curr_node->status = ACL_XML2_S_RGT;
} else
xml->curr_node->status = ACL_XML2_S_LGT;
break;
} else if (IS_SPACE(ch)) {
if (NO_SPACE(xml))
return data;
data++;
xml->curr_node->ltag_size =
END(xml) - xml->curr_node->ltag;
ADD(xml, '\0');
xml->curr_node->status = ACL_XML2_S_ATTR;
xml->curr_node->last_ch = ch;
break;
} else {
if (NO_SPACE(xml))
return data;
data++;
ADD(xml, ch);
xml->curr_node->last_ch = ch;
}
}
return data;
}
static const char *xml_parse_attr(ACL_XML *xml, const char *data)
{
int ch;
ACL_XML_ATTR *attr = xml->curr_node->curr_attr;
if (attr == NULL || LEN(attr->name) == 0) {
SKIP_SPACE(data); /* 略过 ' ', '\t' */
if (*data == 0)
return (NULL);
SKIP_WHILE(*data == '=', data);
if (*data == 0)
return (NULL);
}
if (*data == '>') {
xml->curr_node->status = ACL_XML_S_LGT;
xml_parse_check_self_closed(xml);
xml->curr_node->curr_attr = NULL;
data++;
return (data);
}
xml->curr_node->last_ch = *data;
if (*data == '/') {
data++;
return (data);
}
if (attr == NULL) {
attr = acl_xml_attr_alloc(xml->curr_node);
xml->curr_node->curr_attr = attr;
}
while ((ch = *data) != 0) {
xml->curr_node->last_ch = ch;
if (ch == '=') {
xml->curr_node->status = ACL_XML_S_AVAL;
data++;
break;
}
if (!IS_SPACE(ch))
ACL_VSTRING_ADDCH(attr->name, ch);
data++;
}
ACL_VSTRING_TERMINATE(attr->name);
return (data);
}
static const char *xml_parse_attr_val(ACL_XML2 *xml, const char *data)
{
int ch;
ACL_XML2_ATTR *attr = xml->curr_node->curr_attr;
if (attr->value == xml->addr && !attr->quote) {
SKIP_SPACE(data);
if (IS_QUOTE(*data))
attr->quote = *data++;
}
if (*data == 0)
return data;
if (attr->value == xml->addr)
attr->value = xml->ptr;
while ((ch = *data) != 0) {
if (attr->quote) {
if (ch == attr->quote) {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
attr->value_size = xml->ptr - attr->value;
*xml->ptr++ = 0;
xml->curr_node->status = ACL_XML2_S_ATTR;
xml->curr_node->last_ch = ch;
break;
}
if (xml->len < MIN_LEN)
return data;
xml->len--;
*xml->ptr++ = ch;
xml->curr_node->last_ch = ch;
} else if (ch == '>') {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
attr->value_size = xml->ptr - attr->value;
*xml->ptr++ = 0;
xml_parse_check_self_closed(xml);
if ((xml->curr_node->flag & ACL_XML2_F_SELF_CL)
&& xml->curr_node->last_ch == '/')
{
if (attr->value_size >= 2)
attr->value[attr->value_size - 2] = 0;
xml->curr_node->status = ACL_XML2_S_RGT;
} else
xml->curr_node->status = ACL_XML2_S_LGT;
break;
} else if (IS_SPACE(ch)) {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
attr->value_size = xml->ptr - attr->value;
*xml->ptr++ = 0;
xml->curr_node->status = ACL_XML2_S_ATTR;
xml->curr_node->last_ch = ch;
break;
} else {
if (xml->len < MIN_LEN)
return data;
xml->len--;
*xml->ptr++ = ch;
xml->curr_node->last_ch = ch;
}
data++;
}
/* 说明属性值还未解析完，需要继续解析 */
if (xml->curr_node->status == ACL_XML2_S_AVAL)
return data;
/* 当状态发生改变时，则说明属性值已经完毕 */
if ((xml->flag & ACL_XML2_FLAG_XML_DECODE) && attr->value_size > 1
&& xml->len > 0)
{
const char *val = attr->value;
attr->value = xml->ptr;
(void) acl_xml_decode2(val, &xml->ptr, &xml->len);
attr->value_size = xml->ptr - attr->value - 1;
}
/* 将该标签ID号映射至哈希表中，以便于快速查询 */
if (IS_ID(attr->name) && *attr->value != 0) {
const char *ptr = attr->value;
/* 防止重复ID被插入现象 */
if (acl_htable_find(xml->id_table, ptr) == NULL) {
acl_htable_enter(xml->id_table, ptr, attr);
/* 当该属性被加入哈希表后才会赋于节点 id */
xml->curr_node->id = attr->value;
}
}
/* 必须将该节点的当前属性对象置空，以便于继续解析时
* 可以创建新的属性对象
*/
xml->curr_node->curr_attr = NULL;
return data;
}
static const char *xml_parse_attr(ACL_XML2 *xml, const char *data)
{
int ch;
ACL_XML2_ATTR *attr = xml->curr_node->curr_attr;
if (attr == NULL || attr->name == xml->addr) {
SKIP_SPACE(data);
SKIP_WHILE(*data == '=', data);
}
if (*data == 0)
return data;
if (*data == '>') {
xml_parse_check_self_closed(xml);
if ((xml->curr_node->flag & ACL_XML2_F_SELF_CL)
&& xml->curr_node->last_ch == '/')
{
xml->curr_node->status = ACL_XML2_S_RGT;
} else
xml->curr_node->status = ACL_XML2_S_LGT;
xml->curr_node->curr_attr = NULL;
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
*xml->ptr++ = 0;
return data;
}
xml->curr_node->last_ch = *data;
if (*data == '/') {
data++;
/* 此处返回后会触发本函数再次被调用，当下一个字节为 '>' 时，
* 上面通过调用 xml_parse_check_self_closed 检查是否为自封闭
* 标签: "/>"
*/
return data;
}
if (attr == NULL) {
attr = acl_xml2_attr_alloc(xml->curr_node);
xml->curr_node->curr_attr = attr;
attr->name = xml->ptr;
}
while ((ch = *data) != 0) {
xml->curr_node->last_ch = ch;
if (ch == '=') {
if (xml->len < MIN_LEN)
return data;
data++;
xml->len--;
attr->name_size = xml->ptr - attr->name;
*xml->ptr++ = 0;
xml->curr_node->status = ACL_XML2_S_AVAL;
break;
}
if (!IS_SPACE(ch)) {
if (xml->len < MIN_LEN)
return data;
xml->len--;
*xml->ptr++ = ch;
}
data++;
}
return data;
}
static char *xml_parse_attr_val(ACL_XML3 *xml, char *data)
{
int ch;
ACL_XML3_ATTR *attr = xml->curr_node->curr_attr;
if (attr->value == xml->addr && !attr->quote) {
SKIP_SPACE(data);
if (IS_QUOTE(*data))
attr->quote = *data++;
}
if (*data == 0)
return data;
if (attr->value == xml->addr)
attr->value = data;
while ((ch = *data) != 0) {
if (attr->quote) {
if (ch == attr->quote) {
attr->value_size = data - attr->value;
xml->curr_node->status = ACL_XML3_S_ATTR;
xml->curr_node->last_ch = ch;
*data++ = 0;
break;
}
xml->curr_node->last_ch = ch;
} else if (ch == '>') {
if (attr->value_size == 0)
attr->value_size = data - attr->value;
*data++ = 0;
xml_parse_check_self_closed(xml);
if ((xml->curr_node->flag & ACL_XML3_F_SELF_CL)
&& xml->curr_node->last_ch == '/')
{
if (attr->value_size >= 2)
attr->value[attr->value_size - 2] = 0;
xml->curr_node->status = ACL_XML3_S_RGT;
} else
xml->curr_node->status = ACL_XML3_S_LGT;
break;
} else if (IS_SPACE(ch)) {
attr->value_size = data - attr->value;
xml->curr_node->status = ACL_XML3_S_ATTR;
xml->curr_node->last_ch = ch;
*data++ = 0;
break;
} else
xml->curr_node->last_ch = ch;
data++;
}
/* 当状态发生改变时，则说明属性值已经完毕 */
if (xml->curr_node->status != ACL_XML3_S_AVAL) {
/* 将该标签ID号映射至哈希表中，以便于快速查询 */
if (IS_ID(attr->name) && *attr->value != 0) {
char *ptr = attr->value;
/* 防止重复ID被插入现象 */
if (acl_htable_find(xml->id_table, ptr) == NULL) {
acl_htable_enter(xml->id_table, ptr, attr);
/* 当该属性被加入哈希表后才会赋于节点 id */
xml->curr_node->id = attr->value;
}
}
/* 必须将该节点的当前属性对象置空，以便于继续解析时
* 可以创建新的属性对象
*/
xml->curr_node->curr_attr = NULL;
}
return data;
}
static const char *xml_parse_attr_val(ACL_XML *xml, const char *data)
{
int ch;
ACL_XML_ATTR *attr = xml->curr_node->curr_attr;
if (LEN(attr->value) == 0 && !attr->quote) {
SKIP_SPACE(data);
if (IS_QUOTE(*data)) {
attr->quote = *data++;
}
if (*data == 0) {
return (NULL);
}
}
while ((ch = *data) != 0) {
if (attr->backslash) {
ACL_VSTRING_ADDCH(attr->value, ch);
xml->curr_node->last_ch = ch;
attr->backslash = 0;
} else if (ch == '\\') {
if (attr->part_word) {
ACL_VSTRING_ADDCH(attr->value, ch);
attr->part_word = 0;
}
else
attr->backslash = 1;
} else if (attr->quote) {
if (ch == attr->quote) {
xml->curr_node->status = ACL_XML_S_ATTR;
xml->curr_node->last_ch = ch;
data++;
break;
}
ACL_VSTRING_ADDCH(attr->value, ch);
xml->curr_node->last_ch = ch;
} else if (ch == '>') {
xml->curr_node->status = ACL_XML_S_LGT;
xml_parse_check_self_closed(xml);
data++;
break;
} else if (IS_SPACE(ch)) {
xml->curr_node->status = ACL_XML_S_ATTR;
xml->curr_node->last_ch = ch;
data++;
break;
} else {
ACL_VSTRING_ADDCH(attr->value, ch);
xml->curr_node->last_ch = ch;
if ((xml->flag & ACL_XML_FLAG_PART_WORD)) {
/* 处理半个汉字的情形 */
if (attr->part_word)
attr->part_word = 0;
else if (ch < 0)
attr->part_word = 1;
}
}
data++;
}
ACL_VSTRING_TERMINATE(attr->value);
if (xml->curr_node->status != ACL_XML_S_AVAL) {
/* 将该标签ID号映射至哈希表中，以便于快速查询 */
if (IS_ID(STR(attr->name)) && LEN(attr->value) > 0) {
const char *ptr = STR(attr->value);
/* 防止重复ID被插入现象 */
if (acl_htable_find(xml->id_table, ptr) == NULL) {
acl_htable_enter(xml->id_table, ptr, attr);
/* 只有当该属性被加入哈希表后才会赋于结点的 id */
xml->curr_node->id = attr->value;
}
}
/* 必须将该结点的当前属性对象置空，以便于继续解析时
* 可以创建新的属性对象
*/
xml->curr_node->curr_attr = NULL;
}
return (data);
}
