static tb_object_ref_t tb_object_xplist_reader_func_string(tb_object_xplist_reader_t* reader, tb_size_t event)
{
// check
tb_assert_and_check_return_val(reader && reader->reader && event, tb_null);
// empty?
if (event == TB_XML_READER_EVENT_ELEMENT_EMPTY)
return tb_object_string_init_from_cstr(tb_null);
// done
tb_bool_t leave = tb_false;
tb_object_ref_t string = tb_null;
while (!leave && (event = tb_xml_reader_next(reader->reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_END:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// is end?
if (!tb_stricmp(name, "string"))
{
// empty?
if (!string) string = tb_object_string_init_from_cstr(tb_null);
// leave it
leave = tb_true;
}
}
break;
case TB_XML_READER_EVENT_TEXT:
{
// text
tb_char_t const* text = tb_xml_reader_text(reader->reader);
tb_assert_and_check_break_state(text, leave, tb_true);
tb_trace_d("string: %s", text);
// string
string = tb_object_string_init_from_cstr(text);
tb_assert_and_check_break_state(string, leave, tb_true);
}
break;
default:
break;
}
}
// ok?
return string;
}
static tb_object_ref_t tb_object_xplist_reader_done(tb_stream_ref_t stream)
{
// init reader
tb_object_xplist_reader_t reader = {0};
reader.reader = tb_xml_reader_init();
tb_assert_and_check_return_val(reader.reader, tb_null);
// open reader
tb_object_ref_t object = tb_null;
if (tb_xml_reader_open(reader.reader, stream, tb_false))
{
// done
tb_bool_t leave = tb_false;
tb_size_t event = TB_XML_READER_EVENT_NONE;
while (!leave && !object && (event = tb_xml_reader_next(reader.reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
case TB_XML_READER_EVENT_ELEMENT_BEG:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader.reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// <plist/> ?
if (tb_stricmp(name, "plist"))
{
// func
tb_object_xplist_reader_func_t func = tb_object_xplist_reader_func(name);
tb_assert_and_check_break_state(func, leave, tb_true);
// read
object = func(&reader, event);
}
}
break;
default:
break;
}
}
}
// exit reader
tb_xml_reader_exit(reader.reader);
// ok?
return object;
}
/* //////////////////////////////////////////////////////////////////////////////////////
* implementation
*/
static tb_object_ref_t tb_object_xplist_reader_func_date(tb_object_xplist_reader_t* reader, tb_size_t event)
{
// check
tb_assert_and_check_return_val(reader && reader->reader && event, tb_null);
// empty?
if (event == TB_XML_READER_EVENT_ELEMENT_EMPTY)
return tb_object_date_init_from_time(0);
// done
tb_bool_t leave = tb_false;
tb_object_ref_t date = tb_null;
while (!leave && (event = tb_xml_reader_next(reader->reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_END:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// is end?
if (!tb_stricmp(name, "date"))
{
// empty?
if (!date) date = tb_object_date_init_from_time(0);
// leave it
leave = tb_true;
}
}
break;
case TB_XML_READER_EVENT_TEXT:
{
// text
tb_char_t const* text = tb_xml_reader_text(reader->reader);
tb_assert_and_check_break_state(text, leave, tb_true);
tb_trace_d("date: %s", text);
// done date: %04ld-%02ld-%02ld %02ld:%02ld:%02ld
tb_tm_t tm = {0};
tb_char_t const* p = text;
tb_char_t const* e = text + tb_strlen(text);
// init year
while (p < e && *p && !tb_isdigit(*p)) p++;
tb_assert_and_check_break_state(p < e, leave, tb_true);
tm.year = tb_atoi(p);
// init month
while (p < e && *p && tb_isdigit(*p)) p++;
while (p < e && *p && !tb_isdigit(*p)) p++;
tb_assert_and_check_break_state(p < e, leave, tb_true);
tm.month = tb_atoi(p);
// init day
while (p < e && *p && tb_isdigit(*p)) p++;
while (p < e && *p && !tb_isdigit(*p)) p++;
tb_assert_and_check_break_state(p < e, leave, tb_true);
tm.mday = tb_atoi(p);
// init hour
while (p < e && *p && tb_isdigit(*p)) p++;
while (p < e && *p && !tb_isdigit(*p)) p++;
tb_assert_and_check_break_state(p < e, leave, tb_true);
tm.hour = tb_atoi(p);
// init minute
while (p < e && *p && tb_isdigit(*p)) p++;
while (p < e && *p && !tb_isdigit(*p)) p++;
tb_assert_and_check_break_state(p < e, leave, tb_true);
tm.minute = tb_atoi(p);
// init second
while (p < e && *p && tb_isdigit(*p)) p++;
while (p < e && *p && !tb_isdigit(*p)) p++;
tb_assert_and_check_break_state(p < e, leave, tb_true);
tm.second = tb_atoi(p);
// time
tb_time_t time = tb_mktime(&tm);
tb_assert_and_check_break_state(time >= 0, leave, tb_true);
// date
date = tb_object_date_init_from_time(time);
}
break;
default:
break;
}
}
// ok?
return date;
}
static tb_object_ref_t tb_object_xplist_reader_func_dictionary(tb_object_xplist_reader_t* reader, tb_size_t event)
{
// check
tb_assert_and_check_return_val(reader && reader->reader && event, tb_null);
// empty?
if (event == TB_XML_READER_EVENT_ELEMENT_EMPTY)
return tb_object_dictionary_init(TB_OBJECT_DICTIONARY_SIZE_MICRO, tb_false);
// init key name
tb_static_string_t kname;
tb_char_t kdata[8192];
if (!tb_static_string_init(&kname, kdata, 8192)) return tb_null;
// init dictionary
tb_object_ref_t dictionary = tb_object_dictionary_init(0, tb_false);
tb_assert_and_check_return_val(dictionary, tb_null);
// done
tb_long_t ok = 0;
tb_bool_t key = tb_false;
while (!ok && (event = tb_xml_reader_next(reader->reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_BEG:
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, ok, -1);
tb_trace_d("%s", name);
// is key
if (!tb_stricmp(name, "key")) key = tb_true;
else if (!key)
{
// func
tb_object_xplist_reader_func_t func = tb_object_xplist_reader_func(name);
tb_assert_and_check_break_state(func, ok, -1);
// read
tb_object_ref_t object = func(reader, event);
tb_trace_d("%s => %p", tb_static_string_cstr(&kname), object);
tb_assert_and_check_break_state(object, ok, -1);
// set key & value
if (tb_static_string_size(&kname) && dictionary)
tb_object_dictionary_insert(dictionary, tb_static_string_cstr(&kname), object);
// clear key name
tb_static_string_clear(&kname);
}
}
break;
case TB_XML_READER_EVENT_ELEMENT_END:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, ok, -1);
// is end?
if (!tb_stricmp(name, "dict")) ok = 1;
else if (!tb_stricmp(name, "key")) key = tb_false;
}
break;
case TB_XML_READER_EVENT_TEXT:
{
if (key)
{
// text
tb_char_t const* text = tb_xml_reader_text(reader->reader);
tb_assert_and_check_break_state(text, ok, -1);
// writ key name
tb_static_string_cstrcpy(&kname, text);
}
}
break;
default:
break;
}
}
// failed
if (ok < 0)
{
// exit it
if (dictionary) tb_object_exit(dictionary);
dictionary = tb_null;
}
// exit key name
tb_static_string_exit(&kname);
// ok?
return dictionary;
}
static tb_object_ref_t tb_object_xplist_reader_func_number(tb_object_xplist_reader_t* reader, tb_size_t event)
{
// check
tb_assert_and_check_return_val(reader && reader->reader && event, tb_null);
// empty?
if (event == TB_XML_READER_EVENT_ELEMENT_EMPTY)
return tb_object_number_init_from_uint32(0);
// done
tb_bool_t leave = tb_false;
tb_object_ref_t number = tb_null;
while (!leave && (event = tb_xml_reader_next(reader->reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_END:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// is end?
if (!tb_stricmp(name, "integer") || !tb_stricmp(name, "real")) leave = tb_true;
}
break;
case TB_XML_READER_EVENT_TEXT:
{
// text
tb_char_t const* text = tb_xml_reader_text(reader->reader);
tb_assert_and_check_break_state(text, leave, tb_true);
tb_trace_d("number: %s", text);
// has sign? is float?
tb_size_t s = 0;
tb_size_t f = 0;
tb_char_t const* p = text;
for (; *p; p++)
{
if (!s && *p == '-') s = 1;
if (!f && *p == '.') f = 1;
if (s && f) break;
}
// number
#ifdef TB_CONFIG_TYPE_HAVE_FLOAT
if (f) number = tb_object_number_init_from_double(tb_atof(text));
#else
if (f) tb_trace_noimpl();
#endif
else number = s? tb_object_number_init_from_sint64(tb_stoi64(text)) : tb_object_number_init_from_uint64(tb_stou64(text));
tb_assert_and_check_break_state(number, leave, tb_true);
}
break;
default:
break;
}
}
// ok?
return number;
}
static tb_object_ref_t tb_object_xplist_reader_func_array(tb_object_xplist_reader_t* reader, tb_size_t event)
{
// check
tb_assert_and_check_return_val(reader && reader->reader && event, tb_null);
// empty?
if (event == TB_XML_READER_EVENT_ELEMENT_EMPTY)
return tb_object_array_init(TB_OBJECT_XPLIST_READER_ARRAY_GROW, tb_false);
// init array
tb_object_ref_t array = tb_object_array_init(TB_OBJECT_XPLIST_READER_ARRAY_GROW, tb_false);
tb_assert_and_check_return_val(array, tb_null);
// done
tb_long_t ok = 0;
while (!ok && (event = tb_xml_reader_next(reader->reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_BEG:
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, ok, -1);
tb_trace_d("item: %s", name);
// func
tb_object_xplist_reader_func_t func = tb_object_xplist_reader_func(name);
tb_assert_and_check_break_state(func, ok, -1);
// read
tb_object_ref_t object = func(reader, event);
// append object
if (object) tb_object_array_append(array, object);
}
break;
case TB_XML_READER_EVENT_ELEMENT_END:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, ok, -1);
// is end?
if (!tb_stricmp(name, "array")) ok = 1;
}
break;
default:
break;
}
}
// failed?
if (ok < 0)
{
// exit it
if (array) tb_object_exit(array);
array = tb_null;
}
// ok?
return array;
}
static tb_object_ref_t tb_object_xplist_reader_func_data(tb_object_xplist_reader_t* reader, tb_size_t event)
{
// check
tb_assert_and_check_return_val(reader && reader->reader && event, tb_null);
// empty?
if (event == TB_XML_READER_EVENT_ELEMENT_EMPTY)
return tb_object_data_init_from_data(tb_null, 0);
// done
tb_bool_t leave = tb_false;
tb_char_t* base64 = tb_null;
tb_object_ref_t data = tb_null;
while (!leave && (event = tb_xml_reader_next(reader->reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_END:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader->reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// is end?
if (!tb_stricmp(name, "data"))
{
// empty?
if (!data) data = tb_object_data_init_from_data(tb_null, 0);
// leave it
leave = tb_true;
}
}
break;
case TB_XML_READER_EVENT_TEXT:
{
// text
tb_char_t const* text = tb_xml_reader_text(reader->reader);
tb_assert_and_check_break_state(text, leave, tb_true);
tb_trace_d("data: %s", text);
// base64
base64 = tb_strdup(text);
tb_char_t* p = base64;
tb_char_t* q = p;
for (; *p; p++) if (!tb_isspace(*p)) *q++ = *p;
*q = '\0';
// decode base64 data
tb_char_t const* ib = base64;
tb_size_t in = tb_strlen(base64);
if (in)
{
tb_size_t on = in;
tb_byte_t* ob = tb_malloc0_bytes(on);
tb_assert_and_check_break_state(ob && on, leave, tb_true);
on = tb_base64_decode(ib, in, ob, on);
tb_trace_d("base64: %u => %u", in, on);
// init data
data = tb_object_data_init_from_data(ob, on); tb_free(ob);
}
else data = tb_object_data_init_from_data(tb_null, 0);
tb_assert_and_check_break_state(data, leave, tb_true);
}
break;
default:
break;
}
}
// free
if (base64) tb_free(base64);
// ok?
return data;
}
static tb_size_t tb_demo_spider_parser_get_url(tb_xml_reader_ref_t reader, tb_url_ref_t url)
{
// check
tb_assert_and_check_return_val(reader && url, tb_false);
// done
tb_size_t ok = 0;
tb_size_t event = TB_XML_READER_EVENT_NONE;
while (!ok && (event = tb_xml_reader_next(reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
case TB_XML_READER_EVENT_ELEMENT_BEG:
{
// the element name
tb_char_t const* name = tb_xml_reader_element(reader);
tb_check_break(name);
// <a href="" />?
// <link href="" />
// <img src="" />?
// <script src="" />?
// <source src="" />?
// <frame src="" />?
if ( !tb_stricmp(name, "a")
|| !tb_stricmp(name, "link")
|| !tb_stricmp(name, "img")
|| !tb_stricmp(name, "frame")
|| !tb_stricmp(name, "source"))
{
// walk attributes
tb_xml_node_ref_t attr = (tb_xml_node_ref_t)tb_xml_reader_attributes(reader);
for (; attr; attr = attr->next)
{
// href or src?
if ( tb_string_size(&attr->data)
&& ( !tb_string_cstricmp(&attr->name, "href")
|| !tb_string_cstricmp(&attr->name, "src")))
{
// the url protocol
tb_size_t protocol = tb_url_protocol_probe(tb_string_cstr(&attr->data));
// http?
if(protocol == TB_URL_PROTOCOL_HTTP)
{
// save url
ok = tb_url_set(url, tb_string_cstr(&attr->data));
}
// file?
else if (protocol == TB_URL_PROTOCOL_FILE)
{
// save path
tb_url_path_set(url, tb_string_cstr(&attr->data));
// ok
ok = tb_true;
}
}
}
}
}
break;
default:
break;
}
}
// ok?
return ok;
}
tb_xml_node_ref_t tb_xml_reader_load(tb_xml_reader_ref_t reader)
{
// check
tb_assert_and_check_return_val(reader, tb_null);
// done
tb_bool_t ok = tb_true;
tb_xml_node_ref_t node = tb_null;
tb_size_t event = TB_XML_READER_EVENT_NONE;
while (ok && (event = tb_xml_reader_next(reader)))
{
// init document node
if (!node)
{
node = tb_xml_node_init_document(tb_xml_reader_version(reader), tb_xml_reader_charset(reader));
tb_assert_and_check_break_state(node && !node->parent, ok, tb_false);
}
switch (event)
{
case TB_XML_READER_EVENT_DOCUMENT:
break;
case TB_XML_READER_EVENT_DOCUMENT_TYPE:
{
// init
tb_xml_node_ref_t doctype = tb_xml_node_init_document_type(tb_xml_reader_doctype(reader));
tb_assert_and_check_break_state(doctype, ok, tb_false);
// append
tb_xml_node_append_ctail(node, doctype);
tb_assert_and_check_break_state(doctype->parent, ok, tb_false);
}
break;
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
{
// init
tb_xml_node_ref_t element = tb_xml_node_init_element(tb_xml_reader_element(reader));
tb_assert_and_check_break_state(element, ok, tb_false);
// attributes
tb_xml_node_ref_t attr = tb_xml_reader_attributes(reader);
for (; attr; attr = attr->next)
tb_xml_node_append_atail(element, tb_xml_node_init_attribute(tb_string_cstr(&attr->name), tb_string_cstr(&attr->data)));
// append
tb_xml_node_append_ctail(node, element);
tb_assert_and_check_break_state(element->parent, ok, tb_false);
}
break;
case TB_XML_READER_EVENT_ELEMENT_BEG:
{
// init
tb_xml_node_ref_t element = tb_xml_node_init_element(tb_xml_reader_element(reader));
tb_assert_and_check_break_state(element, ok, tb_false);
// attributes
tb_xml_node_ref_t attr = tb_xml_reader_attributes(reader);
for (; attr; attr = attr->next)
tb_xml_node_append_atail(element, tb_xml_node_init_attribute(tb_string_cstr(&attr->name), tb_string_cstr(&attr->data)));
// append
tb_xml_node_append_ctail(node, element);
tb_assert_and_check_break_state(element->parent, ok, tb_false);
// enter
node = element;
}
break;
case TB_XML_READER_EVENT_ELEMENT_END:
{
// check
tb_assert_and_check_break_state(node, ok, tb_false);
// the parent node
node = node->parent;
}
break;
case TB_XML_READER_EVENT_TEXT:
{
// init
tb_xml_node_ref_t text = tb_xml_node_init_text(tb_xml_reader_text(reader));
tb_assert_and_check_break_state(text, ok, tb_false);
// append
tb_xml_node_append_ctail(node, text);
tb_assert_and_check_break_state(text->parent, ok, tb_false);
}
break;
case TB_XML_READER_EVENT_CDATA:
{
// init
tb_xml_node_ref_t cdata = tb_xml_node_init_cdata(tb_xml_reader_cdata(reader));
tb_assert_and_check_break_state(cdata, ok, tb_false);
// append
tb_xml_node_append_ctail(node, cdata);
tb_assert_and_check_break_state(cdata->parent, ok, tb_false);
}
break;
case TB_XML_READER_EVENT_COMMENT:
{
// init
tb_xml_node_ref_t comment = tb_xml_node_init_comment(tb_xml_reader_comment(reader));
tb_assert_and_check_break_state(comment, ok, tb_false);
// append
tb_xml_node_append_ctail(node, comment);
tb_assert_and_check_break_state(comment->parent, ok, tb_false);
}
break;
default:
break;
}
}
// failed?
if (!ok)
{
// exit it
if (node) tb_xml_node_exit(node);
node = tb_null;
}
// ok
return node;
}
tb_bool_t tb_xml_reader_goto(tb_xml_reader_ref_t reader, tb_char_t const* path)
{
// check
tb_xml_reader_impl_t* impl = (tb_xml_reader_impl_t*)reader;
tb_assert_and_check_return_val(impl && impl->rstream && path, tb_false);
tb_trace_d("goto: %s", path);
// init level
impl->level = 0;
// seek to the stream head
if (!tb_stream_seek(impl->rstream, 0)) return tb_false;
// init
tb_static_string_t s;
tb_char_t data[8192];
if (!tb_static_string_init(&s, data, 8192)) return tb_false;
// save the current offset
tb_hize_t save = tb_stream_offset(impl->rstream);
// done
tb_bool_t ok = tb_false;
tb_bool_t leave = tb_false;
tb_size_t event = TB_XML_READER_EVENT_NONE;
while (!leave && !ok && (event = tb_xml_reader_next(reader)))
{
switch (event)
{
case TB_XML_READER_EVENT_ELEMENT_EMPTY:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// append
tb_size_t n = tb_static_string_size(&s);
tb_static_string_chrcat(&s, '/');
tb_static_string_cstrcat(&s, name);
// ok?
if (!tb_static_string_cstricmp(&s, path)) ok = tb_true;
tb_trace_d("path: %s", tb_static_string_cstr(&s));
// remove
tb_static_string_strip(&s, n);
// restore
if (ok) if (!(ok = tb_stream_seek(impl->rstream, save))) leave = tb_true;
}
break;
case TB_XML_READER_EVENT_ELEMENT_BEG:
{
// name
tb_char_t const* name = tb_xml_reader_element(reader);
tb_assert_and_check_break_state(name, leave, tb_true);
// append
tb_static_string_chrcat(&s, '/');
tb_static_string_cstrcat(&s, name);
// ok?
if (!tb_static_string_cstricmp(&s, path)) ok = tb_true;
tb_trace_d("path: %s", tb_static_string_cstr(&s));
// restore
if (ok) if (!(ok = tb_stream_seek(impl->rstream, save))) leave = tb_true;
}
break;
case TB_XML_READER_EVENT_ELEMENT_END:
{
// remove
tb_long_t p = tb_static_string_strrchr(&s, 0, '/');
if (p >= 0) tb_static_string_strip(&s, p);
// ok?
if (!tb_static_string_cstricmp(&s, path)) ok = tb_true;
tb_trace_d("path: %s", tb_static_string_cstr(&s));
// restore
if (ok) if (!(ok = tb_stream_seek(impl->rstream, save))) leave = tb_true;
}
break;
default:
break;
}
// save
save = tb_stream_offset(impl->rstream);
}
// exit string
tb_static_string_exit(&s);
// clear level
impl->level = 0;
// failed? restore to the stream head
if (!ok) tb_stream_seek(impl->rstream, 0);
// ok?
return ok;
}