void dumpNode( TidyNode tnod, int indent )
{
TidyNode child;
for ( child = tidyGetChild(tnod); child; child = tidyGetNext(child) )
{
ctmbstr name;
switch ( tidyNodeGetType(child) )
{
case TidyNode_Root: name = "Root"; break;
case TidyNode_DocType: name = "DOCTYPE"; break;
case TidyNode_Comment: name = "Comment"; break;
case TidyNode_ProcIns: name = "Processing Instruction"; break;
case TidyNode_Text: name = "Text"; break;
case TidyNode_CDATA: name = "CDATA"; break;
case TidyNode_Section: name = "XML Section"; break;
case TidyNode_Asp: name = "ASP"; break;
case TidyNode_Jste: name = "JSTE"; break;
case TidyNode_Php: name = "PHP"; break;
case TidyNode_XmlDecl: name = "XML Declaration"; break;
case TidyNode_Start:
case TidyNode_End:
case TidyNode_StartEnd:
default:
name = tidyNodeGetName( child );
break;
}
assert( name != NULL );
TRACE( "------%d,%d, %s,%s", indent, indent, " ", name );
dumpNode( child, indent + 4 );
}
}
void CCaHtmlParse::__GetPriceAndRamainTicket(UINT *pPrice, UINT *pRemainTicket, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
*pPrice = 0;
*pRemainTicket = 0;
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
CStringA straKey("</font>");
int iPos = straRet.Find(straKey);
int iEndPos = straRet.Find("</strong>");
int iStartPos = iPos+straKey.GetLength();
CStringA straPrice = straRet.Mid(iStartPos, iEndPos-iStartPos);
straPrice.Remove(0x0d);//去掉回车
straPrice.Remove(0x0a);//去掉换行
*pPrice = atoi(straPrice.GetBuffer(0));
straPrice.ReleaseBuffer();
//剩余座位
straKey = "：";
iPos = straRet.Find(straKey);
iEndPos = straRet.Find("</td>");
iStartPos = iPos+straKey.GetLength();
CStringA straRemainSeat = straRet.Mid(iStartPos, iEndPos-iStartPos);
*pRemainTicket = atoi(straRemainSeat.GetBuffer(0));
straRemainSeat.ReleaseBuffer();
tidyBufFree(&text);
}
DFNode *fromTidyNode(DFDocument *htmlDoc, TidyDoc tdoc, TidyNode tnode)
{
switch (tidyNodeGetType(tnode)) {
case TidyNode_Text: {
char *value = copyTidyNodeValue(tnode,tdoc);
DFNode *result = DFCreateTextNode(htmlDoc,value);
free(value);
return result;
}
case TidyNode_CDATA:
break;
case TidyNode_Comment:
break;
case TidyNode_Root:
printf("Have root\n");
break;
default: {
const char *name = tidyNodeGetName(tnode);
if (name == NULL) {
printf("NULL name for %p, type %d\n",tnode,tidyNodeGetType(tnode));
return NULL;
}
const NamespaceDecl *namespaceDecl = DFNameMapNamespaceForID(htmlDoc->map,NAMESPACE_HTML);
Tag tag = DFNameMapTagForName(htmlDoc->map,namespaceDecl->namespaceURI,name);
DFNode *element = DFCreateElement(htmlDoc,tag);
for (TidyAttr tattr = tidyAttrFirst(tnode); tattr != NULL; tattr = tidyAttrNext(tattr)) {
const char *name = tidyAttrName(tattr);
const char *value = tidyAttrValue(tattr);
if (value == NULL) // Can happen in case of the empty string
value = "";;
Tag attrTag = DFNameMapTagForName(htmlDoc->map,namespaceDecl->namespaceURI,name);
DFSetAttribute(element,attrTag,value);
}
for (TidyNode tchild = tidyGetChild(tnode); tchild != NULL; tchild = tidyGetNext(tchild)) {
DFNode *child = fromTidyNode(htmlDoc,tdoc,tchild);
if (child != NULL)
DFAppendChild(element,child);
}
return element;
}
}
return NULL;
}
/* {{{ proto boolean tidyNode::isPhp()
Returns true if this node is PHP */
static TIDY_NODE_METHOD(isPhp)
{
TIDY_FETCH_ONLY_OBJECT;
if (tidyNodeGetType(obj->node) == TidyNode_Php) {
RETURN_TRUE;
} else {
RETURN_FALSE;
}
}
/* {{{ proto boolean tidyNode::isHtml()
Returns true if this node is part of a HTML document */
static TIDY_NODE_METHOD(isHtml)
{
TIDY_FETCH_ONLY_OBJECT;
if (tidyNodeGetType(obj->node) & (TidyNode_Start | TidyNode_End | TidyNode_StartEnd)) {
RETURN_TRUE;
}
RETURN_FALSE;
}
void CCaHtmlParse::__GetAirPortCode(CStringA & straDCode, CStringA & straACode, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
CStringA straSha("上海虹桥");
CStringA straPvg("上海浦东");
CStringA straPek("北京首都");
CStringA straNay("北京南苑");
straDCode = "";
straACode = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
CStringA straKey("<br />");//<br />后有回车换行符
int iPos = straRet.Find(straKey);
CStringA straDCity = straRet.Mid(4, iPos-4);
if (-1 != straDCity.Find(straSha))
straDCode = "SHA";
else if (-1 != straDCity.Find(straPvg))
straDCode = "PVG";
else if (-1 != straDCity.Find(straPek))
straDCode = "PEK";
else if(-1 != straDCity.Find(straNay))
straDCode = "NAY";
else
{
}
//取到达机场，<br />后有回车换行符
straRet = straRet.Mid(iPos+straKey.GetLength());
straRet.Remove(0x0d);//去掉回车
straRet.Remove(0x0a);//去掉换行
int iEndPos = straRet.Find("</td>");
CStringA straACity = straRet.Left(iEndPos);
if (-1 != straACity.Find(straSha))
straACode = "SHA";
else if (-1 != straACity.Find(straPvg))
straACode = "PVG";
else if (-1 != straACity.Find(straPek))
straACode = "PEK";
else if(-1 != straACity.Find(straNay))
straACode = "NAY";
else
{
}
tidyBufFree(&text);
}
void CCaHtmlParse::__GetFlightStartTime(CStringA & strFlightStartTime, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
strFlightStartTime = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
CStringA straKey("\">");
int iStartPos = straRet.Find(straKey);
int iEndPos = straRet.Find("</strong>");
iStartPos = iStartPos+straKey.GetLength();
strFlightStartTime = straRet.Mid(iStartPos, iEndPos-iStartPos);
tidyBufFree(&text);
}
void CCaHtmlParse::__GetSaleEndDate(CStringA & straEndDate, CStringA & straEndTime, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
straEndDate = "";
straEndTime = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
int iPos = straRet.Find("<br");
straRet = straRet.Mid(4, iPos-4);
straEndDate = straRet.Left(10);
straEndTime = straRet.Mid(11);
straEndTime.Trim();
tidyBufFree(&text);
}
static void traverseNodes(TidyDoc doc, lua_State *L, TidyNode node)
{
TidyNodeType nt = tidyNodeGetType ( node );
switch (nt)
{
case TidyNode_Text:
if ( tidyNodeHasText (doc, node) )
{
if ( push_node_text(L, doc, node) == 1 )
{
lua_tableinsert(L);
}
}
break;
case TidyNode_Start:
case TidyNode_StartEnd:
lua_newtable(L);
if ( push_node_attributes(L, node) == 1 )
{
lua_pushstring(L, "attr");
lua_insert(L,-2);
lua_rawset(L,-3);
}
lua_pushstring(L,"tag");
lua_pushstring(L, tidyNodeGetName (node));
lua_rawset(L,-3);
traverseDeeper(doc, L, node);
lua_tableinsert(L);
break;
case TidyNode_Root:
traverseDeeper(doc, L, node);
break;
default:
// ignore. keep gcc compiler happy.
break;
}
}
void nuiHTMLNode::SetFromNode(const void* _tdoc, const void* _tnod, nglTextEncoding encoding, bool ComputeStyle)
{
Clear();
TidyDoc tdoc = (TidyDoc)_tdoc;
TidyNode tnod = (TidyNode)_tnod;
mName = nglString(tidyNodeGetName(tnod), encoding);
mType = (NodeType)tidyNodeGetType(tnod);
mTagType = (TagType)tidyNodeGetId(tnod);
nglString text;
TidyBuffer buf;
tidyBufInit(&buf);
if (tidyNodeGetValue(tdoc, tnod, &buf))
{
mText.Import((const char*)buf.bp, (int32)buf.size, encoding);
//NGL_OUT(_T("<%s> %s\n"), mName.GetChars(), mText.GetChars());
}
tidyBufFree(&buf);
// Fill the attributes:
TidyAttr tattr;
for (tattr = tidyAttrFirst(tnod); tattr; tattr = tidyAttrNext(tattr))
{
nuiHTMLAttrib* pAttrib = new nuiHTMLAttrib(tattr, encoding);
mAttributes.push_back(pAttrib);
}
if (ComputeStyle)
{
mpStyle = new nuiCSSStyle(this);
nuiHTMLAttrib* pStyle = GetAttribute(nuiHTMLAttrib::eAttrib_STYLE);
if (pStyle)
{
AddStyleSheet(GetSourceURL(), pStyle->GetValue(), true);
}
}
}
void CCaHtmlParse::__GetFlightNoAndFlightStartDate(CStringA & strFlightNo, CStringA & strFlightStartDate, const TidyDoc & tdoc, const TidyNode & tdNode)
{
CStringA straRet;
strFlightNo = "";
strFlightStartDate = "";
TidyBuffer text = {0};
tidyBufInit(&text);
TidyNodeType type = tidyNodeGetType(tdNode);
tidyNodeGetText(tdoc, tdNode, &text);
straRet.Format("%s",text.bp);
straRet.TrimLeft();
strFlightStartDate = straRet.Mid(4, 10);
int iStartPos = straRet.Find('C');//匹配CA，ca，Ca，cA，<br />后会插入回车换行符，
if(-1 == iStartPos)//没找到大写C，匹配小写c
iStartPos = straRet.Find('c');
int iEndPos = straRet.Find("</");
strFlightNo = straRet.Mid(iStartPos, iEndPos-iStartPos);//国航航班号有3位的，还有4位的
strFlightNo = strFlightNo.MakeUpper();
strFlightNo = strFlightNo.Mid(2);//去掉CA
tidyBufFree(&text);
}
static void tidy_add_default_properties(PHPTidyObj *obj, tidy_obj_type type)
{
TidyBuffer buf;
TidyAttr tempattr;
TidyNode tempnode;
zval attribute, children, temp;
PHPTidyObj *newobj;
switch(type) {
case is_node:
if (!obj->std.properties) {
rebuild_object_properties(&obj->std);
}
tidyBufInit(&buf);
tidyNodeGetText(obj->ptdoc->doc, obj->node, &buf);
ADD_PROPERTY_STRINGL(obj->std.properties, value, buf.bp, buf.size ? buf.size-1 : 0);
tidyBufFree(&buf);
ADD_PROPERTY_STRING(obj->std.properties, name, tidyNodeGetName(obj->node));
ADD_PROPERTY_LONG(obj->std.properties, type, tidyNodeGetType(obj->node));
ADD_PROPERTY_LONG(obj->std.properties, line, tidyNodeLine(obj->node));
ADD_PROPERTY_LONG(obj->std.properties, column, tidyNodeColumn(obj->node));
ADD_PROPERTY_BOOL(obj->std.properties, proprietary, tidyNodeIsProp(obj->ptdoc->doc, obj->node));
switch(tidyNodeGetType(obj->node)) {
case TidyNode_Root:
case TidyNode_DocType:
case TidyNode_Text:
case TidyNode_Comment:
break;
default:
ADD_PROPERTY_LONG(obj->std.properties, id, tidyNodeGetId(obj->node));
}
tempattr = tidyAttrFirst(obj->node);
if (tempattr) {
char *name, *val;
array_init(&attribute);
do {
name = (char *)tidyAttrName(tempattr);
val = (char *)tidyAttrValue(tempattr);
if (name && val) {
add_assoc_string(&attribute, name, val);
}
} while((tempattr = tidyAttrNext(tempattr)));
} else {
ZVAL_NULL(&attribute);
}
zend_hash_str_update(obj->std.properties, "attribute", sizeof("attribute") - 1, &attribute);
tempnode = tidyGetChild(obj->node);
if (tempnode) {
array_init(&children);
do {
tidy_instanciate(tidy_ce_node, &temp);
newobj = Z_TIDY_P(&temp);
newobj->node = tempnode;
newobj->type = is_node;
newobj->ptdoc = obj->ptdoc;
newobj->ptdoc->ref_count++;
tidy_add_default_properties(newobj, is_node);
add_next_index_zval(&children, &temp);
} while((tempnode = tidyGetNext(tempnode)));
} else {
ZVAL_NULL(&children);
}
zend_hash_str_update(obj->std.properties, "child", sizeof("child") - 1, &children);
break;
case is_doc:
if (!obj->std.properties) {
rebuild_object_properties(&obj->std);
}
ADD_PROPERTY_NULL(obj->std.properties, errorBuffer);
ADD_PROPERTY_NULL(obj->std.properties, value);
break;
}
}
static void convertNode(TidyNode node, int level, bool opentag)
{
ctmbstr name;
TidyAttr tattr;
struct htmlTag *t;
int nattr; /* number of attributes */
int i;
switch (tidyNodeGetType(node)) {
case TidyNode_Text:
name = "Text";
break;
case TidyNode_Start:
case TidyNode_End:
case TidyNode_StartEnd:
name = tidyNodeGetName(node);
break;
default:
return;
}
t = newTag((char *)name);
if (!t)
return;
if (!opentag) {
t->slash = true;
return;
}
/* if a js script, remember the line number for error messages */
if (t->action == TAGACT_SCRIPT)
t->js_ln = tidyNodeLine(node);
/* this is the open tag, set the attributes */
/* special case for text tag */
if (t->action == TAGACT_TEXT) {
TidyBuffer tnv = { 0 }; /* text-node value */
tidyBufClear(&tnv);
tidyNodeGetValue(tdoc, node, &tnv);
if (tnv.size) {
t->textval = cloneString(tnv.bp);
tidyBufFree(&tnv);
}
}
nattr = 0;
tattr = tidyAttrFirst(node);
while (tattr != NULL) {
++nattr;
tattr = tidyAttrNext(tattr);
}
t->attributes = allocMem(sizeof(char *) * (nattr + 1));
t->atvals = allocMem(sizeof(char *) * (nattr + 1));
i = 0;
tattr = tidyAttrFirst(node);
while (tattr != NULL) {
t->attributes[i] = cloneString(tidyAttrName(tattr));
t->atvals[i] = cloneString(tidyAttrValue(tattr));
++i;
tattr = tidyAttrNext(tattr);
}
t->attributes[i] = 0;
t->atvals[i] = 0;
/* innerHTML, only for certain tags */
if (t->info->bits & TAG_INNERHTML) {
TidyBuffer tnv = { 0 }; /* text-node value */
tidyBufClear(&tnv);
t->innerHTML = emptyString;
tidyNodeGetText(tdoc, node, &tnv);
if (tnv.size) {
/* But it's not the original html, it has been sanitized.
* Put a cap on size, else memory consumed could, theoretically,
* grow as the size of the document squared. */
if (tnv.size <= 4096)
t->innerHTML = cloneString(tnv.bp);
tagStrip(t->innerHTML);
tidyBufFree(&tnv);
}
}
} /* convertNode */
/* this is strictly for debugging, level >= 5 */
static void printNode(TidyNode node, int level, bool opentag)
{
ctmbstr name;
TidyAttr tattr;
if (!opentag) {
puts("}");
return;
}
switch (tidyNodeGetType(node)) {
case TidyNode_Root:
name = "Root";
break;
case TidyNode_DocType:
name = "DOCTYPE";
break;
case TidyNode_Comment:
name = "Comment";
break;
case TidyNode_ProcIns:
name = "Processing Instruction";
break;
case TidyNode_Text:
name = "Text";
break;
case TidyNode_CDATA:
name = "CDATA";
break;
case TidyNode_Section:
name = "XML Section";
break;
case TidyNode_Asp:
name = "ASP";
break;
case TidyNode_Jste:
name = "JSTE";
break;
case TidyNode_Php:
name = "PHP";
break;
case TidyNode_XmlDecl:
name = "XML Declaration";
break;
case TidyNode_Start:
case TidyNode_End:
case TidyNode_StartEnd:
default:
name = tidyNodeGetName(node);
break;
}
assert(name != NULL);
printf("Node(%d): %s {\n", level, ((char *)name));
/* the ifs could be combined with && */
if (stringEqual(((char *)name), "Text")) {
TidyBuffer tnv = { 0 }; /* text-node value */
tidyBufClear(&tnv);
tidyNodeGetValue(tdoc, node, &tnv);
printf("Text: %s\n", tnv.bp);
if (tnv.size)
tidyBufFree(&tnv);
}
/* Get the first attribute for the node */
tattr = tidyAttrFirst(node);
while (tattr != NULL) {
/* Print the node and its attribute */
printf("@%s = %s\n", tidyAttrName(tattr), tidyAttrValue(tattr));
/* Get the next attribute */
tattr = tidyAttrNext(tattr);
}
} /* printNode */
static void parse_html(TidyDoc tdoc, TidyNode tnod, const url_list_t *elem, int indent, FILE *outfile)
{
TidyNode child;
TidyAttr attr;
TidyAttrId attr_id = TidyAttr_UNKNOWN;
TidyNodeType node_type;
TidyTagId node_id;
ctmbstr name;
char *url, *relative_url = NULL;
int found = 0;
int get_html_link = (!option_values.depth || elem->level < option_values.depth);
int get_int_html_link = (!option_values.depth || elem->level < option_values.depth+1);
int get_ext_depends = ((!option_values.depth || elem->level < option_values.depth+1)
&& !option_values.no_html_dependencies);
for (child = tidyGetChild(tnod); child; child = tidyGetNext(child)) {
node_type = tidyNodeGetType(child);
switch (node_type) {
case TidyNode_Start:
case TidyNode_StartEnd:
node_id = tidyNodeGetId(child);
if (get_html_link && (node_id == TidyTag_A || node_id == TidyTag_AREA || node_id == TidyTag_MAP)) {
found = 1;
attr_id = TidyAttr_HREF;
}
else if (get_int_html_link && (node_id == TidyTag_FRAME || node_id == TidyTag_IFRAME)) {
found = 1;
attr_id = TidyAttr_SRC;
}
else if (get_ext_depends) {
if (node_id == TidyTag_LINK) {
found = 1;
attr_id = TidyAttr_HREF;
}
else if (node_id == TidyTag_IMG || node_id == TidyTag_SCRIPT) {
found = 1;
attr_id = TidyAttr_SRC;
}
else {
found = 0;
attr_id = TidyAttr_UNKNOWN;
}
}
else {
found = 0;
attr_id = TidyAttr_UNKNOWN;
}
if (found && (attr = tidyAttrGetById(child, attr_id)) != NULL) {
url = (char *) tidyAttrValue(attr);
string_free(relative_url);
if (url && *url)
add_new_url_and_check(elem, url, outfile ? &relative_url : NULL);
}
if (outfile && (name = tidyNodeGetName(child)) != NULL) {
fprintf(outfile, "%*.*s%s", indent, indent, "<", name);
for (attr = tidyAttrFirst(child); attr; attr = tidyAttrNext(attr)) {
fprintf(outfile, " %s", tidyAttrName(attr));
if (relative_url && (tidyAttrGetId(attr) == attr_id))
fprintf(outfile, "=\"%s\"", relative_url);
else if (tidyAttrValue(attr))
fprintf(outfile, "=\"%s\"", tidyAttrValue(attr) ? tidyAttrValue(attr) : "");
else
fprintf(outfile, "=\"\"");
}
string_free(relative_url);
if (node_type == TidyNode_StartEnd)
fprintf(outfile, "/>\n");
else {
fprintf(outfile, ">\n");
parse_html(tdoc, child, elem, indent + 1, outfile);
fprintf(outfile, "%*.*s%s>\n", indent + 1, indent + 1, "</", name);
}
}
else {
string_free(relative_url);
parse_html(tdoc, child, elem, indent + 1, outfile);
}
break;
case TidyNode_End:
if (outfile) {
if ((name = tidyNodeGetName(child)) != NULL)
fprintf(outfile, "%*.*s/%s>\n", indent, indent, "<", name);
}
break;
case TidyNode_Text:
if (outfile) {
TidyBuffer buf;
TidyTagId parent_node_id = tidyNodeGetId(tnod);
tidyBufInit(&buf);
if (parent_node_id == TidyTag_SCRIPT || parent_node_id == TidyTag_STYLE)
tidyNodeGetValue(tdoc, child, &buf);
else
tidyNodeGetText(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "%s", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
case TidyNode_Comment:
if (outfile) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "<!--%s-->\n", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
case TidyNode_CDATA:
if (outfile) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "<![CDATA[%s]]>\n", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
case TidyNode_DocType:
if (outfile) {
int pub = 0;
fprintf(outfile, "<!DOCTYPE %s", tidyNodeGetName(child));
for (attr = tidyAttrFirst(child); attr; attr = tidyAttrNext(attr)) {
if (!pub) {
fprintf(outfile, " %s", tidyAttrName(attr));
if (!string_casecmp(tidyAttrName(attr), "PUBLIC"))
pub = 1;
}
if (tidyAttrValue(attr))
fprintf(outfile, " \"%s\"", tidyAttrValue(attr));
}
fprintf(outfile, ">\n");
}
break;
default:
if (outfile) {
TidyBuffer buf;
tidyBufInit(&buf);
tidyNodeGetValue(tdoc, child, &buf);
if (buf.bp)
fprintf(outfile, "%s", (char *)buf.bp);
tidyBufFree(&buf);
}
break;
}
}
}