nsresult
nsLoggingSink::OpenNode(const char* aKind, const nsIParserNode& aNode) {
WriteTabs(mOutput,++mLevel);
PR_fprintf(mOutput,"<open container=");
nsHTMLTag nodeType = nsHTMLTag(aNode.GetNodeType());
if ((nodeType >= eHTMLTag_unknown) &&
(nodeType <= nsHTMLTag(NS_HTML_TAG_MAX))) {
const PRUnichar* tag = nsHTMLTags::GetStringValue(nodeType);
PR_fprintf(mOutput, "\"%s\"", NS_ConvertUTF16toUTF8(tag).get());
}
else {
char* text = nsnull;
GetNewCString(aNode.GetText(), &text);
if(text) {
PR_fprintf(mOutput, "\"%s\"", text);
nsMemory::Free(text);
}
}
if (WillWriteAttributes(aNode)) {
PR_fprintf(mOutput, ">\n");
WriteAttributes(aNode);
PR_fprintf(mOutput, "</open>\n");
}
else {
PR_fprintf(mOutput, ">\n");
}
return NS_OK;
}
nsresult
nsLoggingSink::LeafNode(const nsIParserNode& aNode)
{
WriteTabs(mOutput,1+mLevel);
nsHTMLTag nodeType = nsHTMLTag(aNode.GetNodeType());
if ((nodeType >= eHTMLTag_unknown) &&
(nodeType <= nsHTMLTag(NS_HTML_TAG_MAX))) {
const PRUnichar* tag = nsHTMLTags::GetStringValue(nodeType);
if(tag)
PR_fprintf(mOutput, "<leaf tag=\"%s\"", NS_ConvertUTF16toUTF8(tag).get());
else
PR_fprintf(mOutput, "<leaf tag=\"???\"");
if (WillWriteAttributes(aNode)) {
PR_fprintf(mOutput, ">\n");
WriteAttributes(aNode);
PR_fprintf(mOutput, "</leaf>\n");
}
else {
PR_fprintf(mOutput, "/>\n");
}
}
else {
PRInt32 pos;
nsAutoString tmp;
char* str = nsnull;
switch (nodeType) {
case eHTMLTag_whitespace:
case eHTMLTag_text:
GetNewCString(aNode.GetText(), &str);
if(str) {
PR_fprintf(mOutput, "<text value=\"%s\"/>\n", str);
nsMemory::Free(str);
}
break;
case eHTMLTag_newline:
PR_fprintf(mOutput, "<newline/>\n");
break;
case eHTMLTag_entity:
tmp.Append(aNode.GetText());
tmp.Cut(0, 1);
pos = tmp.Length() - 1;
if (pos >= 0) {
tmp.Cut(pos, 1);
}
PR_fprintf(mOutput, "<entity value=\"%s\"/>\n", NS_LossyConvertUTF16toASCII(tmp).get());
break;
default:
NS_NOTREACHED("unsupported leaf node type");
}//switch
}
return NS_OK;
}
NS_IMETHODIMP
mozSanitizingHTMLSerializer::AddLeaf(const nsIParserNode& aNode)
{
eHTMLTags type = (eHTMLTags)aNode.GetNodeType();
const nsAString& text = aNode.GetText();
mParserNode = const_cast<nsIParserNode*>(&aNode);
return DoAddLeaf(type, text);
}
NS_IMETHODIMP
mozSanitizingHTMLSerializer::OpenContainer(const nsIParserNode& aNode)
{
PRInt32 type = aNode.GetNodeType();
mParserNode = const_cast<nsIParserNode *>(&aNode);
return DoOpenContainer(type);
}
nsresult
nsLoggingSink::WriteAttributes(const nsIParserNode& aNode) {
WriteTabs(mOutput,1+mLevel);
nsAutoString tmp;
PRInt32 ac = aNode.GetAttributeCount();
for (PRInt32 i = 0; i < ac; ++i) {
char* key=nsnull;
char* value=nsnull;
const nsAString& k = aNode.GetKeyAt(i);
const nsAString& v = aNode.GetValueAt(i);
GetNewCString(k, &key);
if(key) {
PR_fprintf(mOutput, " <attr key=\"%s\" value=\"", key);
nsMemory::Free(key);
}
tmp.Truncate();
tmp.Append(v);
if(!tmp.IsEmpty()) {
PRUnichar first = tmp.First();
if ((first == '"') || (first == '\'')) {
if (tmp.Last() == first) {
tmp.Cut(0, 1);
PRInt32 pos = tmp.Length() - 1;
if (pos >= 0) {
tmp.Cut(pos, 1);
}
} else {
// Mismatched quotes - leave them in
}
}
GetNewCString(tmp, &value);
if(value) {
PR_fprintf(mOutput, "%s\"/>\n", value);
WriteTabs(mOutput,1+mLevel);
nsMemory::Free(value);
}
}
}
WriteTabs(mOutput,1+mLevel);
return NS_OK;
}
PRBool
nsLoggingSink::WillWriteAttributes(const nsIParserNode& aNode)
{
PRInt32 ac = aNode.GetAttributeCount();
if (0 != ac) {
return PR_TRUE;
}
return PR_FALSE;
}
nsresult FBMLContentSink::AddComment(const nsIParserNode& aNode) {
PRInt32 tagId = aNode.GetNodeType();
#ifdef VERBOSE_LOGGING
printf("AddComment: %d\n", tagId);
string text;
nsString stext(aNode.GetText());
UTF16ToStdString(text, stext);
printf("\tComment Text: [%s]\n", text.c_str());
#endif
if (mPreserveComment) {
OpenContainer(aNode);
PRInt32 lineNo = aNode.GetSourceLineNumber();
fbml_node *node = fbml_node_add_child(mCurrent, tagId, GetFlag(tagId), lineNo);
string text;
nsString stext(aNode.GetText());
UTF16ToStdString(text, stext);
node->text = strdup(text.c_str());
CloseContainer((nsHTMLTag)tagId);
}
return NS_OK;
}
/**
* This gets called by the parser when you want to add
* a PI node to the current container in the content
* model.
*
* @updated gess 3/25/98
* @param
* @return
*/
NS_IMETHODIMP
nsLoggingSink::AddProcessingInstruction(const nsIParserNode& aNode){
#ifdef VERBOSE_DEBUG
DebugDump("<",aNode.GetText(),(mNodeStackPos)*2);
#endif
nsresult theResult=NS_OK;
//then proxy the call to the real sink if you have one.
if(mSink) {
theResult=mSink->AddProcessingInstruction(aNode);
}
return theResult;
}
nsresult FBMLContentSink::OpenContainer(const nsIParserNode& aNode) {
PRInt32 tagId = aNode.GetNodeType();
bool attrSpecial = false;
contexts *back_contexts;
PRInt32 attrId;
#ifdef VERBOSE_LOGGING
printf("OpenContainer: %d [%s]\n", tagId,
GetTagName((nsHTMLTag)tagId).c_str());
#endif
if (mRuleStack.empty()) {
back_contexts = NULL;
} else {
back_contexts = mRuleStack.back();
}
if (tagId == eHTMLTag_userdefined) {
string text;
nsString stext(aNode.GetText());
UTF16ToStdString(text, stext);
PRInt32 lineNo = aNode.GetSourceLineNumber();
char errorMsg[512];
snprintf(errorMsg, sizeof(errorMsg),
"FBML Error (line %d): unknown tag \"%s\"\n",
(int)lineNo, text.c_str());
mError += errorMsg;
} else if (!mSkipSchemaChecking && mCurrent && back_contexts) {
// check illegal children
vector<vector<char> > &rules = back_contexts->tag_rules;
if (tagId >= 0 && tagId < rules.size()) {
for (fbml_node *p = mCurrent; p; p = p->parent) {
unsigned short parentTagId = p->eHTMLTag;
if (parentTagId >= rules.size() || rules[tagId][parentTagId]) {
PRInt32 lineNo = aNode.GetSourceLineNumber();
char errorMsg[512];
snprintf(errorMsg, sizeof(errorMsg),
"FBML Error (line %d): illegal tag \"%s\" under \"%s\"\n",
(int)lineNo,
GetTagName((nsHTMLTag)tagId).c_str(),
GetTagName((nsHTMLTag)parentTagId).c_str());
mError += errorMsg;
tagId = eHTMLTag_userdefined; // nuke it
}
}
}
}
PRInt32 lineNo = aNode.GetSourceLineNumber();
fbml_node *node = fbml_node_add_child(mCurrent, tagId, GetFlag(tagId), lineNo);
if (node) {
int count = aNode.GetAttributeCount();
for (int i = 0; i < count; i++) {
string n; nsString sn(aNode.GetKeyAt(i)); UTF16ToStdString(n, sn);
string v; nsString sv(aNode.GetValueAt(i)); UTF16ToUTF8String(v, sv);
attrId = nsHTMLAttrs::LookupAttr(sn);
char attrFlag = GetAttrFlag(attrId);
// If we are in internal page skip all these checks
if (mSkipSchemaChecking) {
fbml_node_add_attribute(node, (char*)n.c_str(), (char*)v.c_str(), attrFlag, 1);
continue;
}
bool fail = false;
for (char *p = (char*)n.c_str(); *p; p++) {
if (*p >= 'A' && *p <= 'Z') *p |= 0x20;
// make sure attrs are well formed
if (*p != ':' && *p !='_' && *p != '-' && (*p<'a' || *p>'z') && (*p<'0' || *p>'9') ) {
fail = true;
break;
}
}
if (fail) {
PRInt32 lineNo = aNode.GetSourceLineNumber();
char errorMsg[512];
snprintf(errorMsg, sizeof(errorMsg),
"FBML Error (line %d): illegal attr \"%s\" in tag \"%s\". Attribute names can only contain alphanumeric characters, underscores, and hyphens.",
(int)lineNo,
n.c_str(),
GetTagName((nsHTMLTag)tagId).c_str()
);
mError += errorMsg;
}
// check schema
else if (back_contexts && attrId >= 0 &&
attrId < back_contexts->attr_rules.size() &&
back_contexts->attr_rules[attrId]) {
PRInt32 lineNo = aNode.GetSourceLineNumber();
char errorMsg[512];
snprintf(errorMsg, sizeof(errorMsg),
"FBML Error (line %d): illegal attr \"%s\" in tag \"%s\" under \"%s\"\n",
(int)lineNo,
n.c_str(),
GetTagName((nsHTMLTag)tagId).c_str(),
back_contexts->tag_name.c_str());
mError += errorMsg;
} else if (mCSSSanitizer && (attrFlag & FB_FLAG_ATTR_STYLE)) {
char *sanitized_css = NULL;
char *error = NULL;
int ret = fbml_sanitize_css((char*)v.c_str(), 1, lineNo,
mCSSSanitizer, &sanitized_css, &error);
if (sanitized_css) {
fbml_node_add_attribute(node, (char*)n.c_str(), sanitized_css, attrFlag, 0);
}
if (error) {
mError += error;
free(error);
}
} else if (mJSSanitizer && (attrFlag & FB_FLAG_ATTR_SCRIPT)) {
char *sanitized_js = NULL;
char *error = NULL;
int ret = fbml_sanitize_js((char*)v.c_str(), v.length(), 1, lineNo,
mJSSanitizer, &sanitized_js, &error);
if (sanitized_js) {
if (mJSSanitizer->pfunc_eh_translator) {
char *translated = mJSSanitizer->pfunc_eh_translator
(sanitized_js, mJSSanitizer->eh_translate_data);
free(sanitized_js);
sanitized_js = translated;
}
fbml_node_add_attribute(node, (char*)n.c_str(), sanitized_js, attrFlag, 0);
}
if (error) {
mError += error;
free(error);
}
} else if (mRewriter && mRewriter->pfunc_rewriter && (attrFlag & FB_FLAG_ATTR_REWRITE)){
char *rewritten_attr =
mRewriter->pfunc_rewriter((char *)GetTagName((nsHTMLTag)tagId).c_str(),
(char *)n.c_str(),
(char *)v.c_str(), mRewriter->rewrite_data);
fbml_node_add_attribute(node, (char*)n.c_str(), rewritten_attr, attrFlag, 0);
} else {
if (attrFlag & FB_FLAG_ATTR_SPECIAL) {
attrSpecial = true;
}
fbml_node_add_attribute(node, (char*)n.c_str(), (char*)v.c_str(), attrFlag, 1);
}
}
if (attrSpecial) {
fbml_node_add_flags(node, FB_FLAG_HAS_SPECIAL_ATTR);
}
mCurrent = node;
if (!mSkipSchemaChecking && tagId >= 0 && tagId < m_context_rules.size()) {
contexts *rule = &m_context_rules[tagId];
if (rule &&
(!rule->tag_rules.empty() || !rule->attr_rules.empty())) {
mRuleStack.push_back(rule);
} else if (mRuleStack.empty()) {
mRuleStack.push_back(NULL);
} else {
mRuleStack.push_back(mRuleStack.back());
}
} else {
mRuleStack.push_back(NULL);
}
}
return NS_OK;
}
nsresult FBMLContentSink::AddLeaf(const nsIParserNode& aNode) {
PRInt32 tagId = aNode.GetNodeType();
#ifdef VERBOSE_LOGGING
printf("AddLeaf: %d [%s] Attribute Count: [%d]\n", tagId,
GetTagName((nsHTMLTag)tagId).c_str(), aNode.GetAttributeCount());
string text;
nsString stext(aNode.GetText());
UTF16ToStdString(text, stext);
printf("\tLeaf Text: [%s]\n", text.c_str());
#endif
if (tagId == eHTMLTag_title || tagId == eHTMLTag_script ||
(GetFlag(tagId) & FB_FLAG_STYLE)) {
nsCOMPtr<nsIDTD> dtd;
mParser->GetDTD(getter_AddRefs(dtd));
nsAutoString script;
PRInt32 lineNo = 0;
dtd->CollectSkippedContent(tagId, script, lineNo);
string text;
UTF16ToStdString(text, script);
#ifdef VERBOSE_LOGGING
printf("\tCollected Content:: [%s]\n", text.c_str());
#endif
OpenContainer(aNode);
fbml_node *node = fbml_node_add_child(mCurrent, tagId, GetFlag(tagId), lineNo);
if (mCSSSanitizer && (GetFlag(tagId) & FB_FLAG_STYLE)) {
char *sanitized_css = NULL;
char *error = NULL;
int ret = fbml_sanitize_css((char*)text.c_str(), 0, lineNo,
mCSSSanitizer, &sanitized_css, &error);
if (sanitized_css) {
node->text = sanitized_css;
}
if (error) {
mError += error;
free(error);
}
} else if (mJSSanitizer && tagId == eHTMLTag_script) {
char *sanitized_js = NULL;
char *error = NULL;
int ret = fbml_sanitize_js((char*)text.c_str(), text.length(), 0,
lineNo, mJSSanitizer, &sanitized_js,
&error);
if (sanitized_js) {
node->text = sanitized_js;
}
if (error) {
mError += error;
free(error);
}
} else {
node->text = strdup(text.c_str());
}
CloseContainer((nsHTMLTag)tagId);
return NS_OK;
}
switch (aNode.GetTokenType()) {
case eToken_start:
OpenContainer(aNode);
CloseContainer((nsHTMLTag)tagId);
break;
case eToken_text:
case eToken_whitespace:
case eToken_newline:
{
PRInt32 lineNo = aNode.GetSourceLineNumber();
fbml_node *node = fbml_node_add_child(mCurrent, tagId, GetFlag(tagId), lineNo);
string text;
nsString stext(aNode.GetText());
UTF16ToStdString(text, stext);
node->text = strdup(text.c_str());
}
break;
}
return NS_OK;
}
