void HTMLParser :: headTag( TElementShared aParent ) {
cout << "head tag found\n";
// Add to parent
TElementShared element = mDocument->createElement( "head" );
aParent->appendChild( element );
bool insideHead = true;
while ( insideHead ) {
getTag();
if ( isStartTag() ) {
if ( isTitleTag() ||
isStyleTag() ||
isScriptTag() ) {
normalHeadTag( element );
continue;
}
if ( isIsIndexTag() ||
isBaseTag() ||
isMetaTag() ||
isLinkTag() ) {
emptyElementTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "headTag: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( isHeadTag() ) {
cout << "head closing tag found\n";
// End the while loop
insideHead = false;
}
else {
cout << "headTag: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
}
}
void HTMLParser :: htmlTag() {
cout << "html tag found\n";
// Add to DOM tree
TElementShared element = mDocument->createElement( "html" );
mDocument->appendChild( element );
bool insideHtml = true;
while ( insideHtml ) {
getTag();
if ( isStartTag() ) {
if ( isHeadTag() ) {
headTag( element );
continue;
}
if ( isBodyTag() ) {
bodyStyleTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
cout << "htmlTag: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( !mTag.compare( "html" ) ) {
cout << "html closing tag found\n";
insideHtml = false;
}
else {
cout << "htmlTag: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
}
}
void HTMLParser :: normalHeadTag( TElementShared aParent ) {
cout << mTag << " tag found\n";
// Save the tag name
string tag = mTag;
// Add to parent
TElementShared element = mDocument->createElement( tag );
aParent->appendChild( element );
bool insideHeadLevel = true;
string data;
while ( insideHeadLevel ) {
data += getText();
getTag();
if ( !isStartTag() ) {
if ( !mTag.compare( tag ) ) {
cout << mTag << " closing tag found\n";
insideHeadLevel = false;
}
else {
cout << "normalHead: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
else {
if ( isCommentTag() ) {
commentTag( element );
continue;
}
cout << "normalHead: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
}
TTextShared text = mDocument->createText( data );
element->appendChild( text );
cout << "Text is: " << data << endl;
}
void XMLPathTraversalTest()
{
typedef std::map<std::basic_string<wchar_t>,
std::basic_string<wchar_t> > AttributesType;
TagElement<wchar_t> rootTag(L"root");
TagElement<wchar_t> childTag1(L"child1");
TagElement<wchar_t> childTag1_(L"child1");
TagElement<wchar_t> childTag2(L"child2");
StringElement<wchar_t> stringTag(L"string");
CommentElement<wchar_t> commentTag(L"comment");
rootTag.addChild(&childTag1);
rootTag.addChild(&childTag2);
rootTag.addChild(&childTag1_);
childTag1.addChild(&stringTag);
childTag1_.addChild(&commentTag);
CPPUNIT_ASSERT(rootTag.children.size() == 3);
CPPUNIT_ASSERT(dynamic_cast<TagElement<wchar_t>*>
(rootTag.getChildElement(L"child1"))->getTagName() ==
L"child1");
CPPUNIT_ASSERT(dynamic_cast<TagElement<wchar_t>*>
(rootTag.getChildElement(L"child2"))->getTagName() ==
L"child2");
std::vector<Element<wchar_t>*> result;
XMLPath<wchar_t> path_root(L"/root");
result = path_root.evaluate(&rootTag);
CPPUNIT_ASSERT(result.size() == 1);
CPPUNIT_ASSERT(result[0] == &rootTag);
XMLPath<wchar_t> path_roots(L"/root[]");
result = path_roots.evaluate(&rootTag);
CPPUNIT_ASSERT(result.size() == 1);
CPPUNIT_ASSERT(result[0] == &rootTag);
XMLPath<wchar_t> path_child1(L"/root/child1");
result = path_child1.evaluate(&rootTag);
CPPUNIT_ASSERT(result.size() == 1);
CPPUNIT_ASSERT(result[0] == &childTag1);
XMLPath<wchar_t> path_child1s(L"/root/child1[]");
result = path_child1s.evaluate(&rootTag);
CPPUNIT_ASSERT(result.size() == 2);
CPPUNIT_ASSERT(result[0] == &childTag1);
CPPUNIT_ASSERT(result[1] == &childTag1_);
XMLPath<wchar_t> path_text(L"/root/child1[]/#text");
result = path_text.evaluate(&rootTag);
CPPUNIT_ASSERT(result.size() == 1);
CPPUNIT_ASSERT(result[0] == &stringTag);
XMLPath<wchar_t> path_comment(L"/root/child1[]/#comment");
result = path_comment.evaluate(&rootTag);
CPPUNIT_ASSERT(result.size() == 1);
CPPUNIT_ASSERT(result[0] == &commentTag);
rootTag.removeChild(&childTag1);
rootTag.removeChild(&childTag2);
rootTag.removeChild(&childTag1_);
childTag1.removeChild(&stringTag);
childTag1_.removeChild(&commentTag);
}
void HTMLParser :: normalTextTag( TElementShared aParent, bool aConserveSpaces, bool aInsideAnchor ) {
cout << mTag << " tag found\n";
// Save the tag name
string tag = mTag;
// Add to parent
TElementShared element = mDocument->createElement( mTag );
aParent->appendChild( element );
bool insideNormalText = true;
string text;
string attribute;
while ( insideNormalText ) {
string data = getString( aConserveSpaces );
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( !isStartTag() ) {
if ( !mTag.compare( tag ) ) {
cout << tag << " closing tag found\n";
insideNormalText = false;
continue;
}
cout << "normalText: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
else {
if ( isTextLevelTag() ) {
textLevelTag( element, aConserveSpaces, aInsideAnchor );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "normalText: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
break;
}
case TEXT : {
if ( ( data.compare( " " ) && data.compare( "" ) ) || ( aConserveSpaces && data.compare( "" ) ) ) {
cout << "Text is:" << endl << data << endl;
TTextShared text = mDocument->createText( data );
element->appendChild( text );
}
break;
}
}
}
}
void HTMLParser :: mapTag( TElementShared aParent ) {
cout << "map tag found\n";
// Add to parent
TElementShared element = mDocument->createElement( "map" );
aParent->appendChild( element );
bool insideMap = true;
string attribute;
while ( insideMap ) {
string data = getString();
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( isStartTag() ) {
if ( isAreaTag() ) {
emptyElementTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "map: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( isMapTag() ) {
cout << "map closing tag found\n";
// End the while loop
insideMap = false;
}
else {
cout << "map: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
break;
}
case TEXT : {
if ( ( data.compare( " " ) && data.compare( "" ) ) ) {
cout << "Text found in illegal place. Skipping...\n";
}
break;
}
}
}
}
void HTMLParser :: pcDataTag( TElementShared aParent, bool aConserveSpaces ) {
cout << mTag << " tag found\n";
// Save the tag name
string tag = mTag;
// Add to parent
TElementShared element = mDocument->createElement( mTag );
aParent->appendChild( element );
bool insidePcData = true;
string attribute;
while ( insidePcData ) {
string data = getString( aConserveSpaces );
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( isStartTag() ) {
if ( isCommentTag() ) {
commentTag( element );
continue;
}
cout << mTag << " closed implicitly\n";
// End the while loop
insidePcData = false;
backPedal();
continue;
}
else {
if ( !mTag.compare( tag ) ) {
cout << mTag << " closing tag found\n";
// End the while loop
insidePcData = false;
}
else {
cout << mTag << " closed implicitly\n";
// End the while loop
insidePcData = false;
backPedal();
continue;
}
}
break;
}
case TEXT : {
if ( ( data.compare( " " ) && data.compare( "" ) ) || ( aConserveSpaces && data.compare( "" ) ) ) {
cout << "Text is:" << endl << data << endl;
TTextShared text = mDocument->createText( data );
element->appendChild( text );
}
break;
}
}
}
}
void HTMLParser :: preTag( TElementShared aParent ) {
cout << "pre tag found\n";
// Add to parent
TElementShared element = mDocument->createElement( "pre" );
aParent->appendChild( element );
bool insidePre = true;
string attribute;
while ( insidePre ) {
string data = getString( true );
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( !isStartTag() ) {
if ( isPreTag() ) {
cout << "pre closing tag found\n";
insidePre = false;
continue;
}
cout << "pre: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
else {
if ( isFontStylePreTag() ||
isPhraseTag() ) {
normalTextTag( element, true );
continue;
}
if ( isFormFieldTag() ) {
formFieldTag( element, true );
continue;
}
if ( isAnchorTag() ) {
normalTextTag( element, true, true );
continue;
}
if ( isAppletTag() ) {
appletTag( element, true );
continue;
}
if ( isEmptyTextPreTag() ) {
emptyElementTag( element );
continue;
}
if ( isMapTag() ) {
mapTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "pre: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
break;
}
case TEXT : {
if ( data.compare( " " ) && data.compare( "" ) ) {
cout << "Text is:" << endl << data << endl;
TTextShared text = mDocument->createText( data );
element->appendChild( text );
}
break;
}
}
}
}
void HTMLParser :: trTag( TElementShared aParent ) {
cout << "tr tag found\n";
// Add to parent
TElementShared element = mDocument->createElement( "tr" );
aParent->appendChild( element );
bool insideTr = true;
string attribute;
while ( insideTr ) {
// Warning: more possible than a tag only
string data = getString();
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( isStartTag() ) {
if ( isThTag() || isTdTag() ) {
bodyStyleTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "tr: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( isTrTag() ) {
cout << "tr closing tag found\n";
// End the while loop
insideTr = false;
}
else {
cout << "tr: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
break;
}
case TEXT : {
if ( data.compare( " " ) && data.compare( "" ) ) {
cout << "Text found in illegal place. Skipping...\n";
}
break;
}
}
}
}
void HTMLParser :: listTag( TElementShared aParent ) {
cout << mTag << " tag found\n";
// Save the tag name
string tag = mTag;
// To check if this list can have list items or not
bool listItem = false;
if ( isULTag() || isOLTag() ) {
listItem = true;
}
// Add to parent
TElementShared element = mDocument->createElement( mTag );
aParent->appendChild( element );
bool insideList = true;
string attribute;
while ( insideList ) {
// Warning: more possible than a tag only
string data = getString();
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( isStartTag() ) {
if ( ( isLITag() && listItem ) ||
( isDDTag() && !listItem ) ) {
flowLevelTag( element );
continue;
}
if ( isDTTag() && !listItem ) {
textLevelTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "list: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( !mTag.compare( tag ) ) {
cout << mTag << " closing tag found\n";
// End the while loop
insideList = false;
}
else {
cout << "list: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
break;
}
case TEXT : {
if ( data.compare( " " ) && data.compare( "" ) ) {
cout << "Text found in illegal place. Skipping...\n";
}
break;
}
}
}
}
void HTMLParser :: pTag( TElementShared aParent ) {
cout << "p tag found\n";
// Add to parent
TElementShared element = mDocument->createElement( "p" );
aParent->appendChild( element );
bool insideP = true;
string attribute;
while ( insideP ) {
string data = getString();
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( !isStartTag() ) {
if ( isPTag() ) {
cout << "p closing tag found\n";
insideP = false;
continue;
}
if ( isAdressTag() ||
isBodyStyleTag() ||
isFormTag() ||
isListTag() ) {
cout << "p closed implicitly\n";
insideP = false;
backPedal();
continue;
}
cout << "p: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
else {
if ( isBlockLevelTag() ||
isHeadingTag() ||
isLITag() ) {
cout << "p closed implicitly\n";
insideP = false;
backPedal();
continue;
}
if ( isTextLevelTag() ) {
textLevelTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "p: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
break;
}
case TEXT : {
if ( data.compare( " " ) && data.compare( "" ) ) {
cout << "Text is:" << endl << data << endl;
TTextShared text = mDocument->createText( data );
element->appendChild( text );
}
break;
}
}
}
}
void HTMLParser :: adressTag( TElementShared aParent ) {
cout << "adress tag found\n";
// Add to parent
TElementShared element = mDocument->createElement( "adress" );
aParent->appendChild( element );
bool insideAdress = true;
string attribute;
while ( insideAdress ) {
string data = getString();
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( isStartTag() ) {
if ( isPTag() ) {
pTag( element );
continue;
}
if ( isTextLevelTag() ) {
textLevelTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "adress: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( !mTag.compare( "adress" ) ) {
cout << mTag << "closing tag found\n";
// End the while loop
insideAdress = false;
}
else {
cout << "adress: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
break;
}
case TEXT : {
if ( data.compare( " " ) && data.compare( "" ) ) {
cout << "Text is:" << endl << data << endl;
TTextShared text = mDocument->createText( data );
element->appendChild( text );
}
break;
}
}
}
}
void HTMLParser :: bodyStyleTag( TElementShared aParent, bool aInsideForm ) {
cout << mTag << " tag found\n";
// Save the tag name
string tag = mTag;
// Add to parent
TElementShared element = mDocument->createElement( mTag );
aParent->appendChild( element );
bool insideBodyStyle = true;
string attribute;
while ( insideBodyStyle ) {
// Warning: more possible than a tag only
string data = getString();
switch ( mStringType ) {
case ATTR : {
attribute = data;
if ( mAttrNoValue ) {
element->setAttribute( attribute, "" );
attribute = "";
}
break;
}
case ATTRVALUE : {
if ( attribute.compare( "" ) ) {
// Attribute has a name
// I'll declare it legal
element->setAttribute( attribute, data );
attribute = "";
}
break;
}
case TAG : {
if ( isStartTag() ) {
if ( isHeadingTag() ) {
normalTextTag( element );
continue;
}
if ( isAdressTag() ) {
adressTag( element );
continue;
}
if ( isBlockLevelTag() ) {
blockLevelTag( element, aInsideForm );
continue;
}
if ( isTextLevelTag() ) {
textLevelTag( element );
continue;
}
if ( isCommentTag() ) {
commentTag( element );
continue;
}
// Not a known tag
cout << "bodyStyle: Unexpected tag found: " << mTag << ". Skipping...\n";
skipTag();
}
else {
if ( !mTag.compare( tag ) ) {
cout << mTag << " closing tag found\n";
// End the while loop
insideBodyStyle = false;
}
else {
cout << "bodyStyle: Unexpected closing tag found: " << mTag << ". Skipping...\n";
}
}
break;
}
case TEXT : {
if ( data.compare( " " ) && data.compare( "" ) ) {
cout << "Text is:" << endl << data << endl;
TTextShared text = mDocument->createText( data );
element->appendChild( text );
}
break;
}
}
}
}