std::string GumboInterface::prettyprint(GumboNode* node, int lvl, const std::string indent_chars)
{
// special case the document node
if (node->type == GUMBO_NODE_DOCUMENT) {
std::string results = build_doctype(node);
results.append(prettyprint_contents(node,lvl+1,indent_chars));
return results;
}
std::string tagname = get_tag_name(node);
std::string parentname = get_tag_name(node->parent);
bool in_head = (parentname == "head");
bool is_structural = in_set(structural_tags, tagname);
bool is_inline = in_set(nonbreaking_inline, tagname);
// build attr string
std::string atts = "";
bool no_entity_substitution = in_set(no_entity_sub, tagname);
const GumboVector * attribs = &node->v.element.attributes;
for (unsigned int i=0; i< attribs->length; ++i) {
GumboAttribute* at = static_cast<GumboAttribute*>(attribs->data[i]);
atts.append(build_attributes(at, no_entity_substitution));
}
bool is_void_tag = in_set(void_tags, tagname);
// get tag contents
std::string contents = "";
if (!is_void_tag) {
if (is_structural && tagname != "html") {
contents = prettyprint_contents(node, lvl+1, indent_chars);
} else {
contents = prettyprint_contents(node, lvl, indent_chars);
}
}
bool keep_whitespace = in_set(preserve_whitespace, tagname);
if (!keep_whitespace && !is_inline) {
rtrim(contents);
}
bool single = is_void_tag;
// for xhtml serialization that allows non-void tags to be self-closing
// uncomment the following line
// single = single || contents.empty();
char c = indent_chars.at(0);
int n = indent_chars.length();
std::string indent_space = std::string((lvl-1)*n,c);
// handle self-closed tags with no contents first
if (single) {
std::string selfclosetag = "<" + tagname + atts + "/>";
if (is_inline) {
// always add newline after br tags when they are children of structural tags
if ((tagname == "br") && in_set(structural_tags, parentname)) {
selfclosetag.append("\n");
if (!in_head && (tagname != "html")) selfclosetag.append("\n");
}
return selfclosetag;
}
if (!in_head && (tagname != "html")) selfclosetag.append("\n");
return indent_space + selfclosetag + "\n";
}
// Handle the general case
std::string results;
std::string starttag = "<" + tagname + atts + ">";
std::string closetag = "</" + tagname + ">";
if (is_structural) {
results = indent_space + starttag;
if (!contents.empty()) {
results.append("\n" + contents + "\n" + indent_space);
}
results.append(closetag + "\n");
if (!in_head && (tagname != "html")) results.append("\n");
} else if (is_inline) {
results = starttag;
results.append(contents);
results.append(closetag);
} else /** all others */ {
results = indent_space + starttag;
if (!keep_whitespace) {
ltrim(contents);
}
results.append(contents);
results.append(closetag + "\n");
if (!in_head && (tagname != "html")) results.append("\n");
}
return results;
}
std::string GumboInterface::prettyprint(GumboNode* node, int lvl, const std::string indent_chars)
{
// special case the document node
if (node->type == GUMBO_NODE_DOCUMENT) {
std::string results = build_doctype(node);
results.append(prettyprint_contents(node,lvl+1,indent_chars));
return results;
}
std::string close = "";
std::string closeTag = "";
std::string atts = "";
std::string tagname = get_tag_name(node);
std::string parentname = get_tag_name(node->parent);
bool in_head = (parentname == "head");
// bool need_special_handling = in_set(special_handling, tagname);
bool is_empty_tag = in_set(empty_tags, tagname);
bool no_entity_substitution = in_set(no_entity_sub, tagname);
bool keep_whitespace = in_set(preserve_whitespace, tagname);
bool is_inline = in_set(nonbreaking_inline, tagname) && (parentname != "body");
bool is_structural = in_set(structural_tags, tagname);
bool pp_okay = !is_inline && !keep_whitespace;
char c = indent_chars.at(0);
int n = indent_chars.length();
// build attr string
const GumboVector * attribs = &node->v.element.attributes;
for (unsigned int i=0; i< attribs->length; ++i) {
GumboAttribute* at = static_cast<GumboAttribute*>(attribs->data[i]);
atts.append(build_attributes(at, no_entity_substitution));
}
// determine closing tag type
if (is_empty_tag) {
close = "/";
} else {
closeTag = "</" + tagname + ">";
}
std::string indent_space = std::string((lvl-1)*n,c);
std::string contents;
// prettyprint your contents
if (is_structural && tagname != "html") {
contents = prettyprint_contents(node, lvl+1, indent_chars);
} else {
contents = prettyprint_contents(node, lvl, indent_chars);
}
if (is_structural) {
rtrim(contents);
if (!contents.empty()) contents.append("\n");
}
// remove any leading or trailing whitespace form within paragraphs
if (tagname == "p") {
ltrim(contents);
rtrim(contents);
}
char last_char = ' ';
if (!contents.empty()) {
last_char = contents.at(contents.length()-1);
}
// build results
std::string results;
if (!is_inline && !in_set(nonbreaking_inline, parentname)) {
results.append(indent_space);
}
results.append("<"+tagname+atts+close+">");
if (pp_okay && is_structural && !contents.empty()) {
results.append("\n");
}
results.append(contents);
if (pp_okay && (last_char != '\n') && !contents.empty() && is_structural) {
results.append("\n");
}
// handle any indent before structural close tags
if (!is_inline && is_structural && !closeTag.empty() && !contents.empty()) {
results.append(indent_space);
}
results.append(closeTag);
if ((pp_okay || tagname =="br") && !in_set(nonbreaking_inline, parentname)) {
if (!in_head && tagname != "html") {
results.append("\n\n");
} else {
results.append("\n");
}
}
return results;
}
std::string GumboInterface::serialize(GumboNode* node, enum UpdateTypes doupdates) {
// special case the document node
if (node->type == GUMBO_NODE_DOCUMENT) {
std::string results = build_doctype(node);
results.append(serialize_contents(node, doupdates));
return results;
}
std::string close = "";
std::string closeTag = "";
std::string atts = "";
std::string tagname = get_tag_name(node);
bool need_special_handling = in_set(special_handling, tagname);
bool is_void_tag = in_set(void_tags, tagname);
bool no_entity_substitution = in_set(no_entity_sub, tagname);
// bool is_inline = in_set(nonbreaking_inline, tagname);
bool is_href_src_tag = in_set(href_src_tags, tagname);
// build attr string
const GumboVector * attribs = &node->v.element.attributes;
for (unsigned int i=0; i< attribs->length; ++i) {
GumboAttribute* at = static_cast<GumboAttribute*>(attribs->data[i]);
atts.append(build_attributes(at, no_entity_substitution, ((doupdates & SourceUpdates) && is_href_src_tag), (doupdates & StyleUpdates)));
}
// Make sure that the xmlns attribute exists as an html tag attribute
if (tagname == "html") {
if (atts.find("xmlns=") == std::string::npos) {
atts.append(" xmlns=\"http://www.w3.org/1999/xhtml\"");
}
}
// determine closing tag type
if (is_void_tag) {
close = "/";
} else {
closeTag = "</" + tagname + ">";
}
std::string contents;
if ((tagname == "body") && (doupdates & BodyUpdates)) {
contents = m_newbody;
} else {
// serialize your contents
contents = serialize_contents(node, doupdates);
}
if ((doupdates & StyleUpdates) && (tagname == "style") &&
(node->parent->type == GUMBO_NODE_ELEMENT) &&
(node->parent->v.element.tag == GUMBO_TAG_HEAD)) {
contents = update_style_urls(contents);
}
if (need_special_handling) {
ltrimnewlines(contents);
rtrim(contents);
contents.append("\n");
}
// build results
std::string results;
if ((doupdates & LinkUpdates) && (tagname == "link") &&
(node->parent->type == GUMBO_NODE_ELEMENT) &&
(node->parent->v.element.tag == GUMBO_TAG_HEAD)) {
return "";
}
results.append("<"+tagname+atts+close+">");
if (need_special_handling) results.append("\n");
results.append(contents);
if ((doupdates & LinkUpdates) && (tagname == "head")) {
results.append(m_newcsslinks);
}
results.append(closeTag);
if (need_special_handling) results.append("\n");
return results;
}
std::string GumboInterface::serialize(GumboNode* node, enum UpdateTypes doupdates) {
// special case the document node
if (node->type == GUMBO_NODE_DOCUMENT) {
std::string results = build_doctype(node);
results.append(serialize_contents(node, doupdates));
return results;
}
std::string close = "";
std::string closeTag = "";
std::string atts = "";
std::string tagname = get_tag_name(node);
std::string key = "|" + tagname + "|";
bool need_special_handling = special_handling.find(key) != std::string::npos;
bool is_empty_tag = empty_tags.find(key) != std::string::npos;
bool no_entity_substitution = no_entity_sub.find(key) != std::string::npos;
bool is_inline = nonbreaking_inline.find(key) != std::string::npos;
bool is_href_src_tag = href_src_tags.find(key) != std::string::npos;
// build attr string
const GumboVector * attribs = &node->v.element.attributes;
for (int i=0; i< attribs->length; ++i) {
GumboAttribute* at = static_cast<GumboAttribute*>(attribs->data[i]);
atts.append(build_attributes(at, no_entity_substitution, ((doupdates & SourceUpdates) && is_href_src_tag) ));
}
// determine closing tag type
if (is_empty_tag) {
close = "/";
} else {
closeTag = "</" + tagname + ">";
}
std::string contents;
if ((tagname == "body") && (doupdates & BodyUpdates)) {
contents = m_newbody;
} else {
// serialize your contents
contents = serialize_contents(node, doupdates);
}
if (need_special_handling) {
ltrimnewlines(contents);
rtrim(contents);
contents.append("\n");
}
// build results
std::string results;
if ((doupdates & LinkUpdates) && (tagname == "link") &&
(node->parent->type == GUMBO_NODE_ELEMENT) &&
(node->parent->v.element.tag == GUMBO_TAG_HEAD)) {
return "";
}
results.append("<"+tagname+atts+close+">");
if (need_special_handling) results.append("\n");
results.append(contents);
if ((doupdates & LinkUpdates) && (tagname == "head")) {
results.append(m_newcsslinks);
}
results.append(closeTag);
if (need_special_handling) results.append("\n");
return results;
}
