test (ed9e9569) · Commits · jan.koester / blogi

editor/src/htmlimport.cpp

+79 −2

Original line number	Diff line number	Diff line
		@@ -898,8 +898,7 @@ std::string blogi::htmlimport::extractCustomCss(

		std::string blogi::htmlimport::serializeElement(const libhtmlpp::Element *el) {
		if (!el) return "";
		if (el->getType() == libhtmlpp::HtmlEl \|\| el->getType() == libhtmlpp::SvgEL \|\|
		el->getType() == libhtmlpp::TextAreaEL) {
		if (el->getType() == libhtmlpp::HtmlEl \|\| el->getType() == libhtmlpp::TextAreaEL) {
		const libhtmlpp::HtmlElement hel = static_cast<const libhtmlpp::HtmlElement>(el);
		std::string tag = hel->getTagname();
		std::string result = "<" + tag;
		@@ -914,6 +913,33 @@ std::string blogi::htmlimport::serializeElement(const libhtmlpp::Element *el) {
		result += serializeInnerHtml(const_cast<libhtmlpp::HtmlElement*>(hel));
		result += "</" + tag + ">";
		return result;
		} else if (el->getType() == libhtmlpp::SvgEL) {
		// Unlike a plain HtmlElement, SvgElement stores its inner markup as
		// a raw blob (getSvg()) rather than a real child tree -- its own
		// insertChild/appendChild are deleted outright, and firstChild()
		// always returns null. Reusing the HtmlEl branch above (which walks
		// firstChild() via serializeInnerHtml) silently produced an empty
		// "<svg ...></svg>" shell for every icon on this class of
		// page-builder site (icons wrap inline <svg><path>/<polygon>...
		// markup, e.g. the header's location/email/phone icons) -- the
		// outer tag/attributes came through fine, only the actual icon
		// artwork vanished. Mirrors the ScriptEL branch below, which
		// already gets this right for the same reason (getScript()).
		const libhtmlpp::SvgElement sel = static_cast<const libhtmlpp::SvgElement>(el);
		std::string tag = sel->getTagname();
		std::string result = "<" + tag;
		for (const auto *attr = sel->firstAttribute(); attr; attr = attr->nextAttribute()) {
		result += " " + attr->getKey();
		std::string val = attr->getValue();
		if (!val.empty()) {
		result += "=\"" + val + "\"";
		}
		}
		result += ">";
		std::vector<char> svg = const_cast<libhtmlpp::SvgElement*>(sel)->getSvg();
		result.append(svg.begin(), svg.end());
		result += "</" + tag + ">";
		return result;
		} else if (el->getType() == libhtmlpp::TextEl) {
		return const_cast<libhtmlpp::TextElement*>(
		static_cast<const libhtmlpp::TextElement*>(el))->getText();
		@@ -1804,6 +1830,57 @@ void blogi::htmlimport::htmlElementToWidgetXml(
		if (child) childFrames.push_back({child, w, nullptr, elementEnv, childAncestorChain});
		}

		} else if (tag == "svg") {
		// No dedicated widget type for a raw <svg> icon -- it becomes
		// CustomHtml below, same as the generic fallback, EXCEPT: a
		// page-builder icon like this commonly has no width/height
		// attribute of its own (just a viewBox), sized purely by an
		// external class-based CSS rule such as
		// ".con-kit-component-icon svg{width:...;height:...}". That
		// rule can never survive import: htmlimport's own "global
		// CSS" extraction only ever keeps :root/html/body/*-scoped
		// rules, and even per-element CSS matching deliberately
		// rejects a bare-tag target after a combinator like this
		// (see compoundMatches's own doc comment -- a real, previously
		// fixed hazard, not an oversight). Confirmed on a real page:
		// every header icon (location/email/phone) rendered as an
		// empty circle, collapsed to 0x0, with its <path>/<polygon>
		// artwork otherwise intact (see serializeElement's SvgEL
		// branch). Falling back to the svg's own viewBox dimensions
		// when neither a width/height attribute nor a matched
		// "width"/"height" cssProps entry exists at least makes the
		// icon visible -- not pixel-identical to the source page,
		// but far closer than invisible.
		std::string svgHtml = serializeElement(cur);
		if (elemAttrs.count("width") == 0 && elemAttrs.count("height") == 0 &&
		!cssProps.count("width") && !cssProps.count("height")) {
		auto vbIt = elemAttrs.find("viewBox");
		if (vbIt != elemAttrs.end()) {
		std::istringstream vb(vbIt->second);
		double minX, minY, vbWidth, vbHeight;
		if ((vb >> minX >> minY >> vbWidth >> vbHeight) && vbWidth > 0 && vbHeight > 0) {
		auto fmt = [](double v) {
		long long i = static_cast<long long>(v);
		return (v == static_cast<double>(i)) ? std::to_string(i) : std::to_string(v);
		};
		std::string sizeAttrs = " width=\"" + fmt(vbWidth) +
		"\" height=\"" + fmt(vbHeight) + "\"";
		size_t tagEnd = svgHtml.find("<svg");
		if (tagEnd != std::string::npos) svgHtml.insert(tagEnd + 4, sizeAttrs);
		}
		}
		}
		tinyxml2::XMLElement *w = doc.NewElement("CustomHtml");
		tinyxml2::XMLElement *htmlContent = doc.NewElement("HtmlContent");
		htmlContent->SetText(svgHtml.c_str());
		w->InsertEndChild(htmlContent);
		if (!style.empty()) {
		tinyxml2::XMLElement *cssEl = doc.NewElement("CssContent");
		cssEl->SetText(style.c_str());
		w->InsertEndChild(cssEl);
		}
		pXml->InsertEndChild(w);

		} else if (tag == "h1" \|\| tag == "h2" \|\| tag == "h3" \|\| tag == "h4" \|\|
		tag == "h5" \|\| tag == "h6" \|\| tag == "p" \|\| tag == "span" \|\|
		tag == "b" \|\| tag == "i" \|\| tag == "em" \|\| tag == "strong" \|\|

tests/htmlimport_smoke.cpp

+39 −0

Original line number	Diff line number	Diff line
		@@ -509,6 +509,45 @@ int main(){
		"an attribute not referenced by any media rule is not preserved");
		}

		// --- serializeElement: <svg> icon content survives the CustomHtml
		// fallback ---
		// SvgElement stores its inner markup as a raw blob (getSvg()), not a
		// real child tree (its own insertChild/appendChild are deleted, and
		// firstChild() always returns null) -- serializeElement's generic
		// HtmlEl branch walked firstChild() via serializeInnerHtml and silently
		// produced an empty "<svg ...></svg>" shell, dropping every icon's
		// actual <path>/<polygon> artwork (real-world symptom: header
		// location/email/phone icons rendered as bare circles with nothing
		// inside).
		std::cout << "=== <svg> icon content preserved through CustomHtml fallback ===" << std::endl;
		{
		std::string html =
		"<div class=\"icon\"><svg viewBox=\"0 0 32 32\">"
		"<path d=\"M1,2\"></path><polygon points=\"1,2 3,4\"></polygon>"
		"</svg></div>";
		std::string xml = importXml(html);
		check(xml.find("path") != std::string::npos,
		"the <path> child survives import (was silently dropped -- empty <svg></svg> shell)");
		check(xml.find("polygon") != std::string::npos,
		"the <polygon> child survives import");
		check(xml.find("width=\"32\" height=\"32\"") != std::string::npos,
		"an <svg> with no width/height attribute of its own falls back to its viewBox "
		"dimensions (real-world symptom: icons sized purely by an external CSS rule that "
		"can't survive import collapsed to an invisible 0x0 box)");
		}
		{
		// An <svg> that already specifies its own width/height is left
		// alone -- the viewBox fallback only fires when neither is present.
		std::string html =
		"<div class=\"icon\"><svg viewBox=\"0 0 32 32\" width=\"18\" height=\"18\">"
		"<path d=\"M1,2\"></path></svg></div>";
		std::string xml = importXml(html);
		check(xml.find("width=\"18\" height=\"18\"") != std::string::npos,
		"an <svg> with its own explicit width/height keeps it, unmodified");
		check(xml.find("width=\"32\"") == std::string::npos,
		"the viewBox fallback does not also add its own width/height alongside the real one");
		}

		// --- Summary ---
		std::cout << "\n=== " << passCount << "/" << testCount << " tests passed ===" << std::endl;