standardize

func StandardizeContent ¶

func StandardizeContent(element *goquery.Selection, metadata *metadata.Metadata, doc *goquery.Document, debug bool)

StandardizeContent standardizes and cleans up the main content element JavaScript original code:

export function standardizeContent(element: Element, metadata: DefuddleMetadata, doc: Document, debug: boolean = false): void {
	standardizeSpaces(element);

	// Remove HTML comments
	removeHtmlComments(element);

	// Handle H1 elements - remove first one and convert others to H2
	standardizeHeadings(element, metadata.title, doc);

	// Standardize footnotes and citations
	standardizeFootnotes(element);

	// Convert embedded content to standard formats
	standardizeElements(element, doc);

	// If not debug mode, do the full cleanup
	if (!debug) {
		// First pass of div flattening
		flattenWrapperElements(element, doc);

		// Strip unwanted attributes
		stripUnwantedAttributes(element, debug);

		// Remove empty elements
		removeEmptyElements(element);

		// Remove trailing headings
		removeTrailingHeadings(element);

		// Final pass of div flattening after cleanup operations
		flattenWrapperElements(element, doc);

		// Standardize consecutive br elements
		stripExtraBrElements(element);

		// Clean up empty lines
		removeEmptyLines(element, doc);
	} else {
		// In debug mode, still do basic cleanup but preserve structure
		stripUnwantedAttributes(element, debug);
		removeTrailingHeadings(element);
		stripExtraBrElements(element);
		logDebug('Debug mode: Skipping div flattening to preserve structure');
	}
}