User:Wengier/wordcount.js

function getWordCount(html) {
	var str = html.innerHTML.replace(/(<([^>]+)>)/ig,"").trim();

	var wordCount = 0;

	var arr = str.match(/[\u3040-\u309F]|[\u30A0-\u30FF]|[\u4E00-\u9FFF\uF900-\uFAFF\u3400-\u4DBF]|\S+/g);

	if (arr) {
		wordCount = arr.length;
	}

	return wordCount;
}

function getContentDiv() {
	var contentDiv;

	if (mw.config.get('wgAction') == 'edit') {
		contentDiv = document.getElementById('wikiPreview');
	}
	else {
		contentDiv = document.getElementById('mw-content-text');
	}

	return contentDiv;
}

function isValidListNode(node) {
	if (node.parentElement.id == "word-count-stats") {
		return false;
	}

	if (node.className.indexOf("toclevel-") > -1 || 
		node.parentElement.parentElement.id == "toc") {

		return false;
	}

	if (node.parentElement.parentElement.parentElement.className == "catlinks") {
		return false;
	}

	var bodyContent = getContentDiv();
	var curNode = node.parentElement.parentElement;

	while (curNode && (curNode != bodyContent)) {
		if (curNode.className.indexOf("infobox") > -1) {
			return false;
		}
		else if (curNode.className.indexOf("metadata") > -1) {
			return false;
		}
		else if (curNode.className.indexOf("navbox") > -1) {
			return false;
		}
		else {
			curNode = curNode.parentElement;
		}
	}

	return true;
}

function isValidParagraphNode(node) {
	if (node.parentNode.className.indexOf("mw-parser-output") > -1 ||
		node.parentNode.parentNode.className.indexOf("mw-parser-output") > -1 ||
		node.parentNode.nodeName == "BLOCKQUOTE") {

		return true;
	}
	else {
		return false;
	}
}

function isValidReferenceNode(node) {
	var bodyContent = getContentDiv();
	var curNode = node.parentElement;

	while (curNode && (curNode != bodyContent)) {
		if (curNode.classList.contains("references") ||
			curNode.classList.contains("reflist") ||
			curNode.classList.contains("refbegin")) {

			return true;
		}

		curNode = curNode.parentElement;
	}

	return false;
}

function toggleWordCount() {
	if (mw.config.get('wgAction') == 'edit') {
		var wikiPreview = document.getElementById('wikiPreview');

		var wikiPreviewStyle = window.getComputedStyle(wikiPreview);

		if (wikiPreviewStyle.display === 'none') {
			alert("You need to preview the text for the word count script to work in edit mode.");

			return;
		}
	}

	var bodyContent = getContentDiv();

	var output = document.getElementById("word-count-stats");

	if (output) {
		var oldStyle = output.className;

		var i = 0;

		// Cleanup background color
		var pList = bodyContent.getElementsByTagName("p");

		if (pList) {
			for (i=0; i < pList.length; i++){
				if (isValidParagraphNode(pList[i])) {
					pList[i].style.cssText = oldStyle;
				}
			}
		}

		var listTypes = ["li", "dd"];
	
		for (var j = 0; j < listTypes.length; j++) {
			var liList = bodyContent.getElementsByTagName(listTypes[j]);

			if (liList) {
				for (i=0; i < liList.length; i++) {
					liList[i].style.cssText = oldStyle;
				}
			}
		}

		var hList = bodyContent.getElementsByClassName("mw-heading");

		if (hList) {
			for (i=0; i < hList.length; i++) {
				hList[i].style.cssText = oldStyle;
			}
		}

		// Remove nodes
		output.parentNode.removeChild(output);

		var header = document.getElementById("word-count-header");

		header.parentNode.removeChild(header);

		var footer = document.getElementById("word-count-footer");

		footer.parentNode.removeChild(footer);
	}
	else {
		getStatistics(bodyContent);
	}
}

//
// Main counting function
//
function getStatistics(bodyContent) {

	// Statistics
	var output = document.createElement("ul");
	output.id = "word-count-stats";

	var lead_value = document.createElement("li");
	lead_value.id = "lead-stat";
	output.appendChild(lead_value);
	output.className = bodyContent.getElementsByTagName("p").item(0).style.cssText;

	var main_body_value = document.createElement("li");
	main_body_value.id = "main-body-stat";
	output.appendChild(main_body_value);

	var subtotal_value = document.createElement("li");
	subtotal_value.id = "subtotal-stat";
	output.appendChild(subtotal_value);

	var ref_value = document.createElement("li");
	ref_value.id = "ref-stat";
	output.appendChild(ref_value);

	bodyContent.insertBefore(output, bodyContent.firstChild);

	// Header
	var header = document.createElement("span");
	header.id = "word-count-header";
	header.innerHTML = "<b>条目字数统计:</b>";
	bodyContent.insertBefore(header,output);

	var footer = document.createElement("span");
	footer.id = "word-count-footer";
	footer.innerHTML = "<b>总字数:</b><p/>";
	bodyContent.insertBefore(footer,output.nextSibling);

	// Create counters
	var main_body_count = 0;
	var lead_count = 0;
	var ref_count = 0;

	var i = 0, sec = 0, sectop = 0;
	
	// Count within headings
	var hList = bodyContent.getElementsByClassName("mw-heading");

	if (hList && hList.length) {
		const rect = hList[0].getBoundingClientRect();
		sec = 1;
		sectop = rect.top;
	}

	// Count within paragraphs
	var pList = bodyContent.getElementsByTagName("p");

	if (pList) {
		for (i=0; i < pList.length; i++) {
			var para = pList[i];

			if (isValidParagraphNode(para)) {
				var paraCount = getWordCount(para);
				const rect = para.getBoundingClientRect();
				

				if (paraCount > 0) {
					const rect = para.getBoundingClientRect();
					if (sec && rect.top < sectop) {
						lead_count += paraCount;
						para.style.cssText = "background-color:lightgreen";
					} else {
						main_body_count += paraCount;
						para.style.cssText = "background-color:yellow";
					}
				}
			}
		}
	}

	// Count within lists
	var listTypes = ["li", "dd"];

	for (var j = 0; j < listTypes.length; j++) {
		var liList = bodyContent.getElementsByTagName(listTypes[j]);

		if (liList) {
			for (i=0; i < liList.length; i++) {
				var li = liList[i];

				if (isValidReferenceNode(li)) {
					ref_count += getWordCount(li);
					li.style.cssText = "background-color:cyan";
				}
				else if (isValidListNode(li)) {
					main_body_count += getWordCount(li);
					li.style.cssText = "background-color:yellow";
				}
			}
		}
	}

	if (hList) {
		for (i=0; i < hList.length; i++) {
			var h = hList[i];

			if (h.id == "Contents") {
				continue;
			}

			main_body_count += getWordCount(h);
			h.style.cssText = "background-color:yellow";
		}
	}

	lead_value.innerHTML = "序言章节: " + lead_count + "个字";
	main_body_value.innerHTML = "正文内容: " + main_body_count + "个字";
	subtotal_value.innerHTML = "字数总计: " + (main_body_count + lead_count) + "个字";
	ref_value.innerHTML = "参考文献: " + ref_count + "个字";
	footer.innerHTML = "<b>条目总字数: " + (main_body_count + lead_count + ref_count) + "</b><p/>";
}

jQuery(function () {
	mw.loader.using( ['mediawiki.util'], function () {
		if($.inArray(mw.config.get('wgAction'), ['edit', 'view' , 'submit' , 'historysubmit' , 'purge']) !== -1) {
			$( mw.util.addPortletLink('p-tb', '#', '字数统计', 't-word-count', '计算条目字数') )
			.click( toggleWordCount );
		}
	});
});

//</nowiki>