WEB高性能开发之疯狂的HTML压缩

一般我们启动gzip都比较少对html启动gzip，因为现在的html都是动态的，不会使用浏览器缓存，而启用gzip的话每次请求都需要压缩，会比较消耗服务器资源，对js,css启动gzip比较好是因为js,css都会使用缓存。我个人觉得的压缩html的最大好处就是一本万利，只要写好了一次，以后所有程序都可以使用，不会增加任何额外的开发工作。
在“JS、CSS的合并、压缩、缓存管理”一文中说到自己写过的1个自动合并、压缩JS,CSS,并添加版本号的组件。这次把压缩html的功能也加入到该组件中，流程很简单，就是在程序启动(contextInitialized or Application_Start)的时候扫描所有html,jsp(aspx)进行压缩。
压缩的注意事项：
实现的方式主要是用正则表达式去查找，替换。在html压缩的时候，主要要注意下面几点：
1. pre,textarea 标签里面的内容格式需要保留，不能压缩。
2. 去掉html注释的时候，有些注释是不能去掉的，比如：<!--[if IE 6]> ..... <![endif]-->
3. 压缩嵌入式js中的注释要注意，因为可能注释符号会出现在字符串中，比如： var url = "http://www.cnblogs.com"; // 前面的//不是注释
去掉JS换行符的时候，不能直接跟一下行动内容，需要有空格，考虑下面的代码：
else
return;
如果不带空格，则变成elsereturn。
4. jsp(aspx) 中很有可能会使用<% %>嵌入一些服务器代码，这个时候也需要单独处理，里面注释的处理方法跟js的一样。
源代码：
下面是java实现的源代码，也可以 猛击此处 下载该代码，相信大家都看的懂，也很容易改成net代码：
复制代码 代码如下:
import java.io.StringReader;
import java.io.StringWriter;
import java.util.*;
import java.util.regex.*;
/*******************************************
* 压缩jsp,html中的代码，去掉所有空白符、换行符
* @author bearrui(ak-47)
* @version 0.1
* @date 2010-5-13
*******************************************/
public class HtmlCompressor {
private static String tempPreBlock = "%%%HTMLCOMPRESS~PRE&&&";
private static String tempTextAreaBlock = "%%%HTMLCOMPRESS~TEXTAREA&&&";
private static String tempScriptBlock = "%%%HTMLCOMPRESS~SCRIPT&&&";
private static String tempStyleBlock = "%%%HTMLCOMPRESS~STYLE&&&";
private static String tempJspBlock = "%%%HTMLCOMPRESS~JSP&&&";
private static Pattern commentPattern = Pattern.compile("<!--\\s*[^\\[].*?-->", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
private static Pattern itsPattern = Pattern.compile(">\\s+?<", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
private static Pattern prePattern = Pattern.compile("<pre[^>]*?>.*?</pre>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
private static Pattern taPattern = Pattern.compile("<textarea[^>]*?>.*?</textarea>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
private static Pattern jspPattern = Pattern.compile("<%([^-@][\\w\\W]*?)%>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
// <script></script>
private static Pattern scriptPattern = Pattern.compile("(?:<script\\s*>|<script type=['\"]text/javascript['\"]\\s*>)(.*?)</script>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
private static Pattern stylePattern = Pattern.compile("<style[^>()]*?>(.+)</style>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
// 单行注释，
private static Pattern signleCommentPattern = Pattern.compile("//.*");
// 字符串匹配
private static Pattern stringPattern = Pattern.compile("(\"[^\"\\n]*?\"|'[^'\\n]*?')");
// trim去空格和换行符
private static Pattern trimPattern = Pattern.compile("\\n\\s*",Pattern.MULTILINE);
private static Pattern trimPattern2 = Pattern.compile("\\s*\\r",Pattern.MULTILINE);
// 多行注释
private static Pattern multiCommentPattern = Pattern.compile("/\\*.*?\\*/", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
private static String tempSingleCommentBlock = "%%%HTMLCOMPRESS~SINGLECOMMENT&&&"; // //占位符
private static String tempMulitCommentBlock1 = "%%%HTMLCOMPRESS~MULITCOMMENT1&&&"; // /*占位符
private static String tempMulitCommentBlock2 = "%%%HTMLCOMPRESS~MULITCOMMENT2&&&"; // */占位符

public static String compress(String html) throws Exception {
if(html == null || html.length() == 0) {
return html;
}
List<String> preBlocks = new ArrayList<String>();
List<String> taBlocks = new ArrayList<String>();
List<String> scriptBlocks = new ArrayList<String>();
List<String> styleBlocks = new ArrayList<String>();
List<String> jspBlocks = new ArrayList<String>();
String result = html;
//preserve inline java code
Matcher jspMatcher = jspPattern.matcher(result);
while(jspMatcher.find()) {
jspBlocks.add(jspMatcher.group(0));
}
result = jspMatcher.replaceAll(tempJspBlock);
//preserve PRE tags
Matcher preMatcher = prePattern.matcher(result);
while(preMatcher.find()) {
preBlocks.add(preMatcher.group(0));
}
result = preMatcher.replaceAll(tempPreBlock);
//preserve TEXTAREA tags
Matcher taMatcher = taPattern.matcher(result);
while(taMatcher.find()) {
taBlocks.add(taMatcher.group(0));
}
result = taMatcher.replaceAll(tempTextAreaBlock);
//preserve SCRIPT tags
Matcher scriptMatcher = scriptPattern.matcher(result);
while(scriptMatcher.find()) {
scriptBlocks.add(scriptMatcher.group(0));
}
result = scriptMatcher.replaceAll(tempScriptBlock);
// don't process inline css
Matcher styleMatcher = stylePattern.matcher(result);
while(styleMatcher.find()) {
styleBlocks.add(styleMatcher.group(0));
}
result = styleMatcher.replaceAll(tempStyleBlock);
//process pure html
result = processHtml(result);
//process preserved blocks
result = processPreBlocks(result, preBlocks);
result = processTextareaBlocks(result, taBlocks);
result = processScriptBlocks(result, scriptBlocks);
result = processStyleBlocks(result, styleBlocks);
result = processJspBlocks(result, jspBlocks);
preBlocks = taBlocks = scriptBlocks = styleBlocks = jspBlocks = null;
return result.trim();
}
private static String processHtml(String html) {
String result = html;
//remove comments
// if(removeComments) {
result = commentPattern.matcher(result).replaceAll("");
// }
//remove inter-tag spaces
// if(removeIntertagSpaces) {
result = itsPattern.matcher(result).replaceAll("><");
// }
//remove multi whitespace characters
// if(removeMultiSpaces) {
result = result.replaceAll("\\s{2,}"," ");
// }
return result;
}
private static String processJspBlocks(String html, List<String> blocks){
String result = html;
for(int i = 0; i < blocks.size(); i++) {
blocks.set(i, compressJsp(blocks.get(i)));
}
//put preserved blocks back
while(result.contains(tempJspBlock)) {
result = result.replaceFirst(tempJspBlock, Matcher.quoteReplacement(blocks.remove(0)));
}
return result;
}
private static String processPreBlocks(String html, List<String> blocks) throws Exception {
String result = html;
//put preserved blocks back
while(result.contains(tempPreBlock)) {
result = result.replaceFirst(tempPreBlock, Matcher.quoteReplacement(blocks.remove(0)));
}
return result;
}
private static String processTextareaBlocks(String html, List<String> blocks) throws Exception {
String result = html;
//put preserved blocks back
while(result.contains(tempTextAreaBlock)) {
result = result.replaceFirst(tempTextAreaBlock, Matcher.quoteReplacement(blocks.remove(0)));
}
return result;
}
private static String processScriptBlocks(String html, List<String> blocks) throws Exception {
String result = html;
// if(compressJavaScript) {
for(int i = 0; i < blocks.size(); i++) {
blocks.set(i, compressJavaScript(blocks.get(i)));
}
// }
//put preserved blocks back
while(result.contains(tempScriptBlock)) {
result = result.replaceFirst(tempScriptBlock, Matcher.quoteReplacement(blocks.remove(0)));
}
return result;
}
private static String processStyleBlocks(String html, List<String> blocks) throws Exception {
String result = html;
// if(compressCss) {
for(int i = 0; i < blocks.size(); i++) {
blocks.set(i, compressCssStyles(blocks.get(i)));
}
// }
//put preserved blocks back
while(result.contains(tempStyleBlock)) {
result = result.replaceFirst(tempStyleBlock, Matcher.quoteReplacement(blocks.remove(0)));
}
return result;
}
private static String compressJsp(String source) {
//check if block is not empty
Matcher jspMatcher = jspPattern.matcher(source);
if(jspMatcher.find()) {
String result = compressJspJs(jspMatcher.group(1));
return (new StringBuilder(source.substring(0, jspMatcher.start(1))).append(result).append(source.substring(jspMatcher.end(1)))).toString();
} else {
return source;
}
}
private static String compressJavaScript(String source) {
//check if block is not empty
Matcher scriptMatcher = scriptPattern.matcher(source);
if(scriptMatcher.find()) {
String result = compressJspJs(scriptMatcher.group(1));
return (new StringBuilder(source.substring(0, scriptMatcher.start(1))).append(result).append(source.substring(scriptMatcher.end(1)))).toString();
} else {
return source;
}
}
private static String compressCssStyles(String source) {
//check if block is not empty
Matcher styleMatcher = stylePattern.matcher(source);
if(styleMatcher.find()) {
// 去掉注释，换行
String result= multiCommentPattern.matcher(styleMatcher.group(1)).replaceAll("");
result = trimPattern.matcher(result).replaceAll("");
result = trimPattern2.matcher(result).replaceAll("");
return (new StringBuilder(source.substring(0, styleMatcher.start(1))).append(result).append(source.substring(styleMatcher.end(1)))).toString();
} else {
return source;
}
}
private static String compressJspJs(String source){
String result = source;
// 因注释符合有可能出现在字符串中，所以要先把字符串中的特殊符好去掉
Matcher stringMatcher = stringPattern.matcher(result);
while(stringMatcher.find()){
String tmpStr = stringMatcher.group(0);
if(tmpStr.indexOf("//") != -1 || tmpStr.indexOf("/*") != -1 || tmpStr.indexOf("*/") != -1){
String blockStr = tmpStr.replaceAll("//", tempSingleCommentBlock).replaceAll("/\\*", tempMulitCommentBlock1)
.replaceAll("\\*/", tempMulitCommentBlock2);
result = result.replace(tmpStr, blockStr);
}
}
// 去掉注释
result = signleCommentPattern.matcher(result).replaceAll("");
result = multiCommentPattern.matcher(result).replaceAll("");
result = trimPattern2.matcher(result).replaceAll("");
result = trimPattern.matcher(result).replaceAll(" ");
// 恢复替换掉的字符串
result = result.replaceAll(tempSingleCommentBlock, "//").replaceAll(tempMulitCommentBlock1, "/*")
.replaceAll(tempMulitCommentBlock2, "*/");
return result;
}
}

使用注意事项：

使用了上面方法后，再运行程序，是不是发现每个页面查看源代码的时候都变成1行啦，还不错吧，但是在使用的时候还是要注意一些问题：
1. 嵌入js本来想调用yuicompressor来压缩，yuicompressor压缩JS前，会先编译js是否合法，因我们嵌入的js中可能很多会用到一些服务器端代码，比如 var now = <%=DateTime.now %> ，这样的代码会编译不通过，所以无法使用yuicompressor。
最后只能自己写压缩JS代码，自己写的比较粗燥，所以有个问题还解决，就是如果开发人员在一句js代码后面没有加分号的话，压缩成1行就很有可能出问题。所以使用这个需要保证每条语句结束后都必须带分号。

2. 因为是在程序启动的时候压缩所有jsp(aspx)，所以如果是用户请求的时候动态产生的html就无法压缩。