WEB高性能开发：HTML压缩

上一篇随笔中网友 skyaspnet 问我如何压缩HTML，当时回答是推荐他使用gzip，后来想想，要是能把所有的html，jsp(aspx)在运行前都压缩成1行未免不是一件好事啊。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：
      1 import java.io.StringReader;

2 import java.io.StringWriter;
  3 import java.util.*;
  4 import java.util.regex.*;
  5
  6 /*******************************************
  7 * 压缩jsp,html中的代码，去掉所有空白符、换行符
  8 * @author  bearrui(ak-47)
  9 * @version 0.1
10 * @date     2010-5-13
11 *******************************************/
12 public class HtmlCompressor {
13     private static String tempPreBlock = "%%%HTMLCOMPRESS~PRE&&&";
14     private static String tempTextAreaBlock = "%%%HTMLCOMPRESS~TEXTAREA&&&";
15     private static String tempScriptBlock = "%%%HTMLCOMPRESS~SCRIPT&&&";
16     private static String tempStyleBlock = "%%%HTMLCOMPRESS~STYLE&&&";
17     private static String tempJspBlock = "%%%HTMLCOMPRESS~JSP&&&";
18    
19     private static Pattern commentPattern = Pattern.compile("<!--\\s*[^\\[].*?-->", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
20     private static Pattern itsPattern = Pattern.compile(">\\s+?<", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
21     private static Pattern prePattern = Pattern.compile("<pre[^>]*?>.*?</pre>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
22     private static Pattern taPattern = Pattern.compile("<textarea[^>]*?>.*?</textarea>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
23     private static Pattern jspPattern = Pattern.compile("<%([^-@][\\w\\W]*?)%>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
24     // <script></script>
25     private static Pattern scriptPattern = Pattern.compile("(?:<script\\s*>|<script type=['\"]text/javascript['\"]\\s*>)(.*?)</script>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
26     private static Pattern stylePattern = Pattern.compile("<style[^>()]*?>(.+)</style>", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
27
28     // 单行注释，
29     private static Pattern signleCommentPattern = Pattern.compile("//.*");
30     // 字符串匹配
31     private static Pattern stringPattern = Pattern.compile("(\"[^\"\\n]*?\"|'[^'\\n]*?')");
32     // trim去空格和换行符
33     private static Pattern trimPattern = Pattern.compile("\\n\\s*",Pattern.MULTILINE);
34     private static Pattern trimPattern2 = Pattern.compile("\\s*\\r",Pattern.MULTILINE);
35     // 多行注释
36     private static Pattern multiCommentPattern = Pattern.compile("/\\*.*?\\*/", Pattern.DOTALL | Pattern.CASE_INSENSITIVE | Pattern.MULTILINE);
37
38     private static String tempSingleCommentBlock = "%%%HTMLCOMPRESS~SINGLECOMMENT&&&";  // //占位符
39     private static String tempMulitCommentBlock1 = "%%%HTMLCOMPRESS~MULITCOMMENT1&&&";  // /*占位符
40     private static String tempMulitCommentBlock2 = "%%%HTMLCOMPRESS~MULITCOMMENT2&&&";  // */占位符
41    
42    
43     public static String compress(String html) throws Exception {
44         if(html == null || html.length() == 0) {
45             return html;
46         }
47        
48         List<String> preBlocks = new ArrayList<String>();
49         List<String> taBlocks = new ArrayList<String>();
50         List<String> scriptBlocks = new ArrayList<String>();
51         List<String> styleBlocks = new ArrayList<String>();
52         List<String> jspBlocks = new ArrayList<String>();
53        
54         String result = html;
55        
56         //preserve inline java code
57         Matcher jspMatcher = jspPattern.matcher(result);
58         while(jspMatcher.find()) {
59             jspBlocks.add(jspMatcher.group(0));
60         }
61         result = jspMatcher.replaceAll(tempJspBlock);
62        
63         //preserve PRE tags
64         Matcher preMatcher = prePattern.matcher(result);
65         while(preMatcher.find()) {
66             preBlocks.add(preMatcher.group(0));
67         }
68         result = preMatcher.replaceAll(tempPreBlock);
69        
70         //preserve TEXTAREA tags
71         Matcher taMatcher = taPattern.matcher(result);
72         while(taMatcher.find()) {
73             taBlocks.add(taMatcher.group(0));
74         }
75         result = taMatcher.replaceAll(tempTextAreaBlock);
76        
77         //preserve SCRIPT tags
78         Matcher scriptMatcher = scriptPattern.matcher(result);
79         while(scriptMatcher.find()) {
80             scriptBlocks.add(scriptMatcher.group(0));
81         }
82         result = scriptMatcher.replaceAll(tempScriptBlock);
83        
84         // don't process inline css
85         Matcher styleMatcher = stylePattern.matcher(result);
86         while(styleMatcher.find()) {
87             styleBlocks.add(styleMatcher.group(0));
88         }
89         result = styleMatcher.replaceAll(tempStyleBlock);
90        
91         //process pure html
92         result = processHtml(result);
93        
94         //process preserved blocks
95         result = processPreBlocks(result, preBlocks);
96         result = processTextareaBlocks(result, taBlocks);
97         result = processScriptBlocks(result, scriptBlocks);
98         result = processStyleBlocks(result, styleBlocks);
99         result = processJspBlocks(result, jspBlocks);
100        
101         preBlocks = taBlocks = scriptBlocks = styleBlocks = jspBlocks = null;
102        
103         return result.trim();
104     }
105    
106     private static String processHtml(String html) {
107         String result = html;
108        
109         //remove comments
110 //        if(removeComments) {
111             result = commentPattern.matcher(result).replaceAll("");
112 //        }
113        
114         //remove inter-tag spaces
115 //        if(removeIntertagSpaces) {
116             result = itsPattern.matcher(result).replaceAll("><");
117 //        }
118        
119         //remove multi whitespace characters
120 //        if(removeMultiSpaces) {
121             result = result.replaceAll("\\s{2,}"," ");
122 //        }
123                
124         return result;
125     }
126    
127     private static String processJspBlocks(String html, List<String> blocks){
128         String result = html;
129         for(int i = 0; i < blocks.size(); i++) {
130             blocks.set(i, compressJsp(blocks.get(i)));
131         }
132         //put preserved blocks back
133         while(result.contains(tempJspBlock)) {
134             result = result.replaceFirst(tempJspBlock, Matcher.quoteReplacement(blocks.remove(0)));
135         }
136        
137         return result;
138     }
139     private static String processPreBlocks(String html, List<String> blocks) throws Exception {
140         String result = html;
141        
142         //put preserved blocks back
143         while(result.contains(tempPreBlock)) {
144             result = result.replaceFirst(tempPreBlock, Matcher.quoteReplacement(blocks.remove(0)));
145         }
146        
147         return result;
148     }
149    
150     private static String processTextareaBlocks(String html, List<String> blocks) throws Exception {
151         String result = html;
152        
153         //put preserved blocks back
154         while(result.contains(tempTextAreaBlock)) {
155             result = result.replaceFirst(tempTextAreaBlock, Matcher.quoteReplacement(blocks.remove(0)));
156         }
157        
158         return result;
159     }
160    
161     private static String processScriptBlocks(String html, List<String> blocks) throws Exception {
162         String result = html;
163        
164 //        if(compressJavaScript) {
165             for(int i = 0; i < blocks.size(); i++) {
166                 blocks.set(i, compressJavaScript(blocks.get(i)));
167             }
168 //        }
169        
170         //put preserved blocks back
171         while(result.contains(tempScriptBlock)) {
172             result = result.replaceFirst(tempScriptBlock, Matcher.quoteReplacement(blocks.remove(0)));
173         }
174        
175         return result;
176     }
177    
178     private static String processStyleBlocks(String html, List<String> blocks) throws Exception {
179         String result = html;
180        
181 //        if(compressCss) {
182             for(int i = 0; i < blocks.size(); i++) {
183                 blocks.set(i, compressCssStyles(blocks.get(i)));
184             }
185 //        }
186        
187         //put preserved blocks back
188         while(result.contains(tempStyleBlock)) {
189             result = result.replaceFirst(tempStyleBlock, Matcher.quoteReplacement(blocks.remove(0)));
190         }
191        
192         return result;
193     }
194    
195     private static String compressJsp(String source)  {
196         //check if block is not empty
197         Matcher jspMatcher = jspPattern.matcher(source);
198         if(jspMatcher.find()) {
199             String result = compressJspJs(jspMatcher.group(1));
200             return (new StringBuilder(source.substring(0, jspMatcher.start(1))).append(result).append(source.substring(jspMatcher.end(1)))).toString();
201         } else {
202             return source;
203         }
204     }   
205     private static String compressJavaScript(String source)  {
206         //check if block is not empty
207         Matcher scriptMatcher = scriptPattern.matcher(source);
208         if(scriptMatcher.find()) {
209             String result = compressJspJs(scriptMatcher.group(1));
210             return (new StringBuilder(source.substring(0, scriptMatcher.start(1))).append(result).append(source.substring(scriptMatcher.end(1)))).toString();
211         } else {
212             return source;
213         }
214     }
215        
216     private static String compressCssStyles(String source)  {
217         //check if block is not empty
218         Matcher styleMatcher = stylePattern.matcher(source);
219         if(styleMatcher.find()) {
220             // 去掉注释，换行
221             String result= multiCommentPattern.matcher(styleMatcher.group(1)).replaceAll("");
222             result = trimPattern.matcher(result).replaceAll("");
223             result = trimPattern2.matcher(result).replaceAll("");
224             return (new StringBuilder(source.substring(0, styleMatcher.start(1))).append(result).append(source.substring(styleMatcher.end(1)))).toString();
225         } else {
226             return source;
227         }
228     }
229    
230     private static String compressJspJs(String source){
231         String result = source;
232         // 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
233         Matcher stringMatcher = stringPattern.matcher(result);
234         while(stringMatcher.find()){
235             String tmpStr = stringMatcher.group(0);
236            
237             if(tmpStr.indexOf("//") != -1 || tmpStr.indexOf("/*") != -1 || tmpStr.indexOf("*/") != -1){
238                 String blockStr = tmpStr.replaceAll("//", tempSingleCommentBlock).replaceAll("/\\*", tempMulitCommentBlock1)
239                                 .replaceAll("\\*/", tempMulitCommentBlock2);
240                 result = result.replace(tmpStr, blockStr);
241             }
242         }
243         // 去掉注释
244         result = signleCommentPattern.matcher(result).replaceAll("");
245         result = multiCommentPattern.matcher(result).replaceAll("");
246         result = trimPattern2.matcher(result).replaceAll("");
247         result = trimPattern.matcher(result).replaceAll(" ");
248         // 恢复替换掉的字符串
249         result = result.replaceAll(tempSingleCommentBlock, "//").replaceAll(tempMulitCommentBlock1, "/*")
250                 .replaceAll(tempMulitCommentBlock2, "*/");
251        
252         return result;
253     }
254 }
255

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