|
1
37
38 package org.htmlcleaner;
39
40 import java.io.File  ;
41 import java.io.IOException ;
42 import java.net.URL ;
43
44
50 public class CommandLine {
51
52 private static String getArgValue(String [] args, String name) {
53 for (int i = 0; i < args.length; i++) {
54 String curr = args[i];
55 int eqIndex = curr.indexOf('=');
56 if (eqIndex >= 0) {
57 String argName = curr.substring(0, eqIndex).trim();
58 String argValue = curr.substring(eqIndex+1).trim();
59
60 if (argName.toLowerCase().startsWith(name.toLowerCase())) {
61 return argValue;
62 }
63 }
64 }
65
66 return "";
67 }
68
69 private static boolean toBoolean(String s) {
70 return s != null && ( "on".equalsIgnoreCase(s) || "true".equalsIgnoreCase(s) || "yes".equalsIgnoreCase(s) );
71 }
72
73 public static void main(String [] args) throws IOException {
74 String source = getArgValue(args, "src");
75 if ( "".equals(source) ) {
76 System.err.println("Usage: java -jar htmlcleanerXX.jar src = <url | file> [incharset = <charset>] [dest = <file>] [outcharset = <charset>] [options...]");
77 System.err.println("");
78 System.err.println("where options include:");
79 System.err.println(" outputtype = simple | compact | pretty");
80 System.err.println(" advancedxmlescape = true | false");
81 System.err.println(" usecdata = true | false");
82 System.err.println(" specialentities = true | false");
83 System.err.println(" unicodechars = true | false");
84 System.err.println(" omitunknowntags = true | false");
85 System.err.println(" omitdeprtags = true | false");
86 System.err.println(" omitcomments = true | false");
87 System.err.println(" omitxmldecl = true | false");
88 System.err.println(" omitdoctypedecl = true | false");
89 System.err.println(" omitxmlnsatt = true | false");
90 System.err.println(" hyphenreplacement = <string value>");
91 System.exit(1);
92 }
93
94 String inCharset = getArgValue(args, "incharset");
95 if ("".equals(inCharset)) {
96 inCharset = HtmlCleaner.DEFAULT_CHARSET;
97 }
98
99 String outCharset = getArgValue(args, "outcharset");
100 if ("".equals(outCharset)) {
101 outCharset = HtmlCleaner.DEFAULT_CHARSET;
102 }
103
104 String destination = getArgValue(args, "dest");
105 String outputType = getArgValue(args, "outputtype");
106 String advancedXmlEscape = getArgValue(args, "advancedxmlescape");
107 String useCData = getArgValue(args, "usecdata");
108 String translateSpecialEntities = getArgValue(args, "specialentities");
109 String unicodeChars = getArgValue(args, "unicodechars");
110 String omitUnknownTags = getArgValue(args, "omitunknowntags");
111 String omitDeprecatedTags = getArgValue(args, "omitdeprtags");
112 String omitComments = getArgValue(args, "omitcomments");
113 String omitXmlDeclaration = getArgValue(args, "omitxmldecl");
114 String omitDoctypeDeclaration = getArgValue(args, "omitdoctypedecl");
115 String omitXmlnsAttributes = getArgValue(args, "omitxmlnsatt");
116 String commentHyphen = getArgValue(args, "hyphenreplacement");
117
118 HtmlCleaner cleaner = null;
119
120 String src = source.toLowerCase();
121 if ( src.startsWith("http://") || src.startsWith("https://") ) {
122 cleaner = new HtmlCleaner(new URL (src), inCharset);
123 } else {
124 cleaner = new HtmlCleaner(new File (src), inCharset);
125 }
126
127 if ( !"".equals(omitUnknownTags) ) {
128 cleaner.setOmitUnknownTags( toBoolean(omitUnknownTags) );
129 }
130
131 if ( !"".equals(omitDeprecatedTags) ) {
132 cleaner.setOmitDeprecatedTags( toBoolean(omitDeprecatedTags) );
133 }
134
135 if ( !"".equals(advancedXmlEscape) ) {
136 cleaner.setAdvancedXmlEscape( toBoolean(advancedXmlEscape) );
137 }
138
139 if ( !"".equals(useCData) ) {
140 cleaner.setUseCdataForScriptAndStyle( toBoolean(useCData) );
141 }
142
143 if ( !"".equals(translateSpecialEntities) ) {
144 cleaner.setTranslateSpecialEntities( toBoolean(translateSpecialEntities) );
145 }
146
147 if ( !"".equals(unicodeChars) ) {
148 cleaner.setRecognizeUnicodeChars( toBoolean(unicodeChars) );
149 }
150
151 if ( !"".equals(omitComments) ) {
152 cleaner.setOmitComments( toBoolean(omitComments) );
153 }
154
155 if ( !"".equals(omitXmlDeclaration) ) {
156 cleaner.setOmitXmlDeclaration( toBoolean(omitXmlDeclaration) );
157 }
158
159 if ( !"".equals(omitDoctypeDeclaration) ) {
160 cleaner.setOmitDoctypeDeclaration( toBoolean(omitDoctypeDeclaration) );
161 }
162
163 if ( !"".equals(omitXmlnsAttributes) ) {
164 cleaner.setOmitXmlnsAttributes( toBoolean(omitXmlnsAttributes) );
165 }
166
167 if ( !"".equals(commentHyphen) ) {
168 cleaner.setHyphenReplacementInComment(commentHyphen);
169 }
170
171 cleaner.clean();
172
173 if ( "".equals(destination) ) {
174 if ( "compact".equals(outputType) ) {
175 cleaner.writeCompactXmlToStream(System.out, outCharset);
176 } else if ( "pretty".equals(outputType) ) {
177 cleaner.writePrettyXmlToStream(System.out, outCharset);
178 } else {
179 cleaner.writeXmlToStream(System.out, outCharset);
180 }
181 } else {
182 if ( "compact".equals(outputType) ) {
183 cleaner.writeCompactXmlToFile(destination, outCharset);
184 } else if ( "pretty".equals(outputType) ) {
185 cleaner.writePrettyXmlToFile(destination, outCharset);
186 } else {
187 cleaner.writeXmlToFile(destination, outCharset);
188 }
189 }
190 }
191
192 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.