1 | package org.greenstone.gsdl3.gs3build.xpointer;
|
---|
2 |
|
---|
3 | import java.io.File;
|
---|
4 |
|
---|
5 | import java.net.URL;
|
---|
6 |
|
---|
7 | import javax.xml.parsers.*;
|
---|
8 |
|
---|
9 | import org.w3c.dom.*;
|
---|
10 | import org.apache.xpath.*;
|
---|
11 |
|
---|
12 | import org.xml.sax.SAXException;
|
---|
13 | import org.xml.sax.SAXParseException;
|
---|
14 | import org.xml.sax.helpers.DefaultHandler;
|
---|
15 |
|
---|
16 | import org.greenstone.gsdl3.gs3build.doctypes.HTMLDocumentTools;
|
---|
17 | import org.greenstone.gsdl3.gs3build.util.HTMLTidy;
|
---|
18 |
|
---|
19 | public abstract class XPointer
|
---|
20 | {
|
---|
21 | Document document;
|
---|
22 | String xpointer;
|
---|
23 |
|
---|
24 | public XPointer(org.w3c.dom.Document document, String xpointer)
|
---|
25 | { this.document = document;
|
---|
26 | this.xpointer = xpointer;
|
---|
27 | // now process the type of description
|
---|
28 | }
|
---|
29 |
|
---|
30 | public static String expandXPath(String xpath)
|
---|
31 | {
|
---|
32 | int dashAt;
|
---|
33 | String pathElement;
|
---|
34 | StringBuffer expandedPtr = new StringBuffer();
|
---|
35 |
|
---|
36 | if (xpath.startsWith("//")) {
|
---|
37 | xpath = xpath.substring(2);
|
---|
38 | expandedPtr.append("//");
|
---|
39 | }
|
---|
40 | else if (xpath.startsWith("/")) {
|
---|
41 | xpath = xpath.substring(1);
|
---|
42 | expandedPtr.append("/");
|
---|
43 | }
|
---|
44 |
|
---|
45 | while (xpath.length() > 0) {
|
---|
46 | dashAt = xpath.indexOf('/');
|
---|
47 | if (dashAt >= 0) {
|
---|
48 | pathElement = xpath.substring(0, dashAt);
|
---|
49 | xpath = xpath.substring(dashAt+1);
|
---|
50 | }
|
---|
51 | else {
|
---|
52 | pathElement = xpath;
|
---|
53 | xpath = "";
|
---|
54 | }
|
---|
55 |
|
---|
56 | int c = 0;
|
---|
57 | while (c < pathElement.length()) {
|
---|
58 | if (!Character.isDigit(pathElement.charAt(c)))
|
---|
59 | break;
|
---|
60 | c ++;
|
---|
61 | }
|
---|
62 |
|
---|
63 | if (c == pathElement.length()) {
|
---|
64 | expandedPtr.append("*[");
|
---|
65 | expandedPtr.append(pathElement);
|
---|
66 | expandedPtr.append("]");
|
---|
67 | }
|
---|
68 | else {
|
---|
69 | expandedPtr.append(pathElement);
|
---|
70 | }
|
---|
71 |
|
---|
72 | if (dashAt >= 0) {
|
---|
73 | expandedPtr.append('/');
|
---|
74 | }
|
---|
75 | }
|
---|
76 | xpath = expandedPtr.toString();
|
---|
77 |
|
---|
78 | return xpath;
|
---|
79 | }
|
---|
80 |
|
---|
81 | public static XPointer processXPointer(Document document, String xpointer)
|
---|
82 | { // do a "cheap and dirty" detection of range-to
|
---|
83 | if (xpointer.indexOf("range-to") >= 0) {
|
---|
84 | return new RangeToXPointer(document, xpointer);
|
---|
85 | }
|
---|
86 | return new IdentifierXPointer(document, xpointer);
|
---|
87 | }
|
---|
88 |
|
---|
89 | public static XPointer processXPointer(Document document, URL url)
|
---|
90 | { String reference = url.getRef();
|
---|
91 |
|
---|
92 | reference.trim();
|
---|
93 | if (reference.startsWith("xpointer(")) {
|
---|
94 | reference = reference.substring(9, reference.length() - 1);
|
---|
95 | }
|
---|
96 | else {
|
---|
97 | return null;
|
---|
98 | }
|
---|
99 | // System.out.println(reference);
|
---|
100 |
|
---|
101 | return processXPointer(document, reference);
|
---|
102 | }
|
---|
103 |
|
---|
104 | public static void printNode(Node node, StringBuffer to, boolean close)
|
---|
105 | {
|
---|
106 | if (node.getNodeType() == org.w3c.dom.Node.TEXT_NODE) {
|
---|
107 | if (!close) {
|
---|
108 | to.append(node.getNodeValue());
|
---|
109 | to.append(" ");
|
---|
110 | }
|
---|
111 | }
|
---|
112 | else if (node.getNodeType() == org.w3c.dom.Node.ENTITY_NODE) {
|
---|
113 | if (!close) {
|
---|
114 | to.append("&");
|
---|
115 | to.append(node.getNodeValue());
|
---|
116 | to.append(";");
|
---|
117 | to.append(" ");
|
---|
118 | }
|
---|
119 | }
|
---|
120 | else if (node.getNodeType() == org.w3c.dom.Node.COMMENT_NODE) {
|
---|
121 | if (!close) {
|
---|
122 | to.append("<!-- ");
|
---|
123 | to.append(node.getNodeValue());
|
---|
124 | to.append(" -->");
|
---|
125 | }
|
---|
126 | }
|
---|
127 | else if (node.getNodeType() == org.w3c.dom.Node.ELEMENT_NODE) {
|
---|
128 | boolean hasChildren = (node.getChildNodes().getLength() > 0);
|
---|
129 |
|
---|
130 | if (close && !hasChildren) {
|
---|
131 | return;
|
---|
132 | }
|
---|
133 | to.append("<");
|
---|
134 | if (close) {
|
---|
135 | to.append("/");
|
---|
136 | }
|
---|
137 | to.append(node.getNodeName());
|
---|
138 |
|
---|
139 | if (!close) {
|
---|
140 | NamedNodeMap attributes = ((Element) node).getAttributes();
|
---|
141 | for (int a = 0; a < attributes.getLength(); a ++) {
|
---|
142 | Node attributeNode = attributes.item(a);
|
---|
143 | to.append(" ");
|
---|
144 | to.append(attributeNode.getNodeName());
|
---|
145 |
|
---|
146 | String value = attributeNode.getNodeValue();
|
---|
147 | if (value != null && value.length() > 0) {
|
---|
148 | to.append("=\"");
|
---|
149 | to.append(value);
|
---|
150 | to.append("\"");
|
---|
151 | }
|
---|
152 | }
|
---|
153 | }
|
---|
154 |
|
---|
155 | if (!hasChildren) {
|
---|
156 | to.append(" /");
|
---|
157 | }
|
---|
158 | to.append(">");
|
---|
159 | if (close && node.getNodeName()=="Section") {
|
---|
160 | to.append((char) 3);
|
---|
161 | }
|
---|
162 | }
|
---|
163 | }
|
---|
164 |
|
---|
165 | public static Node getNextNode(Node currentNode) {
|
---|
166 |
|
---|
167 | // Try to get the first valid child
|
---|
168 | NodeList children = currentNode.getChildNodes();
|
---|
169 | for (int c = 0; c < children.getLength(); c++) {
|
---|
170 | Node childNode = children.item(c);
|
---|
171 |
|
---|
172 | if (childNode.getNodeType() == org.w3c.dom.Node.ELEMENT_NODE ||
|
---|
173 | childNode.getNodeType() == org.w3c.dom.Node.COMMENT_NODE ||
|
---|
174 | childNode.getNodeType() == org.w3c.dom.Node.TEXT_NODE ||
|
---|
175 | childNode.getNodeType() == org.w3c.dom.Node.ENTITY_NODE) {
|
---|
176 | return childNode;
|
---|
177 | }
|
---|
178 | }
|
---|
179 |
|
---|
180 | // now try a siblings and parents siblings
|
---|
181 | Node siblingNode = null;
|
---|
182 | do {
|
---|
183 | siblingNode = currentNode.getNextSibling();
|
---|
184 | if (siblingNode != null) {
|
---|
185 | // should we be checking type??
|
---|
186 | return siblingNode;
|
---|
187 | }
|
---|
188 |
|
---|
189 | currentNode = currentNode.getParentNode();
|
---|
190 | } while (currentNode != null);
|
---|
191 |
|
---|
192 | return null;
|
---|
193 | }
|
---|
194 |
|
---|
195 |
|
---|
196 | public static Node getNextNode(Node currentNode, StringBuffer buffer)
|
---|
197 | { Node reply;
|
---|
198 |
|
---|
199 | // Try for children only the once...
|
---|
200 | NodeList children = currentNode.getChildNodes();
|
---|
201 | for (int c = 0; c < children.getLength(); c++) {
|
---|
202 | Node childNode = children.item(c);
|
---|
203 |
|
---|
204 | if (childNode.getNodeType() == org.w3c.dom.Node.ELEMENT_NODE ||
|
---|
205 | childNode.getNodeType() == org.w3c.dom.Node.COMMENT_NODE ||
|
---|
206 | childNode.getNodeType() == org.w3c.dom.Node.TEXT_NODE ||
|
---|
207 | childNode.getNodeType() == org.w3c.dom.Node.ENTITY_NODE) {
|
---|
208 | return childNode;
|
---|
209 | }
|
---|
210 | }
|
---|
211 |
|
---|
212 | // Repeat looking for a sibling & going up until done...
|
---|
213 | do {
|
---|
214 | reply = currentNode.getNextSibling();
|
---|
215 | if (reply != null) {
|
---|
216 | return reply;
|
---|
217 | }
|
---|
218 |
|
---|
219 | currentNode = currentNode.getParentNode();
|
---|
220 | if (buffer != null && currentNode != null) {
|
---|
221 | printNode(currentNode, buffer, true);
|
---|
222 | }
|
---|
223 | } while (currentNode != null);
|
---|
224 |
|
---|
225 | return reply;
|
---|
226 | }
|
---|
227 |
|
---|
228 | public abstract Node getStartNode();
|
---|
229 |
|
---|
230 |
|
---|
231 | public static void main(String args[])
|
---|
232 | {
|
---|
233 | try
|
---|
234 | { File file = new File(args[0]);
|
---|
235 |
|
---|
236 | Document document;
|
---|
237 | if (args[0].endsWith(".html") || args[0].endsWith(".htm")) {
|
---|
238 | HTMLTidy tidier = new HTMLTidy(file);
|
---|
239 | document = tidier.getDocument();
|
---|
240 |
|
---|
241 | HTMLDocumentTools docTools = new HTMLDocumentTools(document);
|
---|
242 | docTools.findSections();
|
---|
243 | System.out.println("Sections found");
|
---|
244 | }
|
---|
245 | else {
|
---|
246 | DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
---|
247 | DocumentBuilder builder = factory.newDocumentBuilder();
|
---|
248 | document = builder.parse(file);
|
---|
249 | }
|
---|
250 |
|
---|
251 | XPointer pointer = null;
|
---|
252 | if (args.length < 2) {
|
---|
253 | }
|
---|
254 | else if (args.length == 2) {
|
---|
255 | pointer = XPointer.processXPointer(document, args[1]);
|
---|
256 | }
|
---|
257 | else {
|
---|
258 | pointer = XPointer.processXPointer(document, args[1]+"/range-to("+args[2]+")");
|
---|
259 | }
|
---|
260 | if (pointer != null) {
|
---|
261 | System.out.println(pointer.toString());
|
---|
262 | }
|
---|
263 | }
|
---|
264 | catch (FactoryConfigurationError e) {
|
---|
265 | System.err.println(e);
|
---|
266 | }
|
---|
267 | catch (ParserConfigurationException ex) {
|
---|
268 | System.err.println(ex);
|
---|
269 | }
|
---|
270 | catch (SAXException ex) {
|
---|
271 | System.err.println(ex);
|
---|
272 | }
|
---|
273 | catch (java.io.IOException ex) {
|
---|
274 | System.err.println(ex);
|
---|
275 | }
|
---|
276 | }
|
---|
277 | }
|
---|
278 |
|
---|
279 |
|
---|
280 |
|
---|
281 |
|
---|