Mercurial > hg > digilib-old
annotate servlet/src/digilib/io/XMLMetaLoader.java @ 385:0155f1c3f505
forgot rgb icon
| author | hertzhaft |
|---|---|
| date | Wed, 07 Dec 2005 16:07:29 +0100 |
| parents | d4c0c4d858a9 |
| children | 6dde90968cf1 |
| rev | line source |
|---|---|
| 130 | 1 /* XMLMetaLoader -- Load an XML format metadata into a Hashtable |
| 2 | |
| 3 Digital Image Library servlet components | |
| 4 | |
| 5 Copyright (C) 2003 Robert Casties (robcast@mail.berlios.de) | |
| 6 | |
| 7 This program is free software; you can redistribute it and/or modify it | |
| 8 under the terms of the GNU General Public License as published by the | |
| 9 Free Software Foundation; either version 2 of the License, or (at your | |
| 10 option) any later version. | |
| 11 | |
| 12 Please read license.txt for the full details. A copy of the GPL | |
| 13 may be found at http://www.gnu.org/copyleft/lgpl.html | |
| 14 | |
| 15 You should have received a copy of the GNU General Public License | |
| 16 along with this program; if not, write to the Free Software | |
| 17 Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | |
| 18 | |
| 19 */ | |
| 20 | |
| 21 package digilib.io; | |
| 22 | |
| 23 import java.io.IOException; | |
| 24 import java.util.HashMap; | |
| 25 import java.util.LinkedList; | |
| 273 | 26 import java.util.Map; |
| 130 | 27 |
| 28 import javax.xml.parsers.ParserConfigurationException; | |
| 29 import javax.xml.parsers.SAXParser; | |
| 30 import javax.xml.parsers.SAXParserFactory; | |
| 31 | |
| 181 | 32 import org.apache.log4j.Logger; |
| 130 | 33 import org.xml.sax.Attributes; |
| 34 import org.xml.sax.SAXException; | |
| 35 import org.xml.sax.helpers.DefaultHandler; | |
| 36 | |
| 37 public class XMLMetaLoader { | |
| 38 | |
| 181 | 39 private Logger logger = Logger.getLogger(this.getClass()); |
| 130 | 40 private String outerTag = "resource"; |
| 41 private String metaTag = "meta"; | |
| 42 private String fileTag = "file"; | |
| 43 private String fileNameTag = "name"; | |
| 44 private String filePathTag = "path"; | |
| 233 | 45 private String imgTag = "img"; |
| 46 private String collectTag = "context"; | |
| 130 | 47 |
| 48 public XMLMetaLoader() { | |
| 49 } | |
| 50 | |
| 51 /** | |
| 52 * inner class XMLMetaParser to be called by the parser | |
| 53 */ | |
| 54 private class XMLMetaParser extends DefaultHandler { | |
| 55 | |
| 56 private LinkedList tags; | |
| 273 | 57 private Map files; |
| 58 private Map meta; | |
| 130 | 59 private StringBuffer content; |
| 233 | 60 private boolean collecting; |
| 61 private StringBuffer collectedContent; | |
| 130 | 62 private String fileName; |
| 63 private String filePath; | |
| 64 | |
|
139
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
65 /** |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
66 * extracts the elements name from either localName ln or qName qn. |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
67 * |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
68 * @param ln localName |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
69 * @param qn qName |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
70 * @return element name |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
71 */ |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
72 private String getName(String ln, String qn) { |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
73 if (ln != null) { |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
74 if (ln.length() > 0) { |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
75 return ln; |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
76 } |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
77 } |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
78 // else it's qName (or nothing) |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
79 return qn; |
|
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
80 } |
| 130 | 81 |
| 233 | 82 /** |
| 83 * returns all attributes as a String | |
| 84 * | |
| 85 * @param attrs | |
| 86 * @return | |
| 87 */ | |
| 88 private String getAttrString(Attributes attrs) { | |
| 89 StringBuffer s = new StringBuffer(); | |
| 90 for (int i = 0; i < attrs.getLength(); i++) { | |
| 91 String key = getName(attrs.getLocalName(i), attrs.getQName(i)); | |
| 246 | 92 s.append(" "+key+"=\""+attrs.getValue(i)+"\""); |
| 233 | 93 } |
| 94 return s.toString(); | |
| 95 } | |
| 96 | |
| 97 | |
| 130 | 98 // Parser calls this once at the beginning of a document |
| 99 public void startDocument() throws SAXException { | |
| 100 tags = new LinkedList(); | |
| 101 files = new HashMap(); | |
| 233 | 102 collecting = false; |
| 103 collectedContent = null; | |
| 130 | 104 } |
| 105 | |
| 106 // Parser calls this for each element in a document | |
| 107 public void startElement( | |
| 108 String namespaceURI, | |
| 109 String localName, | |
| 110 String qName, | |
| 111 Attributes atts) | |
| 112 throws SAXException { | |
| 113 | |
|
139
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
114 String name = getName(localName, qName); |
| 130 | 115 // open a new tag |
| 116 tags.addLast(name); | |
| 117 // start new content (no nesting of tags and content) | |
| 118 content = new StringBuffer(); | |
| 119 | |
| 120 if (name.equals(metaTag)) { | |
| 121 // new meta tag | |
| 122 meta = new HashMap(); | |
| 233 | 123 collectedContent = new StringBuffer(); |
| 130 | 124 } else if (name.equals(fileTag)) { |
| 125 // new file tag | |
| 126 fileName = null; | |
| 127 filePath = null; | |
|
139
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
128 meta = new HashMap(); |
| 233 | 129 collectedContent = new StringBuffer(); |
| 130 } else if (name.equals(collectTag)) { | |
| 131 // start collecting | |
| 132 collecting = true; | |
| 133 if (collectedContent == null) { | |
| 134 collectedContent = new StringBuffer(); | |
| 135 } | |
| 136 } | |
| 137 | |
| 138 // record mode | |
| 139 if (collecting) { | |
| 140 collectedContent.append("<"+name); | |
| 141 collectedContent.append(getAttrString(atts)); | |
| 142 collectedContent.append(">"); | |
| 130 | 143 } |
| 144 } | |
| 145 | |
| 146 // parser calls this for all tag content (possibly more than once) | |
| 147 public void characters(char[] ch, int start, int length) | |
| 148 throws SAXException { | |
| 149 // append data to current string buffer | |
| 233 | 150 if (content == null) { |
| 151 content = new StringBuffer(); | |
| 152 } | |
| 130 | 153 content.append(ch, start, length); |
| 154 } | |
| 155 | |
| 156 // parser calls this at the end of each element | |
| 157 public void endElement( | |
| 158 String namespaceURI, | |
| 159 String localName, | |
| 160 String qName) | |
| 161 throws SAXException { | |
| 151 | 162 |
|
139
11cfe4c89fdc
Servlet version 1.11b1 with improved original-size.
robcast
parents:
130
diff
changeset
|
163 String name = getName(localName, qName); |
| 130 | 164 // exit the tag |
| 165 tags.removeLast(); | |
| 233 | 166 String lastTag = (tags.isEmpty()) ? "" : (String) tags.getLast(); |
| 151 | 167 |
| 168 // was it a file/name tag? | |
| 233 | 169 if (name.equals(fileNameTag) && lastTag.equals(fileTag)) { |
| 130 | 170 // save name as filename |
| 151 | 171 if ((content != null) && (content.length() > 0)) { |
| 172 fileName = content.toString().trim(); | |
| 130 | 173 } |
| 233 | 174 content = null; |
| 130 | 175 return; |
| 176 } | |
| 177 | |
| 151 | 178 // was it a file/path tag? |
| 233 | 179 if (name.equals(filePathTag) && lastTag.equals(fileTag)) { |
| 130 | 180 // save path as filepath |
| 151 | 181 if ((content != null) && (content.length() > 0)) { |
| 182 filePath = content.toString().trim(); | |
| 130 | 183 } |
| 233 | 184 content = null; |
| 130 | 185 return; |
| 186 } | |
| 187 | |
| 188 // was it a file tag? | |
| 189 if (name.equals(fileTag)) { | |
| 190 // is there meta to save? | |
| 151 | 191 if ((meta != null) && (meta.size() > 0)) { |
| 192 // file name is (optional file/path) / file/name | |
| 130 | 193 String fn = null; |
| 151 | 194 |
| 195 if (fileName != null) { | |
| 196 if (filePath != null) { | |
| 197 fn = filePath + "/" + fileName; | |
| 198 } else { | |
| 199 fn = fileName; | |
| 200 } | |
| 130 | 201 } else { |
| 202 // no file name, no file | |
| 233 | 203 content = null; |
| 130 | 204 return; |
| 205 } | |
| 206 // save meta in file list | |
| 207 files.put(fn, meta); | |
| 208 } | |
| 233 | 209 content = null; |
| 130 | 210 return; |
| 211 } | |
| 212 | |
| 213 // was it a meta tag outside a file tag? | |
| 214 if (name.equals(metaTag) && !tags.contains(fileTag)) { | |
| 215 // save meta as dir meta | |
| 151 | 216 if ((meta != null) && (meta.size() > 0)) { |
| 130 | 217 files.put("", meta); |
| 218 } | |
| 233 | 219 content = null; |
| 130 | 220 return; |
| 221 } | |
| 222 | |
| 233 | 223 // is this inside an digilib info (=img) tag? |
| 224 if (lastTag.equals(imgTag)) { | |
| 130 | 225 // then add whatever this is |
| 151 | 226 if ((content != null) && (content.length() > 0)) { |
| 227 meta.put(name, content.toString().trim()); | |
| 130 | 228 } |
| 233 | 229 content = null; |
| 230 return; | |
| 130 | 231 } |
| 151 | 232 |
| 233 | 233 // is this the end of collectTag? |
| 234 if (name.equals(collectTag)) { | |
| 235 collecting = false; | |
| 236 collectedContent.append("</"+collectTag+">\n"); | |
| 237 // store collected stuff | |
| 238 meta.put(collectTag, collectedContent.toString()); | |
| 239 //logger.debug("collected: '"+collectedContent+"'"); | |
| 240 content = null; | |
| 241 return; | |
| 242 } | |
| 243 | |
| 244 // write collected content | |
| 245 if (collecting) { | |
| 246 String s = ""; | |
| 247 if ((content != null) && (content.length() > 0)) { | |
| 248 s = content.toString().trim(); | |
| 249 } | |
| 250 //logger.debug("collect:"+name+" = "+s); | |
| 251 collectedContent.append(s); | |
| 252 collectedContent.append("</"+name+">\n"); | |
| 253 content = null; | |
| 254 return; | |
| 255 } | |
| 130 | 256 } |
| 257 | |
| 258 } | |
| 259 | |
| 260 /** | |
| 261 * load and parse a file (as URL) | |
| 262 * returns HashMap with list data | |
| 263 */ | |
| 273 | 264 public Map loadURL(String path) throws SAXException, IOException { |
| 195 | 265 logger.debug("loading meta: "+path); |
| 130 | 266 // Create a JAXP SAXParserFactory and configure it |
| 267 SAXParserFactory spf = SAXParserFactory.newInstance(); | |
| 268 spf.setNamespaceAware(true); | |
| 269 | |
| 270 SAXParser parser = null; | |
| 271 try { | |
| 272 // Create a JAXP SAXParser | |
| 273 parser = spf.newSAXParser(); | |
| 274 | |
| 275 } catch (ParserConfigurationException e) { | |
| 276 throw new SAXException(e); | |
| 277 } | |
| 278 | |
| 279 // create a list parser (keeps the data!) | |
| 280 XMLMetaParser listParser = new XMLMetaParser(); | |
| 281 | |
| 282 // Tell the SAXParser to parse the XML document | |
| 283 parser.parse(path, listParser); | |
| 284 | |
| 285 return listParser.files; | |
| 286 } | |
| 287 | |
| 288 } |