Mercurial > hg > fulltextIndexer

comparison src/de/mpiwg/dwinter/fulltextIndexer/utils/.svn/text-base/ParseOcrDocument.java.svn-base @ 0:dc7622afcfea default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
initial
author dwinter
date Wed, 03 Nov 2010 12:33:16 +0100
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:dc7622afcfea
1 /* */ package de.mpiwg.dwinter.fulltextIndexer.utils;
2 /* */
3 /* */ import de.mpiwg.dwinter.fulltextIndexer.OCRutils.OCRDocument;
4 /* */ import de.mpiwg.dwinter.fulltextIndexer.OCRutils.OCRDocument.OCRLine;
5 /* */ import java.io.File;
6 /* */ import java.io.IOException;
7 /* */ import java.io.PrintStream;
8 /* */ import java.util.ArrayList;
9 /* */ import org.xml.sax.Attributes;
10 /* */ import org.xml.sax.InputSource;
11 /* */ import org.xml.sax.Locator;
12 /* */ import org.xml.sax.SAXException;
13 /* */ import org.xml.sax.helpers.DefaultHandler;
14 /* */
15 /* */ public class ParseOcrDocument extends DefaultHandler
16 /* */ {
17 /* 24 */ public OCRDocument ocrDocument = new OCRDocument();
18 /* */ private int lineCounter;
19 /* */ private OCRDocument.OCRLine currentLine;
20 /* */ private boolean inLine;
21 /* */
22 /* */ public void startDocument()
23 /* */ throws SAXException
24 /* */ {
25 /* */ }
26 /* */
27 /* */ public void endDocument()
28 /* */ throws SAXException
29 /* */ {
30 /* */ }
31 /* */
32 /* */ public void characters(char[] c, int start, int length)
33 /* */ throws SAXException
34 /* */ {
35 /* 38 */ if (this.inLine)
36 /* 39 */ this.currentLine.content += new String(c, start, length);
37 /* */ }
38 /* */
39 /* */ public void ignorableWhitespace(char[] c, int start, int length)
40 /* */ throws SAXException
41 /* */ {
42 /* */ }
43 /* */
44 /* */ public void processingInstruction(String target, String data)
45 /* */ throws SAXException
46 /* */ {
47 /* */ }
48 /* */
49 /* */ public void setDocumentLocator(Locator arg1)
50 /* */ {
51 /* */ }
52 /* */
53 /* */ public void endElement(String uri, String localName, String name)
54 /* */ throws SAXException
55 /* */ {
56 /* 61 */ if ((!name.equals("span")) ||
57 /* 63 */ (!this.inLine)) return;
58 /* 64 */ this.ocrDocument.OCRLines.add(this.currentLine);
59 /* 65 */ this.inLine = false;
60 /* */ }
61 /* */
62 /* */ public void endPrefixMapping(String prefix)
63 /* */ throws SAXException
64 /* */ {
65 /* */ }
66 /* */
67 /* */ public void skippedEntity(String name)
68 /* */ throws SAXException
69 /* */ {
70 /* */ }
71 /* */
72 /* */ public void startElement(String uri, String localName, String name, Attributes attrs)
73 /* */ throws SAXException
74 /* */ {
75 /* 83 */ if ((name.equals("div")) &&
76 /* 84 */ (attrs != null)) {
77 /* 85 */ int length = attrs.getLength();
78 /* */
79 /* 87 */ for (int i = 0; i < length; ++i) {
80 /* 88 */ if ((!attrs.getLocalName(i).equals("class")) ||
81 /* 89 */ (!attrs.getValue(i).equals("ocr_page"))) continue;
82 /* 90 */ doPage(attrs);
83 /* */ }
84 /* */
85 /* */ }
86 /* */
87 /* 96 */ if ((!name.equals("span")) ||
88 /* 97 */ (attrs == null)) return;
89 /* 98 */ int length = attrs.getLength();
90 /* */
91 /* 100 */ for (int i = 0; i < length; ++i) {
92 /* 101 */ if ((!attrs.getLocalName(i).equals("class")) ||
93 /* 102 */ (!attrs.getValue(i).equals("ocr_line"))) continue;
94 /* 103 */ doLine(attrs);
95 /* */ }
96 /* */ }
97 /* */
98 /* */ private void doPage(Attributes attrs)
99 /* */ {
100 /* 111 */ int length = attrs.getLength();
101 /* */
102 /* 113 */ for (int i = 0; i < length; ++i) {
103 /* 114 */ if (attrs.getLocalName(i).equals("title")) {
104 /* 115 */ String title = attrs.getValue(i);
105 /* 116 */ String[] splitted = title.split(" ");
106 /* 117 */ String dimension = splitted[(splitted.length - 2)] + " " +
107 /* 118 */ splitted[(splitted.length - 1)];
108 /* 119 */ this.ocrDocument.pageDimension = dimension;
109 /* */ }
110 /* */ }
111 /* 122 */ this.lineCounter = 0;
112 /* */ }
113 /* */
114 /* */ private void doLine(Attributes attrs)
115 /* */ {
116 /* */
117 OCRDocument doc = new OCRDocument();
118
119 this.currentLine = doc.new OCRLine();
120 /* 128 */ this.inLine = true;
121 /* 129 */ this.currentLine.content = new String();
122 /* */
123 /* 131 */ int length = attrs.getLength();
124 /* 132 */ this.currentLine.lineNumber = String.valueOf(this.lineCounter);
125 /* 133 */ this.lineCounter += 1;
126 /* 134 */ this.currentLine.bbox = "0 0";
127 /* */
128 /* 137 */ for (int i = 0; i < length; ++i)
129 /* 138 */ if (attrs.getLocalName(i).equals("title")) {
130 /* 139 */ String title = attrs.getValue(i);
131 /* */
132 /* 141 */ String dimension = title.replace("bbox ", "");
133 /* 142 */ this.currentLine.bbox = dimension;
134 /* */ }
135 /* */ }
136 /* */
137 /* */ public void startPrefixMapping(String prefix, String uri)
138 /* */ throws SAXException
139 /* */ {
140 /* */ }
141 /* */
142 /* */ public InputSource resolveEntity(String publicId, String systemId)
143 /* */ throws SAXException, IOException
144 /* */ {
145 /* 160 */ File f = new File(
146 /* 161 */ "/Users/dwinter/text-tools/fulltextsearch/catalog/xhtml1-transitional.dtd");
147 /* 162 */ if (!f.exists())
148 /* 163 */ f = new File(
149 /* 164 */ "/usr/local/apache-tomcat-6.0.16/webapps/fulltextsearch/catalog/xhtml1-transitional.dtd");
150 /* 165 */ if (!f.exists()) {
151 /* 166 */ System.err.println("Cant't find xhtml-dtd: MyResolver");
152 /* 167 */ return null;
153 /* */ }
154 /* 169 */ if (publicId.equals("-//W3C//DTD XHTML 1.0 Transitional//EN"))
155 /* 170 */ return new InputSource(f.getAbsolutePath());
156 /* 171 */ return new InputSource(systemId);
157 /* */ }
158 /* */ }
159
160 /* Location: /private/tmp/fulltextIndexer.jar
161 * Qualified Name: de.mpiwg.dwinter.fulltextIndexer.utils.ParseOcrDocument
162 * JD-Core Version: 0.5.4
163 */