Classes in this File | Line Coverage | Branch Coverage | Complexity | ||||
OSISFilter |
|
| 7.25;7.25 |
1 | /** | |
2 | * Distribution License: | |
3 | * JSword is free software; you can redistribute it and/or modify it under | |
4 | * the terms of the GNU Lesser General Public License, version 2.1 or later | |
5 | * as published by the Free Software Foundation. This program is distributed | |
6 | * in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even | |
7 | * the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. | |
8 | * See the GNU Lesser General Public License for more details. | |
9 | * | |
10 | * The License is available on the internet at: | |
11 | * http://www.gnu.org/copyleft/lgpl.html | |
12 | * or by writing to: | |
13 | * Free Software Foundation, Inc. | |
14 | * 59 Temple Place - Suite 330 | |
15 | * Boston, MA 02111-1307, USA | |
16 | * | |
17 | * © CrossWire Bible Society, 2005 - 2016 | |
18 | * | |
19 | */ | |
20 | package org.crosswire.jsword.book.filter.osis; | |
21 | ||
22 | import java.io.IOException; | |
23 | import java.io.StringReader; | |
24 | import java.util.Arrays; | |
25 | import java.util.List; | |
26 | import java.util.concurrent.ArrayBlockingQueue; | |
27 | import java.util.concurrent.BlockingQueue; | |
28 | import java.util.regex.Pattern; | |
29 | ||
30 | import org.crosswire.common.xml.XMLUtil; | |
31 | import org.crosswire.jsword.book.Book; | |
32 | import org.crosswire.jsword.book.DataPolice; | |
33 | import org.crosswire.jsword.book.OSISUtil; | |
34 | import org.crosswire.jsword.book.filter.SourceFilter; | |
35 | import org.crosswire.jsword.passage.Key; | |
36 | import org.jdom2.Content; | |
37 | import org.jdom2.Document; | |
38 | import org.jdom2.Element; | |
39 | import org.jdom2.JDOMException; | |
40 | import org.jdom2.input.SAXBuilder; | |
41 | import org.xml.sax.InputSource; | |
42 | ||
43 | /** | |
44 | * Filter to convert an OSIS XML string to OSIS format. | |
45 | * | |
46 | * @see gnu.lgpl.License The GNU Lesser General Public License for details. | |
47 | * @author Joe Walker | |
48 | */ | |
49 | 0 | public class OSISFilter implements SourceFilter { |
50 | ||
51 | /* (non-Javadoc) | |
52 | * @see org.crosswire.jsword.book.filter.Filter#toOSIS(org.crosswire.jsword.book.Book, org.crosswire.jsword.passage.Key, java.lang.String) | |
53 | */ | |
54 | public List<Content> toOSIS(Book book, Key key, String plain) { | |
55 | 0 | Element ele = null; |
56 | 0 | Exception ex = null; |
57 | 0 | String clean = plain; |
58 | ||
59 | // The following converts simple <div> and </div> to their milestoned versions. | |
60 | // Current versions of osis2mod do this already | |
61 | // Note: if the div element has attributes, it is not seen. | |
62 | 0 | clean = DIV_START.matcher(clean).replaceAll("<div sID=\"xyz\"/>"); |
63 | 0 | clean = DIV_END.matcher(clean).replaceAll("<div eID=\"xyz\"/>"); |
64 | 0 | clean = CHAPTER_END.matcher(clean).replaceAll("<chapter eID=\"xyz\"/>"); |
65 | 0 | clean = SPEECH_START.matcher(clean).replaceAll("<speech sID=\"xyz\"/>"); |
66 | 0 | clean = SPEECH_END.matcher(clean).replaceAll("<speech eID=\"xyz\"/>"); |
67 | ||
68 | // FIXME(dms): this is a major HACK handling a problem with a badly | |
69 | // encoded module. | |
70 | /* if (book.getInitials().startsWith("NET") && plain.endsWith("</div>")) { | |
71 | clean = clean.substring(0, plain.length() - 6); | |
72 | if (clean.matches(".*</div> <chapter eID=\"[A-Za-z0-9.]+\"/>")) { | |
73 | clean = clean.substring(0, clean.lastIndexOf("</div> <chapter")); | |
74 | } | |
75 | } else if (book.getInitials().equals("Kekchi") && plain.endsWith("</div> <lb type=\"x-begin-paragraph\"/>")) { | |
76 | clean = clean.substring(0, clean.length() - 37); | |
77 | } else if (book.getInitials().equals("VietLCCMN")) { | |
78 | int startPos = clean.indexOf("<div>"), endPos = clean.indexOf("</div>"); | |
79 | if (endPos != -1 && (startPos == -1 || startPos > endPos)) { | |
80 | if (clean.startsWith("<l ") || (clean.startsWith("<title ") && clean.contains("</title><l "))) | |
81 | clean = "<lg>"+clean; | |
82 | clean = "<div><div><div><div>"+clean; | |
83 | } | |
84 | } else */ | |
85 | 0 | if ("MapM".equals(book.getInitials())) { |
86 | 0 | for (String tag : Arrays.asList("cell", "row", "table")) { |
87 | 0 | int startPos = clean.indexOf("<" + tag + ">"); |
88 | 0 | int endPos = clean.indexOf("</" + tag + ">"); |
89 | 0 | if (endPos != -1 && (startPos == -1 || startPos > endPos)) { |
90 | 0 | clean = "<" + tag + ">" + clean; |
91 | } | |
92 | 0 | } |
93 | } | |
94 | ||
95 | try { | |
96 | 0 | ele = parse(clean); |
97 | 0 | } catch (JDOMException e) { |
98 | 0 | ex = e; |
99 | 0 | } catch (IOException e) { |
100 | 0 | ex = e; |
101 | 0 | } |
102 | ||
103 | 0 | if (ele == null) { |
104 | // There should be no bad entities in OSIS. | |
105 | 0 | String cleanedEntities = XMLUtil.cleanAllEntities(clean); |
106 | 0 | if (cleanedEntities != null && !cleanedEntities.equals(clean)) { |
107 | 0 | clean = cleanedEntities; |
108 | try { | |
109 | 0 | ele = parse(clean); |
110 | 0 | ex = null; |
111 | 0 | } catch (JDOMException e) { |
112 | 0 | ex = e; |
113 | 0 | } catch (IOException e) { |
114 | 0 | ex = e; |
115 | 0 | } |
116 | } | |
117 | } | |
118 | ||
119 | 0 | if (ele == null) { |
120 | 0 | String reclosed = XMLUtil.recloseTags(clean); |
121 | 0 | if (reclosed != null && !reclosed.equals(clean)) { |
122 | 0 | clean = reclosed; |
123 | try { | |
124 | 0 | ele = parse(clean); |
125 | 0 | ex = null; |
126 | 0 | } catch (JDOMException e) { |
127 | 0 | ex = e; |
128 | 0 | } catch (IOException e) { |
129 | 0 | ex = e; |
130 | 0 | } |
131 | } | |
132 | } | |
133 | ||
134 | 0 | if (ex != null) { |
135 | 0 | DataPolice.report(book, key, "Parse failed: " + ex.getMessage() + "\non: " + clean); |
136 | 0 | ele = cleanTags(book, key, clean); |
137 | } | |
138 | ||
139 | 0 | if (ele == null) { |
140 | 0 | ele = OSISUtil.factory().createP(); |
141 | } | |
142 | ||
143 | 0 | return ele.removeContent(); |
144 | } | |
145 | ||
146 | @Override | |
147 | public OSISFilter clone() { | |
148 | 0 | OSISFilter clone = null; |
149 | try { | |
150 | 0 | clone = (OSISFilter) super.clone(); |
151 | 0 | } catch (CloneNotSupportedException e) { |
152 | 0 | assert false : e; |
153 | 0 | } |
154 | 0 | return clone; |
155 | } | |
156 | ||
157 | private Element cleanTags(Book book, Key key, String plain) { | |
158 | // So just try to strip out all XML looking things | |
159 | 0 | String shawn = XMLUtil.cleanAllTags(plain); |
160 | 0 | Exception ex = null; |
161 | try { | |
162 | 0 | return parse(shawn); |
163 | 0 | } catch (JDOMException e) { |
164 | 0 | ex = e; |
165 | 0 | } catch (IOException e) { |
166 | 0 | ex = e; |
167 | 0 | } |
168 | ||
169 | 0 | DataPolice.report(book, key, "Parse failed: " + ex.getMessage() + "\non: " + shawn); |
170 | ||
171 | 0 | return null; |
172 | } | |
173 | ||
174 | /** | |
175 | * If the string is invalid then we might want to have more than one crack | |
176 | * at parsing it | |
177 | */ | |
178 | private Element parse(String plain) throws JDOMException, IOException { | |
179 | 0 | SAXBuilder builder = saxBuilders.poll(); |
180 | 0 | if (builder == null) { |
181 | //then we have no sax builders available, so let's create a new one and store | |
182 | 0 | builder = new SAXBuilder(); |
183 | // With JDom 1.x it was important to set Fast Reconfigure on re-usable SAXBuilders | |
184 | // This is the default with 2.x and this method does nothing | |
185 | // builder.setFastReconfigure(true); | |
186 | } | |
187 | ||
188 | // create a root element to house our document fragment | |
189 | 0 | StringReader in = null; |
190 | Element div; | |
191 | try { | |
192 | // Need to contain it in something that we remove when returning it to the user. | |
193 | 0 | in = new StringReader("<xxx>" + plain + "</xxx>"); |
194 | 0 | InputSource is = new InputSource(in); |
195 | 0 | Document doc = builder.build(is); |
196 | 0 | div = doc.getRootElement(); |
197 | } finally { | |
198 | 0 | if (in != null) { |
199 | 0 | in.close(); |
200 | } | |
201 | } | |
202 | ||
203 | //return builder to queue, or offer a new one. Ignore return value as we don't care whether the builder is going to be re-used | |
204 | 0 | saxBuilders.offer(builder); |
205 | ||
206 | 0 | return div; |
207 | } | |
208 | ||
209 | // space for 32 re-usable sax builders, but doesn't bound the number available to the callers | |
210 | 0 | private BlockingQueue<SAXBuilder> saxBuilders = new ArrayBlockingQueue<SAXBuilder>(32); |
211 | ||
212 | /** | |
213 | * Pattern to find the start of a div. Used to convert to a milestoned version. | |
214 | */ | |
215 | 0 | private static final Pattern DIV_START = Pattern.compile("<div>", Pattern.LITERAL); |
216 | /** | |
217 | * Pattern to find the end of a div. Used to convert to a milestoned version. | |
218 | */ | |
219 | 0 | private static final Pattern DIV_END = Pattern.compile("</div>", Pattern.LITERAL); |
220 | /** | |
221 | * Pattern to find the end of a chapter. Used to convert to a milestoned version. | |
222 | */ | |
223 | 0 | private static final Pattern CHAPTER_END = Pattern.compile("</chapter>", Pattern.LITERAL); |
224 | /** | |
225 | * Pattern to find the start of a speech. Used to convert to a milestoned version. | |
226 | */ | |
227 | 0 | private static final Pattern SPEECH_START = Pattern.compile("<speech>", Pattern.LITERAL); |
228 | /** | |
229 | * Pattern to find the end of a speech. Used to convert to a milestoned version. | |
230 | */ | |
231 | 0 | private static final Pattern SPEECH_END = Pattern.compile("</speech>", Pattern.LITERAL); |
232 | } |