001 /*
002 * LAPIS lightweight structured text processing system
003 *
004 * Copyright (C) 1998-2002 Carnegie Mellon University,
005 * Copyright (C) 2003 Massachusetts Institute of Technology.
006 * All rights reserved.
007 *
008 * This library is free software; you can redistribute it
009 * and/or modify it under the terms of the GNU General
010 * Public License as published by the Free Software
011 * Foundation, version 2.
012 *
013 * LAPIS homepage: http://graphics.lcs.mit.edu/lapis/
014 */
015
016 package lapis.tools;
017
018 import lapis.*;
019 import lapis.tc.*;
020 import lapisx.util.Str;
021 //import java.io.*;
022 //import java.util.*;
023
024 public class Highlight implements Tool {
025 public static final lapisx.util.Debug debug = lapisx.util.Debug.QUIET;
026
027 static final String TEXT = "text/plain";
028 static final String HTML = "text/html";
029 static final String TCL_LIST = "application/x-tcl-list";
030
031 /**
032 * Invoke tool.
033 */
034 public Document invoke (Arguments args) throws Exception {
035 TC tcItems = null;
036 String contentType = HTML;
037 TC tcUnits = null;
038
039 args.setUsage (
040 "Usage:\n"
041 + " highlight <pattern> highlights regions matching <pattern>.\n"
042 + " In text, highlights >>like this<<\n"
043 + " In HTML, highlights in bold\n"
044 + "\n"
045 + "Options:\n"
046 + " -as [text|html|same] Output type (default is html)\n"
047 + " -units <pattern> Units to round to\n"
048 + " -help Display this message\n"
049 );
050
051 while (args.hasMoreElements ()) {
052 String name = args.nextName ();
053 if ("query".equals (name)
054 || (name == null && tcItems == null)) {
055 tcItems = args.nextPattern ();
056 }
057 else if ("as".equals (name)) {
058 String type = args.nextString ();
059 if ("text".equalsIgnoreCase (type)
060 || TEXT.equalsIgnoreCase (type))
061 contentType = TEXT;
062 else if ("html".equalsIgnoreCase (type)
063 || HTML.equalsIgnoreCase (type))
064 contentType = HTML;
065 else if ("same".equalsIgnoreCase (type))
066 contentType = null;
067 else
068 args.usage ();
069 }
070 else if ("units".equals (name) || "unit".equals (name)) {
071 tcUnits = args.nextPattern ();
072 }
073 else
074 args.consume (name);
075 }
076
077 // Error if no pattern
078 if (tcItems == null)
079 args.usage ();
080
081 if (tcUnits == null)
082 tcUnits = new TC (new TC.Id (".Layout.Line"));
083
084 String endsWith, highlightStart, highlightEnd;
085 if (HTML.equals (contentType)) {
086 endsWith = "<br>\n";
087 highlightStart = "<b>";
088 highlightEnd = "</b>";
089 } else {
090 endsWith = null;
091 highlightStart = ">>";
092 highlightEnd = "<<";
093 }
094
095 Document doc = args.getInput ();
096 DocumentWriter output = args.getOutput ();
097 String docContentType = (String) doc.getProperty (doc.MIMEProperty);
098
099 if (contentType == null)
100 contentType = docContentType;
101
102 BasicDocument.copyVitalProperties (output, doc);
103 if (contentType != null)
104 output.putProperty (doc.MIMEProperty, contentType);
105
106 RegionSet items = tcItems.matchWithWarnings (doc);
107 RegionSet units = tcUnits.matchWithWarnings (doc);
108
109 boolean html2text = (TEXT.equals (contentType)
110 && HTML.equals (docContentType));
111 boolean text2html = (HTML.equals (contentType)
112 && TEXT.equals (docContentType));
113
114 DocumentView view = doc.getDefaultView ();
115 if (html2text) {
116 view = doc.getCookedView();
117 items = view.getVersion().convert (items.getSingleDocRegionSet(doc));
118 units = view.getVersion().convert (units.getSingleDocRegionSet(doc));
119 }
120
121 items = items.flatten ();
122 units = units.flatten ();
123
124 Relation recordItems = new Relation ();
125 MutableRegionSet records = new MutableRegionSet ();
126 RegionEnumeration e = items.regions ();
127 for (Region item = e.firstFast (); item != null; item = e.nextFast ()) {
128 Region record = (Region) item.include (units);
129 if (!recordItems.add (record, item))
130 records.insert (record);
131 }
132
133 e = records.regions ();
134 for (Region record = e.firstFast (); record != null; record = e.nextFast ()) {
135 String rec = view.getText(record);
136 if (html2text)
137 rec = rec.trim ();
138 else if (text2html) {
139 rec = Str.replace (rec, "&", "&");
140 rec = Str.replace (rec, "<", "<");
141 rec = Str.replace (rec, ">", ">");
142 }
143
144 int off = 0; // number of characters inserted in rec
145 RegionEnumeration re = recordItems.get (record).regions ();
146 for (Region item = re.firstFast ();
147 item != null;
148 item = re.nextFast ()) {
149 debug.println ("highlighting " + item + " in " + record);
150 int st = item.getStart () - record.getStart () + off;
151 int en = item.getEnd () - record.getStart () + off;
152 debug.println ("that's [" + st + "," + en + "] relative");
153 rec = rec.substring (0, st)
154 + highlightStart
155 + rec.substring (st, en)
156 + highlightEnd
157 + rec.substring (en);
158 off += highlightStart.length () + highlightEnd.length ();
159 }
160
161 output.print (rec);
162 if (!rec.endsWith ("\n"))
163 output.println ();
164 output.print (endsWith);
165 }
166
167 return output;
168 }
169 }