comparison software/mpdl-services-new/mpiwg-mpdl-cms-web/src/de/mpg/mpiwg/berlin/mpdl/servlets/lt/GetLemmas.java @ 25:e9fe3186670c default tip

letzter Stand eingecheckt
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Tue, 21 May 2013 10:19:32 +0200
parents
children
comparison
equal deleted inserted replaced
23:e845310098ba 25:e9fe3186670c
1 package de.mpg.mpiwg.berlin.mpdl.servlets.lt;
2
3 import java.io.IOException;
4 import java.io.PrintWriter;
5 import java.util.ArrayList;
6 import java.util.Collections;
7 import java.util.Date;
8
9 import javax.servlet.ServletConfig;
10 import javax.servlet.ServletException;
11 import javax.servlet.http.HttpServlet;
12 import javax.servlet.http.HttpServletRequest;
13 import javax.servlet.http.HttpServletResponse;
14
15 import de.mpg.mpiwg.berlin.mpdl.exception.ApplicationException;
16 import de.mpg.mpiwg.berlin.mpdl.lt.dict.db.LexHandler;
17 import de.mpg.mpiwg.berlin.mpdl.lt.general.Language;
18 import de.mpg.mpiwg.berlin.mpdl.lt.morph.app.Form;
19 import de.mpg.mpiwg.berlin.mpdl.lt.morph.app.Lemma;
20 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.Normalizer;
21
22 public class GetLemmas extends HttpServlet {
23 private static final long serialVersionUID = 1L;
24 private LexHandler lexHandler;
25
26 public GetLemmas() {
27 super();
28 }
29
30 public void init(ServletConfig config) throws ServletException {
31 super.init(config);
32 try {
33 lexHandler = LexHandler.getInstance();
34 } catch (ApplicationException e) {
35 throw new ServletException(e);
36 }
37 }
38
39 protected void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
40 Date begin = new Date();
41 request.setCharacterEncoding("utf-8");
42 response.setCharacterEncoding("utf-8");
43 String query = request.getParameter("query");
44 String language = request.getParameter("language");
45 String inputType = request.getParameter("inputType");
46 String outputFormat = request.getParameter("outputFormat");
47 String outputType = request.getParameter("outputType");
48 String normalization = request.getParameter("normalization");
49 if (language == null)
50 language = "eng";
51 if (inputType == null || ! (inputType.equals("form") || inputType.equals("lemma")))
52 inputType = "form";
53 if (outputFormat == null || ! (outputFormat.equals("xml") || outputFormat.equals("html") || outputFormat.equals("string")))
54 outputFormat = "xml";
55 if (outputType == null || ! (outputType.equals("compact") || outputType.equals("full")))
56 outputType = "compact";
57 if (normalization == null || ! (normalization.equals("none") || normalization.equals("reg") || normalization.equals("reg norm")))
58 normalization = "norm";
59 int normalizationType = Normalizer.DICTIONARY;
60 if (normalization != null && normalization.equals("none"))
61 normalizationType = Normalizer.NONE;
62
63 String xmlQueryString = "<query><name>" + query + "</name>" + "<language>" + language + "</language>" + "<inputType>" + inputType + "</inputType>" +
64 "<outputFormat>" + outputFormat + "</outputFormat>" + "<outputType>" + outputType + "</outputType>" + "<normalization>" + normalization + "</normalization>" + "</query>";
65 try {
66 if (outputFormat.equals("xml"))
67 response.setContentType("text/xml");
68 else if (outputFormat.equals("html") || outputFormat.equals("string"))
69 response.setContentType("text/html");
70 else
71 response.setContentType("text/xml");
72 PrintWriter out = response.getWriter();
73 if (query == null || query.isEmpty()) {
74 out.print("request parameter query is empty. Please specify a query.");
75 out.close();
76 return;
77 }
78 ArrayList<Lemma> lemmas = lexHandler.getLemmas(query, inputType, language, normalizationType, true);
79 String baseUrl = getBaseUrl(request);
80 Date end = new Date();
81 String elapsedTime = String.valueOf(end.getTime() - begin.getTime());
82 String result = "";
83 if (outputFormat == null || outputFormat.equals("xml"))
84 result = createXmlOutputString(query, lemmas, outputType, baseUrl, xmlQueryString, elapsedTime);
85 else if (outputFormat.equals("html"))
86 result = createHtmlOutputString(query, lemmas, outputType, elapsedTime);
87 else if (outputFormat.equals("string"))
88 result = createStringOutputString(lemmas);
89 else
90 result = createXmlOutputString(query, lemmas, outputType, baseUrl, xmlQueryString, elapsedTime);
91 out.print(result);
92 out.close();
93 } catch (ApplicationException e) {
94 throw new ServletException(e);
95 }
96 }
97
98 protected void doPost(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException {
99 doGet(request, response);
100 }
101
102 private String getBaseUrl( HttpServletRequest request ) {
103 if (request.getServerPort() == 80 || request.getServerPort() == 443)
104 return request.getScheme() + "://" + request.getServerName() + request.getContextPath();
105 else
106 return request.getScheme() + "://" + request.getServerName() + ":" + request.getServerPort() + request.getContextPath();
107 }
108
109
110 private String createXmlOutputString(String query, ArrayList<Lemma> lemmas, String outputType, String baseUrl, String xmlQueryString, String elapsedTime) {
111 String result = "<result>";
112 result = result + "<provider>" + "MPIWG MPDL language technology service (see: " + "" + baseUrl + "), Max Planck Institute for the History of Science, Berlin." + "</provider>";
113 result = result + xmlQueryString;
114 result = result + "<elapsed-time-ms>" + elapsedTime + "</elapsed-time-ms>";
115 if (lemmas != null && ! lemmas.isEmpty()) {
116 result = result + "<morphology>";
117 for (int i=0; i<lemmas.size(); i++) {
118 Lemma lemma = lemmas.get(i);
119 String lemmaName = lemma.getLemmaName();
120 String language = lemma.getLanguage();
121 result = result + "<lemma>";
122 result = result + "<name>" + lemmaName + "</name>";
123 if (outputType != null && outputType.equals("full")) {
124 String lemmaProvider = lemma.getProvider();
125 result = result + "<provider>" + lemmaProvider + "</provider>";
126 result = result + "<language>" + language + "</language>";
127 }
128 if (Language.getInstance().isArabic(language) || Language.getInstance().isLatin(language)) {
129 String remoteUrl = "http://www.perseus.tufts.edu/hopper/morph?l=" + lemmaName + "&amp;la=" + language;
130 result = result + "<remoteUrl>" + remoteUrl + "</remoteUrl>";
131 } else if (Language.getInstance().isGreek(language)) {
132 String remoteUrl = "http://www.perseus.tufts.edu/hopper/morph?l=" + lemmaName + "&amp;la=" + "greek";
133 result = result + "<remoteUrl>" + remoteUrl + "</remoteUrl>";
134 }
135 if (outputType != null && outputType.equals("full")) {
136 ArrayList<Form> forms = lemma.getFormsList();
137 Collections.sort(forms);
138 if (forms != null && ! forms.isEmpty()) {
139 result = result + "<forms>";
140 for (int j=0; j<forms.size(); j++) {
141 result = result + "<form>";
142 Form f = forms.get(j);
143 String formName = f.getFormName();
144 String formProvider = f.getProvider();
145 result = result + "<provider>" + formProvider + "</provider>";
146 result = result + "<language>" + language + "</language>";
147 result = result + "<name>" + formName + "</name>";
148 result = result + "</form>";
149 }
150 result = result + "</forms>";
151 }
152 }
153 result = result + "</lemma>";
154 }
155 result = result + "</morphology>";
156 }
157 result = result + "</result>";
158 return result;
159 }
160
161 private String createHtmlOutputString(String query, ArrayList<Lemma> lemmas, String outputType, String elapsedTime) {
162 String result = "";
163 result = result + "<html>";
164 result = result + "<head>";
165 result = result + "<title>Lemmas for: \"" + query + "\"</title>";
166 result = result + "</head>";
167 result = result + "<body>";
168 result = result + "<table align=\"right\" valign=\"top\">";
169 result = result + "<td>[<i>This is a MPIWG language technology service</i>] <a href=\"/mpiwg-mpdl-cms-web/lt-services.html\"><img src=\"/mpiwg-mpdl-cms-web/images/info.png\" valign=\"bottom\" width=\"15\" height=\"15\" border=\"0\" alt=\"MPIWG language technology service\"/></a></td>";
170 result = result + "</table>";
171 result = result + "<p/>";
172 result = result + "<h1>Lemmas for: \"" + query + "\"</h1>";
173 if (lemmas != null && ! lemmas.isEmpty()) {
174 result = result + "<h3>Morphology</h3>";
175 result = result + "<ul>";
176 result = result + "<p/>";
177 for (int i=0; i<lemmas.size(); i++) {
178 Lemma lemma = lemmas.get(i);
179 String lemmaName = lemma.getLemmaName();
180 String language = lemma.getLanguage();
181 result = result + "<li>";
182 result = result + lemmaName;
183 if (outputType != null && outputType.equals("full")) {
184 String lemmaProvider = lemma.getProvider();
185 result = result + " (data provider: " + lemmaProvider + ")";
186 }
187 if (Language.getInstance().isArabic(language) || Language.getInstance().isLatin(language))
188 result = result + " (external link: <a href=\"http://www.perseus.tufts.edu/hopper/morph?l=" + lemmaName + "&amp;la=" + language + "\">" + lemmaName + "</a>)";
189 else if (Language.getInstance().isGreek(language))
190 result = result + " (external link: <a href=\"http://www.perseus.tufts.edu/hopper/morph?l=" + lemmaName + "&amp;la=" + "greek" + "\">" + lemmaName + "</a>)";
191 if (outputType != null && outputType.equals("full")) {
192 ArrayList<Form> forms = lemma.getFormsList();
193 Collections.sort(forms);
194 if (forms != null && ! forms.isEmpty()) {
195 result = result + "<ul>";
196 for (int j=0; j<forms.size(); j++) {
197 Form f = forms.get(j);
198 String formName = f.getFormName();
199 String formProvider = f.getProvider();
200 result = result + formName + " (data provider: " + formProvider + "), ";
201 }
202 result = result.substring(0, result.length() - 2); // without last comma and blank
203 result = result + "</ul>";
204 }
205 }
206 result = result + "</li>";
207 }
208 result = result + "</ul>";
209 }
210 result = result + "[* external links may not function]";
211 result = result + "<hr/>";
212 result = result + "<p/>";
213 result = result + "Elapsed time: " + elapsedTime + " ms, see the <a href=\"/mpiwg-mpdl-lt-web/index.html\">service description</a> of this page, if you find a bug <a href=\"https://it-dev.mpiwg-berlin.mpg.de/tracs/mpdl-project-software/newticket\">let us know</a>";
214 result = result + "</body>";
215 result = result + "</html>";
216 return result;
217 }
218
219 private String createStringOutputString(ArrayList<Lemma> lemmas) {
220 String result = "";
221 for (int i=0; i<lemmas.size(); i++) {
222 Lemma l = lemmas.get(i);
223 String lemmaName = l.getLemmaName();
224 result = result + lemmaName + " ";
225 }
226 result = result.substring(0, result.length() - 1); // without last blank
227 return result;
228 }
229
230
231 }