1 | package eu.clarin.sru.fcs.aggregator.search; |
---|
2 | |
---|
3 | import eu.clarin.sru.fcs.aggregator.util.LanguagesISO693; |
---|
4 | import eu.clarin.weblicht.wlfxb.io.WLDObjector; |
---|
5 | import eu.clarin.weblicht.wlfxb.io.WLFormatException; |
---|
6 | import eu.clarin.weblicht.wlfxb.md.xb.MetaData; |
---|
7 | import eu.clarin.weblicht.wlfxb.tc.xb.TextCorpusStored; |
---|
8 | import eu.clarin.weblicht.wlfxb.xb.WLData; |
---|
9 | import java.io.ByteArrayOutputStream; |
---|
10 | import java.io.IOException; |
---|
11 | import java.util.List; |
---|
12 | import java.util.logging.Level; |
---|
13 | import java.util.logging.Logger; |
---|
14 | import org.apache.poi.ss.usermodel.Cell; |
---|
15 | import org.apache.poi.ss.usermodel.CellStyle; |
---|
16 | import org.apache.poi.ss.usermodel.Font; |
---|
17 | import org.apache.poi.ss.usermodel.Row; |
---|
18 | import org.apache.poi.ss.usermodel.Sheet; |
---|
19 | import org.apache.poi.xssf.streaming.SXSSFWorkbook; |
---|
20 | |
---|
21 | /** |
---|
22 | * Utility for representing SearchResult data in different formats. |
---|
23 | * |
---|
24 | * @author Yana Panchenko |
---|
25 | */ |
---|
26 | public class Exports { |
---|
27 | |
---|
28 | private static final Logger LOGGER = Logger.getLogger(Exports.class.getName()); |
---|
29 | |
---|
30 | public static String getExportCSV(List<Result> resultsProcessed, String separator) { |
---|
31 | |
---|
32 | boolean noResult = true; |
---|
33 | StringBuilder csv = new StringBuilder(); |
---|
34 | if (resultsProcessed != null && !resultsProcessed.isEmpty()) { |
---|
35 | String[] headers = new String[]{ |
---|
36 | "LEFT CONTEXT", "KEYWORD", "RIGHT CONTEXT", "PID", "REFERENCE"}; |
---|
37 | for (String header : headers) { |
---|
38 | csv.append("\""); |
---|
39 | csv.append(header); |
---|
40 | csv.append("\""); |
---|
41 | csv.append(separator); |
---|
42 | } |
---|
43 | csv.append("\n"); |
---|
44 | |
---|
45 | for (Result result : resultsProcessed) { |
---|
46 | for (Kwic kwic : result.getKwics()) { |
---|
47 | csv.append("\""); |
---|
48 | csv.append(escapeQuotes(kwic.getLeft())); |
---|
49 | csv.append("\""); |
---|
50 | csv.append(separator); |
---|
51 | csv.append("\""); |
---|
52 | csv.append(escapeQuotes(kwic.getKeyword())); |
---|
53 | csv.append("\""); |
---|
54 | csv.append(separator); |
---|
55 | csv.append("\""); |
---|
56 | csv.append(escapeQuotes(kwic.getRight())); |
---|
57 | csv.append("\""); |
---|
58 | csv.append(separator); |
---|
59 | csv.append("\""); |
---|
60 | if (kwic.getPid() != null) { |
---|
61 | csv.append(escapeQuotes(kwic.getPid())); |
---|
62 | } |
---|
63 | csv.append("\""); |
---|
64 | csv.append(separator); |
---|
65 | csv.append("\""); |
---|
66 | if (kwic.getReference() != null) { |
---|
67 | csv.append(escapeQuotes(kwic.getReference())); |
---|
68 | } |
---|
69 | csv.append("\""); |
---|
70 | csv.append("\n"); |
---|
71 | noResult = false; |
---|
72 | } |
---|
73 | } |
---|
74 | } |
---|
75 | if (noResult) { |
---|
76 | return null; |
---|
77 | } else { |
---|
78 | return csv.toString(); |
---|
79 | } |
---|
80 | } |
---|
81 | |
---|
82 | private static CharSequence escapeQuotes(String text) { |
---|
83 | StringBuilder sb = new StringBuilder(); |
---|
84 | for (int i = 0; i < text.length(); i++) { |
---|
85 | char ch = text.charAt(i); |
---|
86 | if (ch == '"') { |
---|
87 | sb.append('"'); |
---|
88 | } |
---|
89 | sb.append(ch); |
---|
90 | } |
---|
91 | return sb; |
---|
92 | } |
---|
93 | |
---|
94 | public static byte[] getExportExcel(List<Result> resultsProcessed) throws ExportException { |
---|
95 | |
---|
96 | boolean noResult = true; |
---|
97 | SXSSFWorkbook workbook = null; |
---|
98 | ByteArrayOutputStream excelStream = new ByteArrayOutputStream(); |
---|
99 | if (resultsProcessed != null && !resultsProcessed.isEmpty()) { |
---|
100 | try { |
---|
101 | String[] headers = new String[]{ |
---|
102 | "LEFT CONTEXT", "KEYWORD", "RIGHT CONTEXT", "PID", "REFERENCE"}; |
---|
103 | |
---|
104 | workbook = new SXSSFWorkbook(); |
---|
105 | Sheet sheet = workbook.createSheet(); |
---|
106 | |
---|
107 | Font boldFont = workbook.createFont(); |
---|
108 | boldFont.setBoldweight(Font.BOLDWEIGHT_BOLD); |
---|
109 | |
---|
110 | // Header |
---|
111 | CellStyle headerStyle = workbook.createCellStyle(); |
---|
112 | headerStyle.setFont(boldFont); |
---|
113 | |
---|
114 | Row row = sheet.createRow(0); |
---|
115 | |
---|
116 | for (int j = 0; j < headers.length; ++j) { |
---|
117 | Cell cell = row.createCell(j, Cell.CELL_TYPE_STRING); |
---|
118 | cell.setCellValue(headers[j]); |
---|
119 | cell.setCellStyle(headerStyle); |
---|
120 | } |
---|
121 | |
---|
122 | // Body |
---|
123 | Cell cell; |
---|
124 | for (int k = 0; k < resultsProcessed.size(); k++) { |
---|
125 | Result result = resultsProcessed.get(k); |
---|
126 | List<Kwic> kwics = result.getKwics(); |
---|
127 | for (int i = 0; i < kwics.size(); i++) { |
---|
128 | Kwic kwic = kwics.get(i); |
---|
129 | row = sheet.createRow(k + i + 1); |
---|
130 | cell = row.createCell(0, Cell.CELL_TYPE_STRING); |
---|
131 | cell.setCellValue(kwic.getLeft()); |
---|
132 | cell = row.createCell(1, Cell.CELL_TYPE_STRING); |
---|
133 | cell.setCellValue(kwic.getKeyword()); |
---|
134 | cell = row.createCell(2, Cell.CELL_TYPE_STRING); |
---|
135 | cell.setCellValue(kwic.getRight()); |
---|
136 | if (kwic.getPid() != null) { |
---|
137 | cell = row.createCell(3, Cell.CELL_TYPE_STRING); |
---|
138 | cell.setCellValue(kwic.getPid()); |
---|
139 | } |
---|
140 | if (kwic.getReference() != null) { |
---|
141 | cell = row.createCell(3, Cell.CELL_TYPE_STRING); |
---|
142 | cell.setCellValue(kwic.getReference()); |
---|
143 | } |
---|
144 | noResult = false; |
---|
145 | } |
---|
146 | } |
---|
147 | workbook.write(excelStream); |
---|
148 | } catch (IOException ex) { |
---|
149 | LOGGER.log(Level.SEVERE, null, ex); |
---|
150 | throw new ExportException("Exception exporting Excel", ex); |
---|
151 | } finally { |
---|
152 | if (workbook != null) { |
---|
153 | workbook.dispose(); |
---|
154 | } |
---|
155 | } |
---|
156 | } |
---|
157 | if (noResult) { |
---|
158 | return null; |
---|
159 | } else { |
---|
160 | return excelStream.toByteArray(); |
---|
161 | } |
---|
162 | |
---|
163 | } |
---|
164 | |
---|
165 | public static byte[] getExportTCF(List<Result> resultsProcessed, |
---|
166 | String searchLanguage) throws ExportException { |
---|
167 | String text = getExportText(resultsProcessed); |
---|
168 | if (text == null || text.isEmpty()) { |
---|
169 | return null; |
---|
170 | } else { |
---|
171 | WLData data; |
---|
172 | MetaData md = new MetaData(); |
---|
173 | String languageCode = LanguagesISO693.getInstance().code_1ForCode_3(searchLanguage); |
---|
174 | TextCorpusStored tc = new TextCorpusStored(languageCode); |
---|
175 | tc.createTextLayer().addText(text); |
---|
176 | data = new WLData(md, tc); |
---|
177 | ByteArrayOutputStream os = new ByteArrayOutputStream(); |
---|
178 | try { |
---|
179 | WLDObjector.write(data, os); |
---|
180 | } catch (WLFormatException ex) { |
---|
181 | LOGGER.log(Level.SEVERE, "Error exporting TCF {0} {1}", new String[]{ex.getClass().getName(), ex.getMessage()}); |
---|
182 | throw new ExportException("Error exporting TCF", ex); |
---|
183 | } |
---|
184 | return os.toByteArray(); |
---|
185 | } |
---|
186 | } |
---|
187 | |
---|
188 | public static String getExportText(List<Result> resultsProcessed) { |
---|
189 | StringBuilder text = new StringBuilder(); |
---|
190 | if (resultsProcessed != null && !resultsProcessed.isEmpty()) { |
---|
191 | for (Result result : resultsProcessed) { |
---|
192 | for (Kwic kwic : result.getKwics()) { |
---|
193 | int i = kwic.getFragments().size() - 1; |
---|
194 | for (Kwic.TextFragment tf : kwic.getFragments()) { |
---|
195 | text.append(tf.text); |
---|
196 | if (i > 0) { |
---|
197 | text.append(" "); |
---|
198 | } |
---|
199 | i--; |
---|
200 | } |
---|
201 | text.append("\n"); |
---|
202 | } |
---|
203 | } |
---|
204 | } |
---|
205 | if (text.length() == 0) { |
---|
206 | return null; |
---|
207 | } else { |
---|
208 | return text.toString(); |
---|
209 | } |
---|
210 | } |
---|
211 | } |
---|