Révision 3013
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/ReferencePattern.java (revision 3013) | ||
|---|---|---|
| 32 | 32 |
import java.util.HashMap; |
| 33 | 33 |
import java.util.Iterator; |
| 34 | 34 |
import java.util.List; |
| 35 |
import java.util.regex.Matcher; |
|
| 36 |
import java.util.regex.Pattern; |
|
| 35 | 37 |
|
| 36 | 38 |
import org.apache.commons.lang.StringUtils; |
| 37 | 39 |
import org.txm.core.messages.TXMCoreMessages; |
| ... | ... | |
| 40 | 42 |
import org.txm.searchengine.cqp.corpus.StructuralUnit; |
| 41 | 43 |
import org.txm.searchengine.cqp.corpus.StructuralUnitProperty; |
| 42 | 44 |
import org.txm.searchengine.cqp.corpus.VirtualProperty; |
| 45 |
import org.txm.searchengine.cqp.corpus.WordProperty; |
|
| 43 | 46 |
import org.txm.utils.logger.Log; |
| 44 | 47 |
|
| 45 | 48 |
import com.google.gson.Gson; |
| ... | ... | |
| 64 | 67 |
*/ |
| 65 | 68 |
@SuppressWarnings("unlikely-arg-type")
|
| 66 | 69 |
public static ReferencePattern stringToReferencePattern(CQPCorpus corpus, String v) {
|
| 70 |
ReferencePattern ref = fromGSonString(corpus, v); |
|
| 71 |
if (ref == null) {
|
|
| 72 |
ref = fromPatternString(corpus, v); |
|
| 73 |
} |
|
| 74 |
if (ref == null) {
|
|
| 75 |
ref = fromTabulatedString(corpus, v); |
|
| 76 |
} |
|
| 77 |
return ref; |
|
| 78 |
} |
|
| 79 |
|
|
| 80 |
/** |
|
| 81 |
* |
|
| 82 |
* @param corpus |
|
| 83 |
* @param v |
|
| 84 |
* @return a list of word and structure properties read from a String |
|
| 85 |
*/ |
|
| 86 |
@SuppressWarnings("unlikely-arg-type")
|
|
| 87 |
public static ReferencePattern fromGSonString(CQPCorpus corpus, String v) {
|
|
| 67 | 88 |
try {
|
| 68 | 89 |
Gson gson = new Gson(); |
| 69 |
|
|
| 90 |
|
|
| 70 | 91 |
HashMap<?, ?> values = gson.fromJson(v, HashMap.class); |
| 71 | 92 |
Object oProperties = values.get("properties");
|
| 72 | 93 |
Object oFormat = values.get("format");
|
| ... | ... | |
| 104 | 125 |
return props; |
| 105 | 126 |
} |
| 106 | 127 |
catch (Exception e) {
|
| 107 |
System.out.println("Error: " + v + " property not found in " + corpus);
|
|
| 128 |
// System.out.println("Error: " + v + " property not found in " + corpus);
|
|
| 108 | 129 |
return null; |
| 109 | 130 |
} |
| 110 | 131 |
} |
| 111 | 132 |
|
| 112 | 133 |
/** |
| 134 |
* |
|
| 135 |
* @param corpus |
|
| 136 |
* @param v "${strct_prop} like string"
|
|
| 137 |
* @return |
|
| 138 |
*/ |
|
| 139 |
protected static ReferencePattern fromPatternString(CQPCorpus corpus, String v) {
|
|
| 140 |
try {
|
|
| 141 |
ReferencePattern props = new ReferencePattern(); |
|
| 142 |
if (v == null) return props; |
|
| 143 |
if (v.length() == 0) return props; |
|
| 144 |
|
|
| 145 |
Pattern pattern = Pattern.compile("\\$\\{([^\\}]+)\\}");
|
|
| 146 |
Matcher matcher = pattern.matcher(v); |
|
| 147 |
int nPropExpected = 0; |
|
| 148 |
while (matcher.find()) {
|
|
| 149 |
nPropExpected++; |
|
| 150 |
String gg = matcher.group(1).trim(); |
|
| 151 |
String[] split = gg.split("_");
|
|
| 152 |
if (split.length == 2) {
|
|
| 153 |
String sstruct = split[0]; |
|
| 154 |
String sprop = split[1]; |
|
| 155 |
StructuralUnit struct = corpus.getStructuralUnit(sstruct); |
|
| 156 |
if (struct != null) {
|
|
| 157 |
StructuralUnitProperty prop = struct.getProperty(sprop); |
|
| 158 |
if (prop != null) {
|
|
| 159 |
props.addProperty(prop); |
|
| 160 |
} |
|
| 161 |
} |
|
| 162 |
} |
|
| 163 |
else if (split.length == 1) {
|
|
| 164 |
String sprop = split[0]; |
|
| 165 |
WordProperty prop = corpus.getProperty(sprop); |
|
| 166 |
if (prop != null) {
|
|
| 167 |
props.addProperty(prop); |
|
| 168 |
} |
|
| 169 |
} |
|
| 170 |
} |
|
| 171 |
|
|
| 172 |
if (nPropExpected > 0 && props.getProperties().size() == nPropExpected) {
|
|
| 173 |
props.setPattern(matcher.replaceAll("%s"));
|
|
| 174 |
return props; |
|
| 175 |
} |
|
| 176 |
else {
|
|
| 177 |
return null; |
|
| 178 |
} |
|
| 179 |
} |
|
| 180 |
catch (Exception e) {
|
|
| 181 |
// System.out.println("Error: " + v + " property not found in " + corpus);
|
|
| 182 |
return null; |
|
| 183 |
} |
|
| 184 |
} |
|
| 185 |
|
|
| 186 |
protected static ReferencePattern fromTabulatedString(CQPCorpus corpus, String v) {
|
|
| 187 |
try {
|
|
| 188 |
ReferencePattern props = new ReferencePattern(); |
|
| 189 |
if (v == null) return props; |
|
| 190 |
if (v.length() == 0) return props; |
|
| 191 |
|
|
| 192 |
String[] split = v.split("\t");
|
|
| 193 |
for (String s : split) {
|
|
| 194 |
s = s.trim(); |
|
| 195 |
if (s.contains("_")) {
|
|
| 196 |
String[] split2 = s.split("_", 2);
|
|
| 197 |
StructuralUnit su = corpus.getStructuralUnit(split2[0]); |
|
| 198 |
if (su != null) {
|
|
| 199 |
Property prop = su.getProperty(split2[1]); |
|
| 200 |
if (prop != null) {
|
|
| 201 |
props.addProperty(prop); |
|
| 202 |
} |
|
| 203 |
} |
|
| 204 |
} |
|
| 205 |
else {
|
|
| 206 |
Property prop = corpus.getProperty(s); |
|
| 207 |
if (prop != null) {
|
|
| 208 |
props.addProperty(prop); |
|
| 209 |
} |
|
| 210 |
} |
|
| 211 |
} |
|
| 212 |
return props; |
|
| 213 |
} |
|
| 214 |
catch (Exception e) {
|
|
| 215 |
// System.out.println("Error: " + v + " property not found in " + corpus);
|
|
| 216 |
return null; |
|
| 217 |
} |
|
| 218 |
} |
|
| 219 |
|
|
| 220 |
/** |
|
| 113 | 221 |
* Instantiates a new reference pattern. |
| 114 | 222 |
*/ |
| 115 | 223 |
public ReferencePattern() {
|
| ... | ... | |
| 254 | 362 |
} |
| 255 | 363 |
if (title.length() == 0) {
|
| 256 | 364 |
title = TXMCoreMessages.common_reference; |
| 257 |
} else {
|
|
| 365 |
} |
|
| 366 |
else {
|
|
| 258 | 367 |
title = title.substring(0, title.length() - 2); |
| 259 | 368 |
} |
| 260 | 369 |
return title; |
| ... | ... | |
| 267 | 376 |
} |
| 268 | 377 |
|
| 269 | 378 |
Gson gson = new Gson(); |
| 270 |
HashMap<String, Object> values = new HashMap<String, Object>();
|
|
| 379 |
HashMap<String, Object> values = new HashMap<>(); |
|
| 271 | 380 |
values.put("properties", names);
|
| 272 | 381 |
values.put("format", pattern.getPattern());
|
| 273 | 382 |
return gson.toJson(values); |
Formats disponibles : Unified diff