1
|
package eu.dnetlib.functionality.index.query;
|
2
|
|
3
|
import java.io.IOException;
|
4
|
import java.util.ArrayList;
|
5
|
import java.util.HashMap;
|
6
|
import java.util.List;
|
7
|
import java.util.Map;
|
8
|
|
9
|
import org.apache.commons.logging.Log;
|
10
|
import org.apache.commons.logging.LogFactory;
|
11
|
import org.z3950.zing.cql.CQLBooleanNode;
|
12
|
import org.z3950.zing.cql.CQLNode;
|
13
|
import org.z3950.zing.cql.CQLParseException;
|
14
|
import org.z3950.zing.cql.CQLParser;
|
15
|
import org.z3950.zing.cql.CQLPrefixNode;
|
16
|
import org.z3950.zing.cql.CQLSortNode;
|
17
|
import org.z3950.zing.cql.CQLTermNode;
|
18
|
|
19
|
import com.google.common.base.Predicates;
|
20
|
import com.google.common.collect.Iterables;
|
21
|
import com.google.common.collect.Lists;
|
22
|
import com.google.common.collect.Maps;
|
23
|
|
24
|
import eu.dnetlib.miscutils.collections.MappedCollection;
|
25
|
import eu.dnetlib.miscutils.functional.UnaryFunction;
|
26
|
|
27
|
/**
|
28
|
* Use this class to cleanup a CQL tree and obtain all the options
|
29
|
*
|
30
|
* @author marko & claudio
|
31
|
*
|
32
|
*/
|
33
|
public class Pruner {
|
34
|
private static final Log log = LogFactory.getLog(Pruner.class); // NOPMD by marko on 11/24/08 5:02 PM
|
35
|
|
36
|
/**
|
37
|
* All options have to be in this namespace.
|
38
|
*/
|
39
|
public static final String DNET_URI = "NAMESPACE";
|
40
|
|
41
|
private String optionUri = DNET_URI;
|
42
|
|
43
|
/**
|
44
|
* Helper method, parse a given CQL string.
|
45
|
*
|
46
|
* @param cqlQuery
|
47
|
* @return
|
48
|
* @throws CQLParseException
|
49
|
* @throws IOException
|
50
|
*/
|
51
|
CQLNode parse(final String cqlQuery) throws CQLParseException, IOException {
|
52
|
return new CQLParser().parse(cqlQuery);
|
53
|
}
|
54
|
|
55
|
class Result {
|
56
|
private CQLNode node;
|
57
|
private List<String> options;
|
58
|
|
59
|
public Result(final CQLNode node, final List<String> options) {
|
60
|
super();
|
61
|
this.node = node;
|
62
|
this.options = options;
|
63
|
}
|
64
|
|
65
|
public Result(final CQLNode node, final Iterable<String> concat) {
|
66
|
this.node = node;
|
67
|
this.options = Lists.newArrayList(concat);
|
68
|
}
|
69
|
|
70
|
public CQLNode getNode() {
|
71
|
return node;
|
72
|
}
|
73
|
|
74
|
public void setNode(final CQLNode node) {
|
75
|
this.node = node;
|
76
|
}
|
77
|
|
78
|
public List<String> getOptions() {
|
79
|
return options;
|
80
|
}
|
81
|
|
82
|
public void setOptions(final List<String> options) {
|
83
|
this.options = options;
|
84
|
}
|
85
|
|
86
|
public Map<String, List<String>> getOptionMap() {
|
87
|
Map<String, List<String>> res = new HashMap<String, List<String>>();
|
88
|
for (String opt : options) {
|
89
|
String[] k = opt.split("=");
|
90
|
List<String> l = res.get(k[0]);
|
91
|
if(l == null)
|
92
|
l = new ArrayList<String>();
|
93
|
l.add(k[1]);
|
94
|
res.put(k[0], l);
|
95
|
}
|
96
|
return res;
|
97
|
}
|
98
|
}
|
99
|
|
100
|
/**
|
101
|
* Remove all options from a given CQL AST and return all the options.
|
102
|
*
|
103
|
* The CQL tree is modified.
|
104
|
*
|
105
|
* @param root
|
106
|
* cql tree
|
107
|
* @return pair containing a new root node and a list of options
|
108
|
*/
|
109
|
public Result prune(final CQLNode root) {
|
110
|
return prune(new HashMap<String, String>(), root);
|
111
|
}
|
112
|
|
113
|
/**
|
114
|
* Actual recursive implementation, dispatches the implementation to the appropriate overloaded method.
|
115
|
*
|
116
|
* @param prefixes
|
117
|
* @param root
|
118
|
* @return
|
119
|
*/
|
120
|
public Result prune(final Map<String, String> prefixes, final CQLNode root) {
|
121
|
|
122
|
if (root instanceof CQLBooleanNode)
|
123
|
return prune(prefixes, (CQLBooleanNode) root);
|
124
|
|
125
|
if (root instanceof CQLPrefixNode)
|
126
|
return prune(prefixes, (CQLPrefixNode) root);
|
127
|
|
128
|
if (root instanceof CQLSortNode)
|
129
|
return prune(prefixes, (CQLSortNode) root);
|
130
|
|
131
|
return new Result(root, new ArrayList<String>());
|
132
|
}
|
133
|
|
134
|
/**
|
135
|
* If the current node is a cql "sort" node, just return the inner subtree.
|
136
|
*
|
137
|
* @param prefixes
|
138
|
* @param node
|
139
|
* @return
|
140
|
*/
|
141
|
public Result prune(final Map<String, String> prefixes, final CQLSortNode node) {
|
142
|
Result res = prune(prefixes, node.subtree);
|
143
|
node.subtree = res.getNode();
|
144
|
res.setNode(node);
|
145
|
return res;
|
146
|
}
|
147
|
|
148
|
/**
|
149
|
* If the current node is a cql "prefix" node, add his namespace declaration to the current list of namespaces and
|
150
|
* return the pruned inner subtree.
|
151
|
*
|
152
|
* If the prefix node contains only one single option element, we have to return null. (TODO: perhaps there is a
|
153
|
* better solution).
|
154
|
*
|
155
|
* @param prefixes
|
156
|
* @param node
|
157
|
* @return
|
158
|
*/
|
159
|
public Result prune(final Map<String, String> prefixes, final CQLPrefixNode node) {
|
160
|
final HashMap<String, String> subPrefixes = Maps.newHashMap(prefixes);
|
161
|
subPrefixes.put(node.prefix.name, node.prefix.identifier);
|
162
|
|
163
|
if (isOption(subPrefixes, node.subtree))
|
164
|
return new Result(null, Lists.newArrayList(getOption(node.subtree)));
|
165
|
|
166
|
boolean pruneThisPrefix = node.prefix.identifier.equals(optionUri);
|
167
|
if(pruneThisPrefix)
|
168
|
return prune(subPrefixes, node.subtree);
|
169
|
|
170
|
Result res = prune(subPrefixes, node.subtree);
|
171
|
node.subtree = res.getNode();
|
172
|
res.setNode(node);
|
173
|
return res;
|
174
|
|
175
|
}
|
176
|
|
177
|
/**
|
178
|
* boolean prunes are handled in the prune(prefix, node, left, right).
|
179
|
*
|
180
|
* @param prefixes
|
181
|
* @param node
|
182
|
* @return
|
183
|
*/
|
184
|
public Result prune(final Map<String, String> prefixes, final CQLBooleanNode node) {
|
185
|
return prune(prefixes, node, node.left, node.right);
|
186
|
}
|
187
|
|
188
|
/**
|
189
|
* Detects if a left or right side of a boolean node is a option term, and returns the other side (recursively
|
190
|
* pruned). It also returns the accumulated options along the way.
|
191
|
*
|
192
|
* @param prefixes
|
193
|
* @param bool
|
194
|
* @param left
|
195
|
* @param right
|
196
|
* @return
|
197
|
*/
|
198
|
public Result prune(final Map<String, String> prefixes, final CQLBooleanNode bool, final CQLNode left, final CQLNode right) {
|
199
|
|
200
|
if (isOption(prefixes, left) && isOption(prefixes, right)) {
|
201
|
List<Result> r = Lists.newArrayList(trimOption(prefixes, left, right), trimOption(prefixes, right, left));
|
202
|
|
203
|
return new Result(null, Iterables.concat(MappedCollection.map(Iterables.filter(r, Predicates.notNull()),
|
204
|
new UnaryFunction<Iterable<String>, Result>() {
|
205
|
@Override
|
206
|
public Iterable<String> evaluate(Result res) {
|
207
|
return res.getOptions();
|
208
|
}
|
209
|
})));
|
210
|
}
|
211
|
|
212
|
Result res = anyNotNull(trimOption(prefixes, left, right), trimOption(prefixes, right, left));
|
213
|
|
214
|
if (res != null)
|
215
|
return res;
|
216
|
|
217
|
final Result leftResult = prune(prefixes, left);
|
218
|
final Result rightResult = prune(prefixes, right);
|
219
|
|
220
|
bool.left = leftResult.getNode();
|
221
|
bool.right = rightResult.getNode();
|
222
|
return new Result(clean(bool), Iterables.concat(leftResult.getOptions(), rightResult.getOptions()));
|
223
|
}
|
224
|
|
225
|
public <T> T anyNotNull(T a, T b) {
|
226
|
if (a != null)
|
227
|
return a;
|
228
|
return b;
|
229
|
}
|
230
|
|
231
|
/**
|
232
|
* Trims an option from a boolean node if one if it's sides is an option term.
|
233
|
*
|
234
|
* Intended to be used once for each sides and then swap.
|
235
|
*
|
236
|
* @param prefixes
|
237
|
* @param a
|
238
|
* @param b
|
239
|
* @return
|
240
|
*/
|
241
|
public Result trimOption(final Map<String, String> prefixes, final CQLNode a, final CQLNode b) {
|
242
|
log.debug("trim option?" + prefixes + " a " + a.toCQL());
|
243
|
if (isOption(prefixes, a)) {
|
244
|
log.debug("IS OPTION...");
|
245
|
return trimOption(prefixes, prefixFromOption(a), getOption(a), b);
|
246
|
}
|
247
|
log.debug("IS NOT OPTION");
|
248
|
return null;
|
249
|
}
|
250
|
|
251
|
/**
|
252
|
* prune(prefixes, bool, left, right) uses this helper method to do the dirty job:
|
253
|
*
|
254
|
* we have to detect if a term node is a term option node. by checking the namespace uri associated with the term
|
255
|
* prefix according the the current namespace prefix scope (held in prefixes, which is passed down recursively by
|
256
|
* copy).
|
257
|
*
|
258
|
* @param prefixes
|
259
|
* @param ns
|
260
|
* @param o
|
261
|
* @param subtree
|
262
|
* @return
|
263
|
*/
|
264
|
public Result trimOption(final Map<String, String> prefixes, final String ns, final String o, final CQLNode subtree) {
|
265
|
log.debug("trimming " + prefixes + " ns " + ns + " o " + o);
|
266
|
|
267
|
final String namespaceUri = prefixes.get(ns);
|
268
|
|
269
|
if (!optionUri.equals(namespaceUri)) {
|
270
|
return null;
|
271
|
}
|
272
|
|
273
|
final Result res = prune(prefixes, subtree);
|
274
|
return new Result(res.getNode(), Iterables.concat(Lists.newArrayList(o), res.getOptions()));
|
275
|
}
|
276
|
|
277
|
/**
|
278
|
* Drop a boolean node (and, or etc) if one of the sides has been dropped.
|
279
|
*
|
280
|
* @param bool
|
281
|
* @return
|
282
|
*/
|
283
|
private CQLNode clean(final CQLBooleanNode bool) {
|
284
|
if (bool.left == null)
|
285
|
return bool.right;
|
286
|
if (bool.right == null)
|
287
|
return bool.left;
|
288
|
return bool;
|
289
|
}
|
290
|
|
291
|
////////////////// helpers
|
292
|
|
293
|
public String getOption(final CQLNode node) {
|
294
|
return indexFromOption(node) + "=" + termFromOption(node);
|
295
|
}
|
296
|
|
297
|
private String indexFromOption(final CQLNode node) {
|
298
|
return ((CQLTermNode) node).getIndex().replaceAll("[a-z]*\\.(.+)", "$1");
|
299
|
}
|
300
|
|
301
|
private String termFromOption(final CQLNode node) {
|
302
|
return ((CQLTermNode) node).getTerm();
|
303
|
}
|
304
|
|
305
|
public String prefixFromOption(final String option) {
|
306
|
return option.replaceAll("([a-z]*)\\..+", "$1");
|
307
|
}
|
308
|
|
309
|
public String prefixFromOption(final CQLNode node) {
|
310
|
if (node instanceof CQLTermNode)
|
311
|
return prefixFromOption(((CQLTermNode) node).getIndex());
|
312
|
|
313
|
return null;
|
314
|
}
|
315
|
|
316
|
public boolean isOption(final Map<String, String> prefixes, final String option) {
|
317
|
return prefixes.containsKey(prefixFromOption(option)) && prefixes.get(prefixFromOption(option)).equals(getOptionUri());
|
318
|
}
|
319
|
|
320
|
public boolean isOption(final Map<String, String> prefixes, final CQLNode node) {
|
321
|
if (node instanceof CQLTermNode)
|
322
|
return isOption(prefixes, ((CQLTermNode) node).getIndex());
|
323
|
|
324
|
return false;
|
325
|
}
|
326
|
|
327
|
public String getOptionUri() {
|
328
|
return optionUri;
|
329
|
}
|
330
|
|
331
|
public void setOptionUri(String optionUri) {
|
332
|
this.optionUri = optionUri;
|
333
|
}
|
334
|
}
|