Revision 31954
Added by Alessia Bardi over 9 years ago
| TermNode.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.functionality.index.parse; |
| 2 | 2 |
|
| 3 |
import java.text.ParseException; |
|
| 4 |
import java.text.SimpleDateFormat; |
|
| 3 | 5 |
import java.util.List; |
| 4 | 6 |
import java.util.Map; |
| 5 | 7 |
import java.util.Map.Entry; |
| 6 | 8 |
import java.util.StringTokenizer; |
| 7 | 9 |
|
| 8 |
import org.apache.lucene.queryparser.classic.QueryParser; |
|
| 10 |
import org.apache.lucene.queryparser.classic.QueryParserBase;
|
|
| 9 | 11 |
|
| 10 | 12 |
import com.google.common.collect.BiMap; |
| 11 | 13 |
import com.google.common.collect.HashBiMap; |
| ... | ... | |
| 14 | 16 |
public class TermNode extends Node {
|
| 15 | 17 |
|
| 16 | 18 |
public static final String dnetDefaultField = "__all"; |
| 17 |
|
|
| 19 |
|
|
| 18 | 20 |
private String field; |
| 19 | 21 |
private Relation rel; |
| 20 | 22 |
private String value; |
| 21 | 23 |
private Map<String, List<String>> options = Maps.newHashMap(); |
| 22 | 24 |
private BiMap<String, String> aliases = HashBiMap.create(); |
| 23 | 25 |
private Map<String, String> weights = Maps.newHashMap(); |
| 24 |
|
|
| 25 |
public TermNode(String field, Relation rel, String value) {
|
|
| 26 |
private SimpleDateFormat simpleDateFormat = new SimpleDateFormat("yyyy-MM-dd");
|
|
| 27 |
|
|
| 28 |
public TermNode(final String field, final Relation rel, final String value) {
|
|
| 26 | 29 |
this.field = field; |
| 27 | 30 |
this.rel = rel; |
| 28 | 31 |
this.value = value; |
| 32 |
simpleDateFormat.setLenient(false); |
|
| 29 | 33 |
} |
| 30 |
|
|
| 31 |
public TermNode(String field, Relation rel, String value, Map<String, List<String>> options, BiMap<String, String> aliases, Map<String, String> weights) {
|
|
| 34 |
|
|
| 35 |
public TermNode(final String field, final Relation rel, final String value, final Map<String, List<String>> options, final BiMap<String, String> aliases, |
|
| 36 |
final Map<String, String> weights) {
|
|
| 32 | 37 |
this(field, rel, value); |
| 33 | 38 |
this.options = options; |
| 34 | 39 |
this.aliases = aliases; |
| 35 | 40 |
this.weights = weights; |
| 36 |
}
|
|
| 41 |
} |
|
| 37 | 42 |
|
| 38 | 43 |
@Override |
| 39 | 44 |
public String toString() {
|
| ... | ... | |
| 44 | 49 |
public String toLucene() {
|
| 45 | 50 |
StringTokenizer termTokenizer = new StringTokenizer(value, " "); |
| 46 | 51 |
StringTokenizer weightTokenizer = new StringTokenizer(value, " "); |
| 47 |
|
|
| 52 |
|
|
| 48 | 53 |
switch (rel) {
|
| 49 |
case EXACT: |
|
| 50 |
return getFieldName() + ":" + "\"" + value + "\"" + weight() + " " + expand(value); |
|
| 51 |
case ALL: |
|
| 52 |
return "(" + handleTokens(termTokenizer, "+") + " " + expandTokens(weightTokenizer) + ")";
|
|
| 53 |
case EQUAL: |
|
| 54 |
case ANY: |
|
| 55 |
return "(" + handleTokens(termTokenizer, "") + " " + expandTokens(weightTokenizer) + ")";
|
|
| 56 |
case NOT: |
|
| 57 |
return "-" + field + ":" + "\"" + value + "\""; |
|
| 58 |
case LT: |
|
| 59 |
return field + ":" + "{* TO " + value + "}" + weight();
|
|
| 60 |
case GT: |
|
| 61 |
return field + ":" + "{" + value + " TO *}" + weight();
|
|
| 62 |
case LTE: |
|
| 63 |
return field + ":" + "[* TO " + value + "]" + weight(); |
|
| 64 |
case GTE: |
|
| 65 |
return field + ":" + "[" + value + " TO *]" + weight(); |
|
| 66 |
case WITHIN: |
|
| 67 |
String lower = checkDate(value.split(" ")[0]);
|
|
| 68 |
String upper = checkDate(value.split(" ")[1]);
|
|
| 69 |
return field + ":[" + lower + " TO " + upper + "]" + weight(); |
|
| 70 |
default: |
|
| 71 |
throw new RuntimeException("unable to serialize: " + toString());
|
|
| 54 |
case EXACT: |
|
| 55 |
return getFieldName() + ":" + "\"" + value + "\"" + weight() + " " + expand(value); |
|
| 56 |
case ALL: |
|
| 57 |
return "(" + handleTokens(termTokenizer, "+") + " " + expandTokens(weightTokenizer) + ")";
|
|
| 58 |
case EQUAL: |
|
| 59 |
case ANY: |
|
| 60 |
return "(" + handleTokens(termTokenizer, "") + " " + expandTokens(weightTokenizer) + ")";
|
|
| 61 |
case NOT: |
|
| 62 |
return "-" + field + ":" + "\"" + value + "\""; |
|
| 63 |
case LT: |
|
| 64 |
if (isDate(value)) {
|
|
| 65 |
value = checkDate(value); |
|
| 66 |
} |
|
| 67 |
return field + ":" + "{* TO " + value + "}" + weight();
|
|
| 68 |
case GT: |
|
| 69 |
if (isDate(value)) {
|
|
| 70 |
value = checkDate(value); |
|
| 71 |
} |
|
| 72 |
return field + ":" + "{" + value + " TO *}" + weight();
|
|
| 73 |
case LTE: |
|
| 74 |
if (isDate(value)) {
|
|
| 75 |
value = checkDate(value); |
|
| 76 |
} |
|
| 77 |
return field + ":" + "[* TO " + value + "]" + weight(); |
|
| 78 |
case GTE: |
|
| 79 |
if (isDate(value)) {
|
|
| 80 |
value = checkDate(value); |
|
| 81 |
} |
|
| 82 |
return field + ":" + "[" + value + " TO *]" + weight(); |
|
| 83 |
case WITHIN: |
|
| 84 |
String lowerValue = value.split(" ")[0];
|
|
| 85 |
String upperValue = value.split(" ")[1];
|
|
| 86 |
if (isDate(lowerValue)) {
|
|
| 87 |
lowerValue = checkDate(lowerValue); |
|
| 88 |
} |
|
| 89 |
if (isDate(upperValue)) {
|
|
| 90 |
upperValue = checkDate(upperValue); |
|
| 91 |
} |
|
| 92 |
return field + ":[" + lowerValue + " TO " + upperValue + "]" + weight(); |
|
| 93 |
default: |
|
| 94 |
throw new RuntimeException("unable to serialize: " + toString());
|
|
| 72 | 95 |
} |
| 73 | 96 |
} |
| 74 | 97 |
|
| ... | ... | |
| 77 | 100 |
} |
| 78 | 101 |
|
| 79 | 102 |
private String weight() {
|
| 80 |
return weights != null && weights.get(field) != null ? "^" + weights.get(field) : "";
|
|
| 103 |
return (weights != null) && (weights.get(field) != null) ? "^" + weights.get(field) : "";
|
|
| 81 | 104 |
} |
| 82 |
|
|
| 83 |
private String expandTokens(StringTokenizer tokenizer) {
|
|
| 105 |
|
|
| 106 |
private String expandTokens(final StringTokenizer tokenizer) {
|
|
| 84 | 107 |
String ret = ""; |
| 85 | 108 |
while (tokenizer.hasMoreTokens()) {
|
| 86 | 109 |
String token = tokenizer.nextToken(); |
| ... | ... | |
| 91 | 114 |
} |
| 92 | 115 |
return ret.trim(); |
| 93 | 116 |
} |
| 94 |
|
|
| 95 |
private String expand(String token) {
|
|
| 117 |
|
|
| 118 |
private String expand(final String token) {
|
|
| 96 | 119 |
String ret = ""; |
| 97 | 120 |
if (!weights.keySet().contains(field)) {
|
| 98 |
for(Entry<String, String> e : weights.entrySet()) {
|
|
| 121 |
for (Entry<String, String> e : weights.entrySet()) {
|
|
| 99 | 122 |
ret += e.getKey() + ":\"" + checkEscaping(token) + "\"^" + e.getValue() + " "; |
| 100 | 123 |
} |
| 101 | 124 |
} |
| 102 | 125 |
return ret; |
| 103 |
}
|
|
| 126 |
} |
|
| 104 | 127 |
|
| 105 |
private String handleTokens(StringTokenizer tokenizer, String op) {
|
|
| 128 |
private String handleTokens(final StringTokenizer tokenizer, final String op) {
|
|
| 106 | 129 |
String ret = ""; |
| 107 | 130 |
while (tokenizer.hasMoreTokens()) {
|
| 108 | 131 |
String token = tokenizer.nextToken(); |
| ... | ... | |
| 110 | 133 |
} |
| 111 | 134 |
return ret.trim(); |
| 112 | 135 |
} |
| 113 |
|
|
| 136 |
|
|
| 114 | 137 |
private String checkEscaping(String token) {
|
| 115 | 138 |
boolean isWildcard = token.contains("*") || token.contains("?");
|
| 116 |
boolean isWildcardEnabled = (options.get("wildcard") != null && options.get("wildcard").contains("true")) || token.equals("*");
|
|
| 117 |
|
|
| 118 |
if (!(isWildcard & isWildcardEnabled) ) {
|
|
| 119 |
token = QueryParser.escape(token); |
|
| 139 |
boolean isWildcardEnabled = ((options.get("wildcard") != null) && options.get("wildcard").contains("true")) || token.equals("*");
|
|
| 140 |
|
|
| 141 |
if (!(isWildcard & isWildcardEnabled)) {
|
|
| 142 |
token = QueryParserBase.escape(token);
|
|
| 120 | 143 |
} |
| 121 | 144 |
return token; |
| 122 | 145 |
} |
| 123 | 146 |
|
| 147 |
private boolean isDate(final String aPossibleDate) {
|
|
| 148 |
try {
|
|
| 149 |
simpleDateFormat.parse(aPossibleDate); |
|
| 150 |
} catch (ParseException pe) {
|
|
| 151 |
return false; |
|
| 152 |
} |
|
| 153 |
return true; |
|
| 154 |
} |
|
| 155 |
|
|
| 124 | 156 |
private String checkDate(final String date) {
|
| 125 |
if (!date.endsWith("Z")) {
|
|
| 126 |
return date + "T00:00:00Z"; |
|
| 127 |
} |
|
| 157 |
if (!date.endsWith("Z")) return date + "T00:00:00Z";
|
|
| 128 | 158 |
return date; |
| 129 | 159 |
} |
| 130 |
|
|
| 160 |
|
|
| 131 | 161 |
public String getField() {
|
| 132 | 162 |
return field; |
| 133 | 163 |
} |
| ... | ... | |
| 138 | 168 |
|
| 139 | 169 |
public String getValue() {
|
| 140 | 170 |
return value; |
| 141 |
}
|
|
| 171 |
} |
|
| 142 | 172 |
|
| 143 | 173 |
} |
Also available in: Unified diff
Added check for date terms in queries. Solr requires dates to be in full ISO.