| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 1 | package de.ids_mannheim.korap.utils; |
| 2 | |
| margaretha | 7677aa4 | 2024-01-05 10:26:06 +0100 | [diff] [blame] | 3 | import java.util.Arrays; |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 4 | import java.util.HashMap; |
| margaretha | b786411 | 2017-06-29 16:36:13 +0200 | [diff] [blame] | 5 | import java.util.List; |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 6 | import java.util.Map; |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 7 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 8 | import com.fasterxml.jackson.databind.JsonNode; |
| 9 | import com.fasterxml.jackson.databind.node.ArrayNode; |
| 10 | import com.fasterxml.jackson.databind.node.ObjectNode; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 11 | |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 12 | import de.ids_mannheim.korap.exceptions.KustvaktException; |
| margaretha | b786411 | 2017-06-29 16:36:13 +0200 | [diff] [blame] | 13 | import de.ids_mannheim.korap.exceptions.StatusCodes; |
| 14 | import de.ids_mannheim.korap.query.serialize.CollectionQueryProcessor; |
| 15 | import de.ids_mannheim.korap.response.Notifications; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 16 | |
| 17 | /** |
| 18 | * convenience builder class for collection query |
| 19 | * |
| margaretha | b786411 | 2017-06-29 16:36:13 +0200 | [diff] [blame] | 20 | * @author hanl, margaretha |
| 21 | * @date 29/06/2017 |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 22 | */ |
| 23 | public class KoralCollectionQueryBuilder { |
| 24 | |
| 25 | public enum EQ { |
| 26 | EQUAL, UNEQUAL |
| 27 | } |
| 28 | |
| 29 | private boolean verbose; |
| 30 | private JsonNode base; |
| 31 | private StringBuilder builder; |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 32 | private String mergeOperator; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 33 | |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 34 | public KoralCollectionQueryBuilder () { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 35 | this(false); |
| 36 | } |
| 37 | |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 38 | public KoralCollectionQueryBuilder (boolean verbose) { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 39 | this.verbose = verbose; |
| 40 | this.builder = new StringBuilder(); |
| 41 | this.base = null; |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 42 | this.mergeOperator = null; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 43 | } |
| 44 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 45 | /** |
| 46 | * raw method for field - value pair adding. Supports all |
| 47 | * operators (leq, geq, contains, etc.) |
| 48 | * |
| 49 | * @param field |
| 50 | * @param op |
| 51 | * @param value |
| 52 | * @return |
| 53 | */ |
| Michael Hanl | 9a24216 | 2016-06-03 11:18:06 +0200 | [diff] [blame] | 54 | public KoralCollectionQueryBuilder with (String field, String op, |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 55 | String value) { |
| Michael Hanl | 9a24216 | 2016-06-03 11:18:06 +0200 | [diff] [blame] | 56 | //String end = this.builder.substring(this.builder.length() - 4, |
| 57 | // this.builder.length() - 1); |
| 58 | //if (this.builder.length() != 0 |
| 59 | // && (!end.contains("&") | !end.contains("|"))) |
| 60 | // throw new RuntimeException("no join operator given!"); |
| 61 | this.with(field + op + value); |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 62 | return this; |
| 63 | } |
| 64 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 65 | /** |
| 66 | * element can be a more complex sub query like |
| 67 | * (textClass=freizeit & Attributes.CORPUS_SIGLE=WPD) |
| 68 | * |
| 69 | * @param query |
| 70 | * will be enclosed by parenthesis in order to make sub |
| 71 | * query |
| 72 | * element |
| 73 | * @return |
| 74 | */ |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 75 | public KoralCollectionQueryBuilder with (String query) { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 76 | if (!query.startsWith("(") && !query.endsWith(")")) |
| 77 | query = "(" + query + ")"; |
| 78 | this.builder.append(query); |
| 79 | return this; |
| 80 | } |
| 81 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 82 | public KoralCollectionQueryBuilder and () { |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 83 | if (this.builder.length() != 0) |
| 84 | this.builder.append(" & "); |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 85 | if (this.base != null && this.mergeOperator == null) |
| 86 | this.mergeOperator = "AND"; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 87 | return this; |
| 88 | } |
| 89 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 90 | public KoralCollectionQueryBuilder or () { |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 91 | if (this.builder.length() != 0) |
| 92 | this.builder.append(" | "); |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 93 | if (this.base != null && this.mergeOperator == null) |
| 94 | this.mergeOperator = "OR"; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 95 | return this; |
| 96 | } |
| 97 | |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 98 | public JsonNode rebaseCollection () throws KustvaktException { |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 99 | if (this.builder.length() == 0 && this.base == null) |
| 100 | return null; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 101 | |
| 102 | JsonNode request = null; |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 103 | if (this.builder.length() != 0) { |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 104 | CollectionQueryProcessor tree = new CollectionQueryProcessor( |
| 105 | this.verbose); |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 106 | tree.process(this.builder.toString()); |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 107 | if (tree.getErrors().size() > 0) { |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 108 | // legacy support |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 109 | for (List<Object> e : tree.getErrors()) { |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 110 | if (e.get(1) instanceof String[]) { |
| 111 | Notifications notif = new Notifications(); |
| 112 | int code = (int) e.get(0); |
| margaretha | b786411 | 2017-06-29 16:36:13 +0200 | [diff] [blame] | 113 | notif.addError(code, (String[]) e.get(1)); |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 114 | String notificationStr = notif.toJsonString(); |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 115 | throw new KustvaktException( |
| 116 | StatusCodes.SERIALIZATION_FAILED, |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 117 | notificationStr, true); |
| 118 | } |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 119 | else { |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 120 | break; |
| margaretha | b786411 | 2017-06-29 16:36:13 +0200 | [diff] [blame] | 121 | } |
| 122 | } |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 123 | // -- end of legacy support |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 124 | |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 125 | Map<String, Object> map = new HashMap<>(); |
| 126 | map.put("errors", tree.getErrors()); |
| 127 | String errors = JsonUtils.toJSON(map); |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 128 | throw new KustvaktException(StatusCodes.SERIALIZATION_FAILED, |
| margaretha | 2558a7c | 2019-02-18 16:48:54 +0100 | [diff] [blame] | 129 | errors, true); |
| margaretha | b786411 | 2017-06-29 16:36:13 +0200 | [diff] [blame] | 130 | } |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 131 | request = JsonUtils.valueToTree(tree.getRequestMap()); |
| 132 | } |
| 133 | |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 134 | if (this.base != null) { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 135 | // check that collection non empty |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 136 | JsonNode tmp = this.base.deepCopy(); |
| Michael Hanl | 9a24216 | 2016-06-03 11:18:06 +0200 | [diff] [blame] | 137 | if (request != null) |
| 138 | request = mergeWith(request); |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 139 | else |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 140 | request = tmp; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 141 | } |
| 142 | return request; |
| 143 | } |
| 144 | |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 145 | public JsonNode mergeWith (JsonNode node) { |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 146 | if (this.base != null) { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 147 | if (node != null) { |
| 148 | JsonNode tobase = node.at("/collection"); |
| 149 | JsonNode base = this.base.deepCopy(); |
| Michael Hanl | cb2d3f9 | 2016-06-02 17:34:06 +0200 | [diff] [blame] | 150 | JsonNode result = base.at("/collection"); |
| 151 | |
| 152 | if (result.isMissingNode() && !tobase.isMissingNode()) |
| 153 | result = tobase; |
| 154 | else if (result.isMissingNode() && tobase.isMissingNode()) |
| 155 | return base; |
| 156 | else { |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 157 | result = JsonBuilder |
| 158 | .buildDocGroup(this.mergeOperator != null |
| 159 | ? this.mergeOperator.toLowerCase() |
| 160 | : "and", result, tobase); |
| Michael Hanl | cb2d3f9 | 2016-06-02 17:34:06 +0200 | [diff] [blame] | 161 | } |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 162 | ((ObjectNode) base).put("collection", result); |
| 163 | return base; |
| 164 | } |
| Michael Hanl | cb2d3f9 | 2016-06-02 17:34:06 +0200 | [diff] [blame] | 165 | return this.base; |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 166 | } |
| Michael Hanl | cb2d3f9 | 2016-06-02 17:34:06 +0200 | [diff] [blame] | 167 | throw new RuntimeException("no query found to merge with!"); |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 168 | } |
| 169 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 170 | /** |
| 171 | * sets base query. All consequent queries are added to the first |
| 172 | * koral:docGroup within the collection base query |
| 173 | * If no group in base query, consequent queries are skipped. |
| 174 | * |
| 175 | * @param query |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 176 | * @throws KustvaktException |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 177 | */ |
| margaretha | 35e1ca2 | 2023-11-16 22:00:01 +0100 | [diff] [blame] | 178 | public KoralCollectionQueryBuilder setBaseQuery (String query) |
| 179 | throws KustvaktException { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 180 | this.base = JsonUtils.readTree(query); |
| 181 | return this; |
| 182 | } |
| Michael Hanl | 33829ec | 2016-05-28 17:03:38 +0200 | [diff] [blame] | 183 | |
| Michael Hanl | cedf721 | 2016-05-28 10:43:09 +0200 | [diff] [blame] | 184 | public KoralCollectionQueryBuilder setBaseQuery (JsonNode query) { |
| 185 | this.base = query; |
| 186 | return this; |
| 187 | } |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 188 | |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 189 | public String toJSON () throws KustvaktException { |
| Michael Hanl | 9a24216 | 2016-06-03 11:18:06 +0200 | [diff] [blame] | 190 | return JsonUtils.toJSON(rebaseCollection()); |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 191 | } |
| 192 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 193 | @Override |
| 194 | public String toString () { |
| 195 | return this.builder.toString(); |
| 196 | } |
| 197 | |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 198 | private static class JsonBuilder { |
| 199 | |
| 200 | public static ObjectNode buildDoc (String key, String value) { |
| 201 | ObjectNode node = JsonUtils.createObjectNode(); |
| 202 | node.put("@type", "koral:doc"); |
| 203 | // eq.equals(EQ.EQUAL) ? "match:eq" : "match:ne" |
| 204 | node.put("match", "match:eq"); |
| 205 | node.put("key", key); |
| 206 | node.put("value", value); |
| 207 | return node; |
| 208 | } |
| 209 | |
| margaretha | 894a7d7 | 2017-11-08 19:24:20 +0100 | [diff] [blame] | 210 | public static ObjectNode buildDocGroup (String op, |
| 211 | JsonNode ... groups) { |
| Michael Hanl | 173ba4b | 2016-05-24 20:25:05 +0200 | [diff] [blame] | 212 | ObjectNode node = JsonUtils.createObjectNode(); |
| 213 | node.put("@type", "koral:docGroup"); |
| 214 | node.put("operation", "operation:" + op); |
| 215 | ArrayNode ops = JsonUtils.createArrayNode(); |
| 216 | ops.addAll(Arrays.asList(groups)); |
| 217 | node.put("operands", ops); |
| 218 | return node; |
| 219 | } |
| 220 | |
| 221 | } |
| 222 | |
| 223 | } |