| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1 | import static org.junit.Assert.*; |
| 2 | |
| 3 | import org.junit.Test; |
| 4 | |
| 5 | import de.ids_mannheim.korap.query.serialize.PoliqarpPlusTree; |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 6 | import de.ids_mannheim.korap.util.QueryException; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 7 | |
| 8 | public class PoliqarpPlusTreeTest { |
| 9 | |
| 10 | PoliqarpPlusTree ppt; |
| 11 | String map; |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 12 | String expected; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 13 | String metaExpected; |
| 14 | String metaMap; |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 15 | String query; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 16 | |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 17 | private boolean equalsQueryContent(String res, String query) throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 18 | res = res.replaceAll(" ", ""); |
| 19 | ppt = new PoliqarpPlusTree(query); |
| 20 | String queryMap = ppt.getRequestMap().get("query").toString().replaceAll(" ", ""); |
| 21 | return res.equals(queryMap); |
| 22 | } |
| 23 | |
| Joachim Bingel | 8181263 | 2014-02-18 08:55:22 +0000 | [diff] [blame] | 24 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 25 | public void testContext() throws QueryException { |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 26 | String contextString = "http://ids-mannheim.de/ns/KorAP/json-ld/v0.2/context.jsonld"; |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 27 | ppt = new PoliqarpPlusTree("Test"); |
| 28 | assertEquals(contextString.replaceAll(" ", ""), ppt.getRequestMap().get("@context").toString().replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 29 | } |
| 30 | |
| 31 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 32 | public void testSingleTokens() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 33 | // [base=Mann] |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 34 | String token1 = "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 35 | assertTrue(equalsQueryContent(token1, "[base=Mann]")); |
| 36 | |
| 37 | // [orth!=Frau] |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 38 | String token2 = "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Frau, match=match:ne}}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 39 | assertTrue(equalsQueryContent(token2, "[orth!=Frau]")); |
| 40 | |
| 41 | // [!p=NN] |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 42 | query = "[!p=NN]"; |
| 43 | String token3 = "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:ne}}"; |
| 44 | ppt = new PoliqarpPlusTree(query); |
| 45 | map = ppt.getRequestMap().get("query").toString(); |
| 46 | assertEquals(token3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 47 | |
| 48 | // [!p!=NN] |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 49 | query = "[!p!=NN]"; |
| 50 | String token4 = "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:eq}}"; |
| 51 | ppt = new PoliqarpPlusTree(query); |
| 52 | map = ppt.getRequestMap().get("query").toString(); |
| 53 | assertEquals(token4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 54 | |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 55 | query = "[base=schland/x]"; |
| Joachim Bingel | a83f8cc | 2014-08-05 14:12:59 +0000 | [diff] [blame] | 56 | String token5 = "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=.*?schland.*?, match=match:eq, type=type:regex}}"; |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 57 | ppt = new PoliqarpPlusTree(query); |
| 58 | map = ppt.getRequestMap().get("query").toString(); |
| 59 | assertEquals(token5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 60 | } |
| 61 | |
| 62 | @Test |
| 63 | public void testValue() throws QueryException { |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 64 | query = "[mate/m=temp:pres]"; |
| 65 | String value1 = "{@type=korap:token, wrap={@type=korap:term, foundry=mate, layer=m, key=temp, value=pres, match=match:eq}}"; |
| 66 | ppt = new PoliqarpPlusTree(query); |
| 67 | map = ppt.getRequestMap().get("query").toString(); |
| 68 | assertEquals(value1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 69 | } |
| 70 | |
| 71 | @Test |
| Joachim Bingel | 41e112e | 2014-02-12 10:46:18 +0000 | [diff] [blame] | 72 | public void testRegex() throws QueryException { |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 73 | query = "[orth=\"M(a|ä)nn(er)?\"]"; |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 74 | String re1 = "{@type=korap:token, wrap={@type=korap:term, layer=orth, type=type:regex, key=M(a|ä)nn(er)?, match=match:eq}}"; |
| Joachim Bingel | 41e112e | 2014-02-12 10:46:18 +0000 | [diff] [blame] | 75 | ppt = new PoliqarpPlusTree(query); |
| 76 | map = ppt.getRequestMap().get("query").toString(); |
| 77 | assertEquals(re1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 78 | |
| 79 | query = "[orth=\"M(a|ä)nn(er)?\"/x]"; |
| 80 | String re2 = "{@type=korap:token, wrap={@type=korap:term, layer=orth, type=type:regex, key=.*?M(a|ä)nn(er)?.*?, match=match:eq}}"; |
| 81 | ppt = new PoliqarpPlusTree(query); |
| 82 | map = ppt.getRequestMap().get("query").toString(); |
| 83 | assertEquals(re2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 84 | |
| 85 | query = "\"M(a|ä)nn(er)?\"/x"; |
| 86 | String re3 = "{@type=korap:token, wrap={@type=korap:term, type=type:regex, layer=orth, key=.*?M(a|ä)nn(er)?.*?, match=match:eq}}"; |
| 87 | ppt = new PoliqarpPlusTree(query); |
| 88 | map = ppt.getRequestMap().get("query").toString(); |
| 89 | assertEquals(re3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 90 | |
| 91 | query = "schland/x"; |
| Joachim Bingel | a83f8cc | 2014-08-05 14:12:59 +0000 | [diff] [blame] | 92 | String re4 = "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=.*?schland.*?, match=match:eq, type=type:regex}}"; |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 93 | ppt = new PoliqarpPlusTree(query); |
| 94 | map = ppt.getRequestMap().get("query").toString(); |
| 95 | assertEquals(re4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 96 | |
| 97 | query = "\".*?Mann.*?\""; |
| 98 | String re5 = "{@type=korap:token, wrap={@type=korap:term, type=type:regex, layer=orth, key=.*?Mann.*?, match=match:eq}}"; |
| 99 | ppt = new PoliqarpPlusTree(query); |
| 100 | map = ppt.getRequestMap().get("query").toString(); |
| 101 | assertEquals(re5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | a83f8cc | 2014-08-05 14:12:59 +0000 | [diff] [blame] | 102 | |
| 103 | query = "z.B./x"; |
| 104 | String re6 = "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=.*?z\\.B\\..*?, match=match:eq, type=type:regex}}"; |
| 105 | ppt = new PoliqarpPlusTree(query); |
| 106 | map = ppt.getRequestMap().get("query").toString(); |
| 107 | assertEquals(re6.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 41e112e | 2014-02-12 10:46:18 +0000 | [diff] [blame] | 108 | } |
| 109 | |
| 110 | @Test |
| Joachim Bingel | 0207d5e | 2014-02-12 14:18:41 +0000 | [diff] [blame] | 111 | public void testCaseSensitivityFlag() throws QueryException { |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 112 | query="[orth=deutscher/i]"; |
| Joachim Bingel | 0207d5e | 2014-02-12 14:18:41 +0000 | [diff] [blame] | 113 | String cs1 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 114 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=deutscher, match=match:eq, caseInsensitive=true}}"; |
| Joachim Bingel | 0207d5e | 2014-02-12 14:18:41 +0000 | [diff] [blame] | 115 | ppt = new PoliqarpPlusTree(query); |
| 116 | map = ppt.getRequestMap().get("query").toString(); |
| 117 | assertEquals(cs1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 118 | |
| 119 | query="[orth=deutscher/i][orth=Bundestag]"; |
| 120 | String cs2 = |
| 121 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 122 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=deutscher, match=match:eq, caseInsensitive=true}}," + |
| 123 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Bundestag, match=match:eq}}" + |
| Joachim Bingel | 0207d5e | 2014-02-12 14:18:41 +0000 | [diff] [blame] | 124 | "]}"; |
| 125 | ppt = new PoliqarpPlusTree(query); |
| 126 | map = ppt.getRequestMap().get("query").toString(); |
| 127 | assertEquals(cs2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 128 | |
| 129 | query="deutscher/i"; |
| 130 | String cs3 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 131 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=deutscher, match=match:eq, caseInsensitive=true}}"; |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 132 | ppt = new PoliqarpPlusTree(query); |
| 133 | map = ppt.getRequestMap().get("query").toString(); |
| 134 | assertEquals(cs3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 135 | |
| 136 | query="deutscher/i Bundestag"; |
| 137 | String cs4 = |
| 138 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 139 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=deutscher, match=match:eq, caseInsensitive=true}}," + |
| 140 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Bundestag, match=match:eq}}" + |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 141 | "]}"; |
| 142 | ppt = new PoliqarpPlusTree(query); |
| 143 | map = ppt.getRequestMap().get("query").toString(); |
| 144 | assertEquals(cs4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 145 | |
| 146 | query="deutscher Bundestag/i"; |
| 147 | String cs5 = |
| 148 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 149 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=deutscher, match=match:eq}}," + |
| 150 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Bundestag, match=match:eq, caseInsensitive=true}}" + |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 151 | "]}"; |
| 152 | ppt = new PoliqarpPlusTree(query); |
| 153 | map = ppt.getRequestMap().get("query").toString(); |
| 154 | assertEquals(cs5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 0207d5e | 2014-02-12 14:18:41 +0000 | [diff] [blame] | 155 | } |
| 156 | |
| 157 | @Test |
| Joachim Bingel | 3a41a44 | 2014-07-22 12:16:16 +0000 | [diff] [blame] | 158 | public void testSpans() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 159 | // <s> |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 160 | String elem1 = "{@type=korap:span, key=s}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 161 | assertTrue(equalsQueryContent(elem1, "<s>")); |
| 162 | |
| 163 | // <vp> |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 164 | String elem2 = "{@type=korap:span, key=vp}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 165 | assertTrue(equalsQueryContent(elem2, "<vp>")); |
| Joachim Bingel | fc1fb35 | 2014-02-26 14:40:27 +0000 | [diff] [blame] | 166 | |
| 167 | // <cnx/c=vp> |
| 168 | query = "<c=vp>"; |
| 169 | String span3 = "{@type=korap:span, layer=c, key=vp}"; |
| 170 | ppt = new PoliqarpPlusTree(query); |
| 171 | map = ppt.getRequestMap().get("query").toString(); |
| 172 | assertEquals(span3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 173 | |
| 174 | // <cnx/c=vp> |
| 175 | query = "<cnx/c=vp>"; |
| 176 | String span4 = "{@type=korap:span, foundry=cnx, layer=c, key=vp}"; |
| 177 | ppt = new PoliqarpPlusTree(query); |
| 178 | map = ppt.getRequestMap().get("query").toString(); |
| 179 | assertEquals(span4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 3a41a44 | 2014-07-22 12:16:16 +0000 | [diff] [blame] | 180 | |
| 181 | // span negation |
| 182 | query = "<cnx/c!=vp>"; |
| 183 | expected = "{@type=korap:span, foundry=cnx, layer=c, key=vp, match=match:ne}"; |
| 184 | ppt = new PoliqarpPlusTree(query); |
| 185 | map = ppt.getRequestMap().get("query").toString(); |
| 186 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 187 | |
| 188 | // span negation |
| 189 | query = "<cnx/c!=vp>"; |
| 190 | expected = "{@type=korap:span, foundry=cnx, layer=c, key=vp, match=match:ne}"; |
| 191 | ppt = new PoliqarpPlusTree(query); |
| 192 | map = ppt.getRequestMap().get("query").toString(); |
| 193 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 194 | |
| 195 | query = "<cnx/c=vp class!=header>"; |
| 196 | expected = "{@type=korap:span, foundry=cnx, layer=c, key=vp, attr={@type=korap:term, key=class, value=header, match=match:ne}}"; |
| 197 | ppt = new PoliqarpPlusTree(query); |
| 198 | map = ppt.getRequestMap().get("query").toString(); |
| 199 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 200 | |
| 201 | query = "<cnx/c=vp !(class=header&id=7)>"; |
| 202 | expected = |
| 203 | "{@type=korap:span, foundry=cnx, layer=c, key=vp, attr=" + |
| 204 | "{@type=korap:termGroup, relation=relation:and, operands=[" + |
| 205 | "{@type=korap:term, key=class, value=header, match=match:ne}," + |
| 206 | "{@type=korap:term, key=id, value=7, match=match:ne}" + |
| 207 | "]}" + |
| 208 | "}"; |
| 209 | ppt = new PoliqarpPlusTree(query); |
| 210 | map = ppt.getRequestMap().get("query").toString(); |
| 211 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 212 | |
| 213 | query = "<cnx/c!=vp (class=header&id=7)>"; |
| 214 | expected = |
| 215 | "{@type=korap:span, foundry=cnx, layer=c, key=vp, match=match:ne, attr=" + |
| 216 | "{@type=korap:termGroup, relation=relation:and, operands=[" + |
| 217 | "{@type=korap:term, key=class, value=header, match=match:eq}," + |
| 218 | "{@type=korap:term, key=id, value=7, match=match:eq}" + |
| 219 | "]}" + |
| 220 | "}"; |
| 221 | ppt = new PoliqarpPlusTree(query); |
| 222 | map = ppt.getRequestMap().get("query").toString(); |
| 223 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 224 | |
| 225 | query = "<cnx/c=vp !!class=header>"; |
| 226 | expected = "{@type=korap:span, foundry=cnx, layer=c, key=vp, attr={@type=korap:term, key=class, value=header, match=match:eq}}"; |
| 227 | ppt = new PoliqarpPlusTree(query); |
| 228 | map = ppt.getRequestMap().get("query").toString(); |
| 229 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | a0ef08f | 2014-07-24 08:27:57 +0000 | [diff] [blame] | 230 | |
| 231 | query = "<cnx/c!=vp (foundry/class=header&id=7)>"; |
| 232 | expected = |
| 233 | "{@type=korap:span, foundry=cnx, layer=c, key=vp, match=match:ne, attr=" + |
| 234 | "{@type=korap:termGroup, relation=relation:and, operands=[" + |
| 235 | "{@type=korap:term, foundry=foundry, key=class, value=header, match=match:eq}," + |
| 236 | "{@type=korap:term, key=id, value=7, match=match:eq}" + |
| 237 | "]}" + |
| 238 | "}"; |
| 239 | ppt = new PoliqarpPlusTree(query); |
| 240 | map = ppt.getRequestMap().get("query").toString(); |
| 241 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 242 | } |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 243 | |
| 244 | @Test |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 245 | public void testDistances() throws QueryException { |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 246 | // [base=der][][base=Mann] |
| 247 | String et1 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 248 | "{@type=korap:group, operation=operation:sequence, " + |
| Joachim Bingel | 11d5b15 | 2014-02-11 21:33:47 +0000 | [diff] [blame] | 249 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 250 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 251 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 252 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 5a146cd | 2014-07-24 14:44:57 +0000 | [diff] [blame] | 253 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=2, max=2}, min=2, max=2}" + |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 254 | "]}"; |
| 255 | ppt = new PoliqarpPlusTree("[base=der][][base=Mann]"); |
| 256 | map = ppt.getRequestMap().get("query").toString(); |
| 257 | assertEquals(et1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 258 | |
| 259 | // [base=der][][][base=Mann] |
| 260 | String et2 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 261 | "{@type=korap:group, operation=operation:sequence, " + |
| Joachim Bingel | 11d5b15 | 2014-02-11 21:33:47 +0000 | [diff] [blame] | 262 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 263 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 264 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 265 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 5a146cd | 2014-07-24 14:44:57 +0000 | [diff] [blame] | 266 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=3, max=3}, min=3, max=3}" + |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 267 | "]}"; |
| Joachim Bingel | ffd65e3 | 2014-01-22 14:22:57 +0000 | [diff] [blame] | 268 | ppt = new PoliqarpPlusTree("[base=der][][][base=Mann]"); |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 269 | map = ppt.getRequestMap().get("query").toString(); |
| 270 | assertEquals(et2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 271 | |
| 272 | // [base=der][][]?[base=Mann] |
| 273 | String et3 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 274 | "{@type=korap:group, operation=operation:sequence, " + |
| Joachim Bingel | 11d5b15 | 2014-02-11 21:33:47 +0000 | [diff] [blame] | 275 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 276 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 277 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 278 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 5a146cd | 2014-07-24 14:44:57 +0000 | [diff] [blame] | 279 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=2, max=3}, min=2, max=3}" + |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 280 | "]}"; |
| 281 | ppt = new PoliqarpPlusTree("[base=der][][]?[base=Mann]"); |
| 282 | map = ppt.getRequestMap().get("query").toString(); |
| 283 | assertEquals(et3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 284 | |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 285 | |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 286 | |
| Joachim Bingel | ffd65e3 | 2014-01-22 14:22:57 +0000 | [diff] [blame] | 287 | |
| Joachim Bingel | 11d5b15 | 2014-02-11 21:33:47 +0000 | [diff] [blame] | 288 | // [base=der][]{2,5}[base=Mann][]?[][base=Frau] nested distances= |
| Joachim Bingel | ffd65e3 | 2014-01-22 14:22:57 +0000 | [diff] [blame] | 289 | String et5 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 290 | "{@type=korap:group, operation=operation:sequence," + |
| Joachim Bingel | 11d5b15 | 2014-02-11 21:33:47 +0000 | [diff] [blame] | 291 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 292 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 293 | "{@type=korap:group, operation=operation:sequence, " + |
| Joachim Bingel | 11d5b15 | 2014-02-11 21:33:47 +0000 | [diff] [blame] | 294 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 295 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}," + |
| 296 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Frau, match=match:eq}}" + |
| 297 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 5a146cd | 2014-07-24 14:44:57 +0000 | [diff] [blame] | 298 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=2, max=3}, min=2, max=3}" + |
| Joachim Bingel | ffd65e3 | 2014-01-22 14:22:57 +0000 | [diff] [blame] | 299 | "]}" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 300 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 5a146cd | 2014-07-24 14:44:57 +0000 | [diff] [blame] | 301 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=3, max=6}, min=3, max=6}" + |
| Joachim Bingel | ffd65e3 | 2014-01-22 14:22:57 +0000 | [diff] [blame] | 302 | "]}"; |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 303 | ppt = new PoliqarpPlusTree("[base=der][]{2,5}[base=Mann][]?[][base=Frau]"); |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 304 | map = ppt.getRequestMap().get("query").toString(); |
| 305 | assertEquals(et5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 306 | |
| Joachim Bingel | c15720f | 2014-05-07 14:37:23 +0000 | [diff] [blame] | 307 | // [base=der][]*[base=Mann] |
| 308 | String et6 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 309 | "{@type=korap:group, operation=operation:sequence, " + |
| Joachim Bingel | c15720f | 2014-05-07 14:37:23 +0000 | [diff] [blame] | 310 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 311 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 312 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 313 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 314 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=1}, min=1}" + |
| Joachim Bingel | c15720f | 2014-05-07 14:37:23 +0000 | [diff] [blame] | 315 | "]}"; |
| 316 | ppt = new PoliqarpPlusTree("[base=der][]*[base=Mann]"); |
| 317 | map = ppt.getRequestMap().get("query").toString(); |
| 318 | assertEquals(et6.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 319 | |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 320 | // [base=der][]+[base=Mann] |
| Joachim Bingel | c15720f | 2014-05-07 14:37:23 +0000 | [diff] [blame] | 321 | String et7 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 322 | "{@type=korap:group, operation=operation:sequence, " + |
| Joachim Bingel | c15720f | 2014-05-07 14:37:23 +0000 | [diff] [blame] | 323 | "operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 324 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 325 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 326 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 327 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=2}, min=2}" + |
| Joachim Bingel | c15720f | 2014-05-07 14:37:23 +0000 | [diff] [blame] | 328 | "]}"; |
| 329 | ppt = new PoliqarpPlusTree("[base=der][]+[base=Mann]"); |
| 330 | map = ppt.getRequestMap().get("query").toString(); |
| 331 | assertEquals(et7.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 332 | |
| 333 | // [base=der][]+[base=Mann] |
| 334 | String et8 = |
| 335 | "{@type=korap:group, operation=operation:sequence, " + |
| 336 | "operands=[" + |
| 337 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 338 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 339 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 340 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=2, max=103}, min=2, max=103}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 341 | "]}"; |
| 342 | ppt = new PoliqarpPlusTree("[base=der][]{1,102}[base=Mann]"); |
| 343 | map = ppt.getRequestMap().get("query").toString(); |
| 344 | assertEquals(et8.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 345 | |
| 346 | // [base=geht][base=der][]*[base=Mann] |
| 347 | String et9 = |
| 348 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 349 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=geht, match=match:eq}}," + |
| 350 | "{@type=korap:group, operation=operation:sequence, " + |
| 351 | "operands=[" + |
| 352 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 353 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 354 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 355 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=1}, min=1}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 356 | "]}" + |
| 357 | "]}"; |
| 358 | ppt = new PoliqarpPlusTree("[base=geht][base=der][]*[base=Mann]"); |
| 359 | map = ppt.getRequestMap().get("query").toString(); |
| 360 | assertEquals(et9.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 361 | |
| 362 | query = "[base=geht][base=der][]*[base=Mann][base=da]"; |
| 363 | expected = |
| 364 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 365 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=geht, match=match:eq}}," + |
| 366 | "{@type=korap:group, operation=operation:sequence, " + |
| 367 | "operands=[" + |
| 368 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 369 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 370 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}," + |
| 371 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=da, match=match:eq}}" + |
| 372 | "]}" + |
| 373 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 374 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=1}, min=1}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 375 | "]}" + |
| 376 | "]}"; |
| 377 | ppt = new PoliqarpPlusTree(query); |
| 378 | map = ppt.getRequestMap().get("query").toString(); |
| 379 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 380 | |
| 381 | query = "[base=geht][base=der][]*contains(<s>,<np>)"; |
| 382 | expected = |
| 383 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 384 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=geht, match=match:eq}}," + |
| 385 | "{@type=korap:group, operation=operation:sequence, " + |
| 386 | "operands=[" + |
| 387 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 388 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 389 | "{@type=korap:span, key=s}," + |
| 390 | "{@type=korap:span, key=np}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 391 | "], frame=frame:contains}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 392 | "], inOrder=true, distances=[" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 393 | "{@type=korap:distance, key=w, boundary={@type=korap:boundary, min=1}, min=1}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 394 | "]}" + |
| 395 | "]}"; |
| 396 | ppt = new PoliqarpPlusTree(query); |
| 397 | map = ppt.getRequestMap().get("query").toString(); |
| 398 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 87480d0 | 2014-01-17 14:07:46 +0000 | [diff] [blame] | 399 | } |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 400 | |
| 401 | @Test |
| Joachim Bingel | bd38636 | 2014-10-02 12:03:18 +0000 | [diff] [blame] | 402 | public void testDistancesWithClass() throws QueryException { |
| 403 | query = "[base=der]{1:[]}[base=Mann]"; |
| 404 | expected = |
| 405 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 406 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 407 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | bd38636 | 2014-10-02 12:03:18 +0000 | [diff] [blame] | 408 | "{@type=korap:token}" + |
| 409 | "]}," + |
| 410 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 411 | "]}"; |
| 412 | ppt = new PoliqarpPlusTree(query); |
| 413 | map = ppt.getRequestMap().get("query").toString(); |
| 414 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 415 | |
| 416 | query = "{1:[]}[base=der][base=Mann]"; |
| 417 | expected = |
| 418 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 419 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | bd38636 | 2014-10-02 12:03:18 +0000 | [diff] [blame] | 420 | "{@type=korap:token}" + |
| 421 | "]}," + |
| 422 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 423 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 424 | "]}"; |
| 425 | ppt = new PoliqarpPlusTree(query); |
| 426 | map = ppt.getRequestMap().get("query").toString(); |
| 427 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 428 | } |
| 429 | |
| 430 | @Test |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 431 | public void testLeadingTrailingEmptyTokens() throws QueryException { |
| 432 | // startswith(<s>, [][base=Mann] |
| 433 | String et1 = |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 434 | "{@type=korap:group, operation=operation:position, frames=[frames:startswith], operands=[" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 435 | "{@type=korap:span, key=s}," + |
| 436 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 437 | "{@type=korap:token}," + |
| 438 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 439 | "]}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 440 | "], frame=frame:startswith}"; |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 441 | ppt = new PoliqarpPlusTree("startswith(<s>, [][base=Mann])"); |
| 442 | map = ppt.getRequestMap().get("query").toString(); |
| 443 | assertEquals(et1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 444 | |
| 445 | query = "[][base=Mann]"; |
| 446 | expected = |
| 447 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 448 | "{@type=korap:token}," + |
| 449 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 450 | "]}"; |
| 451 | ppt = new PoliqarpPlusTree(query); |
| 452 | map = ppt.getRequestMap().get("query").toString(); |
| 453 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 454 | |
| 455 | query = "[][][base=Mann]"; |
| 456 | expected = |
| 457 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 458 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 459 | "{@type=korap:token}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 460 | "], boundary={@type=korap:boundary, min=2, max=2}, min=2, max=2}," + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 461 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 462 | "]}"; |
| 463 | ppt = new PoliqarpPlusTree(query); |
| 464 | map = ppt.getRequestMap().get("query").toString(); |
| 465 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 466 | |
| 467 | query = "[][]*[base=Mann]"; |
| 468 | expected = |
| 469 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 470 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 471 | "{@type=korap:token}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 472 | "], boundary={@type=korap:boundary, min=1}, min=1}," + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 473 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| 474 | "]}"; |
| 475 | ppt = new PoliqarpPlusTree(query); |
| 476 | map = ppt.getRequestMap().get("query").toString(); |
| 477 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 478 | |
| 479 | query = "[][]*[base=Mann][][]"; |
| 480 | expected = |
| 481 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 482 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 483 | "{@type=korap:token}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 484 | "], boundary={@type=korap:boundary, min=1}, min=1}," + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 485 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}," + |
| 486 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 487 | "{@type=korap:token}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 488 | "], boundary={@type=korap:boundary, min=2, max=2}, min=2, max=2}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 489 | "]}"; |
| 490 | ppt = new PoliqarpPlusTree(query); |
| 491 | map = ppt.getRequestMap().get("query").toString(); |
| 492 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 493 | |
| 494 | query = "[][]*contains(<s>, <np>)[][]"; |
| 495 | expected = |
| 496 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 497 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 498 | "{@type=korap:token}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 499 | "], boundary={@type=korap:boundary, min=1}, min=1}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 500 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 501 | "{@type=korap:span, key=s}," + |
| 502 | "{@type=korap:span, key=np}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 503 | "], frame=frame:contains}," + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 504 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 505 | "{@type=korap:token}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 506 | "], boundary={@type=korap:boundary, min=2, max=2}, min=2, max=2}" + |
| Joachim Bingel | 14239d8 | 2014-07-22 09:55:04 +0000 | [diff] [blame] | 507 | "]}"; |
| 508 | ppt = new PoliqarpPlusTree(query); |
| 509 | map = ppt.getRequestMap().get("query").toString(); |
| 510 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 511 | } |
| 512 | |
| 513 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 514 | public void testCoordinatedFields() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 515 | // [base=Mann&(cas=N|cas=A)] |
| 516 | String cof1 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 517 | "{@type=korap:token, wrap=" + |
| Joachim Bingel | 2980bda | 2014-04-04 12:09:46 +0000 | [diff] [blame] | 518 | "{@type=korap:termGroup, relation=relation:and, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 519 | "{@type=korap:term, layer=lemma, key=Mann, match=match:eq}," + |
| Joachim Bingel | 2980bda | 2014-04-04 12:09:46 +0000 | [diff] [blame] | 520 | "{@type=korap:termGroup, relation=relation:or, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 521 | "{@type=korap:term, layer=cas, key=N, match=match:eq}," + |
| 522 | "{@type=korap:term, layer=cas, key=A, match=match:eq}" + |
| Joachim Bingel | ee3b21d | 2014-02-12 12:34:59 +0000 | [diff] [blame] | 523 | "]}" + |
| 524 | "]}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 525 | "}"; |
| 526 | ppt = new PoliqarpPlusTree("[base=Mann&(cas=N|cas=A)]"); |
| 527 | map = ppt.getRequestMap().get("query").toString(); |
| 528 | assertEquals(cof1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Nils Diewald | 7d486c2 | 2013-12-13 16:32:18 +0000 | [diff] [blame] | 529 | |
| 530 | |
| 531 | assertEquals( |
| 532 | new PoliqarpPlusTree(" [ base=Mann & ( cas=N | cas=A)] ").getRequestMap().get("query").toString(), |
| 533 | new PoliqarpPlusTree("[base=Mann &(cas=N|cas=A)]").getRequestMap().get("query").toString() |
| 534 | ); |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 535 | |
| 536 | // [base=Mann&cas=N&gen=m] |
| 537 | String cof2 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 538 | "{@type=korap:token, wrap=" + |
| Joachim Bingel | 2980bda | 2014-04-04 12:09:46 +0000 | [diff] [blame] | 539 | "{@type=korap:termGroup, relation=relation:and, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 540 | "{@type=korap:term, layer=lemma, key=Mann, match=match:eq}," + |
| 541 | "{@type=korap:termGroup, relation=relation:and, operands=[" + |
| 542 | "{@type=korap:term, layer=cas, key=N, match=match:eq}," + |
| 543 | "{@type=korap:term, layer=gen, key=m, match=match:eq}" + |
| 544 | "]}" + |
| Joachim Bingel | ee3b21d | 2014-02-12 12:34:59 +0000 | [diff] [blame] | 545 | "]}" + |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 546 | "}"; |
| 547 | ppt = new PoliqarpPlusTree("[base=Mann&cas=N&gen=m]"); |
| 548 | map = ppt.getRequestMap().get("query").toString(); |
| 549 | assertEquals(cof2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 550 | } |
| 551 | |
| 552 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 553 | public void testOccurrence() throws QueryException { |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 554 | // [base=foo]* |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 555 | String occ1 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 556 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 557 | "], boundary={@type=korap:boundary, min=0}, min=0}"; |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 558 | ppt = new PoliqarpPlusTree("[base=foo]*"); |
| 559 | map = ppt.getRequestMap().get("query").toString(); |
| 560 | assertEquals(occ1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 561 | |
| 562 | // [base=foo]*[base=bar] |
| 563 | String occ2 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 564 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 565 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 566 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 567 | "], boundary={@type=korap:boundary, min=0}, min=0 }," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 568 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}" + |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 569 | "]}"; |
| 570 | ppt = new PoliqarpPlusTree("[base=foo]*[base=bar]"); |
| 571 | map = ppt.getRequestMap().get("query").toString(); |
| 572 | assertEquals(occ2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 573 | |
| 574 | // [base=bar][base=foo]* |
| 575 | String occ3 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 576 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 577 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}," + |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 578 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 579 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 580 | "], boundary={@type=korap:boundary, min=0}, min=0 }" + |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 581 | "]}"; |
| 582 | ppt = new PoliqarpPlusTree("[base=bar][base=foo]*"); |
| 583 | map = ppt.getRequestMap().get("query").toString(); |
| 584 | assertEquals(occ3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 585 | |
| 586 | // ([base=bar][base=foo])* |
| 587 | String occ4 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 588 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 589 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 590 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}," + |
| 591 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 592 | "]}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 593 | "], boundary={@type=korap:boundary, min=0}, min=0}" ; |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 594 | ppt = new PoliqarpPlusTree("([base=bar][base=foo])*"); |
| 595 | map = ppt.getRequestMap().get("query").toString(); |
| 596 | assertEquals(occ4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 597 | |
| 598 | // <s>([base=bar][base=foo])* |
| 599 | String occ5 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 600 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 601 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 602 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 603 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 604 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}," + |
| 605 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 606 | "]}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 607 | "], boundary={@type=korap:boundary, min=0}, min=0 }" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 608 | "]}" ; |
| 609 | ppt = new PoliqarpPlusTree("<s>([base=bar][base=foo])*"); |
| 610 | map = ppt.getRequestMap().get("query").toString(); |
| 611 | assertEquals(occ5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 612 | |
| 613 | // <s><np>([base=bar][base=foo])* |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 614 | String occ6 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 615 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 616 | "{@type=korap:span, key=s}," + |
| 617 | "{@type=korap:span, key=np}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 618 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 619 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 620 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}," + |
| 621 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 622 | "]}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 623 | "], boundary={@type=korap:boundary, min=0}, min=0 }" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 624 | "]}" ; |
| 625 | ppt = new PoliqarpPlusTree("<s><np>([base=bar][base=foo])*"); |
| 626 | map = ppt.getRequestMap().get("query").toString(); |
| 627 | assertEquals(occ6.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 628 | |
| 629 | // <s><np>([base=bar][base=foo])*[p=NN] |
| 630 | // comment: embedded sequence shouldn't really be here, but does not really hurt, either. (?) |
| 631 | // really hard to get this behaviour out of the PQPlus grammar... |
| 632 | String occ7 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 633 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 634 | "{@type=korap:span, key=s}," + |
| 635 | "{@type=korap:span, key=np}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 636 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 637 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 638 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}," + |
| 639 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 640 | "]}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 641 | "], boundary={@type=korap:boundary, min=0}, min=0 }," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 642 | "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 643 | "]}" ; |
| 644 | ppt = new PoliqarpPlusTree("<s><np>([base=bar][base=foo])*[p=NN]"); |
| 645 | map = ppt.getRequestMap().get("query").toString(); |
| 646 | assertEquals(occ7.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 647 | |
| 648 | // ([base=bar][base=foo])*[p=NN] |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 649 | String occ8 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 650 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 651 | "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 652 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 653 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=bar, match=match:eq}}," + |
| 654 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 655 | "]}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 656 | "], boundary={@type=korap:boundary, min=0}, min=0 }," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 657 | "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | b4da702 | 2013-12-09 23:17:24 +0000 | [diff] [blame] | 658 | "]}" ; |
| 659 | ppt = new PoliqarpPlusTree("([base=bar][base=foo])*[p=NN]"); |
| 660 | map = ppt.getRequestMap().get("query").toString(); |
| 661 | assertEquals(occ8.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 662 | |
| 663 | // [base=foo]+ |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 664 | String occ9 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 665 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 666 | "], boundary={@type=korap:boundary, min=1}, min=1}"; |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 667 | ppt = new PoliqarpPlusTree("[base=foo]+"); |
| 668 | map = ppt.getRequestMap().get("query").toString(); |
| 669 | assertEquals(occ9.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 670 | |
| 671 | // [base=foo]? |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 672 | String occ10 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 673 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 674 | "], boundary={@type=korap:boundary, min=0, max=1}, min=0, max=1}"; |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 675 | ppt = new PoliqarpPlusTree("[base=foo]?"); |
| 676 | map = ppt.getRequestMap().get("query").toString(); |
| 677 | assertEquals(occ10.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 678 | |
| 679 | // [base=foo]{2,5} |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 680 | String occ11 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 681 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 682 | "], boundary={@type=korap:boundary, min=2, max=5}, min=2, max=5}"; |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 683 | ppt = new PoliqarpPlusTree("[base=foo]{2,5}"); |
| 684 | map = ppt.getRequestMap().get("query").toString(); |
| 685 | assertEquals(occ11.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 2980bda | 2014-04-04 12:09:46 +0000 | [diff] [blame] | 686 | |
| 687 | // [base=foo]{2} |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 688 | String occ12 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 689 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 690 | "], boundary={@type=korap:boundary, min=2, max=2}, min=2, max=2}"; |
| Joachim Bingel | 2980bda | 2014-04-04 12:09:46 +0000 | [diff] [blame] | 691 | ppt = new PoliqarpPlusTree("[base=foo]{2}"); |
| 692 | map = ppt.getRequestMap().get("query").toString(); |
| 693 | assertEquals(occ12.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | b2aa1c7 | 2014-07-01 12:48:46 +0000 | [diff] [blame] | 694 | |
| 695 | // [base=foo]{2} |
| 696 | String occ13 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 697 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 698 | "], boundary={@type=korap:boundary, min=2}, min=2}"; |
| Joachim Bingel | b2aa1c7 | 2014-07-01 12:48:46 +0000 | [diff] [blame] | 699 | ppt = new PoliqarpPlusTree("[base=foo]{2,}"); |
| 700 | map = ppt.getRequestMap().get("query").toString(); |
| 701 | assertEquals(occ13.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 702 | |
| 703 | // [base=foo]{2} |
| 704 | String occ14 = "{@type=korap:group, operation=operation:repetition, operands=[" + |
| 705 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=foo, match=match:eq}}" + |
| Joachim Bingel | 6163156 | 2014-07-24 14:26:02 +0000 | [diff] [blame] | 706 | "], boundary={@type=korap:boundary, min=0, max=2}, min=0, max=2}"; |
| Joachim Bingel | b2aa1c7 | 2014-07-01 12:48:46 +0000 | [diff] [blame] | 707 | ppt = new PoliqarpPlusTree("[base=foo]{,2}"); |
| 708 | map = ppt.getRequestMap().get("query").toString(); |
| 709 | assertEquals(occ14.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 710 | } |
| 711 | |
| 712 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 713 | public void testTokenSequence() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 714 | // [base=Mann][orth=Frau] |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 715 | String seq1 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 716 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}, " + |
| 717 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Frau, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 718 | "]}"; |
| 719 | assertTrue(equalsQueryContent(seq1, "[base=Mann][orth=Frau]")); |
| 720 | |
| 721 | // [base=Mann][orth=Frau][p=NN] |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 722 | String seq2 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 723 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}, " + |
| 724 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Frau, match=match:eq}}, " + |
| 725 | "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 726 | "]}"; |
| 727 | assertTrue(equalsQueryContent(seq2, "[base=Mann][orth=Frau][p=NN]")); |
| 728 | } |
| 729 | |
| 730 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 731 | public void testDisjSegments() throws QueryException { |
| Joachim Bingel | 94a1ccd | 2013-12-10 10:37:29 +0000 | [diff] [blame] | 732 | // ([base=der]|[base=das])[base=Schild] |
| 733 | String disj1 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 734 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 735 | "{@type=korap:group, operation=operation:or, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 736 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 737 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=das, match=match:eq}}" + |
| Joachim Bingel | 94a1ccd | 2013-12-10 10:37:29 +0000 | [diff] [blame] | 738 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 739 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Schild, match=match:eq}}" + |
| Joachim Bingel | 94a1ccd | 2013-12-10 10:37:29 +0000 | [diff] [blame] | 740 | "]}"; |
| 741 | ppt = new PoliqarpPlusTree("([base=der]|[base=das])[base=Schild]"); |
| 742 | map = ppt.getRequestMap().get("query").toString(); |
| 743 | assertEquals(disj1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 744 | |
| 745 | // [base=Schild]([base=der]|[base=das]) |
| 746 | String disj2 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 747 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 748 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Schild, match=match:eq}}," + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 749 | "{@type=korap:group, operation=operation:or, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 750 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| 751 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=das, match=match:eq}}" + |
| Joachim Bingel | 94a1ccd | 2013-12-10 10:37:29 +0000 | [diff] [blame] | 752 | "]}" + |
| 753 | "]}"; |
| 754 | ppt = new PoliqarpPlusTree("[base=Schild]([base=der]|[base=das])"); |
| 755 | map = ppt.getRequestMap().get("query").toString(); |
| 756 | assertEquals(disj2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 757 | |
| 758 | // "([orth=der][base=katze])|([orth=eine][base=baum])" |
| 759 | String disj3 = |
| 760 | "{@type=korap:group, operation=operation:or, operands=[" + |
| 761 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 762 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| 763 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=katze, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 764 | "]}," + |
| 765 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 766 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=eine, match=match:eq}}," + |
| 767 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=baum, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 768 | "]}" + |
| 769 | "]}"; |
| 770 | ppt = new PoliqarpPlusTree("([orth=der][base=katze])|([orth=eine][base=baum])"); |
| 771 | map = ppt.getRequestMap().get("query").toString(); |
| 772 | assertEquals(disj3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 773 | |
| 774 | // "[orth=der][base=katze]|[orth=eine][base=baum]" |
| 775 | String disj4 = |
| 776 | "{@type=korap:group, operation=operation:or, operands=[" + |
| 777 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 778 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| 779 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=katze, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 780 | "]}," + |
| 781 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 782 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=eine, match=match:eq}}," + |
| 783 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=baum, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 784 | "]}" + |
| 785 | "]}"; |
| 786 | ppt = new PoliqarpPlusTree("[orth=der][base=katze]|[orth=eine][base=baum]"); |
| 787 | map = ppt.getRequestMap().get("query").toString(); |
| 788 | assertEquals(disj4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 789 | |
| 790 | PoliqarpPlusTree ppt1 = new PoliqarpPlusTree("[orth=der][base=katze]|[orth=eine][base=baum]"); |
| 791 | PoliqarpPlusTree ppt2 = new PoliqarpPlusTree("([orth=der][base=katze])|([orth=eine][base=baum])"); |
| 792 | assertEquals(ppt1.getRequestMap().toString(), ppt2.getRequestMap().toString()); |
| 793 | |
| 794 | // "[orth=der][base=katze]|[orth=der][base=hund]|[orth=der][base=baum]" |
| 795 | String disj5 = |
| 796 | "{@type=korap:group, operation=operation:or, operands=[" + |
| 797 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 798 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| 799 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=katze, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 800 | "]}," + |
| 801 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 802 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| 803 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=hund, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 804 | "]}," + |
| 805 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 806 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| 807 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=baum, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 808 | "]}" + |
| 809 | "]}"; |
| 810 | ppt = new PoliqarpPlusTree("[orth=der][base=katze]|[orth=der][base=hund]|[orth=der][base=baum]"); |
| 811 | map = ppt.getRequestMap().get("query").toString(); |
| 812 | assertEquals(disj5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 813 | |
| 814 | // [orth=der]([base=katze]|[base=hund]|[base=baum]) |
| 815 | String disj6 = |
| 816 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 817 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 818 | "{@type=korap:group, operation=operation:or, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 819 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=katze, match=match:eq}}," + |
| 820 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=hund, match=match:eq}}," + |
| 821 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=baum, match=match:eq}}" + |
| Joachim Bingel | 76b498d | 2014-06-06 12:06:46 +0000 | [diff] [blame] | 822 | "]}" + |
| 823 | "]}"; |
| 824 | ppt = new PoliqarpPlusTree("[orth=der]([base=katze]|[base=hund]|[base=baum])"); |
| 825 | map = ppt.getRequestMap().get("query").toString(); |
| 826 | assertEquals(disj6.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 94a1ccd | 2013-12-10 10:37:29 +0000 | [diff] [blame] | 827 | } |
| 828 | |
| 829 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 830 | public void testTokenElemSequence() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 831 | // [base=Mann]<vp> |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 832 | String seq1 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 833 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}, " + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 834 | "{@type=korap:span, key=vp}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 835 | "]}"; |
| 836 | assertTrue(equalsQueryContent(seq1, "[base=Mann]<vp>")); |
| 837 | |
| 838 | // <vp>[base=Mann] |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 839 | String seq2 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 840 | "{@type=korap:span, key=vp}, "+ |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 841 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}} " + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 842 | "]}"; |
| 843 | assertTrue(equalsQueryContent(seq2, "<vp>[base=Mann]")); |
| 844 | |
| 845 | // <vp>[base=Mann]<pp> |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 846 | String seq3 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 847 | "{@type=korap:span, key=vp}, "+ |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 848 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}, " + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 849 | "{@type=korap:span, key=pp} "+ |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 850 | "]}"; |
| 851 | assertTrue(equalsQueryContent(seq3, "<vp>[base=Mann]<pp>")); |
| 852 | } |
| 853 | |
| 854 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 855 | public void testElemSequence() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 856 | // <np><vp> |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 857 | String seq1 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 858 | "{@type=korap:span, key=np}," + |
| 859 | "{@type=korap:span, key=vp}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 860 | "]}"; |
| 861 | assertTrue(equalsQueryContent(seq1, "<np><vp>")); |
| 862 | |
| 863 | // <np><vp><pp> |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 864 | String seq2 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 865 | "{@type=korap:span, key=np}," + |
| 866 | "{@type=korap:span, key=vp}," + |
| 867 | "{@type=korap:span, key=pp}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 868 | "]}"; |
| 869 | assertTrue(equalsQueryContent(seq2, "<np><vp><pp>")); |
| 870 | } |
| 871 | |
| 872 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 873 | public void testClasses() throws QueryException { |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 874 | String query; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 875 | // {[base=Mann]} |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 876 | String cls1 = "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 877 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 878 | "]}"; |
| Joachim Bingel | 8c640e4 | 2014-02-07 16:20:47 +0000 | [diff] [blame] | 879 | ppt = new PoliqarpPlusTree("{[base=Mann]}"); |
| 880 | map = ppt.getRequestMap().get("query").toString(); |
| 881 | assertEquals(cls1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 882 | |
| 883 | // {[base=Mann][orth=Frau]} |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 884 | query = "{[base=Mann][orth=Frau]}"; |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 885 | String cls2 = "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 886 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 887 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}," + |
| 888 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Frau, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 889 | "]}" + |
| 890 | "]}"; |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 891 | ppt = new PoliqarpPlusTree(query); |
| 892 | map = ppt.getRequestMap().get("query").toString(); |
| 893 | assertEquals(cls2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 894 | |
| 895 | // [p=NN]{[base=Mann][orth=Frau]} |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 896 | String cls3 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 897 | "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:eq}}," + |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 898 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 899 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 900 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}," + |
| 901 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Frau, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 902 | "]}" + |
| 903 | "]}" + |
| 904 | "]}"; |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 905 | ppt = new PoliqarpPlusTree("[p=NN]{[base=Mann][orth=Frau]}"); |
| 906 | map = ppt.getRequestMap().get("query").toString(); |
| 907 | assertEquals(cls3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 908 | |
| 909 | // {[base=Mann][orth=Frau]}[p=NN] |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 910 | String cls4 = "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 911 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 912 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 913 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}," + |
| 914 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Frau, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 915 | "]}" + |
| 916 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 917 | "{@type=korap:token, wrap={@type=korap:term, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 918 | "]}"; |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 919 | ppt = new PoliqarpPlusTree("{[base=Mann][orth=Frau]}[p=NN]"); |
| 920 | map = ppt.getRequestMap().get("query").toString(); |
| 921 | assertEquals(cls4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 922 | |
| 923 | // {2:{1:[tt/p=ADJA]}[mate/p=NN]}" |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 924 | String cls5 = "{@type=korap:group, operation=operation:class, class=2, classOut=2, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 925 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 926 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 927 | "{@type=korap:token, wrap={@type=korap:term, foundry=tt, layer=p, key=ADJA, match=match:eq}}" + |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 928 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 929 | "{@type=korap:token, wrap={@type=korap:term, foundry=mate, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 930 | "]}" + |
| 931 | "]}"; |
| 932 | ppt = new PoliqarpPlusTree("{2: {1:[tt/p=ADJA]}[mate/p=NN]}"); |
| 933 | map = ppt.getRequestMap().get("query").toString(); |
| 934 | assertEquals(cls5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 935 | } |
| 936 | |
| 937 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 938 | public void testPositions() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 939 | // contains(<s>,<np>) |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 940 | String pos1 = "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 941 | "{@type=korap:span, key=s}," + |
| 942 | "{@type=korap:span, key=np}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 943 | "], frame=frame:contains}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 944 | assertTrue(equalsQueryContent(pos1, "contains(<s>,<np>)")); |
| 945 | |
| 946 | // contains(<s>,[base=Mann]) |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 947 | String pos2 = "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 948 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 949 | "{@type=korap:token, wrap= {@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 950 | "], frame=frame:contains}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 951 | assertTrue(equalsQueryContent(pos2, "contains(<s>,[base=Mann])")); |
| 952 | |
| 953 | // contains(<s>,[orth=der][orth=Mann]) |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 954 | String pos3 = "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 955 | "{@type=korap:span, key=s}," + |
| 956 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 957 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| 958 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 959 | "]}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 960 | "], frame=frame:contains}"; |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 961 | ppt = new PoliqarpPlusTree("contains(<s>,[orth=der][orth=Mann])"); |
| 962 | map = ppt.getRequestMap().get("query").toString(); |
| 963 | assertEquals(pos3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 964 | |
| 965 | // [base=Auto]contains(<s>,[base=Mann]) |
| 966 | String pos4 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 967 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 968 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Auto, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 969 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 970 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 971 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 972 | "], frame=frame:contains}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 973 | "]}"; |
| 974 | ppt = new PoliqarpPlusTree("[base=Auto]contains(<s>,[base=Mann])"); |
| 975 | map = ppt.getRequestMap().get("query").toString(); |
| 976 | assertEquals(pos4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 977 | |
| 978 | // contains(<s>,[pos=N]*) |
| 979 | String pos5 = |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 980 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 981 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 982 | "{@type=korap:group, operation=operation:repetition, " + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 983 | "operands=[{@type=korap:token, wrap={@type=korap:term, layer=pos, key=N, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 984 | "], boundary={@type=korap:boundary, min=0}, min=0" + |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 985 | "}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 986 | "], frame=frame:contains}"; |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 987 | ppt = new PoliqarpPlusTree("contains(<s>,[pos=N]*)"); |
| 988 | map = ppt.getRequestMap().get("query").toString(); |
| 989 | assertEquals(pos5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 990 | |
| 991 | // [base=Auto]contains(<s>,[pos=N]*) |
| 992 | String pos6 = |
| 993 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 994 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Auto, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 995 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 996 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 33bd45f | 2014-06-25 15:00:54 +0000 | [diff] [blame] | 997 | "{@type=korap:group, operation=operation:repetition, " + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 998 | "operands=[{@type=korap:token, wrap={@type=korap:term, layer=pos, key=N, match=match:eq}}" + |
| Joachim Bingel | 687e4d4 | 2014-07-30 09:34:18 +0000 | [diff] [blame] | 999 | "], boundary={@type=korap:boundary, min=0}, min=0" + |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 1000 | "}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1001 | "], frame=frame:contains}" + |
| Joachim Bingel | ef7059b | 2014-04-22 13:44:19 +0000 | [diff] [blame] | 1002 | "]}"; |
| 1003 | ppt = new PoliqarpPlusTree("[base=Auto]contains(<s>,[pos=N]*)"); |
| 1004 | map = ppt.getRequestMap().get("query").toString(); |
| 1005 | assertEquals(pos6.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1006 | } |
| 1007 | |
| 1008 | @Test |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1009 | public void testNestedPositions() throws QueryException { |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1010 | // contains(<s>,startswith(<np>,[orth=Der])) |
| 1011 | String npos1 = |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1012 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1013 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1014 | "{@type=korap:group, operation=operation:position, frames=[frames:startswith], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1015 | "{@type=korap:span, key=np}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1016 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1017 | "], frame=frame:startswith}" + |
| 1018 | "], frame=frame:contains}"; |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 1019 | ppt = new PoliqarpPlusTree("contains(<s>, startswith(<np>,[orth=Der]))"); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1020 | map = ppt.getRequestMap().get("query").toString(); |
| 1021 | assertEquals(npos1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1022 | } |
| 1023 | |
| 1024 | @Test |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1025 | public void testFocusSplit() throws QueryException { |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1026 | // focus([orth=Der]{[orth=Mann]}) |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1027 | String shr1 = |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 1028 | "{@type=korap:reference, operation=operation:focus, classRef=[1], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1029 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1030 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}," + |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 1031 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1032 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1033 | "]}" + |
| 1034 | "]}" + |
| 1035 | "]}"; |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 1036 | ppt = new PoliqarpPlusTree("focus([orth=Der]{[orth=Mann]})"); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1037 | map = ppt.getRequestMap().get("query").toString(); |
| 1038 | assertEquals(shr1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1039 | |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1040 | // focus([orth=Der]{[orth=Mann][orth=geht]}) |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1041 | String shr2 = |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 1042 | "{@type=korap:reference, operation=operation:focus, classRef=[1], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1043 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1044 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}," + |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 1045 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1046 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1047 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Mann, match=match:eq}}," + |
| 1048 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=geht, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1049 | "]}" + |
| 1050 | "]}" + |
| 1051 | "]}" + |
| 1052 | "]}"; |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 1053 | ppt = new PoliqarpPlusTree("focus([orth=Der]{[orth=Mann][orth=geht]})"); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1054 | map = ppt.getRequestMap().get("query").toString(); |
| 1055 | assertEquals(shr2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1056 | |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1057 | // focus(1:[orth=Der]{1:[orth=Mann][orth=geht]}) |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1058 | String shr3 = |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 1059 | "{@type=korap:reference, operation=operation:focus, classRef=[1], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1060 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1061 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1062 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1063 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1064 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Mann, match=match:eq}}," + |
| 1065 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=geht, match=match:eq}}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1066 | "]}" + |
| 1067 | "]}" + |
| 1068 | "]}" + |
| 1069 | "]}"; |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 1070 | ppt = new PoliqarpPlusTree("focus(1:[orth=Der]{1:[orth=Mann][orth=geht]})"); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1071 | map = ppt.getRequestMap().get("query").toString(); |
| 1072 | assertEquals(shr3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1073 | |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1074 | // focus(1:startswith(<s>,{1:<np>})) |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1075 | String shr4 = |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 1076 | "{@type=korap:reference, operation=operation:focus, classRef=[1], operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1077 | "{@type=korap:group, operation=operation:position, frames=[frames:startswith], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1078 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1079 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1080 | "{@type=korap:span, key=np}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1081 | "]}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1082 | "], frame=frame:startswith}" + |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1083 | "]}"; |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 1084 | ppt = new PoliqarpPlusTree("focus(1:startswith(<s>,{1:<np>}))"); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1085 | map = ppt.getRequestMap().get("query").toString(); |
| 1086 | assertEquals(shr4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 1087 | |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1088 | // focus(3: startswith(<s>, {3:[base=der]{1:[mate/p=ADJA]{2:[tt/p=NN]}}})) |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 1089 | String shr5 = |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 1090 | "{@type=korap:reference, operation=operation:focus, classRef=[3], operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1091 | "{@type=korap:group, operation=operation:position, frames=[frames:startswith], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1092 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1093 | "{@type=korap:group, operation=operation:class, class=3, classOut=3, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1094 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1095 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1096 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1097 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1098 | "{@type=korap:token, wrap={@type=korap:term, foundry=mate, layer=p, key=ADJA, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1099 | "{@type=korap:group, operation=operation:class, class=2, classOut=2, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1100 | "{@type=korap:token, wrap={@type=korap:term, foundry=tt, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 1101 | "]}" + |
| 1102 | "]}" + |
| 1103 | "]}" + |
| 1104 | "]}" + |
| 1105 | "]}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1106 | "], frame=frame:startswith}" + |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 1107 | "]}"; |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 1108 | ppt = new PoliqarpPlusTree("focus(3:startswith(<s>,{3:[base=der]{1:[mate/p=ADJA]{2:[tt/p=NN]}}})) "); |
| Joachim Bingel | cd9ed33 | 2013-12-09 21:01:35 +0000 | [diff] [blame] | 1109 | map = ppt.getRequestMap().get("query").toString(); |
| 1110 | assertEquals(shr5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 1111 | |
| 1112 | // split(3: startswith(<s>, {3:[base=der]{1:[mate/p=ADJA]{2:[tt/p=NN]}}})) |
| 1113 | String shr6 = |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 1114 | "{@type=korap:reference, operation=operation:split, classRef=[3], operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1115 | "{@type=korap:group, operation=operation:position, frames=[frames:startswith], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1116 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1117 | "{@type=korap:group, operation=operation:class, class=3, classOut=3, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1118 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1119 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1120 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1121 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1122 | "{@type=korap:token, wrap={@type=korap:term, foundry=mate, layer=p, key=ADJA, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1123 | "{@type=korap:group, operation=operation:class, class=2, classOut=2, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1124 | "{@type=korap:token, wrap={@type=korap:term, foundry=tt, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 1125 | "]}" + |
| 1126 | "]}" + |
| 1127 | "]}" + |
| 1128 | "]}" + |
| 1129 | "]}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1130 | "], frame=frame:startswith}" + |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 1131 | "]}"; |
| 1132 | ppt = new PoliqarpPlusTree("split(3:startswith(<s>,{3:[base=der]{1:[mate/p=ADJA]{2:[tt/p=NN]}}})) "); |
| 1133 | map = ppt.getRequestMap().get("query").toString(); |
| 1134 | assertEquals(shr6.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 8c640e4 | 2014-02-07 16:20:47 +0000 | [diff] [blame] | 1135 | |
| 1136 | // split(2|3: startswith(<s>, {3:[base=der]{1:[mate/p=ADJA]{2:[tt/p=NN]}}})) |
| 1137 | String shr7 = |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 1138 | "{@type=korap:reference, operation=operation:split, classRef=[2, 3], classRefOp=classRefOp:intersection, operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1139 | "{@type=korap:group, operation=operation:position, frames=[frames:startswith], operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1140 | "{@type=korap:span, key=s}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1141 | "{@type=korap:group, operation=operation:class, class=3, classOut=3, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1142 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1143 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1144 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1145 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1146 | "{@type=korap:token, wrap={@type=korap:term, foundry=mate, layer=p, key=ADJA, match=match:eq}}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1147 | "{@type=korap:group, operation=operation:class, class=2, classOut=2, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1148 | "{@type=korap:token, wrap={@type=korap:term, foundry=tt, layer=p, key=NN, match=match:eq}}" + |
| Joachim Bingel | 8c640e4 | 2014-02-07 16:20:47 +0000 | [diff] [blame] | 1149 | "]}" + |
| 1150 | "]}" + |
| 1151 | "]}" + |
| 1152 | "]}" + |
| 1153 | "]}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1154 | "], frame=frame:startswith}" + |
| Joachim Bingel | 8c640e4 | 2014-02-07 16:20:47 +0000 | [diff] [blame] | 1155 | "]}"; |
| 1156 | ppt = new PoliqarpPlusTree("split(2|3:startswith(<s>,{3:[base=der]{1:[mate/p=ADJA]{2:[tt/p=NN]}}})) "); |
| 1157 | map = ppt.getRequestMap().get("query").toString(); |
| 1158 | assertEquals(shr7.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 5146ebe | 2014-06-12 08:49:21 +0000 | [diff] [blame] | 1159 | |
| 1160 | |
| 1161 | String shr8 = |
| Joachim Bingel | d1a3e71 | 2014-07-16 08:02:05 +0000 | [diff] [blame] | 1162 | "{@type=korap:reference, operation=operation:focus, classRef=[1], operands=[" + |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 1163 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 4c0c201 | 2014-10-22 09:35:22 +0000 | [diff] [blame^] | 1164 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1165 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=der, match=match:eq}}" + |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 1166 | "]}," + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1167 | "{@type=korap:group, operation=operation:class, class=1, classOut=1, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1168 | "{@type=korap:token, wrap={@type=korap:term, layer=pos, key=ADJA, match=match:eq}}" + |
| Joachim Bingel | 5146ebe | 2014-06-12 08:49:21 +0000 | [diff] [blame] | 1169 | "]}" + |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 1170 | "]}" + |
| 1171 | "]}"; |
| Joachim Bingel | f8dfcf8 | 2014-06-25 09:57:44 +0000 | [diff] [blame] | 1172 | ppt = new PoliqarpPlusTree("focus(1:{[base=der]}{1:[pos=ADJA]})"); |
| Joachim Bingel | add0fb1 | 2014-06-24 08:54:14 +0000 | [diff] [blame] | 1173 | map = ppt.getRequestMap().get("query").toString(); |
| 1174 | assertEquals(shr8.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 5146ebe | 2014-06-12 08:49:21 +0000 | [diff] [blame] | 1175 | |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1176 | } |
| 1177 | |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1178 | @Test |
| 1179 | public void testSubspan() throws QueryException { |
| Joachim Bingel | 9bbd4fc | 2014-08-11 14:56:48 +0000 | [diff] [blame] | 1180 | query = "submatch(1,:<s>)"; |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1181 | expected = |
| 1182 | "{@type=korap:reference, operation=operation:focus, operands=[" + |
| 1183 | "{@type=korap:span, key=s}" + |
| 1184 | "], spanRef=[1]" + |
| 1185 | "}"; |
| 1186 | ppt = new PoliqarpPlusTree(query); |
| 1187 | map = ppt.getRequestMap().get("query").toString(); |
| 1188 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1189 | |
| Joachim Bingel | 9bbd4fc | 2014-08-11 14:56:48 +0000 | [diff] [blame] | 1190 | query = "submatch(1,4:<s>)"; |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1191 | expected = |
| 1192 | "{@type=korap:reference, operation=operation:focus, operands=[" + |
| 1193 | "{@type=korap:span, key=s}" + |
| 1194 | "], spanRef=[1,4]" + |
| 1195 | "}"; |
| 1196 | ppt = new PoliqarpPlusTree(query); |
| 1197 | map = ppt.getRequestMap().get("query").toString(); |
| 1198 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1199 | |
| Joachim Bingel | 9bbd4fc | 2014-08-11 14:56:48 +0000 | [diff] [blame] | 1200 | query = "submatch(1,4:contains(<s>,[base=Haus]))"; |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1201 | expected = |
| 1202 | "{@type=korap:reference, operation=operation:focus, operands=[" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1203 | "{@type=korap:group, operation=operation:position, frames=[frames:contains], operands=[" + |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1204 | "{@type=korap:span, key=s}," + |
| 1205 | "{@type=korap:token, wrap= {@type=korap:term, layer=lemma, key=Haus, match=match:eq}}" + |
| Joachim Bingel | 6b52d76 | 2014-09-30 15:29:11 +0000 | [diff] [blame] | 1206 | "], frame=frame:contains}" + |
| Joachim Bingel | 23c31ad | 2014-08-11 09:44:46 +0000 | [diff] [blame] | 1207 | "], spanRef=[1,4]" + |
| 1208 | "}"; |
| 1209 | ppt = new PoliqarpPlusTree(query); |
| 1210 | map = ppt.getRequestMap().get("query").toString(); |
| 1211 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1212 | } |
| 1213 | |
| 1214 | @Test |
| 1215 | public void testRelations() throws QueryException { |
| 1216 | query = "relatesTo(<s>,<np>)"; |
| 1217 | expected = |
| 1218 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1219 | "{@type=korap:span, key=s}," + |
| 1220 | "{@type=korap:span, key=np}" + |
| 1221 | "], relation={@type=korap:relation}" + |
| 1222 | "}"; |
| 1223 | ppt = new PoliqarpPlusTree(query); |
| 1224 | map = ppt.getRequestMap().get("query").toString(); |
| 1225 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1226 | |
| 1227 | query = "relatesTo([base=Baum],<np>)"; |
| 1228 | expected = |
| 1229 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1230 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Baum, match=match:eq}}," + |
| 1231 | "{@type=korap:span, key=np}" + |
| 1232 | "], relation={@type=korap:relation}" + |
| 1233 | "}"; |
| 1234 | ppt = new PoliqarpPlusTree(query); |
| 1235 | map = ppt.getRequestMap().get("query").toString(); |
| 1236 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1237 | |
| 1238 | query = "dominates(<np>,[base=Baum])"; |
| 1239 | expected = |
| 1240 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1241 | "{@type=korap:span, key=np}," + |
| 1242 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Baum, match=match:eq}}" + |
| 1243 | "], relation={@type=korap:relation, layer=c}" + |
| 1244 | "}"; |
| 1245 | ppt = new PoliqarpPlusTree(query); |
| 1246 | map = ppt.getRequestMap().get("query").toString(); |
| 1247 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1248 | |
| 1249 | query = "dominates(cnx/c:<np>,[base=Baum])"; |
| 1250 | expected = |
| 1251 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1252 | "{@type=korap:span, key=np}," + |
| 1253 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Baum, match=match:eq}}" + |
| 1254 | "], relation={@type=korap:relation, layer=c, foundry=cnx}" + |
| 1255 | "}"; |
| 1256 | ppt = new PoliqarpPlusTree(query); |
| 1257 | map = ppt.getRequestMap().get("query").toString(); |
| 1258 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1259 | |
| 1260 | query = "dominates(cnx/c*:<np>,[base=Baum])"; |
| 1261 | expected = |
| 1262 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1263 | "{@type=korap:span, key=np}," + |
| 1264 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Baum, match=match:eq}}" + |
| 1265 | "], relation={@type=korap:relation, layer=c, foundry=cnx, boundary={@type=korap:boundary, min=0}}" + |
| 1266 | "}"; |
| 1267 | ppt = new PoliqarpPlusTree(query); |
| 1268 | map = ppt.getRequestMap().get("query").toString(); |
| 1269 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1270 | |
| 1271 | query = "dominates(cnx/c{1,5}:<np>,[base=Baum])"; |
| 1272 | expected = |
| 1273 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1274 | "{@type=korap:span, key=np}," + |
| 1275 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Baum, match=match:eq}}" + |
| 1276 | "], relation={@type=korap:relation, layer=c, foundry=cnx, boundary={@type=korap:boundary, min=1, max=5}}" + |
| 1277 | "}"; |
| 1278 | ppt = new PoliqarpPlusTree(query); |
| 1279 | map = ppt.getRequestMap().get("query").toString(); |
| 1280 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1281 | |
| 1282 | query = "relatesTo(mate/d=HEAD:<np>,[base=Baum])"; |
| 1283 | expected = |
| 1284 | "{@type=korap:group, operation=operation:relation, operands=[" + |
| 1285 | "{@type=korap:span, key=np}," + |
| 1286 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=Baum, match=match:eq}}" + |
| 1287 | "], relation={@type=korap:relation, foundry=mate, layer=d, key=HEAD}" + |
| 1288 | "}"; |
| 1289 | ppt = new PoliqarpPlusTree(query); |
| 1290 | map = ppt.getRequestMap().get("query").toString(); |
| 1291 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1292 | |
| 1293 | } |
| 1294 | |
| 1295 | |
| Joachim Bingel | 0388270 | 2013-12-31 19:53:05 +0000 | [diff] [blame] | 1296 | |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1297 | @Test |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 1298 | public void testFoundries() throws QueryException { |
| 1299 | // [tt/base=Mann] |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1300 | String layer1 = "{@type=korap:token, wrap={@type=korap:term, foundry=tt, layer=lemma, key=Mann, match=match:eq}}"; |
| Joachim Bingel | ba9a0ab | 2014-01-29 10:12:25 +0000 | [diff] [blame] | 1301 | ppt = new PoliqarpPlusTree("[tt/base=Mann]"); |
| 1302 | map = ppt.getRequestMap().get("query").toString(); |
| 1303 | assertEquals(layer1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1304 | |
| 1305 | } |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1306 | |
| 1307 | @Test |
| Joachim Bingel | 8c640e4 | 2014-02-07 16:20:47 +0000 | [diff] [blame] | 1308 | public void testAlign() throws QueryException { |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1309 | // [orth=der]^[orth=Mann] |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1310 | query = "[orth=der]^[orth=Mann]"; |
| 1311 | expected = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1312 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1313 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1314 | "{@type=korap:group, operation=operation:class, class=129, classOut=129, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1315 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1316 | "]}" + |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1317 | "]}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1318 | metaExpected = |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1319 | "{alignment=129}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1320 | ppt = new PoliqarpPlusTree(query); |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1321 | map = ppt.getRequestMap().get("query").toString(); |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1322 | metaMap = ppt.getRequestMap().get("meta").toString(); |
| 1323 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1324 | assertEquals(metaExpected.replaceAll(" ", ""), metaMap.replaceAll(" ", "")); |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1325 | |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1326 | // [orth=der]^[orth=große][orth=Mann] |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1327 | query = "[orth=der]^[orth=große][orth=Mann]"; |
| 1328 | String expected = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1329 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1330 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=der, match=match:eq}}," + |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1331 | "{@type=korap:group, operation=operation:class, class=129, classOut=129, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1332 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1333 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=große, match=match:eq}}," + |
| 1334 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Mann, match=match:eq}}" + |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1335 | "]}" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1336 | "]}" + |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1337 | "]}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1338 | metaExpected = |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1339 | "{alignment=129}"; |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1340 | ppt = new PoliqarpPlusTree(query); |
| 1341 | map = ppt.getRequestMap().get("query").toString(); |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1342 | metaMap = ppt.getRequestMap().get("meta").toString(); |
| 1343 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1344 | assertEquals(metaExpected.replaceAll(" ", ""), metaMap.replaceAll(" ", "")); |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1345 | |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1346 | query = "([base=a]^[base=b])|[base=c]"; |
| 1347 | expected = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1348 | "{@type=korap:group, operation=operation:or, operands=[" + |
| 1349 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1350 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=a, match=match:eq}}," + |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1351 | "{@type=korap:group, operation=operation:class, class=129, classOut=129, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1352 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=b, match=match:eq}}" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1353 | "]}" + |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1354 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1355 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=c, match=match:eq}}" + |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1356 | "]}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1357 | metaExpected = |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1358 | "{alignment=129}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1359 | ppt = new PoliqarpPlusTree(query); |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1360 | map = ppt.getRequestMap().get("query").toString(); |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1361 | metaMap = ppt.getRequestMap().get("meta").toString(); |
| 1362 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1363 | assertEquals(metaExpected.replaceAll(" ", ""), metaMap.replaceAll(" ", "")); |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1364 | |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1365 | query = "([base=a]^[base=b][base=c])|[base=d]"; |
| 1366 | expected = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1367 | "{@type=korap:group, operation=operation:or, operands=[" + |
| 1368 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1369 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=a, match=match:eq}}," + |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1370 | "{@type=korap:group, operation=operation:class, class=129, classOut=129, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1371 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1372 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=b, match=match:eq}}," + |
| 1373 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=c, match=match:eq}}" + |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1374 | "]}" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1375 | "]}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1376 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1377 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=d, match=match:eq}}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1378 | "]}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1379 | metaExpected = |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1380 | "{alignment=129}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1381 | ppt = new PoliqarpPlusTree(query); |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1382 | map = ppt.getRequestMap().get("query").toString(); |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1383 | metaMap = ppt.getRequestMap().get("meta").toString(); |
| 1384 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1385 | assertEquals(metaExpected.replaceAll(" ", ""), metaMap.replaceAll(" ", "")); |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1386 | |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1387 | query = "([base=a]^[base=b]^[base=c])|[base=d]"; |
| 1388 | expected = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1389 | "{@type=korap:group, operation=operation:or, operands=[" + |
| 1390 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1391 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=a, match=match:eq}}," + |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1392 | "{@type=korap:group, operation=operation:class, class=129, classOut=129, operands=[" + |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1393 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1394 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=b, match=match:eq}}," + |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1395 | "{@type=korap:group, operation=operation:class, class=130, classOut=130, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1396 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=c, match=match:eq}}" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1397 | "]}" + |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1398 | "]}" + |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1399 | "]}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1400 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1401 | "{@type=korap:token, wrap={@type=korap:term, layer=lemma, key=d, match=match:eq}}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1402 | "]}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1403 | metaExpected = |
| Joachim Bingel | 6b491c2 | 2014-10-21 08:35:18 +0000 | [diff] [blame] | 1404 | "{alignment=[129,130]}"; |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1405 | ppt = new PoliqarpPlusTree(query); |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1406 | map = ppt.getRequestMap().get("query").toString(); |
| Joachim Bingel | 832800e | 2014-10-17 14:46:39 +0000 | [diff] [blame] | 1407 | metaMap = ppt.getRequestMap().get("meta").toString(); |
| 1408 | assertEquals(expected.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1409 | assertEquals(metaExpected.replaceAll(" ", ""), metaMap.replaceAll(" ", "")); |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1410 | |
| 1411 | |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1412 | } |
| 1413 | |
| 1414 | @Test |
| Joachim Bingel | 8c640e4 | 2014-02-07 16:20:47 +0000 | [diff] [blame] | 1415 | public void testSimpleQueries() throws QueryException { |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1416 | // Baum |
| 1417 | String simple1 = |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1418 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq}}"; |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1419 | ppt = new PoliqarpPlusTree("Baum"); |
| 1420 | map = ppt.getRequestMap().get("query").toString(); |
| 1421 | assertEquals(simple1.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1422 | |
| 1423 | // Baum/i |
| 1424 | String simple1b = |
| 1425 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq, caseInsensitive=true}}"; |
| 1426 | ppt = new PoliqarpPlusTree("Baum/i"); |
| 1427 | map = ppt.getRequestMap().get("query").toString(); |
| 1428 | assertEquals(simple1b.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1429 | |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1430 | // Der Baum |
| 1431 | String simple2 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1432 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1433 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}, " + |
| 1434 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq}}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1435 | "]}"; |
| 1436 | ppt = new PoliqarpPlusTree("Der Baum"); |
| 1437 | map = ppt.getRequestMap().get("query").toString(); |
| 1438 | assertEquals(simple2.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1439 | |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1440 | // Der Baum/i |
| 1441 | String simple2b = |
| 1442 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 1443 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}, " + |
| 1444 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq, caseInsensitive=true}}" + |
| 1445 | "]}"; |
| 1446 | ppt = new PoliqarpPlusTree("Der Baum/i"); |
| 1447 | map = ppt.getRequestMap().get("query").toString(); |
| 1448 | assertEquals(simple2b.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1449 | |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1450 | // Der große Baum |
| 1451 | String simple3 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1452 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1453 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Der, match=match:eq}}, " + |
| 1454 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=große, match=match:eq}}, " + |
| 1455 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq}}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1456 | "]}"; |
| 1457 | ppt = new PoliqarpPlusTree("Der große Baum"); |
| 1458 | map = ppt.getRequestMap().get("query").toString(); |
| 1459 | assertEquals(simple3.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| 1460 | |
| 1461 | // Baum | Stein |
| 1462 | String simple4 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1463 | "{@type=korap:group, operation=operation:or, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1464 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq}}, " + |
| 1465 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Stein, match=match:eq}}" + |
| Joachim Bingel | a67e6a3 | 2014-01-02 18:35:24 +0000 | [diff] [blame] | 1466 | "]}"; |
| 1467 | ppt = new PoliqarpPlusTree("Baum | Stein"); |
| 1468 | map = ppt.getRequestMap().get("query").toString(); |
| 1469 | assertEquals(simple4.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1470 | |
| 1471 | // Baum | Stein Haus |
| 1472 | String query = "(Baum | Stein) Haus"; |
| 1473 | String simple5 = |
| Joachim Bingel | 2daf986 | 2014-02-12 10:18:54 +0000 | [diff] [blame] | 1474 | "{@type=korap:group, operation=operation:sequence, operands=[" + |
| 1475 | "{@type=korap:group, operation=operation:or, operands=[" + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1476 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Baum, match=match:eq}}, " + |
| 1477 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Stein, match=match:eq}}" + |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1478 | "]}," + |
| Joachim Bingel | 0900a89 | 2014-06-30 16:26:21 +0000 | [diff] [blame] | 1479 | "{@type=korap:token, wrap={@type=korap:term, layer=orth, key=Haus, match=match:eq}} " + |
| Joachim Bingel | 84e33df | 2014-01-31 14:02:46 +0000 | [diff] [blame] | 1480 | "]}"; |
| 1481 | ppt = new PoliqarpPlusTree(query); |
| 1482 | map = ppt.getRequestMap().get("query").toString(); |
| 1483 | assertEquals(simple5.replaceAll(" ", ""), map.replaceAll(" ", "")); |
| Joachim Bingel | 16da4e1 | 2013-12-17 09:48:12 +0000 | [diff] [blame] | 1484 | } |
| Joachim Bingel | 53333e6 | 2013-12-09 19:25:52 +0000 | [diff] [blame] | 1485 | } |
| 1486 | |