[3743] | 1 | /*
|
---|
| 2 | * MGWrapperImpl.c
|
---|
| 3 | * Copyright (C) 2002 New Zealand Digital Library, http://www.nzdl.org
|
---|
| 4 | *
|
---|
| 5 | * This program is free software; you can redistribute it and/or modify
|
---|
| 6 | * it under the terms of the GNU General Public License as published by
|
---|
| 7 | * the Free Software Foundation; either version 2 of the License, or
|
---|
| 8 | * (at your option) any later version.
|
---|
| 9 | *
|
---|
| 10 | * This program is distributed in the hope that it will be useful,
|
---|
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 13 | * GNU General Public License for more details.
|
---|
| 14 | *
|
---|
| 15 | * You should have received a copy of the GNU General Public License
|
---|
| 16 | * along with this program; if not, write to the Free Software
|
---|
| 17 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 18 | */
|
---|
| 19 |
|
---|
| 20 |
|
---|
| 21 | #include "MGWrapperImpl.h"
|
---|
| 22 | #include <jni.h>
|
---|
| 23 | #include "org_greenstone_mg_MGWrapper.h"
|
---|
| 24 |
|
---|
| 25 | #include "backend.h"
|
---|
| 26 | #include "environment.h"
|
---|
[7629] | 27 | #include "text_get.h"
|
---|
| 28 | #include "stemmer.h"
|
---|
[3743] | 29 |
|
---|
| 30 | /*************************************************************************
|
---|
| 31 | NOTES
|
---|
| 32 |
|
---|
| 33 | - Features supported by MGPP but not by MG (AFAIK)
|
---|
| 34 |
|
---|
| 35 | - Sorting results by rank.
|
---|
| 36 | Done here as a post-processing operation. Could be more efficient in
|
---|
| 37 | some cases: the current solution is not very good when the number of
|
---|
| 38 | matching documents is large and the number of desired matches is
|
---|
| 39 | small. In this case it would be better to iterate through the array
|
---|
| 40 | picking out the best documents rather than sorting them all.
|
---|
| 41 |
|
---|
| 42 | - Asking for query term frequencies to be returned.
|
---|
| 43 | This cannot be turned off in MG. If the term frequencies are not
|
---|
[4714] | 44 | required, they are simply not passed back to the Java side.
|
---|
[3743] | 45 |
|
---|
| 46 | - Choosing the index for queries.
|
---|
| 47 | It is possible for MG to build Section and Document indexes (for
|
---|
| 48 | example), but these are two separate indexes, and must be loaded
|
---|
| 49 | separately. This module can load more than one index at a time, thus
|
---|
| 50 | processing queries to different indexes more quickly.
|
---|
| 51 | NOTE: This replaces TWO options in the MGPP version: returnLevel and
|
---|
| 52 | queryLevel.
|
---|
| 53 |
|
---|
| 54 | *************************************************************************/
|
---|
| 55 |
|
---|
| 56 |
|
---|
[3981] | 57 | #define MAX_INDEXES_CACHED 3
|
---|
[3743] | 58 |
|
---|
[8919] | 59 | /* copied from mgquery, needed to convert paragraph numbers to document numbers
|
---|
| 60 | for greenstone */
|
---|
| 61 | #if defined(PARADOCNUM) || defined(NZDL)
|
---|
| 62 | static int GetDocNumFromParaNum(query_data *qd, int paranum) {
|
---|
| 63 | int Documents = qd->td->cth.num_of_docs;
|
---|
| 64 | int *Paragraph = qd->paragraph;
|
---|
| 65 | int low = 1, high = Documents;
|
---|
| 66 | int mid = (low+high)/2;
|
---|
[3743] | 67 |
|
---|
[8919] | 68 | while ((mid = (low+high)/2) >=1 && mid <= Documents)
|
---|
| 69 | {
|
---|
| 70 | if (paranum > Paragraph[mid])
|
---|
| 71 | low = mid+1;
|
---|
| 72 | else if (paranum <= Paragraph[mid-1])
|
---|
| 73 | high = mid-1;
|
---|
| 74 | else
|
---|
| 75 | return mid;
|
---|
| 76 | }
|
---|
| 77 | FatalError(1, "Bad paragraph number.\n");
|
---|
| 78 | return 0;
|
---|
| 79 | }
|
---|
| 80 | #endif
|
---|
| 81 |
|
---|
| 82 |
|
---|
[3743] | 83 | /*********************************************
|
---|
| 84 | initialisation stuff
|
---|
| 85 | *********************************************/
|
---|
| 86 |
|
---|
| 87 | /* cached ids for java stuff */
|
---|
| 88 | jfieldID FID_mg_data = NULL; /* MGWrapperData */
|
---|
| 89 | jfieldID FID_query_result = NULL; /* MGQueryResult */
|
---|
| 90 | jmethodID MID_addDoc = NULL; /* MGQueryResult.addDoc() */
|
---|
| 91 | jmethodID MID_addTerm = NULL; /* MGQueryResult.addTerm() */
|
---|
[3791] | 92 | jmethodID MID_addEquivTerm = NULL; /* MGQueryResult.addEquivTerm() */
|
---|
[3743] | 93 | jmethodID MID_setTotalDocs = NULL; /* MGQueryResult.setTotalDocs() */
|
---|
| 94 | jmethodID MID_clearResult = NULL; /* MGQueryResult.clear() */
|
---|
| 95 |
|
---|
| 96 |
|
---|
| 97 | /* to access objects and methods on java side, need their field/method ids -
|
---|
| 98 | this initialises them at the start to avoid recalculating them each time they
|
---|
| 99 | are needed
|
---|
| 100 | Note: the descriptors need to be exactly right, otherwise you get an error
|
---|
| 101 | saying "no such field" but no reference to the fact that it has the right
|
---|
| 102 | name but the wrong type.
|
---|
| 103 | Note: apparently the jclass is a local ref and should only work
|
---|
| 104 | in the method that created it. It seems to work ok, but I'll make it
|
---|
| 105 | global cos the book said I should, and it may avoid future hassles.
|
---|
| 106 | */
|
---|
| 107 | JNIEXPORT void JNICALL
|
---|
| 108 | Java_org_greenstone_mg_MGWrapper_initIDs(JNIEnv *j_env, jclass j_cls)
|
---|
| 109 | {
|
---|
| 110 | jclass JC_MGQueryResult;
|
---|
| 111 |
|
---|
| 112 | /* a long-"J" */
|
---|
| 113 | FID_mg_data = (*j_env)->GetFieldID(j_env, j_cls, "mg_data_ptr_", "J");
|
---|
| 114 | assert(FID_mg_data != NULL);
|
---|
| 115 |
|
---|
| 116 | /* an object -"L<class name>;" */
|
---|
| 117 | FID_query_result = (*j_env)->GetFieldID(j_env, j_cls, "mg_query_result_",
|
---|
| 118 | "Lorg/greenstone/mg/MGQueryResult;");
|
---|
| 119 | assert(FID_query_result != NULL);
|
---|
| 120 |
|
---|
| 121 | /* the methods we want to use */
|
---|
| 122 | JC_MGQueryResult = (*j_env)->FindClass(j_env, "org/greenstone/mg/MGQueryResult");
|
---|
| 123 |
|
---|
| 124 | /* addDoc(long doc, float rank) */
|
---|
| 125 | MID_addDoc = (*j_env)->GetMethodID(j_env, JC_MGQueryResult, "addDoc", "(JF)V");
|
---|
| 126 | assert(MID_addDoc != NULL);
|
---|
| 127 |
|
---|
[3791] | 128 | /* addTerm(String term, int stem) */
|
---|
| 129 | MID_addTerm = (*j_env)->GetMethodID(j_env, JC_MGQueryResult, "addTerm", "(Ljava/lang/String;I)V");
|
---|
[3743] | 130 | assert(MID_addTerm != NULL);
|
---|
| 131 |
|
---|
[3791] | 132 | /* addEquivTerm(String term, String equivTerm, long match, long freq) */
|
---|
| 133 | MID_addEquivTerm = (*j_env)->GetMethodID(j_env, JC_MGQueryResult, "addEquivTerm", "(Ljava/lang/String;Ljava/lang/String;JJ)V");
|
---|
| 134 | assert(MID_addEquivTerm != NULL);
|
---|
| 135 |
|
---|
[3743] | 136 | /* setTotalDocs(long) */
|
---|
| 137 | MID_setTotalDocs = (*j_env)->GetMethodID(j_env, JC_MGQueryResult, "setTotalDocs", "(J)V");
|
---|
| 138 | assert(MID_setTotalDocs != NULL);
|
---|
| 139 |
|
---|
| 140 | /* clear(void) */
|
---|
| 141 | MID_clearResult = (*j_env)->GetMethodID(j_env, JC_MGQueryResult, "clear", "()V");
|
---|
| 142 | assert(MID_clearResult != NULL);
|
---|
| 143 | }
|
---|
| 144 |
|
---|
| 145 |
|
---|
| 146 | /* the java side MGWrapper has a pointer to a C object - MGWrapperData
|
---|
| 147 | initialise this and set the pointer
|
---|
| 148 | */
|
---|
| 149 | JNIEXPORT jboolean JNICALL
|
---|
| 150 | Java_org_greenstone_mg_MGWrapper_initCSide(JNIEnv *j_env, jobject j_obj)
|
---|
| 151 | {
|
---|
| 152 | /* Allocate a MGWrapperData object to store query parameters */
|
---|
| 153 | MGWrapperData* data = (MGWrapperData*) malloc(sizeof(MGWrapperData));
|
---|
| 154 | assert(data != NULL);
|
---|
| 155 |
|
---|
| 156 | /* Set default values - no stemming, no case-folding, boolean OR queries */
|
---|
| 157 | data->defaultStemMethod = 0;
|
---|
| 158 | data->defaultBoolCombine = 0;
|
---|
| 159 |
|
---|
| 160 | /* Allocate a QueryInfo object to store more query parameters */
|
---|
| 161 | data->queryInfo = (QueryInfo*) malloc(sizeof(QueryInfo));
|
---|
| 162 | assert(data->queryInfo != NULL);
|
---|
| 163 |
|
---|
| 164 | /* Set default values - 50 documents max, return term freqs, sort by rank */
|
---|
| 165 | data->queryInfo->index = NULL;
|
---|
| 166 | data->queryInfo->maxDocs = 50;
|
---|
| 167 | data->queryInfo->needTermFreqs = 1;
|
---|
| 168 |
|
---|
| 169 | /* Save the object on the Java side */
|
---|
[25244] | 170 | (*j_env)->SetLongField(j_env, j_obj, FID_mg_data, (long) data);
|
---|
[3743] | 171 |
|
---|
| 172 | /* Initialise MG environment variables */
|
---|
| 173 | InitEnv();
|
---|
| 174 | SetEnv("expert", "true", NULL);
|
---|
| 175 | SetEnv("mode", "docnums", NULL);
|
---|
[13288] | 176 |
|
---|
[3743] | 177 | return 1; /* true - no errors */
|
---|
| 178 | }
|
---|
| 179 |
|
---|
| 180 |
|
---|
| 181 | /*******************************************
|
---|
| 182 | Index caching
|
---|
| 183 | *******************************************/
|
---|
| 184 |
|
---|
| 185 | query_data* cached_indexes[MAX_INDEXES_CACHED] = { NULL };
|
---|
| 186 |
|
---|
| 187 |
|
---|
| 188 | /* Get the index data necessary to perform a query or document retrieval */
|
---|
| 189 | query_data*
|
---|
| 190 | loadIndexData(char* base_dir, char* index_path, char* text_path)
|
---|
| 191 | {
|
---|
| 192 | char* index_path_name;
|
---|
| 193 | char* text_path_name;
|
---|
| 194 | query_data* qd;
|
---|
| 195 | int i = 0;
|
---|
| 196 |
|
---|
| 197 | /* Form the path name of the desired indexes */
|
---|
| 198 | index_path_name = (char*) malloc(strlen(base_dir) + strlen(index_path) + 1);
|
---|
| 199 | assert(index_path_name != NULL);
|
---|
| 200 | strcpy(index_path_name, base_dir);
|
---|
| 201 | strcat(index_path_name, index_path);
|
---|
| 202 | printf("Index pathname: %s\n", index_path_name);
|
---|
| 203 |
|
---|
| 204 | text_path_name = (char*) malloc(strlen(base_dir) + strlen(text_path) + 1);
|
---|
| 205 | assert(text_path_name != NULL);
|
---|
| 206 | strcpy(text_path_name, base_dir);
|
---|
| 207 | strcat(text_path_name, text_path);
|
---|
| 208 | printf("Text pathname: %s\n", text_path_name);
|
---|
| 209 |
|
---|
| 210 | /* Search through the cached indexes for the desired one */
|
---|
| 211 | while (i < MAX_INDEXES_CACHED && cached_indexes[i] != NULL) {
|
---|
| 212 | printf("(Cached) Pathname: %s\n", cached_indexes[i]->pathname);
|
---|
| 213 | printf("(Cached) Textpathname: %s\n", cached_indexes[i]->textpathname);
|
---|
| 214 |
|
---|
| 215 | /* Check if the index has already been loaded */
|
---|
| 216 | if ((strcmp(index_path_name, cached_indexes[i]->pathname) == 0) &&
|
---|
| 217 | (strcmp(text_path_name, cached_indexes[i]->textpathname) == 0)) {
|
---|
| 218 | /* Index has already been loaded and cached, so return it */
|
---|
| 219 | printf("Found index!\n");
|
---|
| 220 | free(index_path_name);
|
---|
| 221 | free(text_path_name);
|
---|
| 222 | return cached_indexes[i];
|
---|
| 223 | }
|
---|
| 224 |
|
---|
| 225 | i++;
|
---|
| 226 | }
|
---|
| 227 |
|
---|
| 228 | /* Text strings no longer needed */
|
---|
| 229 | free(index_path_name);
|
---|
| 230 | free(text_path_name);
|
---|
| 231 |
|
---|
| 232 | /* The index is not cached, so load it now */
|
---|
| 233 | qd = InitQuerySystem(base_dir, index_path, text_path, NULL);
|
---|
| 234 | if (!qd) {
|
---|
| 235 | printf("Error: Could not InitQuerySystem()...\n");
|
---|
| 236 | return NULL;
|
---|
| 237 | }
|
---|
| 238 |
|
---|
| 239 | /* The index loaded OK, so cache it */
|
---|
| 240 | /* This could be a little more sophisticated, eg. replace least frequently used index */
|
---|
| 241 | if (i >= MAX_INDEXES_CACHED)
|
---|
| 242 | i = MAX_INDEXES_CACHED - 1;
|
---|
| 243 |
|
---|
| 244 | /* Free the index being replaced */
|
---|
| 245 | if (cached_indexes[i] != NULL)
|
---|
| 246 | FinishQuerySystem(cached_indexes[i]);
|
---|
| 247 |
|
---|
| 248 | /* Cache the loaded index, and return it */
|
---|
| 249 | cached_indexes[i] = qd;
|
---|
| 250 | return cached_indexes[i];
|
---|
| 251 | }
|
---|
| 252 |
|
---|
| 253 |
|
---|
| 254 | /* Clean up by unloading all cached indexes */
|
---|
| 255 | JNIEXPORT jboolean JNICALL
|
---|
| 256 | Java_org_greenstone_mg_MGWrapper_unloadIndexData(JNIEnv* j_env, jobject j_obj)
|
---|
| 257 | {
|
---|
| 258 | /* Free all the loaded indexes */
|
---|
| 259 | int i = 0;
|
---|
| 260 | while (i < MAX_INDEXES_CACHED && cached_indexes[i] != NULL) {
|
---|
| 261 | FinishQuerySystem(cached_indexes[i]);
|
---|
| 262 | cached_indexes[i] = NULL;
|
---|
| 263 | i++;
|
---|
| 264 | }
|
---|
| 265 |
|
---|
| 266 | return 1; /* true - no errors */
|
---|
| 267 | }
|
---|
| 268 |
|
---|
| 269 |
|
---|
| 270 | /****************************************************
|
---|
| 271 | retrieve a document
|
---|
| 272 | ****************************************************/
|
---|
| 273 |
|
---|
| 274 | /* Returns a document from mg as a string */
|
---|
| 275 | JNIEXPORT jstring JNICALL
|
---|
| 276 | Java_org_greenstone_mg_MGWrapper_getDocument(JNIEnv *j_env, jobject j_obj,
|
---|
| 277 | jstring j_base_dir, jstring j_text_path,
|
---|
| 278 | jlong j_docnum)
|
---|
| 279 | {
|
---|
[25244] | 280 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 281 |
|
---|
| 282 | char* index_path;
|
---|
| 283 | const char* base_dir;
|
---|
| 284 | const char* text_path;
|
---|
| 285 | query_data* qd;
|
---|
| 286 |
|
---|
[25147] | 287 | mg_u_long pos, len;
|
---|
[3743] | 288 | u_char* c_buffer = NULL;
|
---|
| 289 | u_char* uc_buffer = NULL;
|
---|
| 290 | int ULen;
|
---|
| 291 |
|
---|
| 292 | jstring result;
|
---|
| 293 |
|
---|
| 294 | /* Make sure an index has been specified */
|
---|
| 295 | index_path = data->queryInfo->index;
|
---|
| 296 | assert(index_path != NULL);
|
---|
| 297 |
|
---|
| 298 | /* Obtain C versions of the two string parameters */
|
---|
| 299 | base_dir = (*j_env)->GetStringUTFChars(j_env, j_base_dir, NULL);
|
---|
| 300 | if (base_dir == NULL) {
|
---|
| 301 | return NULL;
|
---|
| 302 | }
|
---|
| 303 | text_path = (*j_env)->GetStringUTFChars(j_env, j_text_path, NULL);
|
---|
| 304 | if (text_path == NULL) {
|
---|
| 305 | (*j_env)->ReleaseStringUTFChars(j_env, j_base_dir, base_dir);
|
---|
| 306 | return NULL;
|
---|
| 307 | }
|
---|
| 308 |
|
---|
| 309 | /* Load the appropriate index for satisfying this request */
|
---|
[3791] | 310 | printf("Document retrieval, index path: %s\n", index_path);
|
---|
[3743] | 311 | qd = loadIndexData((char*) base_dir, (char*) index_path, (char*) text_path);
|
---|
| 312 |
|
---|
| 313 | /* The C text strings are no longer needed */
|
---|
| 314 | (*j_env)->ReleaseStringUTFChars(j_env, j_base_dir, base_dir);
|
---|
| 315 | (*j_env)->ReleaseStringUTFChars(j_env, j_text_path, text_path);
|
---|
| 316 |
|
---|
| 317 | /* Check that the index was loaded successfully */
|
---|
[9874] | 318 | if (qd==NULL) {
|
---|
| 319 | return NULL;
|
---|
| 320 | }
|
---|
| 321 | /*assert(qd != NULL);*/
|
---|
[3743] | 322 |
|
---|
| 323 | /* Get the document position and length in the text file */
|
---|
[25147] | 324 | printf("Fetching document number %d...\n", (mg_u_long) j_docnum);
|
---|
| 325 | FetchDocStart(qd, (mg_u_long) j_docnum, &pos, &len);
|
---|
| 326 | printf("Fetched document start. Pos: %d, Len: %d\n", pos, len);
|
---|
[3743] | 327 |
|
---|
| 328 | /* Allocate memory for the document text (from mg/src/text/mgquery.c:RawDocOutput()) */
|
---|
| 329 | c_buffer = (u_char*) malloc(len);
|
---|
| 330 | assert(c_buffer != NULL);
|
---|
| 331 | uc_buffer = (u_char*) malloc((int) (qd->td->cth.ratio * 1.01 * len) + 100);
|
---|
| 332 | assert(uc_buffer != NULL);
|
---|
| 333 |
|
---|
| 334 | /* Seek to the correct position in the file and read the document text */
|
---|
| 335 | Fseek (qd->td->TextFile, pos, 0);
|
---|
| 336 | Fread (c_buffer, 1, len, qd->td->TextFile);
|
---|
| 337 |
|
---|
[3791] | 338 | /* Decompress the document text into another buffer, and terminate it */
|
---|
[3743] | 339 | DecodeText (qd->cd, c_buffer, len, uc_buffer, &ULen);
|
---|
| 340 | uc_buffer[ULen] = '\0';
|
---|
| 341 |
|
---|
| 342 | /* Load the document text into a Java string */
|
---|
| 343 | result = (*j_env)->NewStringUTF(j_env, uc_buffer);
|
---|
| 344 | assert(result != NULL);
|
---|
| 345 |
|
---|
| 346 | /* Free C buffers */
|
---|
| 347 | free(c_buffer);
|
---|
| 348 | free(uc_buffer);
|
---|
| 349 |
|
---|
| 350 | /* Return the document text */
|
---|
| 351 | return result;
|
---|
| 352 | }
|
---|
| 353 |
|
---|
| 354 |
|
---|
| 355 | /*******************************************
|
---|
| 356 | do a query
|
---|
| 357 | *******************************************/
|
---|
| 358 |
|
---|
| 359 | /* do the actual query - the results are written to query_result held on the Java side */
|
---|
| 360 | JNIEXPORT void JNICALL
|
---|
| 361 | Java_org_greenstone_mg_MGWrapper_runQuery(JNIEnv *j_env, jobject j_obj,
|
---|
| 362 | jstring j_base_dir, jstring j_text_path,
|
---|
| 363 | jstring j_query)
|
---|
| 364 | {
|
---|
[25244] | 365 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 366 |
|
---|
| 367 | char* index_path;
|
---|
| 368 | const char* base_dir;
|
---|
| 369 | const char* text_path;
|
---|
| 370 | query_data* qd;
|
---|
| 371 |
|
---|
| 372 | jobject result_ptr;
|
---|
| 373 | char* query;
|
---|
| 374 | int i, j;
|
---|
| 375 |
|
---|
| 376 | jthrowable exc;
|
---|
[9874] | 377 | /* First of all, clear the previous result */
|
---|
| 378 | /* The result to write to */
|
---|
| 379 | result_ptr = (*j_env)->GetObjectField(j_env, j_obj, FID_query_result);
|
---|
| 380 | assert(result_ptr != NULL);
|
---|
| 381 |
|
---|
| 382 | /* Clear any previous query results */
|
---|
| 383 | (*j_env)->CallVoidMethod(j_env, result_ptr, MID_clearResult);
|
---|
| 384 | exc = (*j_env)->ExceptionOccurred(j_env);
|
---|
| 385 | if (exc) {
|
---|
| 386 | (*j_env)->ExceptionDescribe(j_env);
|
---|
| 387 | return;
|
---|
| 388 | }
|
---|
| 389 |
|
---|
[3743] | 390 | /* Make sure an index has been specified */
|
---|
| 391 | index_path = data->queryInfo->index;
|
---|
[11021] | 392 | if (index_path == NULL) {
|
---|
| 393 | return;
|
---|
| 394 | }
|
---|
[3743] | 395 |
|
---|
| 396 | /* Obtain C versions of the two string parameters */
|
---|
| 397 | base_dir = (*j_env)->GetStringUTFChars(j_env, j_base_dir, NULL);
|
---|
| 398 | if (base_dir == NULL) {
|
---|
| 399 | return;
|
---|
| 400 | }
|
---|
| 401 | text_path = (*j_env)->GetStringUTFChars(j_env, j_text_path, NULL);
|
---|
| 402 | if (text_path == NULL) {
|
---|
| 403 | (*j_env)->ReleaseStringUTFChars(j_env, j_base_dir, base_dir);
|
---|
| 404 | return;
|
---|
| 405 | }
|
---|
| 406 |
|
---|
| 407 | /* Load the appropriate index for satisfying this request */
|
---|
| 408 | qd = loadIndexData((char*) base_dir, (char*) index_path, (char*) text_path);
|
---|
| 409 |
|
---|
| 410 | /* The C text strings are no longer needed */
|
---|
| 411 | (*j_env)->ReleaseStringUTFChars(j_env, j_base_dir, base_dir);
|
---|
| 412 | (*j_env)->ReleaseStringUTFChars(j_env, j_text_path, text_path);
|
---|
| 413 |
|
---|
| 414 | /* Check that the index was loaded successfully */
|
---|
[9874] | 415 | if (qd == NULL) {
|
---|
| 416 | return;
|
---|
| 417 | }
|
---|
[11021] | 418 |
|
---|
[3743] | 419 | /* Remove anything hanging around from last time */
|
---|
| 420 | FreeQueryDocs(qd);
|
---|
| 421 |
|
---|
| 422 | /* Obtain a C version of the query string */
|
---|
| 423 | query = (char*) (*j_env)->GetStringUTFChars(j_env, j_query, NULL);
|
---|
[11021] | 424 | if (query == NULL) {
|
---|
| 425 | return;
|
---|
| 426 | }
|
---|
[3743] | 427 | printf("Searching for query \"%s\"...\n", query);
|
---|
| 428 |
|
---|
| 429 | /* Make sure the query isn't empty */
|
---|
| 430 | if (strlen(query) == 0) {
|
---|
| 431 | printf("Warning: Empty query.\n");
|
---|
| 432 | return;
|
---|
| 433 | }
|
---|
| 434 |
|
---|
[3791] | 435 | /* "Some" queries are done as ranked queries */
|
---|
[3743] | 436 | if (data->defaultBoolCombine == 0) {
|
---|
[3791] | 437 | RankedQueryInfo rqi;
|
---|
| 438 | rqi.QueryFreqs = 1; /* Use the frequency of each query term in the query - OK? */
|
---|
| 439 | rqi.Exact = 1; /* Perform exact ranking */
|
---|
| 440 | rqi.MaxDocsToRetrieve = data->queryInfo->maxDocs; /* Get only the desired number */
|
---|
| 441 | rqi.MaxParasToRetrieve = rqi.MaxDocsToRetrieve; /* OK? */
|
---|
[8919] | 442 | /* we may need to get more paragraphs to get enough documents. I copied the following from mgquery. it seems to work, not sure why - kjdon */
|
---|
| 443 | if (qd->id->ifh.InvfLevel == 3 && GetEnv ("maxparas")) {
|
---|
| 444 | rqi.MaxParasToRetrieve = atoi (GetEnv ("maxparas"));
|
---|
| 445 | }
|
---|
| 446 |
|
---|
[3791] | 447 | rqi.Sort = 1; /* Sort the query terms by frequency before ranking */
|
---|
| 448 | rqi.AccumMethod = 'L'; /* Use a list when accumulating (has bugs though...) */
|
---|
[7629] | 449 | /* rqi.MaxAccums = -1; */ /* Use as many accumulators as necessary - CRASHES with list */
|
---|
[3791] | 450 | rqi.MaxAccums = 100000;
|
---|
| 451 | rqi.MaxTerms = -1; /* Use all the query terms */
|
---|
[7629] | 452 | /* rqi.StopAtMaxAccum = 0;*/ /* Don't care (using as many accumulators as necessary) */
|
---|
[3791] | 453 | rqi.StopAtMaxAccum = 1;
|
---|
| 454 | rqi.HashTblSize = 1000; /* Don't care (not using a hash table) */
|
---|
| 455 | rqi.skip_dump = NULL; /* Don't dump skip information */
|
---|
| 456 |
|
---|
| 457 | /* RankedQuery() reads 'casefold' and 'stem' parameters from the environment */
|
---|
| 458 | SetEnv("casefold", ((data->defaultStemMethod & 1) ? "on" : "off"), NULL);
|
---|
| 459 | SetEnv("stem", ((data->defaultStemMethod & 2) ? "on" : "off"), NULL);
|
---|
| 460 |
|
---|
| 461 | RankedQuery(qd, query, &rqi);
|
---|
[3743] | 462 | }
|
---|
[3791] | 463 | /* "All" queries are done as boolean queries */
|
---|
| 464 | else {
|
---|
| 465 | BooleanQueryInfo bqi;
|
---|
| 466 | bqi.MaxDocsToRetrieve = data->queryInfo->maxDocs;
|
---|
[3743] | 467 |
|
---|
[3791] | 468 | /* Had to add "words$o" to LIB_OBJS in mg/src/text/Makefile and recompile mg for this */
|
---|
| 469 | BooleanQuery(qd, query, &bqi, data->defaultStemMethod);
|
---|
| 470 | }
|
---|
[3743] | 471 |
|
---|
| 472 | /* Finished with the C query string */
|
---|
| 473 | (*j_env)->ReleaseStringUTFChars(j_env, j_query, query);
|
---|
| 474 |
|
---|
| 475 | /* Check the query was processed successfully */
|
---|
[11021] | 476 | if (qd->DL == NULL || qd->QTL == NULL || qd->TL == NULL) {
|
---|
| 477 | return;
|
---|
| 478 | }
|
---|
[3743] | 479 |
|
---|
| 480 | /* Record the total number of matching documents */
|
---|
| 481 | (*j_env)->CallVoidMethod(j_env, result_ptr, MID_setTotalDocs, (jlong) qd->DL->num);
|
---|
| 482 | exc = (*j_env)->ExceptionOccurred(j_env);
|
---|
| 483 | if (exc) {
|
---|
| 484 | (*j_env)->ExceptionDescribe(j_env);
|
---|
| 485 | return;
|
---|
| 486 | }
|
---|
| 487 |
|
---|
| 488 | /* Record the matching documents, but only the number requested */
|
---|
| 489 | printf("Number of matching documents: %d\n", qd->DL->num);
|
---|
[8919] | 490 |
|
---|
[3743] | 491 | for (i = 0; (i < qd->DL->num && i < data->queryInfo->maxDocs); i++) {
|
---|
| 492 | int doc_num = qd->DL->DE[i].DocNum;
|
---|
[8920] | 493 | float doc_weight = qd->DL->DE[i].Weight;
|
---|
| 494 |
|
---|
[8919] | 495 | #if defined(PARADOCNUM) || defined(NZDL)
|
---|
| 496 | if (qd->id->ifh.InvfLevel == 3) {
|
---|
| 497 | /* pararaph level, need to convert to doc level*/
|
---|
| 498 | doc_num = GetDocNumFromParaNum(qd, doc_num);
|
---|
| 499 | }
|
---|
| 500 | #endif
|
---|
[8920] | 501 |
|
---|
| 502 |
|
---|
[3743] | 503 | /* Call the addDoc function (Java side) to record a matching document */
|
---|
| 504 | (*j_env)->CallVoidMethod(j_env, result_ptr, MID_addDoc,
|
---|
| 505 | (jlong) doc_num, (jfloat) doc_weight);
|
---|
| 506 | exc = (*j_env)->ExceptionOccurred(j_env);
|
---|
| 507 | if (exc) {
|
---|
| 508 | (*j_env)->ExceptionDescribe(j_env);
|
---|
| 509 | return;
|
---|
| 510 | }
|
---|
| 511 | }
|
---|
| 512 |
|
---|
| 513 | /* Record the term information, if desired */
|
---|
| 514 | if (data->queryInfo->needTermFreqs) {
|
---|
[3791] | 515 | /* The following code is a lot more complicated than it could be, but it is necessary
|
---|
| 516 | to compensate for an oddity in MG. */
|
---|
| 517 | unsigned char** stemmed_terms = malloc(sizeof(unsigned char*) * qd->TL->num);
|
---|
| 518 |
|
---|
[3743] | 519 | printf("Number of terms: %d\n", qd->TL->num);
|
---|
| 520 | printf("Number of query terms: %d\n", qd->QTL->num);
|
---|
| 521 |
|
---|
[3791] | 522 | /* Generate the stemmed form of each of the relevant terms */
|
---|
| 523 | for (i = 0; i < qd->TL->num; i++) {
|
---|
| 524 | u_char* raw_term = qd->TL->TE[i].Word;
|
---|
| 525 | unsigned int term_length = raw_term[0];
|
---|
| 526 |
|
---|
| 527 | u_char* raw_stemmed_term = malloc(term_length + 1);
|
---|
| 528 | unsigned int stemmed_term_length;
|
---|
| 529 |
|
---|
| 530 | /* Copy the term, and stem it */
|
---|
| 531 | for (j = 0; j <= term_length; j++)
|
---|
| 532 | raw_stemmed_term[j] = raw_term[j];
|
---|
| 533 | stemmer(data->defaultStemMethod, qd->sd->sdh.stemmer_num, raw_stemmed_term);
|
---|
| 534 |
|
---|
| 535 | /* Allocate memory to store the stemmed term, and fill it */
|
---|
| 536 | stemmed_term_length = raw_stemmed_term[0];
|
---|
| 537 | stemmed_terms[i] = malloc(stemmed_term_length + 1);
|
---|
| 538 | assert(stemmed_terms[i] != NULL);
|
---|
| 539 | strncpy(stemmed_terms[i], &(raw_stemmed_term[1]), stemmed_term_length);
|
---|
| 540 | stemmed_terms[i][stemmed_term_length] = '\0';
|
---|
| 541 | }
|
---|
| 542 |
|
---|
| 543 | /* Record every query term, along with their equivalent terms */
|
---|
[3743] | 544 | for (i = 0; i < qd->QTL->num; i++) {
|
---|
[3791] | 545 | u_char* raw_query_term = qd->QTL->QTE[i].Term;
|
---|
| 546 | unsigned int query_term_length = raw_query_term[0];
|
---|
[3743] | 547 | unsigned char* query_term;
|
---|
| 548 | jstring j_query_term;
|
---|
| 549 |
|
---|
[3791] | 550 | u_char* raw_stemmed_query_term = malloc(query_term_length + 1);
|
---|
| 551 | unsigned int stemmed_query_term_length;
|
---|
| 552 | unsigned char* stemmed_query_term;
|
---|
| 553 |
|
---|
| 554 | /* Allocate memory to store the query term, and fill it */
|
---|
| 555 | query_term = malloc(query_term_length + 1);
|
---|
[3743] | 556 | assert(query_term != NULL);
|
---|
[3791] | 557 | strncpy(query_term, &(raw_query_term[1]), query_term_length);
|
---|
[3743] | 558 | query_term[query_term_length] = '\0';
|
---|
| 559 |
|
---|
| 560 | /* Allocate a new jstring for the query term */
|
---|
| 561 | j_query_term = (*j_env)->NewStringUTF(j_env, query_term);
|
---|
| 562 | assert(j_query_term != NULL);
|
---|
| 563 |
|
---|
[3791] | 564 | /* Call the addTerm function (Java side) to record the query term */
|
---|
| 565 | (*j_env)->CallVoidMethod(j_env, result_ptr, MID_addTerm,
|
---|
| 566 | j_query_term, (jint) data->defaultStemMethod);
|
---|
| 567 | exc = (*j_env)->ExceptionOccurred(j_env);
|
---|
| 568 | if (exc) {
|
---|
| 569 | (*j_env)->ExceptionDescribe(j_env);
|
---|
| 570 | return;
|
---|
| 571 | }
|
---|
| 572 |
|
---|
| 573 | /* Copy the query term, and stem it */
|
---|
| 574 | for (j = 0; j <= query_term_length; j++)
|
---|
| 575 | raw_stemmed_query_term[j] = raw_query_term[j];
|
---|
| 576 | stemmer(data->defaultStemMethod, qd->sd->sdh.stemmer_num, raw_stemmed_query_term);
|
---|
| 577 |
|
---|
| 578 | /* Allocate memory to store the stemmed query term, and fill it */
|
---|
| 579 | stemmed_query_term_length = raw_stemmed_query_term[0];
|
---|
| 580 | stemmed_query_term = malloc(stemmed_query_term_length + 1);
|
---|
| 581 | assert(stemmed_query_term != NULL);
|
---|
| 582 | strncpy(stemmed_query_term, &(raw_stemmed_query_term[1]), stemmed_query_term_length);
|
---|
| 583 | stemmed_query_term[stemmed_query_term_length] = '\0';
|
---|
| 584 |
|
---|
| 585 | /* Find all the terms equivalent to the query term */
|
---|
[3743] | 586 | for (j = 0; j < qd->TL->num; j++) {
|
---|
[3791] | 587 | /* Check if the stemmed query term matches the stemmed term */
|
---|
| 588 | if (strcmp(stemmed_query_term, stemmed_terms[j]) == 0) {
|
---|
| 589 | u_char* raw_term = qd->TL->TE[j].Word;
|
---|
| 590 | unsigned int term_length = raw_term[0];
|
---|
| 591 | unsigned char* term;
|
---|
| 592 | jstring j_term;
|
---|
[3743] | 593 |
|
---|
[3791] | 594 | /* Allocate memory to store the query term, and fill it */
|
---|
| 595 | term = malloc(term_length + 1);
|
---|
| 596 | assert(term != NULL);
|
---|
| 597 | strncpy(term, &(raw_term[1]), term_length);
|
---|
| 598 | term[term_length] = '\0';
|
---|
[3743] | 599 |
|
---|
[3791] | 600 | /* Allocate a new jstring for the query term */
|
---|
| 601 | j_term = (*j_env)->NewStringUTF(j_env, term);
|
---|
| 602 | assert(j_term != NULL);
|
---|
| 603 |
|
---|
| 604 | /* Call the addEquivTerm function (Java side) to record the equivalent term */
|
---|
| 605 | (*j_env)->CallVoidMethod(j_env, result_ptr, MID_addEquivTerm,
|
---|
| 606 | j_query_term, j_term,
|
---|
| 607 | (jlong) qd->TL->TE[j].WE.doc_count,
|
---|
| 608 | (jlong) qd->TL->TE[j].WE.count);
|
---|
[3743] | 609 | exc = (*j_env)->ExceptionOccurred(j_env);
|
---|
| 610 | if (exc) {
|
---|
| 611 | (*j_env)->ExceptionDescribe(j_env);
|
---|
| 612 | return;
|
---|
| 613 | }
|
---|
| 614 | }
|
---|
| 615 | }
|
---|
| 616 | }
|
---|
| 617 | }
|
---|
| 618 | }
|
---|
| 619 |
|
---|
| 620 |
|
---|
| 621 | /*******************************************
|
---|
| 622 | set query options
|
---|
| 623 | *******************************************/
|
---|
| 624 |
|
---|
| 625 | /* Turn casefolding on or off */
|
---|
| 626 | JNIEXPORT void JNICALL
|
---|
| 627 | Java_org_greenstone_mg_MGWrapper_setCase(JNIEnv *j_env, jobject j_obj,
|
---|
| 628 | jboolean j_on)
|
---|
| 629 | {
|
---|
[25244] | 630 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 631 |
|
---|
| 632 | if (j_on) {
|
---|
| 633 | data->defaultStemMethod |= 1;
|
---|
| 634 | } else {
|
---|
| 635 | data->defaultStemMethod &= 0xe;
|
---|
| 636 | }
|
---|
| 637 | }
|
---|
| 638 |
|
---|
| 639 |
|
---|
| 640 | /* Turn stemming on or off */
|
---|
| 641 | JNIEXPORT void JNICALL
|
---|
| 642 | Java_org_greenstone_mg_MGWrapper_setStem(JNIEnv *j_env, jobject j_obj,
|
---|
| 643 | jboolean j_on)
|
---|
| 644 | {
|
---|
[25244] | 645 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 646 |
|
---|
| 647 | if (j_on) {
|
---|
| 648 | data->defaultStemMethod |= 2;
|
---|
| 649 | } else {
|
---|
| 650 | data->defaultStemMethod &= 0xd;
|
---|
| 651 | }
|
---|
| 652 | }
|
---|
| 653 |
|
---|
| 654 |
|
---|
| 655 | /* Set the maximum number of documents to return from a query */
|
---|
| 656 | JNIEXPORT void JNICALL
|
---|
| 657 | Java_org_greenstone_mg_MGWrapper_setMaxDocs(JNIEnv *j_env, jobject j_obj,
|
---|
| 658 | jint j_max)
|
---|
| 659 | {
|
---|
[25244] | 660 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 661 | data->queryInfo->maxDocs = j_max;
|
---|
| 662 | }
|
---|
| 663 |
|
---|
[13288] | 664 | /* set the maximum number of numeric to split*/
|
---|
| 665 | JNIEXPORT void JNICALL
|
---|
| 666 | Java_org_greenstone_mg_MGWrapper_setMaxNumeric (JNIEnv *j_env,
|
---|
| 667 | jobject j_obj,
|
---|
| 668 | jint j_max) {
|
---|
[3743] | 669 |
|
---|
[13288] | 670 | char text[20];
|
---|
| 671 | char* maxnumeric;
|
---|
| 672 | sprintf(text,"%d",j_max);
|
---|
| 673 | maxnumeric = text;
|
---|
| 674 | SetEnv("maxnumeric",maxnumeric, NULL);
|
---|
| 675 | }
|
---|
| 676 |
|
---|
| 677 |
|
---|
[3743] | 678 | /* Turn term frequency recording on or off */
|
---|
| 679 | JNIEXPORT void JNICALL
|
---|
| 680 | Java_org_greenstone_mg_MGWrapper_setReturnTerms(JNIEnv *j_env, jobject j_obj,
|
---|
| 681 | jboolean j_on)
|
---|
| 682 | {
|
---|
[25244] | 683 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 684 | data->queryInfo->needTermFreqs = j_on;
|
---|
| 685 | }
|
---|
| 686 |
|
---|
| 687 |
|
---|
| 688 | /* Choose MG index to search */
|
---|
| 689 | JNIEXPORT void JNICALL
|
---|
| 690 | Java_org_greenstone_mg_MGWrapper_setIndex(JNIEnv *j_env, jobject j_obj,
|
---|
| 691 | jstring j_index)
|
---|
| 692 | {
|
---|
[25244] | 693 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 694 |
|
---|
| 695 | /* Get the index name as a C string */
|
---|
| 696 | const char* index = (*j_env)->GetStringUTFChars(j_env, j_index, NULL);
|
---|
| 697 | assert(index != NULL);
|
---|
| 698 | printf("Choosing index %s...\n", index);
|
---|
| 699 |
|
---|
| 700 | /* Free the previous index name */
|
---|
| 701 | if (data->queryInfo->index)
|
---|
| 702 | free(data->queryInfo->index);
|
---|
| 703 |
|
---|
| 704 | /* Allocate memory for the index name, and fill it */
|
---|
| 705 | data->queryInfo->index = (char*) malloc(strlen(index) + 1);
|
---|
| 706 | assert(data->queryInfo->index != NULL);
|
---|
| 707 | strcpy(data->queryInfo->index, index);
|
---|
| 708 |
|
---|
| 709 | /* Release the index string */
|
---|
| 710 | (*j_env)->ReleaseStringUTFChars(j_env, j_index, index);
|
---|
| 711 | }
|
---|
| 712 |
|
---|
| 713 |
|
---|
| 714 | /* Choose boolean AND or boolean OR queries */
|
---|
| 715 | JNIEXPORT void JNICALL
|
---|
| 716 | Java_org_greenstone_mg_MGWrapper_setMatchMode(JNIEnv *j_env, jobject j_obj,
|
---|
| 717 | jint j_mode)
|
---|
| 718 | {
|
---|
[25244] | 719 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 720 | data->defaultBoolCombine = j_mode;
|
---|
| 721 | }
|
---|
| 722 |
|
---|
| 723 |
|
---|
| 724 | /* Get a text representation of the current parameter values */
|
---|
| 725 | JNIEXPORT jstring JNICALL
|
---|
| 726 | Java_org_greenstone_mg_MGWrapper_getQueryParams(JNIEnv *j_env, jobject j_obj)
|
---|
| 727 | {
|
---|
[25244] | 728 | MGWrapperData* data = (MGWrapperData*) (*j_env)->GetLongField(j_env, j_obj, FID_mg_data);
|
---|
[3743] | 729 | char result[512]; /* Assume this is big enough */
|
---|
| 730 |
|
---|
| 731 | /* Print the data to a character array */
|
---|
[25147] | 732 | sprintf(result, "Query params:\nindex\t\t%s\ncasefold\t%d\nstem\t\t%d\nquery type\t%s\nmax docs\t%d\n",
|
---|
[3743] | 733 | (data->queryInfo->index == NULL ? "<none loaded>" : data->queryInfo->index),
|
---|
| 734 | (data->defaultStemMethod & 1),
|
---|
| 735 | (data->defaultStemMethod & 2),
|
---|
| 736 | (data->defaultBoolCombine == 1 ? "all" : "some"),
|
---|
| 737 | (data->queryInfo->maxDocs));
|
---|
| 738 |
|
---|
| 739 | /* Convert to a jstring, and return */
|
---|
| 740 | return (*j_env)->NewStringUTF(j_env, result);
|
---|
| 741 | }
|
---|