source: trunk/mgpp/jni/MGPPWrapperImpl.cpp@ 9778

#ifdef __WIN32__
#include <WIN32cfg.h>
#endif

#include <sstream>

#include <jni.h>
#include "org_greenstone_mgpp_MGPPWrapper.h"
#include "MGPPWrapperImpl.h"
#include "TextGet.h"
#include "GSDLQueryParser.h"
#include "MGQuery.h"
  
MGPPWrapperData::MGPPWrapperData() {
  indexData = new IndexData();
  queryInfo = new QueryInfo();

  if (queryInfo==NULL) {
    cerr<<"couldn't allocate new query info\n";
    if (indexData!=NULL) {
      delete indexData;
    }
  }

  // set all the default params
  SetCStr(queryInfo->docLevel, "Document"); // the level to search at
  queryInfo->maxDocs = 50;
  queryInfo->sortByRank = true;
  queryInfo->exactWeights = false;
  queryInfo->needRankInfo = true;
  queryInfo->needTermFreqs = true;

  UCArrayClear(level);
  SetCStr(level, "Document"); // the level to return docs at
  defaultStemMethod=0;
  defaultBoolCombine=0;
  
}

MGPPWrapperData::~MGPPWrapperData() { 
  if (indexData !=NULL) {
    delete indexData;
  }
  if (queryInfo !=NULL) {
    delete queryInfo;
  }
}

// ********************************************
// initialisation stuff
// ********************************************

// cached ids for java stuff 
jfieldID FID_mgpp_data = NULL; // MGPPWrapperData
jfieldID FID_query_result = NULL; // MGPPQueryResult
jmethodID MID_addDoc=NULL; // MGPPQueryResult.addDoc()
jmethodID MID_addTerm=NULL; // MGPPQueryResult.addTerm()
jmethodID MID_setTotalDocs=NULL; // MGPPQueryResult.setTotalDocs()
jmethodID MID_clearResult=NULL; //MGPPQueryResult.clear()
jmethodID MID_setSyntaxError=NULL; // MGPPQueryResult.setSyntaxError()
jclass CID_String=NULL; // class ID of String

/* to access objects and methods on java side, need their field/method ids - 
 this initialises them at the start to avoid recalculating them each time they
 are needed 
Note: the descriptors need to be exactly right, otherwise you get an error 
saying "no such field" but no reference to the fact that it has the right 
name but the wrong type.
Note: apparently the jclass is a local ref and should only work 
in the method that created it. It seems to work ok, but I'll make it
 global cos the book said I should, and it may avoid future hassles.
*/
JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_initIDs (JNIEnv *j_env, jclass j_cls) {
  
  FID_mgpp_data = j_env->GetFieldID(j_cls, "mgpp_data_ptr_", "J"); //a long-"J"
  if (FID_mgpp_data==NULL) {
      cerr <<"MGPP JNI: field mgpp_data_ptr_ not found"<<endl;
  } 

  FID_query_result = j_env->GetFieldID(j_cls, "mgpp_query_result_", "Lorg/greenstone/mgpp/MGPPQueryResult;"); // an object -"L<class name>;"
  if (FID_query_result==NULL) {
      cerr <<"MGPP JNI: field mgpp_query_result_ not found"<<endl;
  }
  // the methods we want to use

  // addDoc(long doc, float rank)
  jclass JC_MGPPQueryResult = j_env->FindClass("org/greenstone/mgpp/MGPPQueryResult");
  MID_addDoc = j_env->GetMethodID(JC_MGPPQueryResult, "addDoc", "(JF)V");
  if (MID_addDoc==NULL) {
      cerr <<"MGPP JNI: addDoc method not found"<<endl;
  }
  // addTerm(String term, String tag, int stem_method, long match_docs, 
  // long term_freq, String[] equiv_terms)
  MID_addTerm = j_env->GetMethodID(JC_MGPPQueryResult, "addTerm", "(Ljava/lang/String;Ljava/lang/String;IJJ[Ljava/lang/String;)V");
  if (MID_addTerm==NULL) {
      cerr <<"MGPP JNI: method addTerm not found"<<endl;
  }
  
  // setTotalDocs(long)
  MID_setTotalDocs = j_env->GetMethodID(JC_MGPPQueryResult, "setTotalDocs", "(J)V");
  if (MID_setTotalDocs==NULL) {
      cerr <<"MGPP JNI: method setTotalDocs not found"<<endl;
  }
  
  MID_clearResult = j_env->GetMethodID(JC_MGPPQueryResult, "clear", "()V");
  if (MID_clearResult==NULL) {
      cerr <<"MGPP JNI: method clear not found"<<endl;
  }
  MID_setSyntaxError = j_env->GetMethodID(JC_MGPPQueryResult, "setSyntaxError", "(Z)V");
  if (MID_clearResult==NULL) {
      cerr <<"MGPP JNI: method setSyntaxError not found"<<endl;
  }

  // get the class for String to use in NewObjectArray in runQuery()
  // FindClass returns a local reference - have to convert it to a global one
  jclass local_CID_String = j_env->FindClass("java/lang/String");
  if (local_CID_String==NULL) {
    cerr <<"MGPP JNI: java String class not found"<<endl;
  } else { 
    /* create a global ref */
    CID_String = (jclass)j_env->NewGlobalRef(local_CID_String);
    /* The local reference is no longer useful */
    j_env->DeleteLocalRef(local_CID_String);
 
    /* Is the global reference created successfully? */
    if (CID_String == NULL) {
      return; /* out of memory exception thrown */
    }
  }
  
}

/* the java side MGPPWrapper has a pointer to a C++ object - MGPPWrapperData
   initialise this and set the pointer 
*/
JNIEXPORT jboolean JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_initCppSide (JNIEnv *j_env, jobject j_obj){

  MGPPWrapperData * data = new MGPPWrapperData();
  j_env->SetIntField(j_obj, FID_mgpp_data, (long)data);

  return true;

}

//****************************************************
// retrieve a document
//****************************************************

/* returns a document from mgpp as a string
Note: TextData isn't cached - just reloaded each time
*/
JNIEXPORT jstring JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_getDocument (JNIEnv *j_env, 
 jobject j_obj, jstring j_text_name, jstring j_level, jlong j_docnum) {
  
#ifdef __WIN32__
  const char* base_dir = "";
#else
  const char* base_dir = "/";
#endif

  const char * text_name = j_env->GetStringUTFChars(j_text_name, NULL);
  if (text_name==NULL) {
   return NULL;
  }

  const char * level = j_env->GetStringUTFChars( j_level, NULL);
  if (level==NULL) {
   j_env->ReleaseStringUTFChars(j_text_name, text_name);
   return NULL;
  }
    
  // does this work alright? j_docnum is a long (64 bit)
  unsigned long docnum = j_docnum;
  TextData td;

  // cast to char* otherwise complains about const
  td.LoadData((char *)base_dir, (char *)text_name);

  UCArray mg_level;
  SetCStr(mg_level, level);
  UCArray docText;
  docText.clear();
  // get the actual text
  if (!GetDocText(td, mg_level, docnum, docText)) {
    cerr <<"MGPP JNI: couldn't retrieve doc text"<<endl;
  }

  td.UnloadData();
  
  char * doc = GetCStr(docText); // do I need to free this char *??
  jstring result = j_env->NewStringUTF(doc);
  // release any gets
  j_env->ReleaseStringUTFChars(j_text_name, text_name);
  j_env->ReleaseStringUTFChars(j_level, level);

  // free any C++ stuff
  delete doc;

  return result;
 
}

//******************************************
// do a query
// ****************************************

/* load the IndexData - cached for querying
 */
JNIEXPORT jboolean JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_loadIndexData (JNIEnv *j_env, jobject j_obj,  jstring j_index_name) {

  jint data_ptr = j_env->GetIntField(j_obj, FID_mgpp_data);
  MGPPWrapperData * data = (MGPPWrapperData *)data_ptr;

#ifdef __WIN32__
  const char* base_dir = "";
#else
  const char* base_dir = "/";
#endif

  const char * index_name = j_env->GetStringUTFChars( j_index_name, NULL);
  if (index_name==NULL) {
    return false;
  }
  
  jboolean j_result=false;
  
  // why doesn't this complain about const??
  if (data->indexData->LoadData(base_dir, index_name)) {
    j_result=true;
  }
  
  // release any gets
  j_env->ReleaseStringUTFChars(j_index_name, index_name);
  
  return j_result;
}

/* unload the data 
 */
JNIEXPORT jboolean JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_unloadIndexData (JNIEnv *j_env, jobject j_obj) {

  jint data_ptr = j_env->GetIntField(j_obj, FID_mgpp_data);
  MGPPWrapperData * data = (MGPPWrapperData *)data_ptr;

  data->indexData->UnloadData();
  return true;

}

/* do the actual query - the results are written to query_result held on the 
   java side */
JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_runQuery (JNIEnv *j_env, jobject j_obj, jstring j_query){
  
  jthrowable exc; // an exception - check if something funny has happened
  const char *query = j_env->GetStringUTFChars(j_query, NULL);
  if (query==NULL) {
    return; // exception already thrown
  }
  // turn to UCArray for mgpp and then release the string
  UCArray queryArray;
  SetCStr(queryArray, query);
  j_env->ReleaseStringUTFChars(j_query, query);
  
  // the query data
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  
  // the result to write to
  jobject result_ptr = j_env->GetObjectField(j_obj, FID_query_result);
  if (result_ptr==NULL) {
    cerr <<"couldn't access the result to write to"<<endl;
    return;
  }
  
  // clear the result
  j_env->CallVoidMethod(result_ptr, MID_clearResult);
  exc = j_env->ExceptionOccurred(); // this catches teh exception I think - it
  //wont be thrown any further
  if (exc) {
    j_env->ExceptionDescribe();
    return;
  }
  // the mgpp QueryResult that we will use
  ExtQueryResult queryResult;

  QueryNode * queryTree = NULL;
  // parse the query string into a tree structure
  queryTree = ParseQuery(queryArray, data->defaultBoolCombine, 
             data->defaultStemMethod);
  if (queryTree == NULL) {
    // invalid syntax
    j_env->CallVoidMethod(result_ptr, MID_setSyntaxError, true);
    cerr << "MGPPWrapperImpl: invalid query syntax!!\n";
    return; 
  }
  // print the query
  PrintNode (cout, queryTree);
  // finally, do the query
  MGQuery(*(data->indexData), *(data->queryInfo), queryTree, queryResult, data->level);

  delete queryTree;

  // convert queryResult to the java side version
  // use levels rather than docs of ExtQueryResult
  // CallVoidMethod(obj, method id, args to method)
  for (int i=0; i<queryResult.levels.size(); i++) {
    jlong doc = queryResult.levels[i];
    jfloat rank = queryResult.ranks[i];
    j_env->CallVoidMethod(result_ptr, MID_addDoc, doc, rank);
    exc = j_env->ExceptionOccurred();
    if (exc) {
      j_env->ExceptionDescribe();
      return;
    }
    
  }

  // actual num of docs
  jlong total = queryResult.actualNumDocs;
  j_env->CallVoidMethod(result_ptr, MID_setTotalDocs, total);
  exc = j_env->ExceptionOccurred();
  if (exc) {
    j_env->ExceptionDescribe();
    return;
  }

  // the terms
  for (int j=0; j<queryResult.termFreqs.size(); j++) {
    
    TermFreqData tf = queryResult.termFreqs[j];
    jstring term = j_env->NewStringUTF(GetCStr(tf.term));
    jstring tag = j_env->NewStringUTF(GetCStr(tf.tag));
    jint stem = tf.stemMethod;
    jlong match = tf.matchDocs;
    jlong freq = tf.termFreq;
    
    jobjectArray equivs=NULL;
    jstring empty = j_env->NewStringUTF(""); // the initial object to fill the array
    jint num_equivs = tf.equivTerms.size();
    equivs = j_env->NewObjectArray(num_equivs, CID_String, empty);
    if (equivs==NULL) {
      cerr<<"couldn't create object array"<<endl;
      
    } else {
      for (int k=0; k<num_equivs;k++) {
    jstring equiv = j_env->NewStringUTF(GetCStr(tf.equivTerms[k]));
    j_env->SetObjectArrayElement(equivs, k, equiv);
      }
      
      
      j_env->CallVoidMethod(result_ptr, MID_addTerm, term, tag, stem, match, freq, equivs);
      exc = j_env->ExceptionOccurred();
      if (exc) {
    j_env->ExceptionDescribe();
    return;
      }
    }
  }


}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setStem (JNIEnv *j_env, 
                          jobject j_obj, 
                          jboolean j_on) {
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  if (j_on) {
    data->defaultStemMethod |= 2;
  } else {
    data->defaultStemMethod &= 0xd;
  }

}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setCase (JNIEnv *j_env, 
                          jobject j_obj, 
                          jboolean j_on) {
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);

  if (j_on) {
    data->defaultStemMethod |= 1;
  } else {
    data->defaultStemMethod &= 0xe;
  }
}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setMaxDocs (JNIEnv *j_env, 
                         jobject j_obj, 
                         jint j_max) {
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  data->queryInfo->maxDocs=j_max;
}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setSortByRank (JNIEnv *j_env, 
                            jobject j_obj, 
                            jboolean j_on) {
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  
  data->queryInfo->sortByRank=j_on;
}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setReturnTerms(JNIEnv *j_env, 
                            jobject j_obj, 
                            jboolean j_on) {
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  data->queryInfo->needTermFreqs = j_on;
 
}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setQueryLevel(JNIEnv *j_env, 
                           jobject j_obj, 
                           jstring j_level){
  
  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);

  const char * level = j_env->GetStringUTFChars(j_level, NULL);
  if (level==NULL) {
    return; // exception already thrown
  }
  
  data->queryInfo->docLevel.clear();
  SetCStr(data->queryInfo->docLevel, level);
  
  // release the java stuff
  j_env->ReleaseStringUTFChars(j_level, level);
 
}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setReturnLevel(JNIEnv *j_env, 
                            jobject j_obj, 
                            jstring j_level){

  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);

  const char * level = j_env->GetStringUTFChars(j_level, NULL);
  if (level==NULL) {
    return; // exception already thrown
  }
  
  data->level.clear();
  SetCStr(data->level, level);
  
  // release the java stuff
  j_env->ReleaseStringUTFChars(j_level, level);
    

}

JNIEXPORT void JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_setMatchMode (JNIEnv *j_env,
                           jobject j_obj,
                           jint j_mode){

  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  data->defaultBoolCombine=j_mode;

}

JNIEXPORT jstring JNICALL 
Java_org_greenstone_mgpp_MGPPWrapper_getQueryParams (JNIEnv *j_env,
                             jobject j_obj){

  MGPPWrapperData * data = (MGPPWrapperData *)j_env->GetIntField(j_obj, FID_mgpp_data);
  
  // print the data to a stringstream, then convert to char*, then to 
  //jstring

  stringstream output;
  output << "Query params:"<<endl
    // need to change this to use platform specific separator for niceness
     << "index\t\t"<<data->indexData->basePath<<"/"<<data->indexData->filename<<endl
     <<"search level\t"<<GetCStr(data->queryInfo->docLevel)<<endl
     <<"result level\t"<<GetCStr(data->level)<<endl
     <<"casefold\t"<<(data->defaultStemMethod&1)<<endl
     <<"stem\t\t"<<(data->defaultStemMethod&2)<<endl
     <<"order by rank\t"<<data->queryInfo->sortByRank<<endl
     <<"query type\t"<<(data->defaultBoolCombine==1?"all":"some")<<endl
         <<"max docs\t"<<data->queryInfo->maxDocs<<endl<<ends;

  const char *result = output.str().c_str();
  jstring j_result = j_env->NewStringUTF(result);
  delete (char *)result;
  return j_result;
}