diff -c2 irsearch.c.original irsearch.c *** irsearch.c.original Sun May 10 17:44:33 1992 --- irsearch.c Fri Dec 4 16:11:45 1992 *************** *** 14,17 **** --- 14,21 ---- * Handle searches on multiple databases */ + + /* Changes by David H Post, Design Research Institute, post@cs.cornell.edu + * Fall 1992: Added code to call SMART for queries. + */ /* Change Log: *************** *** 113,117 **** #define TEST_SEARCH false /* set to TRUE to allow printing to console */ ! char *server_name = NULL; long tcp_port = 0; --- 117,121 ---- #define TEST_SEARCH false /* set to TRUE to allow printing to console */ ! char *server_name = NULL; /* set in server.c */ long tcp_port = 0; *************** *** 125,128 **** --- 129,133 ---- #endif + extern char* smart_cmd; /* defined in server.c, used in run_search */ /*----------------------------------------------------------------------*/ *************** *** 689,695 **** - /*----------------------------------------------------------------------*/ boolean run_search(aSearch, headers, diags, index_directory, seed_words_used, waisProtocolVersion, headerNum) --- 694,891 ---- + /*=========================================================================*/ + /* [dri] Mods to call SMART added here. (Functions called in run_search.) */ + /* Call SMART with query_string. SMART writes response to file. */ + /* Prefix and suffix simplify various uses (search, retrieval). */ + /* Returns false if unsuccessful. */ + + boolean smart_query(db_name, query_string, response_filename) + char* db_name; + char* query_string; + char* response_filename; + { + char* tmpFileName; + FILE* sqfile; /* SMART query file */ + char cmdbufr[200]; + + /* Construct query file to send to SMART */ + tmpFileName = tempnam("/tmp/", "smq"); + sqfile = s_fopen(tmpFileName, "w"); + if (sqfile == NULL) { + waislog(WLOG_HIGH, WLOG_ERROR, "Could not create SMART query file %s", + tmpFileName); + return(false); + } + fprintf(sqfile, "%s\n", query_string); + s_fclose(sqfile); + + /* Send query to SMART, pipe results into response_file */ + sprintf(cmdbufr, "%s %s <%s >%s", + smart_cmd, /* set by -smart cmd-line option */ + db_name, + tmpFileName, + response_filename); + if(system(cmdbufr) != 0) { + /* + waislog(WLOG_HIGH, WLOG_ERROR, "Error %d executing system command: %s", + errno, cmdbufr); + return(false); + */ + } + + unlink (tmpFileName); /* remove temporary file */ + return(true); + + } /* smart_query */ + + /*-------------------------------------------------------------------------*/ + + /* Construct header from document number, score, title, etc. */ + + WAISDocumentHeader* + smart_make_header(waisProtocolVersion, server, db, + docnum, score, length, date, headline) + long waisProtocolVersion; /* IN */ + char* server; /* IN */ + database* db; /* IN */ + long docnum; /* IN: SMART doc number */ + long score; /* IN: (normalized) doc score */ + long length; /* IN: document size in chars */ + char* date; /* IN */ + char* headline; /* IN: doc title */ + { + char local_id[20]; + char** type = NULL; + DocID* theDocID = NULL; + char* originName = db->database_file; + WAISDocumentHeader* header; + + /* use "1 0 " for local_id */ + sprintf(local_id, "1 0 %ld", docnum); + + if (waisProtocolVersion >= '2') { + type = (char**)s_malloc((size_t)(sizeof(char*) * 2)); + type[0] = s_strdup("TEXT"); + type[1] = NULL; + } + else + type = NULL; + + theDocID = makeDocID(); + theDocID->distributorServer = stringToAny(server); + theDocID->originalServer = stringToAny(server); + theDocID->distributorDatabase = stringToAny(originName); + theDocID->originalDatabase = stringToAny(originName); + theDocID->distributorLocalID = stringToAny(local_id); + theDocID->originalLocalID = stringToAny(local_id); + + header = makeWAISDocumentHeader(anyFromDocID(theDocID), + UNUSED, /* version */ + (long)score, + UNUSED, /* best_line */ + length, + UNUSED, /* # of lines */ + type, + s_strdup(originName), + s_strdup(date), + s_strdup(headline), + NULL); /* origin city */ + freeDocID(theDocID); + return(header); + + } /* smart_make_header */ + + /*---------------------------------------------------------------------------*/ + + /* Read results of SMART query from response file, write doc list to headers. */ + /* Returns false if unsuccessful. */ + + boolean smart_get_doclist(response_filename, flag, + waisProtocolVersion, server, db, + headers, headerNum) + + char* response_filename; /* IN: doclist file */ + char* flag; /* IN: skip to here */ + long waisProtocolVersion; /* IN */ + char* server; /* IN */ + database* db; /* IN */ + WAISDocumentHeader** headers; /* OUT: document list */ + long* headerNum; /* IN: initialized to 0, + OUT: number of documents */ + { + FILE* srfile; /* SMART response file */ + char line[100]; + long docnum; + long score, maxRawScore; + long length; + char date[10]; + char title[100]; + + /* Open response file */ + srfile = s_fopen(response_filename, "r"); + if (srfile == NULL) { + waislog(WLOG_HIGH, WLOG_ERROR, "Could not open SMART result file %s", + response_filename); + return(false); + } + + /* Read & discard lines until we find a line starting with flag (or EOF) */ + fgets(line, 100, srfile); + while (strncmp(line, flag, strlen(flag)) != 0) { + fgets(line, 100, srfile); + if (feof(srfile)) { /* probably "No documents to display" */ + s_fclose(srfile); + return(false); + } + } + + maxRawScore = 0; + + /* Read one document per line (while 5 fields are read) */ + while (5 == fscanf(srfile, "%ld 0.%ld %ld %s %[^\n]\n", + &docnum, &score, &length, date, title)) + { + /* save first score, which is highest (doclist already sorted) */ + if (maxRawScore == 0) maxRawScore = score; + + if ( _BE_normalized ) + ; /* score is fine as is */ + else { /* normalize score */ + score = (long)floor( ( ((double)score) / ((double)maxRawScore) ) * + (MAX_NORMAL_SCORE + 1) ); + if (score > MAX_NORMAL_SCORE) score = MAX_NORMAL_SCORE; + } + + if (strcmp(date, "00/00/00") == 0) + strcpy(date, "0"); /* "0" suppresses display of unknown date */ + + /* Create header, add it to list of headers */ + { + WAISDocumentHeader* header = + smart_make_header(waisProtocolVersion, server, db, + docnum, score, length, date, title); + if(NULL != header) { + headers[(*headerNum)++] = header; + headers[*headerNum] = NULL; + } + } + + } /* while */ + + /* + waislog(WLOG_LOW, WLOG_INFO, "Returned %ld document headers from SMART", + (*headerNum) ); + */ + + s_fclose(srfile); + return( (*headerNum) > 0L ); + + } /* smart_get_doclist */ + + /*=======================================================================*/ + + boolean run_search(aSearch, headers, diags, index_directory, seed_words_used, waisProtocolVersion, headerNum) *************** *** 734,738 **** dbName[strlen(dbName) - strlen(".src")] = '\0'; ! if(server_name != NULL) sprintf(server, "%s:%d", server_name, tcp_port); else --- 930,934 ---- dbName[strlen(dbName) - strlen(".src")] = '\0'; ! if(server_name != NULL) /* set in server.c */ sprintf(server, "%s:%d", server_name, tcp_port); else *************** *** 769,795 **** search_result = false; init_search_word(db); #ifdef RELEVANCE_FEEDBACK ! if(wais_search->Docs != NULL) { ! DocObj* doc = NULL; ! boolean res; ! /* assemble the elements and construct a response */ ! for (i = 0, doc = wais_search->Docs[i]; ! doc != NULL; ! doc = wais_search->Docs[++i]){ ! search_result |= ! search_for_words_in_document(doc,i+1,db,diags,&num_diags); ! } ! if (*diags != NULL) { ! num_diags++; ! *diags = (diagnosticRecord**)s_realloc(*diags,(size_t)(sizeof(diagnosticRecord*) * num_diags)); ! (*diags)[num_diags-1] = NULL; } ! } #endif /* RELEVANT_FEEDBACK */ ! ! search_result |= search_for_words(wais_search->SeedWords, db, 0); ! ! if (search_result == true){ /* the search went ok */ hit best_hit; finished_search_word(db); --- 965,1041 ---- search_result = false; init_search_word(db); + + + /* [dri] Mod to call SMART added here ------------------------------------*/ + + if (smart_cmd != NULL) /* if we are a SMART server */ + { + /* SMART query */ + char query[1000]; /* string of SMART commands */ + char fb_docs[200]; /* string of RF document numbers */ + char* doclist_filename = tempnam("/tmp/", "smdl"); + fb_docs[0] = '\0'; + + if (wais_search->Docs != NULL) { /* if any relevance feedback docs */ + DocObj* doc; + char* local_id; + + for (i = 0, doc = wais_search->Docs[i]; + doc != NULL; + doc = wais_search->Docs[++i]) { + local_id = anyToString(GetLocalID(docIDFromAny(doc->DocumentID))); + local_id = strrchr(local_id, ' '); /* remove start & end ptrs */ + /* append space & document number to list (if room) */ + if (strlen(fb_docs) + strlen(local_id) < 200) + strcat(fb_docs, local_id); + else + waislog(WLOG_HIGH, WLOG_ERROR, + "Too many relevance feedback docs, ignoring %s", local_id); + } + + waislog(WLOG_LOW, WLOG_INFO, "Relevance feedback docs: %s", fb_docs); + } + + /* "run_fb" is search command, "\n.\n" terminates search words, + "titles" lists document titles found. */ + sprintf(query, "run_fb %s\n %s\n.\ntitles", + fb_docs, + wais_search->SeedWords); + + /* call SMART */ + search_result = smart_query(dbName, query, doclist_filename); + + if (search_result == true) + search_result = smart_get_doclist(doclist_filename, "Num ", + waisProtocolVersion, server, db, + headers, headerNum); + unlink(doclist_filename); + } + + else { /* normal WAIS search */ + #ifdef RELEVANCE_FEEDBACK ! if(wais_search->Docs != NULL) { ! DocObj* doc = NULL; ! boolean res; ! /* assemble the elements and construct a response */ ! for (i = 0, doc = wais_search->Docs[i]; ! doc != NULL; ! doc = wais_search->Docs[++i]){ ! search_result |= ! search_for_words_in_document(doc,i+1,db,diags,&num_diags); ! } ! if (*diags != NULL) { ! num_diags++; ! *diags = (diagnosticRecord**)s_realloc(*diags,(size_t)(sizeof(diagnosticRecord*) * num_diags)); ! (*diags)[num_diags-1] = NULL; } ! } #endif /* RELEVANT_FEEDBACK */ ! ! search_result |= search_for_words(wais_search->SeedWords, db, 0); ! ! if (search_result == true){ /* the search went ok */ hit best_hit; finished_search_word(db); *************** *** 811,827 **** } } ! else ! { /* something went awry in the search */ ! num_diags++; ! diag = makeDiag(true,D_PermanentSystemError, ! "Serious error in server"); ! *diags = (diagnosticRecord**) ! s_realloc(*diags, (size_t)(sizeof(diagnosticRecord*) * num_diags)); ! (*diags)[num_diags-2] = diag; ! (*diags)[num_diags-1] = NULL; ! } ! finished_best_hit(db); ! /* free everything */ closeDatabase(db); #ifdef GET_QUERY_TIMING ftime(&e_time); --- 1057,1077 ---- } } ! else ! { /* something went awry in the search */ ! num_diags++; ! diag = makeDiag(true,D_PermanentSystemError, ! "Serious error in server"); ! *diags = (diagnosticRecord**) ! s_realloc(*diags, (size_t)(sizeof(diagnosticRecord*) * num_diags)); ! (*diags)[num_diags-2] = diag; ! (*diags)[num_diags-1] = NULL; ! } ! ! finished_best_hit(db); /* free everything */ ! ! } /* else: normal WAIS search */ ! closeDatabase(db); + #ifdef GET_QUERY_TIMING ftime(&e_time);