X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fretrieve.c;h=364bd23fca2449cb1151afcef7662de85c82c26a;hb=3731bdaf94aeba2550fc553aebe34831c203dc36;hp=1d4f4ab25de0dc90985c08e777892bbb5a840a65;hpb=1872e3fc60b482771bbd1cb4b0290b8d6a9ef5d0;p=idzebra-moved-to-github.git diff --git a/index/retrieve.c b/index/retrieve.c index 1d4f4ab..364bd23 100644 --- a/index/retrieve.c +++ b/index/retrieve.c @@ -1,4 +1,4 @@ -/* $Id: retrieve.c,v 1.68 2007-04-16 08:44:32 adam Exp $ +/* $Id: retrieve.c,v 1.76 2007-11-30 12:19:08 adam Exp $ Copyright (C) 1995-2007 Index Data ApS @@ -34,14 +34,15 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include "index.h" #include +#include #include #include #define ZEBRA_XML_HEADER_STR "reg->zei, *rec); @@ -61,8 +62,8 @@ static int zebra_create_record_stream(ZebraHandle zh, else strcpy(full_rep, (*rec)->info[recInfo_filename]); - if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1){ - yaz_log (YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s", + if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1){ + yaz_log(YLOG_WARN|YLOG_ERRNO, "Retrieve fail; missing file: %s", full_rep); rec_free(rec); return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; @@ -121,8 +122,8 @@ static int parse_zebra_elem(const char *elem, int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, const char *elemsetname, - const int *input_format, - const int **output_format, + const Odr_oid *input_format, + const Odr_oid **output_format, char **rec_bufp, int *rec_lenp) { const char *retrieval_index; @@ -130,11 +131,12 @@ int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, const char *retrieval_type; size_t retrieval_type_len; char retrieval_index_cstr[256]; + char retrieval_type_cstr[256]; int ord; /* only accept XML and SUTRS requests */ - if (oid_oidcmp(input_format, yaz_oid_xml()) - && oid_oidcmp(input_format, yaz_oid_sutrs())) + if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) + && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", elemsetname); @@ -160,9 +162,12 @@ int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len); retrieval_index_cstr[retrieval_index_len] = '\0'; + memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len); + retrieval_type_cstr[retrieval_type_len] = '\0'; + ord = zebraExplain_lookup_attr_str(zh->reg->zei, zinfo_index_category_sort, - retrieval_type[0], + retrieval_type_cstr, retrieval_index_cstr); if (ord == -1) return -1; /* is not a sort index */ @@ -170,7 +175,7 @@ int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, { char dst_buf[IT_MAX_WORD]; char str[IT_MAX_WORD]; - int index_type; + const char *index_type; const char *db = 0; const char *string_index = 0; WRBUF wrbuf = wrbuf_alloc(); @@ -182,11 +187,10 @@ int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index); zebra_term_untrans(zh, index_type, dst_buf, str); - - if (!oid_oidcmp(input_format, yaz_oid_xml())) + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { - *output_format = yaz_oid_xml(); + *output_format = yaz_oid_recsyn_xml; wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR " sysno=\"" ZINT_FORMAT "\"" " set=\"zebra::index%s/\">\n", @@ -194,16 +198,16 @@ int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, wrbuf_printf(wrbuf, " ", index_type); + wrbuf_printf(wrbuf, " type=\"%s\">", index_type); wrbuf_xmlputs(wrbuf, dst_buf); wrbuf_printf(wrbuf, "\n"); wrbuf_printf(wrbuf, "\n"); } - else if (!oid_oidcmp(input_format, yaz_oid_sutrs())) + else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { - *output_format = yaz_oid_sutrs(); + *output_format = yaz_oid_recsyn_sutrs; - wrbuf_printf(wrbuf, "%s %c %s\n", string_index, index_type, + wrbuf_printf(wrbuf, "%s %s %s\n", string_index, index_type, dst_buf); } *rec_lenp = wrbuf_len(wrbuf); @@ -217,8 +221,8 @@ int zebra_special_sort_fetch(ZebraHandle zh, zint sysno, ODR odr, int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, Record rec, const char *elemsetname, - const int *input_format, - const int **output_format, + const Odr_oid *input_format, + const Odr_oid **output_format, char **rec_bufp, int *rec_lenp) { const char *retrieval_index; @@ -227,13 +231,14 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, size_t retrieval_type_len; zebra_rec_keys_t keys; int ret_code = 0; + char retrieval_type_cstr[256]; /* set output variables before processing possible error states */ /* *rec_lenp = 0; */ /* only accept XML and SUTRS requests */ - if (oid_oidcmp(input_format, yaz_oid_xml()) - && oid_oidcmp(input_format, yaz_oid_sutrs())) + if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) + && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", elemsetname); @@ -246,11 +251,12 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, &retrieval_type, &retrieval_type_len)) return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; - if (retrieval_type_len != 0 && retrieval_type_len != 1) + if (retrieval_type_len) { - return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + memcpy(retrieval_type_cstr, retrieval_type, retrieval_type_len); + retrieval_type_cstr[retrieval_type_len] = '\0'; } - + if (retrieval_index_len) { char retrieval_index_cstr[256]; @@ -262,8 +268,8 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, if (zebraExplain_lookup_attr_str(zh->reg->zei, zinfo_index_category_index, - (retrieval_type_len == 0 ? -1 : - retrieval_type[0]), + (retrieval_type_len == 0 ? 0 : + retrieval_type_cstr), retrieval_index_cstr) == -1) return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; } @@ -284,7 +290,7 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, struct it_key key_in; WRBUF wrbuf = wrbuf_alloc(); - if (!oid_oidcmp(input_format, yaz_oid_xml())) + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { *output_format = input_format; wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR @@ -292,14 +298,14 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, " set=\"zebra::index%s/\">\n", sysno, elemsetname); } - else if (!oid_oidcmp(input_format, yaz_oid_sutrs())) + else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) *output_format = input_format; while (zebra_rec_keys_read(keys, &str, &slen, &key_in)) { int i; int ord = CAST_ZINT_TO_INT(key_in.mem[0]); - int index_type; + const char *index_type; const char *db = 0; const char *string_index = 0; size_t string_index_len; @@ -318,18 +324,17 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, { /* process only if type is not defined, or is matching */ if (retrieval_type == 0 - || (retrieval_type_len == 1 - && retrieval_type[0] == index_type)) + || !strcmp(retrieval_type_cstr, index_type)) { zebra_term_untrans(zh, index_type, dst_buf, str); if (strlen(dst_buf)) { - if (!oid_oidcmp(input_format, yaz_oid_xml())) + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { wrbuf_printf(wrbuf, " ", key_in.mem[key_in.len -1]); @@ -341,7 +346,7 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, { wrbuf_printf(wrbuf, "%s ", string_index); - wrbuf_printf(wrbuf, "%c", index_type); + wrbuf_printf(wrbuf, "%s", index_type); for (i = 1; i < key_in.len; i++) wrbuf_printf(wrbuf, " " ZINT_FORMAT, @@ -357,7 +362,7 @@ int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr, } } } - if (!oid_oidcmp(input_format, yaz_oid_xml())) + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) wrbuf_printf(wrbuf, "\n"); *rec_lenp = wrbuf_len(wrbuf); *rec_bufp = odr_malloc(odr, *rec_lenp); @@ -399,10 +404,141 @@ static void retrieve_puts_int(WRBUF wrbuf, const char *name, wrbuf_printf(wrbuf, "%s %i\n", name, value); } -int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, + +static void snippet_xml_record(ZebraHandle zh, WRBUF wrbuf, zebra_snippets *doc) +{ + const zebra_snippet_word *doc_w; + int mark_state = 0; + + wrbuf_printf(wrbuf, "%s>\n", ZEBRA_XML_HEADER_STR); + for (doc_w = zebra_snippets_constlist(doc); doc_w; doc_w = doc_w->next) + { + if (doc_w->mark) + { + const char *index_type; + const char *db = 0; + const char *string_index = 0; + + zebraExplain_lookup_ord(zh->reg->zei, doc_w->ord, + &index_type, &db, &string_index); + + if (mark_state == 0) + { + wrbuf_printf(wrbuf, " ", index_type); + } + if (doc_w->match) + wrbuf_puts(wrbuf, ""); + /* not printing leading ws */ + if (mark_state || !doc_w->ws || doc_w->match) + wrbuf_xmlputs(wrbuf, doc_w->term); + if (doc_w->match) + wrbuf_puts(wrbuf, ""); + } + else if (mark_state == 1) + { + wrbuf_puts(wrbuf, "\n"); + } + mark_state = doc_w->mark; + } + if (mark_state == 1) + { + wrbuf_puts(wrbuf, "\n"); + } + wrbuf_printf(wrbuf, ""); +} + +int zebra_get_rec_snippets(ZebraHandle zh, zint sysno, + zebra_snippets *snippets) +{ + int return_code = 0; + Record rec = rec_get(zh->reg->records, sysno); + if (!rec) + { + yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno); + return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + } + else + { + const char *file_type = rec->info[recInfo_fileType]; + void *recTypeClientData; + RecType rt = recType_byName(zh->reg->recTypes, zh->res, + file_type, &recTypeClientData); + + if (!rt) + return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + else + { + struct ZebraRecStream stream; + return_code = zebra_create_record_stream(zh, &rec, &stream); + if (return_code == 0) + { + extract_snippet(zh, snippets, &stream, + rt, recTypeClientData); + + stream.destroy(&stream); + } + } + rec_free(&rec); + } + return return_code; +} + +int zebra_special_snippet_fetch(ZebraHandle zh, const char *setname, + zint sysno, ODR odr, + const char *elemsetname, + const Odr_oid *input_format, + const Odr_oid **output_format, + char **rec_bufp, int *rec_lenp) +{ + zebra_snippets *rec_snippets = zebra_snippets_create(); + int return_code = zebra_get_rec_snippets(zh, sysno, rec_snippets); + + if (!return_code) + { + WRBUF wrbuf = wrbuf_alloc(); + zebra_snippets *hit_snippet = zebra_snippets_create(); + + zebra_snippets_hit_vector(zh, setname, sysno, hit_snippet); + +#if 0 + /* for debugging purposes */ + yaz_log(YLOG_LOG, "---------------------------"); + yaz_log(YLOG_LOG, "REC SNIPPET:"); + zebra_snippets_log(rec_snippet, YLOG_LOG, 1); + yaz_log(YLOG_LOG, "---------------------------"); + yaz_log(YLOG_LOG, "HIT SNIPPET:"); + zebra_snippets_log(hit_snippet, YLOG_LOG, 1); +#endif + + zebra_snippets_ring(rec_snippets, hit_snippet, 5, 5); + +#if 0 + yaz_log(YLOG_LOG, "---------------------------"); + yaz_log(YLOG_LOG, "RING SNIPPET:"); + zebra_snippets_log(rec_snippets, YLOG_LOG, 1); +#endif + snippet_xml_record(zh, wrbuf, rec_snippets); + + *output_format = yaz_oid_recsyn_xml; + + if (return_code == 0) + { + *rec_lenp = wrbuf_len(wrbuf); + *rec_bufp = odr_strdup(odr, wrbuf_cstr(wrbuf)); + } + wrbuf_destroy(wrbuf); + zebra_snippets_destroy(hit_snippet); + } + zebra_snippets_destroy(rec_snippets); + return return_code; +} + +int zebra_special_fetch(ZebraHandle zh, const char *setname, + zint sysno, int score, ODR odr, const char *elemsetname, - const int *input_format, - const int **output_format, + const Odr_oid *input_format, + const Odr_oid **output_format, char **rec_bufp, int *rec_lenp) { Record rec; @@ -411,18 +547,25 @@ int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, /* *rec_lenp = 0; */ + if (elemsetname && 0 == strcmp(elemsetname, "snippet")) + { + return zebra_special_snippet_fetch(zh, setname, sysno, odr, + elemsetname + 7, + input_format, output_format, + rec_bufp, rec_lenp); + } /* processing zebra::meta::sysno elemset without fetching binary data */ if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno")) { int ret = 0; WRBUF wrbuf = wrbuf_alloc(); - if (!oid_oidcmp(input_format, yaz_oid_sutrs())) + if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { wrbuf_printf(wrbuf, ZINT_FORMAT, sysno); *output_format = input_format; } - else if (!oid_oidcmp(input_format, yaz_oid_xml())) + else if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { wrbuf_printf(wrbuf, ZEBRA_XML_HEADER_STR " sysno=\"" ZINT_FORMAT "\"/>\n", @@ -475,8 +618,8 @@ int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, } /* only accept XML and SUTRS requests from now */ - if (oid_oidcmp(input_format, yaz_oid_xml()) - && oid_oidcmp(input_format, yaz_oid_sutrs())) + if (oid_oidcmp(input_format, yaz_oid_recsyn_xml) + && oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", elemsetname); @@ -491,7 +634,7 @@ int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, WRBUF wrbuf = wrbuf_alloc(); RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); - if (!oid_oidcmp(input_format, yaz_oid_xml())) + if (!oid_oidcmp(input_format, yaz_oid_recsyn_xml)) { *output_format = input_format; @@ -511,7 +654,7 @@ int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, recordAttr->recordSize, elemsetname); } - else if (!oid_oidcmp(input_format, yaz_oid_sutrs())) + else if (!oid_oidcmp(input_format, yaz_oid_recsyn_sutrs)) { *output_format = input_format; wrbuf_printf(wrbuf, "sysno " ZINT_FORMAT "\n", sysno); @@ -558,10 +701,11 @@ int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, ODR odr, } -int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, - zebra_snippets *hit_snippet, ODR odr, - const int *input_format, Z_RecordComposition *comp, - const int **output_format, +int zebra_record_fetch(ZebraHandle zh, const char *setname, + zint sysno, int score, + ODR odr, + const Odr_oid *input_format, Z_RecordComposition *comp, + const Odr_oid **output_format, char **rec_bufp, int *rec_lenp, char **basenamep, char **addinfo) { @@ -572,14 +716,22 @@ int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, RecordAttr *recordAttr; void *clientData; int return_code = 0; + zint sysnos[10]; + int no_sysnos = 10; + ZEBRA_RES res; + + res = zebra_result_recid_to_sysno(zh, setname, sysno, sysnos, &no_sysnos); + if (res != ZEBRA_OK) + return ZEBRA_FAIL; + sysno = sysnos[0]; *basenamep = 0; *addinfo = 0; elemsetname = yaz_get_esn(comp); /* processing zebra special elementset names of form 'zebra:: */ if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7)) - return zebra_special_fetch(zh, sysno, score, odr, + return zebra_special_fetch(zh, setname, sysno, score, odr, elemsetname + 7, input_format, output_format, rec_bufp, rec_lenp); @@ -600,8 +752,8 @@ int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, file_type = rec->info[recInfo_fileType]; fname = rec->info[recInfo_filename]; basename = rec->info[recInfo_databaseName]; - *basenamep = (char *) odr_malloc (odr, strlen(basename)+1); - strcpy (*basenamep, basename); + *basenamep = (char *) odr_malloc(odr, strlen(basename)+1); + strcpy(*basenamep, basename); yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d", sysno, score); @@ -610,8 +762,6 @@ int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, if (rec) { - zebra_snippets *snippet; - zebra_rec_keys_t reckeys = zebra_rec_keys_open(); RecType rt; struct recRetrieveCtrl retrieveCtrl; @@ -631,31 +781,6 @@ int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, retrieveCtrl.res = zh->res; retrieveCtrl.rec_buf = 0; retrieveCtrl.rec_len = -1; - retrieveCtrl.hit_snippet = hit_snippet; - retrieveCtrl.doc_snippet = zebra_snippets_create(); - - zebra_rec_keys_set_buf(reckeys, - rec->info[recInfo_delKeys], - rec->size[recInfo_delKeys], - 0); - zebra_rec_keys_to_snippets(zh, reckeys, retrieveCtrl.doc_snippet); - zebra_rec_keys_close(reckeys); - -#if 0 - /* for debugging purposes */ - yaz_log(YLOG_LOG, "DOC SNIPPET:"); - zebra_snippets_log(retrieveCtrl.doc_snippet, YLOG_LOG); - yaz_log(YLOG_LOG, "HIT SNIPPET:"); - zebra_snippets_log(retrieveCtrl.hit_snippet, YLOG_LOG); -#endif - snippet = zebra_snippets_window(retrieveCtrl.doc_snippet, - retrieveCtrl.hit_snippet, - 10); -#if 0 - /* for debugging purposes */ - yaz_log(YLOG_LOG, "WINDOW SNIPPET:"); - zebra_snippets_log(snippet, YLOG_LOG); -#endif if (!(rt = recType_byName(zh->reg->recTypes, zh->res, file_type, &clientData))) @@ -679,9 +804,6 @@ int zebra_record_fetch(ZebraHandle zh, zint sysno, int score, *addinfo = retrieveCtrl.addinfo; } - zebra_snippets_destroy(snippet); - zebra_snippets_destroy(retrieveCtrl.doc_snippet); - stream.destroy(&stream); rec_free(&rec); }