X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fretrieve.c;h=a9bda9e649edfe7bac4db1922ff215fcec8f9a0e;hb=bf647b34fd7224275cb2c1020925897f0d28fe01;hp=ca3cc745e357ae3a566f9ab5eb750c673a4e6b84;hpb=d4451321ac90adc2c6c52d7d07fadc5af632a59a;p=idzebra-moved-to-github.git diff --git a/index/retrieve.c b/index/retrieve.c index ca3cc74..a9bda9e 100644 --- a/index/retrieve.c +++ b/index/retrieve.c @@ -1,4 +1,4 @@ -/* $Id: retrieve.c,v 1.46 2006-11-09 14:39:24 adam Exp $ +/* $Id: retrieve.c,v 1.53 2006-11-16 21:58:54 marc Exp $ Copyright (C) 1995-2006 Index Data ApS @@ -36,49 +36,214 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include -int zebra_storedata_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, + + + +static int zebra_create_record_stream(ZebraHandle zh, + Record *rec, + struct ZebraRecStream *stream){ + + RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, *rec); + + if ((*rec)->size[recInfo_storeData] > 0){ + zebra_create_stream_mem(stream, (*rec)->info[recInfo_storeData], + (*rec)->size[recInfo_storeData]); + } + else { + char full_rep[1024]; + int fd; + + if (zh->path_reg && !yaz_is_abspath((*rec)->info[recInfo_filename])){ + strcpy(full_rep, zh->path_reg); + strcat(full_rep, "/"); + strcat(full_rep, (*rec)->info[recInfo_filename]); + } + 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", + full_rep); + rec_free(rec); + return 14; + } + zebra_create_stream_fd(stream, fd, recordAttr->recordOffset); + } + return 0; +} + + + +static void parse_zebra_elem(const char *elem, + const char **index, size_t *index_len, + const char **type, size_t *type_len) +{ + *type = 0; + *type_len = 0; + + *index = 0; + *index_len = 0; + + if (elem && *elem) + { + const char *cp = strchr(elem, ':'); + + if (!cp) /* no colon */ + { + *index = elem; + *index_len = strlen(elem); + } + else if (cp[1] == '\0') /* 'index:' */ + { + *index = elem; + *index_len = cp - elem; + } + else + { + *index = elem; + *index_len = cp - elem; + *type = cp+1; + *type_len = strlen(cp+1); + } + } +} + + +int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, Record rec, - const char *element_set, + const char *elemsetname, oid_value input_format, oid_value *output_format, char **rec_bufp, int *rec_lenp) { - WRBUF wrbuf = wrbuf_alloc(); - zebra_rec_keys_t keys = zebra_rec_keys_open(); - zebra_rec_keys_set_buf(keys, - rec->info[recInfo_delKeys], - rec->size[recInfo_delKeys], - 0); - if (zebra_rec_keys_rewind(keys)) + const char *retrieval_index; + size_t retrieval_index_len; + const char *retrieval_type; + size_t retrieval_type_len; + WRBUF wrbuf = 0; + zebra_rec_keys_t keys; + + /* set output variables before processing possible error states */ + *rec_lenp = 0; + + /* only accept XML and SUTRS requests */ + if (input_format != VAL_TEXT_XML + && input_format != VAL_SUTRS){ + yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", + elemsetname); + *output_format = VAL_NONE; + return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; + } + + parse_zebra_elem(elemsetname, + &retrieval_index, &retrieval_index_len, + &retrieval_type, &retrieval_type_len); + + if (retrieval_type_len != 0 && retrieval_type_len != 1) { + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } + + if (retrieval_index_len) + { + char retrieval_index_cstr[256]; + + if (retrieval_index_len < sizeof(retrieval_index_cstr) -1) + { + memcpy(retrieval_index_cstr, retrieval_index, retrieval_index_len); + retrieval_index_cstr[retrieval_index_len] = '\0'; + + if (zebraExplain_lookup_attr_str(zh->reg->zei, + zinfo_index_category_index, + (retrieval_type_len == 0 ? -1 : + retrieval_type[0]), + retrieval_index_cstr) == -1) + { + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; + } + } + } + + keys = zebra_rec_keys_open(); + zebra_rec_keys_set_buf(keys, rec->info[recInfo_delKeys], + rec->size[recInfo_delKeys], 0); + + wrbuf = wrbuf_alloc(); + if (zebra_rec_keys_rewind(keys)){ size_t slen; const char *str; struct it_key key_in; - while(zebra_rec_keys_read(keys, &str, &slen, &key_in)) - { + + if (input_format == VAL_TEXT_XML){ + *output_format = VAL_TEXT_XML; + /*wrbuf_printf(wrbuf, + "\n");*/ + + wrbuf_printf(wrbuf, + "\n", + sysno, elemsetname); + } + else if (input_format == VAL_SUTRS) + *output_format = VAL_SUTRS; + + + 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 *db = 0; const char *string_index = 0; + size_t string_index_len; char dst_buf[IT_MAX_WORD]; zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, &string_index); - - if (string_index) - wrbuf_printf(wrbuf, "%s", string_index); - - zebra_term_untrans(zh, index_type, dst_buf, str); - wrbuf_printf(wrbuf, " %s", dst_buf); - - for (i = 1; i < key_in.len; i++) - wrbuf_printf(wrbuf, " " ZINT_FORMAT, key_in.mem[i]); - wrbuf_printf(wrbuf, "\n"); - + string_index_len = strlen(string_index); + if (retrieval_index == 0 + || (string_index_len == retrieval_index_len + && !memcmp(string_index, retrieval_index, + string_index_len))){ + + if (retrieval_type == 0 + || (retrieval_type_len == 1 + && retrieval_type[0] == index_type)){ + + if (input_format == VAL_TEXT_XML){ + wrbuf_printf(wrbuf, " ", + key_in.mem[key_in.len -1]); + + zebra_term_untrans(zh, index_type, dst_buf, str); + wrbuf_xmlputs(wrbuf, dst_buf); + wrbuf_printf(wrbuf, "\n"); + } + else if (input_format == VAL_SUTRS){ + wrbuf_printf(wrbuf, "%s ", string_index); + + wrbuf_printf(wrbuf, "%c", index_type); + + for (i = 1; i < key_in.len; i++) + wrbuf_printf(wrbuf, " " ZINT_FORMAT, + key_in.mem[i]); + + zebra_term_untrans(zh, index_type, dst_buf, str); + wrbuf_printf(wrbuf, " %s", dst_buf); + + wrbuf_printf(wrbuf, "\n"); + } + } + } } - } - *output_format = VAL_SUTRS; + if (input_format == VAL_TEXT_XML) + wrbuf_printf(wrbuf, "\n"); + } *rec_lenp = wrbuf_len(wrbuf); *rec_bufp = odr_malloc(odr, *rec_lenp); memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp); @@ -86,6 +251,90 @@ int zebra_storedata_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, zebra_rec_keys_close(keys); return 0; } + + +int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr, + const char *elemsetname, + oid_value input_format, + oid_value *output_format, + char **rec_bufp, int *rec_lenp) +{ + Record rec; + + /* set output variables before processing possible error states */ + *rec_lenp = 0; + + /* only accept XML and SUTRS requests */ + if (input_format != VAL_TEXT_XML + && input_format != VAL_SUTRS){ + yaz_log(YLOG_WARN, "unsupported format for element set zebra::%s", + elemsetname); + return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST; + } + + /* processing zebra::meta::sysno elemset without fetching binary data */ + if (elemsetname && 0 == strcmp(elemsetname, "meta::sysno")) + { + char rec_str[128]; + if (input_format == VAL_SUTRS){ + sprintf(rec_str, ZINT_FORMAT, sysno); + *output_format = VAL_SUTRS; + } + else if (input_format == VAL_TEXT_XML){ + sprintf(rec_str, "\n", + sysno, elemsetname); + *output_format = VAL_TEXT_XML; + } + *rec_lenp = strlen(rec_str); + if (*rec_lenp){ + *rec_bufp = odr_strdup(odr, rec_str); + return 0; + } else { + return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + } + } + + /* fetching binary record up for all other display elementsets */ + rec = rec_get(zh->reg->records, sysno); + if (!rec){ + yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno); + return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + } + + /* processing special elementsetnames zebra::index:: */ + if (elemsetname && 0 == strncmp(elemsetname, "index::", 7)){ + + int ret = zebra_special_index_fetch(zh, sysno, odr, rec, + elemsetname + 7, + input_format, output_format, + rec_bufp, rec_lenp); + + rec_free(&rec); + return ret; + } + + /* processing special elementsetnames zebra::data */ + if (elemsetname && 0 == strcmp(elemsetname, "data")){ + struct ZebraRecStream stream; + RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); + zebra_create_record_stream(zh, &rec, &stream); + *output_format = input_format; + *rec_lenp = recordAttr->recordSize; + *rec_bufp = (char *) odr_malloc(odr, *rec_lenp); + stream.readf(&stream, *rec_bufp, *rec_lenp); + stream.destroy(&stream); + rec_free(&rec); + return 0; + } + + if (rec) + rec_free(&rec); + return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_; +} + int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, zebra_snippets *hit_snippet, ODR odr, @@ -96,33 +345,36 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, { Record rec; char *fname, *file_type, *basename; + const char *elemsetname; struct ZebraRecStream stream; RecordAttr *recordAttr; void *clientData; - int raw_mode = 0; int return_code = 0; *basenamep = 0; *addinfo = 0; - if (comp && comp->which == Z_RecordComp_simple && - comp->u.simple->which == Z_ElementSetNames_generic && - !strcmp (comp->u.simple->u.generic, "_sysno_")) - { - char rec_str[60]; - sprintf(rec_str, ZINT_FORMAT, sysno); - *output_format = VAL_SUTRS; - *rec_lenp = strlen(rec_str); - *rec_bufp = odr_strdup(odr, rec_str); - return 0; - } - rec = rec_get (zh->reg->records, sysno); + elemsetname = yaz_get_esn(comp); + + /* processing zebra special elementset names of form 'zebra:: */ + /* SUGGESTION: do not check elemset nema here, buuuut ... */ + if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7)) + return zebra_special_fetch(zh, sysno, odr, + elemsetname + 7, + input_format, output_format, + rec_bufp, rec_lenp); + + + /* processing all other element set names */ + rec = rec_get(zh->reg->records, sysno); if (!rec) { - yaz_log (YLOG_DEBUG, "rec_get fail on sysno=" ZINT_FORMAT, sysno); + yaz_log(YLOG_WARN, "rec_get fail on sysno=" ZINT_FORMAT, sysno); *basenamep = 0; return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; } - recordAttr = rec_init_attr (zh->reg->zei, rec); + + + recordAttr = rec_init_attr(zh->reg->zei, rec); file_type = rec->info[recInfo_fileType]; fname = rec->info[recInfo_filename]; @@ -130,64 +382,11 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, *basenamep = (char *) odr_malloc (odr, strlen(basename)+1); strcpy (*basenamep, basename); - if (comp && comp->which == Z_RecordComp_simple - && comp->u.simple->which == Z_ElementSetNames_generic - && strncmp(comp->u.simple->u.generic, "zebra:", 6) == 0) - { - int r = zebra_storedata_fetch(zh, sysno, odr, rec, - comp->u.simple->u.generic, - input_format, output_format, - rec_bufp, rec_lenp); - - rec_free(&rec); - return r; - } + yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d", + sysno, score); - if (comp && comp->which == Z_RecordComp_simple && - comp->u.simple->which == Z_ElementSetNames_generic && - !strcmp (comp->u.simple->u.generic, "R")) - { - raw_mode = 1; - } - yaz_log (YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d", - sysno, score); - if (rec->size[recInfo_storeData] > 0) - { - zebra_create_stream_mem(&stream, rec->info[recInfo_storeData], - rec->size[recInfo_storeData]); - } - else - { - char full_rep[1024]; - int fd; - - if (zh->path_reg && !yaz_is_abspath (fname)) - { - strcpy (full_rep, zh->path_reg); - strcat (full_rep, "/"); - strcat (full_rep, fname); - } - else - strcpy (full_rep, fname); - - 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 14; - } - zebra_create_stream_fd(&stream, fd, recordAttr->recordOffset); - } - - if (raw_mode) - { - *output_format = VAL_SUTRS; - *rec_lenp = recordAttr->recordSize; - *rec_bufp = (char *) odr_malloc(odr, *rec_lenp); - stream.readf(&stream, *rec_bufp, *rec_lenp); - } - else + zebra_create_record_stream(zh, &rec, &stream); + { /* snippets code */ zebra_snippets *snippet; @@ -237,13 +436,25 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, zebra_snippets_log(snippet, YLOG_LOG); #endif - if (!(rt = recType_byName (zh->reg->recTypes, zh->res, - file_type, &clientData))) + /* SUGGESTION: do not check elemset name here, buuuut ... + add another recType Struct with zebra internal stuff here, + which overrides the Alvis/GRS-1/Safari filters .... + */ + if (!(rt = recType_byName(zh->reg->recTypes, zh->res, + file_type, &clientData))) { return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; } else { + /* SUGGESTION: do not check elemset name here, buuuut ... + add another recType Struct with zebra internal stuff here, + which overrides the Alvis/GRS-1/Safari filters .... + (*rt->retrieve) method to make the correct encoded, etc, retrieval, + where all needed info already is found in the &retrieveCtr + parameter. This way, we do not need to re-code/dublicate a lot of + logic. + */ (*rt->retrieve)(clientData, &retrieveCtrl); return_code = retrieveCtrl.diagnostic; @@ -252,9 +463,15 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score, *rec_lenp = retrieveCtrl.rec_len; *addinfo = retrieveCtrl.addinfo; } + /* another SUGGESTION: throw out all this snippet stuff in this + file, and do it correctly - either inside filters, or in another + new (*rt->retrieve)(clientData, &retrieveCtrl); type 'snippet' + */ + zebra_snippets_destroy(snippet); zebra_snippets_destroy(retrieveCtrl.doc_snippet); - } + } + stream.destroy(&stream); rec_free(&rec);