Use size of rec->info stuff in meta record fetch. Failing that will
[idzebra-moved-to-github.git] / index / retrieve.c
index 04f0b3a..d981ec5 100644 (file)
@@ -1,4 +1,4 @@
-/* $Id: retrieve.c,v 1.50 2006-11-15 14:09:43 marc Exp $
+/* $Id: retrieve.c,v 1.57 2006-11-23 14:01:21 adam Exp $
    Copyright (C) 1995-2006
    Index Data ApS
 
@@ -74,31 +74,40 @@ static int zebra_create_record_stream(ZebraHandle zh,
     
 
 
-static void parse_zebra_elem(const char *elem,
+static int 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;
 
+    *type = 0;
+    *type_len = 0;
+
     if (elem && *elem)
     {
-        const char *cp = strchr(elem, ':');
+        char *cp;
+        /* verify that '::' is in the beginning of *elem 
+           and something more follows */
+        if (':' != *elem
+            || !(elem +1) || ':' != *(elem +1)
+            || !(elem +2) || '\0' == *(elem +2))
+            return 0;
+        /* pick out info from string after '::' */
+        elem = elem + 2;
+        cp = strchr(elem, ':');
 
-        if (!cp) /* no colon */
+        if (!cp) /* index, no colon, no type */
         {
             *index = elem;
             *index_len = strlen(elem);
         }
-        else if (cp[1] == '\0') /* 'index:' */
+        else if (cp[1] == '\0') /* colon, but no following type */
         {
-            *index = elem;
-            *index_len = cp - elem;
+            return 0;
         }
-        else
+        else  /* index, colon and type */
         {
             *index = elem;
             *index_len = cp - elem;
@@ -106,10 +115,11 @@ static void parse_zebra_elem(const char *elem,
             *type_len = strlen(cp+1);
         }
     }
+    return 1;
 }
 
 
-int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
+int zebra_special_index_fetch(ZebraHandle zh, zint sysno, ODR odr,
                           Record rec,
                           const char *elemsetname,
                           oid_value input_format,
@@ -120,29 +130,54 @@ int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
     size_t retrieval_index_len; 
     const char *retrieval_type;
     size_t retrieval_type_len;
-   
-    /* int return_code = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; */
-
-    WRBUF wrbuf = wrbuf_alloc();
+    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_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+        return YAZ_BIB1_NO_SYNTAXES_AVAILABLE_FOR_THIS_REQUEST;
+    }
+
+    if (!parse_zebra_elem(elemsetname,
+                     &retrieval_index, &retrieval_index_len,
+                     &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)
+    {
+        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);
 
-    parse_zebra_elem(elemsetname,
-                     &retrieval_index, &retrieval_index_len,
-                     &retrieval_type,  &retrieval_type_len);
-
+    wrbuf = wrbuf_alloc();
     if (zebra_rec_keys_rewind(keys)){
         size_t slen;
         const char *str;
@@ -157,7 +192,7 @@ int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
                          "<record xmlns="
                          "\"http://www.indexdata.com/zebra/\""
                          " sysno=\"" ZINT_FORMAT "\""
-                         " set=\"zebra::index::%s\">\n",
+                         " set=\"zebra::index%s/\">\n",
                          sysno, elemsetname);
         }
         else if (input_format == VAL_SUTRS)
@@ -176,49 +211,57 @@ int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db,
                                     &string_index);
             string_index_len = strlen(string_index);
+
+            /* process only if index is not defined, 
+               or if defined and matching */
             if (retrieval_index == 0 
                 || (string_index_len == retrieval_index_len 
                     && !memcmp(string_index, retrieval_index,
                                string_index_len))){
-                
+               
+                /* process only if type is not defined, or is matching */
                 if (retrieval_type == 0 
                     || (retrieval_type_len == 1 
                         && retrieval_type[0] == index_type)){
                     
-                    if (input_format == VAL_TEXT_XML){
-                        wrbuf_printf(wrbuf, "  <index name=\"%s\"", 
-                                     string_index);
-                        
-                        wrbuf_printf(wrbuf, " type=\"%c\"", index_type);
+
+                    zebra_term_untrans(zh, index_type, dst_buf, str);
+                    if (strlen(dst_buf)){
+
+                        if (input_format == VAL_TEXT_XML){
+                            wrbuf_printf(wrbuf, "  <index name=\"%s\"", 
+                                         string_index);
+                            
+                            wrbuf_printf(wrbuf, " type=\"%c\"", index_type);
+                            
+                            wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">", 
+                                         key_in.mem[key_in.len -1]);
                         
-                        wrbuf_printf(wrbuf, " seq=\"" ZINT_FORMAT "\">", 
-                                     key_in.mem[key_in.len -1]);
+                            wrbuf_xmlputs(wrbuf, dst_buf);
+                            wrbuf_printf(wrbuf, "</index>\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);
                         
-                        zebra_term_untrans(zh, index_type, dst_buf, str);
-                        wrbuf_xmlputs(wrbuf, dst_buf);
-                        wrbuf_printf(wrbuf, "</index>\n");
+                            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");
-                    }
                 }
             }
         }
         if (input_format == VAL_TEXT_XML)
             wrbuf_printf(wrbuf, "</record>\n");
      }
-    
     *rec_lenp = wrbuf_len(wrbuf);
     *rec_bufp = odr_malloc(odr, *rec_lenp);
     memcpy(*rec_bufp, wrbuf_buf(wrbuf), *rec_lenp);
@@ -228,22 +271,26 @@ int zebra_special_index_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
 }
 
 
-int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
+
+int zebra_special_fetch(ZebraHandle zh, zint sysno, int score, 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_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+        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"))
     {
@@ -254,10 +301,10 @@ int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
         } 
         else if (input_format == VAL_TEXT_XML){
             sprintf(rec_str, "<record xmlns="
-                             "\"http://www.indexdata.com/zebra/\""
+                    "\"http://www.indexdata.com/zebra/\""
                              " sysno=\"" ZINT_FORMAT "\""
-                             " set=\"zebra::%s\">\n",
-                             sysno, elemsetname);
+                             " set=\"zebra::%s\"/>\n",
+                    sysno, elemsetname);
             *output_format = VAL_TEXT_XML;
         }
        *rec_lenp = strlen(rec_str);
@@ -272,15 +319,78 @@ int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
     /* fetching binary record up for all other display elementsets */
     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);
         return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
     }
 
+    /* processing special elementsetnames zebra::meta:: */
+    if (elemsetname && 0 == strcmp(elemsetname, "meta")){
+        int ret = 0;
+        char rec_str[1024];
+        RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); 
+
+        if (input_format == VAL_TEXT_XML){
+            *output_format = VAL_TEXT_XML;
+
+             sprintf(rec_str, 
+                     "<record xmlns="
+                     "\"http://www.indexdata.com/zebra/\""
+                     " sysno=\"" ZINT_FORMAT "\""
+                     " base=\"%.*s\""
+                     " file=\"%.*s\""
+                     " type=\"%.*s\""
+                     " score=\"%i\""
+                     " rank=\"" ZINT_FORMAT "\""
+                     " size=\"%i\""
+                     " set=\"zebra::%s/\">\n",
+                     sysno, 
+                     rec->size[recInfo_databaseName], rec->info[recInfo_databaseName],
+                     rec->size[recInfo_filename], rec->info[recInfo_filename],
+                     rec->size[recInfo_fileType], rec->info[recInfo_fileType],
+                     score,
+                     recordAttr->staticrank,
+                     recordAttr->recordSize,
+                     elemsetname);
+        }
+        else if (input_format == VAL_SUTRS){
+            *output_format = VAL_SUTRS;
+             sprintf(rec_str, 
+                     "sysno " ZINT_FORMAT "\n"
+                     "base %.*s\n"
+                     "file %.*s\n"
+                     "type %.*s\n"
+                     "score %i\n"
+                     "rank " ZINT_FORMAT "\n"
+                     "size %i\n"
+                     "set zebra::%s\n",
+                     sysno, 
+                     rec->size[recInfo_databaseName], rec->info[recInfo_databaseName],
+                     rec->size[recInfo_filename], rec->info[recInfo_filename],
+                     rec->size[recInfo_fileType], rec->info[recInfo_fileType],
+                     score,
+                     recordAttr->staticrank,
+                     recordAttr->recordSize,
+                     elemsetname);
+        }
+        
+        
+       *rec_lenp = strlen(rec_str);
+        if (*rec_lenp){
+            *rec_bufp = odr_strdup(odr, rec_str);
+            ret = 0;
+        } else {
+            ret = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+        }
+
+        rec_free(&rec);
+        return ret;
+    }
+
     /* processing special elementsetnames zebra::index:: */
-    if (elemsetname && 0 == strncmp(elemsetname, "index::", 7)){
+    if (elemsetname && 0 == strncmp(elemsetname, "index", 5)){
         
         int ret = zebra_special_index_fetch(zh, sysno, odr, rec,
-                                            elemsetname + 7,
+                                            elemsetname + 5,
                                             input_format, output_format,
                                             rec_bufp, rec_lenp);
         
@@ -293,7 +403,7 @@ int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
         struct ZebraRecStream stream;
         RecordAttr *recordAttr = rec_init_attr(zh->reg->zei, rec); 
         zebra_create_record_stream(zh, &rec, &stream);
-        *output_format = VAL_SUTRS;
+        *output_format = input_format;
         *rec_lenp = recordAttr->recordSize;
         *rec_bufp = (char *) odr_malloc(odr, *rec_lenp);
         stream.readf(&stream, *rec_bufp, *rec_lenp);
@@ -304,11 +414,11 @@ int zebra_special_fetch(ZebraHandle zh, SYSNO sysno, ODR odr,
 
     if (rec)
         rec_free(&rec);
-    return YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS;
+    return YAZ_BIB1_SPECIFIED_ELEMENT_SET_NAME_NOT_VALID_FOR_SPECIFIED_;
 }
 
                           
-int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
+int zebra_record_fetch(ZebraHandle zh, zint sysno, int score,
                        zebra_snippets *hit_snippet, ODR odr,
                        oid_value input_format, Z_RecordComposition *comp,
                        oid_value *output_format,
@@ -327,21 +437,9 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
     *addinfo = 0;
     elemsetname = yaz_get_esn(comp);
 
-    /*
-    yaz_log(YLOG_LOG, "ELEMENTSET: '%s'", elemsetname);
-    if (comp && comp->which)
-        {
-            yaz_log(YLOG_LOG, "%i %i", comp, comp->which);
-        }
-        If SRU, comp->which ==2 , and yaz_get_esn(comp) does not work correctly
-        IF Z3055, comp->which ==1 , and everything is fine - except reord 
-        encodning
-    */
-
     /* 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,
+    if (elemsetname && 0 == strncmp(elemsetname, "zebra::", 7))
+        return  zebra_special_fetch(zh, sysno, score, odr,
                                     elemsetname + 7,
                                     input_format, output_format,
                                     rec_bufp, rec_lenp);
@@ -351,7 +449,7 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
     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;
     }
@@ -365,9 +463,8 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
     *basenamep = (char *) odr_malloc (odr, strlen(basename)+1);
     strcpy (*basenamep, basename);
 
-
-    yaz_log (YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
-             sysno, score);
+    yaz_log(YLOG_DEBUG, "retrieve localno=" ZINT_FORMAT " score=%d",
+            sysno, score);
 
     zebra_create_record_stream(zh, &rec, &stream);
     
@@ -420,10 +517,6 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
        zebra_snippets_log(snippet, YLOG_LOG);
 #endif
 
-        /* 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)))
         {
@@ -431,14 +524,6 @@ int zebra_record_fetch(ZebraHandle zh, SYSNO sysno, int score,
         }
         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;
 
@@ -447,10 +532,6 @@ 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);