Towards snippets/high-light of matched terms in records. Added

zebra_get_hit_vector that gets hit terms for record with given
ID in a result set. Added print_rec_keys which dumps indexed terms.

diff --git a/index/extract.c b/index/extract.c
index 8b15b89..02ee660 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.182 2005-04-29 10:33:53 adam Exp $
+/* $Id: extract.c,v 1.183 2005-05-31 13:01:36 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -116,7 +116,7 @@ static const char **searchRecordKey (ZebraHandle zh,
        int attrSet, attrUse;
 
        iscz1_decode(decode_handle, &dst, &src);
-       assert(key.len < 4 && key.len > 2);
+       assert(key.len <= 4 && key.len > 2);
 
        attrSet = (int) key.mem[0] >> 16;
        attrUse = (int) key.mem[0] & 65535;
@@ -1473,6 +1473,45 @@ void extract_add_it_key (ZebraHandle zh,
     keys->buf_used = dst - keys->buf;
 }
 
+void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
+{
+    void *decode_handle = iscz1_start();
+    int off = 0;
+    int seqno = 0;
+    NMEM nmem = nmem_create();
+
+    yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
+           reckeys->buf_used);
+    assert(reckeys->buf);
+    while (off < reckeys->buf_used)
+    {
+        const char *src = reckeys->buf + off;
+        struct it_key key;
+       char *dst = (char*) &key;
+       int attrSet, attrUse;
+       char dst_buf[IT_MAX_WORD];
+       char *dst_term = dst_buf;
+
+       iscz1_decode(decode_handle, &dst, &src);
+       assert(key.len <= 4 && key.len > 2);
+
+       attrSet = (int) key.mem[0] >> 16;
+       attrUse = (int) key.mem[0] & 65535;
+       seqno = (int) key.mem[key.len-1];
+
+       zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
+       
+       yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
+               key.mem[0], seqno, dst_term); 
+        while (*src++)
+           ;
+        off = src - reckeys->buf;
+       nmem_reset(nmem);
+    }
+    nmem_destroy(nmem);
+    iscz1_stop(decode_handle);
+}
+
 void extract_add_index_string (RecWord *p, const char *str, int length)
 {
     struct it_key key;

diff --git a/index/index.h b/index/index.h
index 61fe3f4..33ab460 100644 (file)
--- a/index/index.h
+++ b/index/index.h
@@ -1,4 +1,4 @@
-/* $Id: index.h,v 1.138 2005-05-11 12:39:36 adam Exp $
+/* $Id: index.h,v 1.139 2005-05-31 13:01:37 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -463,6 +463,8 @@ off_t zebra_record_int_tell (void *fh);
 int zebra_record_int_read (void *fh, char *buf, size_t count);
 void zebra_record_int_end (void *fh, off_t offset);
 
+void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys);
+
 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
                               int cmd, struct recKeys *reckeys);
 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
@@ -494,6 +496,11 @@ Dict dict_open_res (BFiles bfs, const char *name, int cache, int rw,
 void zebra_setError(ZebraHandle zh, int code, const char *addinfo);
 void zebra_setError_zint(ZebraHandle zh, int code, zint i);
 
+void zebra_term_untrans_iconv(ZebraHandle zh, NMEM stream, int reg_type,
+                             char **dst, const char *src);
+
+ZEBRA_RES zebra_get_hit_vector(ZebraHandle zh, const char *setname, zint sysno);
+
 YAZ_END_CDECL
 
 #endif

diff --git a/index/retrieve.c b/index/retrieve.c
index b623437..acdbb79 100644 (file)
--- a/index/retrieve.c
+++ b/index/retrieve.c
@@ -1,4 +1,4 @@
-/* $Id: retrieve.c,v 1.29 2005-04-14 12:02:58 adam Exp $
+/* $Id: retrieve.c,v 1.30 2005-05-31 13:01:37 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -211,7 +211,15 @@ int zebra_record_fetch (ZebraHandle zh, SYSNO sysno, int score, ODR stream,
     retrieveCtrl.res = zh->res;
     retrieveCtrl.rec_buf = 0;
     retrieveCtrl.rec_len = -1;
-    
+
+    if (1)
+    {
+       struct recKeys reckeys;
+       reckeys.buf = rec->info[recInfo_delKeys];
+       reckeys.buf_used = rec->size[recInfo_delKeys];
+
+       print_rec_keys(zh, &reckeys);
+    }
     (*rt->retrieve)(clientData, &retrieveCtrl);
     *output_format = retrieveCtrl.output_format;
     *rec_bufp = (char *) retrieveCtrl.rec_buf;

diff --git a/index/zebraapi.c b/index/zebraapi.c
index 7c4fccd..f822138 100644 (file)
--- a/index/zebraapi.c
+++ b/index/zebraapi.c
@@ -1,4 +1,4 @@
-/* $Id: zebraapi.c,v 1.170 2005-05-30 13:27:08 adam Exp $
+/* $Id: zebraapi.c,v 1.171 2005-05-31 13:01:37 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -944,6 +944,9 @@ ZEBRA_RES zebra_records_retrieve(ZebraHandle zh, ODR stream,
            {
                char *buf;
                int len;
+
+               zebra_get_hit_vector(zh, setname, poset[i].sysno);
+
                recs[i].errCode =
                    zebra_record_fetch(zh, poset[i].sysno, poset[i].score,
                                       stream, input_format, comp,

diff --git a/index/zrpn.c b/index/zrpn.c
index 71c99d5..951f83f 100644 (file)
--- a/index/zrpn.c
+++ b/index/zrpn.c
@@ -1,4 +1,4 @@
-/* $Id: zrpn.c,v 1.192 2005-05-24 11:35:42 adam Exp $
+/* $Id: zrpn.c,v 1.193 2005-05-31 13:01:37 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -173,8 +173,8 @@ struct grep_info {
     ZebraSet termset;
 };        
 
-static void term_untrans(ZebraHandle zh, int reg_type,
-                           char *dst, const char *src)
+void zebra_term_untrans(ZebraHandle zh, int reg_type,
+                       char *dst, const char *src)
 {
     int len = 0;
     while (*src)
@@ -238,7 +238,7 @@ static void add_isam_p(const char *name, const char *info,
         int su_code = 0;
         int len = key_SU_decode (&su_code, name);
         
-        term_untrans  (p->zh, p->reg_type, term_tmp, name+len+1);
+        zebra_term_untrans  (p->zh, p->reg_type, term_tmp, name+len+1);
         yaz_log(log_level_rpn, "grep: %d %c %s", su_code, name[len], term_tmp);
         zebraExplain_lookup_ord (p->zh->reg->zei,
                                  su_code, &db, &set, &use);
@@ -2634,13 +2634,13 @@ static int scan_handle (char *name, const char *info, int pos, void *client)
     return 0;
 }
 
-static void scan_term_untrans (ZebraHandle zh, NMEM stream, int reg_type,
-                               char **dst, const char *src)
+void zebra_term_untrans_iconv(ZebraHandle zh, NMEM stream, int reg_type,
+                             char **dst, const char *src)
 {
     char term_src[IT_MAX_WORD];
     char term_dst[IT_MAX_WORD];
     
-    term_untrans (zh, reg_type, term_src, src);
+    zebra_term_untrans (zh, reg_type, term_src, src);
 
     if (zh->iconv_from_utf8 != 0)
     {
@@ -2926,8 +2926,8 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
        if (lo >= 0)
        {
            /* get result set for first term */
-           scan_term_untrans(zh, stream->mem, reg_id,
-                             &glist[lo].term, mterm);
+           zebra_term_untrans_iconv(zh, stream->mem, reg_id,
+                                    &glist[lo].term, mterm);
            rset = rset_trunc(zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
                              glist[lo].term, strlen(glist[lo].term),
                              NULL, 0, zapt->term->which, rset_nmem, 
@@ -3015,13 +3015,13 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
        if (j0 == -1)
            break;
        
-       scan_term_untrans (zh, stream->mem, reg_id,
-                          &glist[lo].term, mterm);
+       zebra_term_untrans_iconv(zh, stream->mem, reg_id,
+                                &glist[lo].term, mterm);
        
        rset = rset_trunc
            (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
             glist[lo].term, strlen(glist[lo].term),
-            NULL, 0, zapt->term->which,rset_nmem,
+            NULL, 0, zapt->term->which, rset_nmem,
             kc, kc->scope);
        
        ptr[j0]++;

diff --git a/index/zsets.c b/index/zsets.c
index d729f93..9726903 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,4 +1,4 @@
-/* $Id: zsets.c,v 1.84 2005-05-31 07:29:10 adam Exp $
+/* $Id: zsets.c,v 1.85 2005-05-31 13:01:37 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -1032,3 +1032,54 @@ ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname,
     }
     return ZEBRA_FAIL;
 }
+
+ZEBRA_RES zebra_get_hit_vector(ZebraHandle zh, const char *setname,
+                              zint sysno)
+{
+    ZebraSet sset = resultSetGet(zh, setname);
+    yaz_log(YLOG_LOG, "zebra_get_hit_vector setname=%s zysno=" ZINT_FORMAT,
+           setname, sysno);
+    if (!sset)
+       return ZEBRA_FAIL;
+    else
+    {
+       struct rset_key_control *kc = zebra_key_control_create(zh);
+       NMEM nmem = nmem_create();
+       struct it_key key;
+       RSET rsets[2], rset_comb;
+       RSET rset_temp = rstemp_create(nmem, kc, kc->scope, 
+                                      res_get (zh->res, "setTmpDir"),0 );
+       
+       TERMID termid;
+       RSFD rsfd = rset_open(rset_temp, RSETF_WRITE);
+       
+       key.mem[0] = sysno;
+       key.mem[1] = 0;
+       key.mem[2] = 0;
+       key.mem[3] = 0;
+       key.len = 2;
+       rset_write (rsfd, &key);
+       rset_close (rsfd);
+
+       rsets[0] = rset_temp;
+       rsets[1] = rset_dup(sset->rset);
+       
+       rset_comb = rsmulti_and_create(nmem, kc, kc->scope, 2, rsets);
+
+       rsfd = rset_open(rset_comb, RSETF_READ);
+
+       while (rset_read(rsfd, &key, &termid))
+       {
+           if (termid)
+           {
+               key_logdump_txt(YLOG_LOG, &key, termid->name);
+               yaz_log(YLOG_LOG, "   type=%d", termid->type);
+           }
+       }
+       rset_close(rsfd);
+       
+       rset_delete(rset_comb);
+    }
+    return ZEBRA_OK;
+}
+