X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcclfind.c;h=07e94dc75fc50b20b137c1077612965b1e81a803;hp=5d7f821448d4772098be300afd5de7f32975e98f;hb=240156e69d3a6390699644756e367fffd42891d1;hpb=df60350f9737923dce347e59b9c4cafa789ab7aa

diff --git a/src/cclfind.c b/src/cclfind.c
index 5d7f821..07e94dc 100644
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -38,23 +38,27 @@ static int qual_val_type(ccl_qualifier_t *qa, int type, int value,
                          char **attset)
 {
     int i;
-
     if (!qa)
         return 0;
     for (i = 0; qa[i]; i++)
     {
+        int got_type = 0;
         struct ccl_rpn_attr *q = ccl_qual_get_attr(qa[i]);
-        while (q)
+        for (; q; q = q->next)
         {
-            if (q->type == type && q->kind == CCL_RPN_ATTR_NUMERIC &&
-                q->value.numeric == value)
+            if (q->type == type && q->kind == CCL_RPN_ATTR_NUMERIC)
             {
-                if (attset)
-                    *attset = q->set;
-                return 1;
+                got_type = 1;
+                if (q->value.numeric == value)
+                {
+                    if (attset)
+                        *attset = q->set;
+                    return 1;
+                }
             }
-            q = q->next;
         }
+        if (got_type)
+            return 0;
     }
     return 0;
 }
@@ -113,6 +117,23 @@ struct ccl_rpn_node *ccl_rpn_node_create(enum ccl_rpn_kind kind)
     return p;
 }
 
+static struct ccl_rpn_node *ccl_rpn_node_mkbool(struct ccl_rpn_node *l,
+                                                struct ccl_rpn_node *r,
+                                                enum ccl_rpn_kind op)
+{
+    if (l && r)
+    {
+        struct ccl_rpn_node *tmp = ccl_rpn_node_create(op);
+        tmp->u.p[0] = l;
+        tmp->u.p[1] = r;
+        tmp->u.p[2] = 0;
+        return tmp;
+    }
+    else if (r)
+        return r;
+    return l;
+}
+
 /**
  * ccl_rpn_delete: Delete RPN tree.
  * rpn:   Pointer to tree.
@@ -197,6 +218,24 @@ void ccl_add_attr_numeric(struct ccl_rpn_node *p, const char *set,
     n->value.numeric = value;
 }
 
+void ccl_set_attr_numeric(struct ccl_rpn_node *p, const char *set,
+                          int type, int value)
+{
+    struct ccl_rpn_attr *n;
+    for (n = p->u.t.attr_list; n; n = n->next)
+        if (n->type == type)
+        {
+            xfree(n->set);
+            n->set = set ? xstrdup(set) : 0;
+            if (n->kind == CCL_RPN_ATTR_STRING)
+                xfree(n->value.str);
+            n->kind = CCL_RPN_ATTR_NUMERIC;
+            n->value.numeric = value;
+            return;
+        }
+    ccl_add_attr_numeric(p, set, type, value);
+}
+
 void ccl_add_attr_string(struct ccl_rpn_node *p, const char *set,
                          int type, char *value)
 {
@@ -321,19 +360,16 @@ static int append_term(CCL_parser cclp, const char *src_str, size_t src_len,
 
 
 static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
+                                             struct ccl_token *lookahead0,
                                              struct ccl_rpn_attr *attr_use,
                                              ccl_qualifier_t *qa,
-                                             size_t no, int term_len,
+                                             size_t no,
                                              int is_phrase,
                                              int auto_group)
 {
     struct ccl_rpn_node *p;
     size_t i;
-    int relation_value = -1;
-    int position_value = -1;
     int structure_value = -1;
-    int truncation_value = -1;
-    int completeness_value = -1;
 
     int left_trunc = 0;
     int right_trunc = 0;
@@ -341,11 +377,12 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
     int z3958_trunc = 0;
     int is_ccl_masked = 0;
     char *attset;
-    struct ccl_token *lookahead = cclp->look_token;
+    struct ccl_token *lookahead = lookahead0;
     const char **truncation_aliases;
     const char *t_default[2];
     const char **mask_aliases;
     const char *m_default[2];
+    int term_len = 0;
 
     truncation_aliases =
         ccl_qual_search_special(cclp->bibset, "truncation");
@@ -369,9 +406,11 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
                             truncation_aliases,
                             mask_aliases))
             is_ccl_masked = 1;
+
+        term_len += 1 + lookahead->len + lookahead->ws_prefix_len;
         lookahead = lookahead->next;
     }
-    lookahead = cclp->look_token;
+    lookahead = lookahead0;
 
     p = ccl_rpn_node_create(CCL_RPN_TERM);
     p->u.t.attr_list = NULL;
@@ -387,7 +426,9 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
     {
         struct ccl_rpn_attr *attr;
         for (attr = ccl_qual_get_attr(qa[i]); attr; attr = attr->next)
-            if (attr->type != 1 || attr == attr_use)
+            if (attr->type == 1 && attr_use && attr != attr_use)
+                continue;
+            else
             {
                 switch (attr->kind)
                 {
@@ -400,31 +441,11 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
                     {   /* deal only with REAL attributes (positive) */
                         switch (attr->type)
                         {
-                        case CCL_BIB1_REL:
-                            if (relation_value != -1)
-                                continue;
-                            relation_value = attr->value.numeric;
-                            break;
-                        case CCL_BIB1_POS:
-                            if (position_value != -1)
-                                continue;
-                            position_value = attr->value.numeric;
-                            break;
                         case CCL_BIB1_STR:
                             if (structure_value != -1)
                                 continue;
                             structure_value = attr->value.numeric;
                             break;
-                        case CCL_BIB1_TRU:
-                            if (truncation_value != -1)
-                                continue;
-                            truncation_value = attr->value.numeric;
-                            break;
-                        case CCL_BIB1_COM:
-                            if (completeness_value != -1)
-                                continue;
-                            completeness_value = attr->value.numeric;
-                            break;
                         }
                         ccl_add_attr_numeric(p, attr->set, attr->type,
                                              attr->value.numeric);
@@ -530,6 +551,111 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
     return p;
 }
 
+static struct ccl_rpn_node *ccl_term_multi_use(CCL_parser cclp,
+                                               struct ccl_token *lookahead0,
+                                               ccl_qualifier_t *qa,
+                                               size_t no,
+                                               int is_phrase,
+                                               int auto_group)
+{
+    struct ccl_rpn_node *p = 0;
+    int i;
+    for (i = 0; qa && qa[i]; i++)
+    {
+        struct ccl_rpn_attr *attr;
+        for (attr = ccl_qual_get_attr(qa[i]); attr; attr = attr->next)
+            if (attr->type == 1 && i == 0)
+            {
+                struct ccl_rpn_node *tmp2;
+                tmp2 = ccl_term_one_use(cclp, lookahead0,
+                                        attr, qa, no,
+                                        is_phrase, auto_group);
+                if (!tmp2)
+                {
+                    ccl_rpn_delete(p);
+                    return 0;
+                }
+                p = ccl_rpn_node_mkbool(p, tmp2, CCL_RPN_OR);
+            }
+    }
+    if (!p)
+        p = ccl_term_one_use(cclp, lookahead0,
+                             0 /* attr: no use */, qa, no,
+                             is_phrase, auto_group);
+    return p;
+}
+
+static struct ccl_rpn_node *split_recur(CCL_parser cclp, ccl_qualifier_t *qa,
+                                        struct ccl_token **ar, size_t sz,
+                                        size_t sub_len)
+{
+    size_t l;
+    struct ccl_rpn_node *p_top = 0;
+    assert(sz > 0);
+    for (l = 1; l <= sz && l <= sub_len; l++)
+    {
+        struct ccl_rpn_node *p2 = ccl_term_multi_use(cclp, ar[0],
+                                                     qa, l,
+                                                     l > 1,
+                                                     /* auto_group */0);
+        if (!p2)
+        {
+            ccl_rpn_delete(p_top);
+            return 0;
+        }
+        if (sz > l)
+        {
+            struct ccl_rpn_node *p1 = split_recur(cclp, qa, ar + l, sz - l,
+                sub_len);
+            if (!p1)
+            {
+                ccl_rpn_delete(p2);
+                return 0;
+            }
+            p2 = ccl_rpn_node_mkbool(p2, p1, CCL_RPN_AND);
+        }
+        p_top = ccl_rpn_node_mkbool(p_top, p2, CCL_RPN_OR);
+    }
+    assert(p_top);
+    return p_top;
+}
+
+static struct ccl_rpn_node *search_term_split_list(CCL_parser cclp,
+                                                   ccl_qualifier_t *qa,
+                                                   int *term_list, int multi)
+{
+    struct ccl_rpn_node *p;
+    struct ccl_token **ar;
+    struct ccl_token *lookahead = cclp->look_token;
+    size_t i, sz, sub_len;
+    for (sz = 0; is_term_ok(lookahead->kind, term_list); sz++)
+        lookahead = lookahead->next;
+    if (sz == 0)
+    {
+        cclp->error_code = CCL_ERR_TERM_EXPECTED;
+        return 0;
+    }
+    ar = (struct ccl_token **) xmalloc(sizeof(*lookahead) * sz);
+    lookahead = cclp->look_token;
+    for (i = 0; is_term_ok(lookahead->kind, term_list); i++)
+    {
+        ar[i] = lookahead;
+        lookahead = lookahead->next;
+    }
+    /* choose sub phrase carefully to avoid huge expansions */
+    if (sz >= 7)
+        sub_len = 1;
+    else if (sz >= 5)
+        sub_len = 2;
+    else
+        sub_len = 3;
+    p = split_recur(cclp, qa, ar, sz, sub_len);
+    xfree(ar);
+    for (i = 0; i < sz; i++)
+        ADVANCE;
+    return p;
+}
+
 /**
  * search_term: Parse CCL search term.
  * cclp:   CCL Parser
@@ -554,11 +680,14 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
         auto_group = 1;
     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_OR_LIST, 0))
         or_list = 1;
+    if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_SPLIT_LIST, 0))
+    {
+        return search_term_split_list(cclp, qa, term_list, multi);
+    }
     while (1)
     {
         struct ccl_rpn_node *p = 0;
         size_t no, i;
-        int len = 0;
         int is_phrase = 0;
         size_t max = 200;
         if (and_list || or_list || !multi)
@@ -585,67 +714,18 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
             }
             else if (this_is_phrase || no > 0)
                 is_phrase = 1;
-            len += 1+lookahead->len+lookahead->ws_prefix_len;
             lookahead = lookahead->next;
         }
 
-        if (len == 0)
+        if (no == 0)
             break;      /* no more terms . stop . */
-
-        /* go through all attributes and add them to the attribute list */
-        for (i = 0; qa && qa[i]; i++)
-        {
-            struct ccl_rpn_attr *attr;
-
-            for (attr = ccl_qual_get_attr(qa[i]); attr; attr = attr->next)
-                if (attr->type == 1)
-                {
-                    struct ccl_rpn_node *tmp2;
-                    tmp2 = ccl_term_one_use(cclp, attr, qa, no, len,
-                                            is_phrase, auto_group);
-                    if (!tmp2)
-                    {
-                        ccl_rpn_delete(p);
-                        return 0;
-                    }
-                    if (!p)
-                        p = tmp2;
-                    else
-                    {
-                        struct ccl_rpn_node *tmp1;
-                        tmp1 = ccl_rpn_node_create(CCL_RPN_OR);
-                        tmp1->u.p[0] = p;
-                        tmp1->u.p[1] = tmp2;
-                        p = tmp1;
-                    }
-                }
-        }
-        if (!p)
-            p = ccl_term_one_use(cclp, 0 /* attr: no use */, qa, no, len,
-                                 is_phrase, auto_group);
+        p = ccl_term_multi_use(cclp, cclp->look_token, qa, no,
+                               is_phrase, auto_group);
         for (i = 0; i < no; i++)
             ADVANCE;
         if (!p)
             return 0;
-        /* make the top node point to us.. */
-        if (p_top)
-        {
-            struct ccl_rpn_node *tmp;
-
-            if (or_list)
-                tmp = ccl_rpn_node_create(CCL_RPN_OR);
-            else if (and_list)
-                tmp = ccl_rpn_node_create(CCL_RPN_AND);
-            else
-                tmp = ccl_rpn_node_create(CCL_RPN_AND);
-            tmp->u.p[0] = p_top;
-            tmp->u.p[1] = p;
-
-            p_top = tmp;
-        }
-        else
-            p_top = p;
-
+        p_top = ccl_rpn_node_mkbool(p_top, p, or_list ? CCL_RPN_OR : CCL_RPN_AND);
         if (!multi)
             break;
     }
@@ -816,14 +896,14 @@ struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
             }
             p = ccl_rpn_node_create(CCL_RPN_AND);
             p->u.p[0] = p1;
-            ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
+            ccl_set_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
             p->u.p[1] = p2;
-            ccl_add_attr_numeric(p2, attset, CCL_BIB1_REL, 2);
+            ccl_set_attr_numeric(p2, attset, CCL_BIB1_REL, 2);
             return p;
         }
         else                       /* = term -    */
         {
-            ccl_add_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
+            ccl_set_attr_numeric(p1, attset, CCL_BIB1_REL, 4);
             return p1;
         }
     }
@@ -834,7 +914,7 @@ struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
         ADVANCE;
         if (!(p = search_term(cclp, ap)))
             return NULL;
-        ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, 2);
+        ccl_set_attr_numeric(p, attset, CCL_BIB1_REL, 2);
         return p;
     }
     else
@@ -843,7 +923,7 @@ struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
             return NULL;
         if (rel != 3 ||
             !qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_OMIT_EQUALS, 0))
-            ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, rel);
+            ccl_set_attr_numeric(p, attset, CCL_BIB1_REL, rel);
         return p;
     }
     return NULL;
@@ -933,16 +1013,7 @@ static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
                     xfree(ap);
                     return 0;
                 }
-                if (node)
-                {
-                    struct ccl_rpn_node *node_this =
-                        ccl_rpn_node_create(CCL_RPN_OR);
-                    node_this->u.p[0] = node;
-                    node_this->u.p[1] = node_sub;
-                    node = node_this;
-                }
-                else
-                    node = node_sub;
+                node = ccl_rpn_node_mkbool(node, node_sub, CCL_RPN_OR);
                 seq++;
             }
             if (seq == 0)
@@ -1005,16 +1076,7 @@ static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
                 ccl_rpn_delete(node);
                 break;
             }
-            if (node)
-            {
-                struct ccl_rpn_node *node_this =
-                    ccl_rpn_node_create(CCL_RPN_OR);
-                node_this->u.p[0] = node;
-                node_this->u.p[1] = node_sub;
-                node = node_this;
-            }
-            else
-                node = node_sub;
+            node = ccl_rpn_node_mkbool(node, node_sub, CCL_RPN_OR);
             seq++;
         }
     }
@@ -1147,17 +1209,7 @@ static struct ccl_rpn_node *search_elements(CCL_parser cclp,
                 ccl_rpn_delete(node);
                 return 0;
             }
-            if (node)
-            {
-                struct ccl_rpn_node *node_this =
-                    ccl_rpn_node_create(CCL_RPN_OR);
-                node_this->u.p[0] = node;
-                node_this->u.p[1] = node_sub;
-                node_this->u.p[2] = 0;
-                node = node_this;
-            }
-            else
-                node = node_sub;
+            node = ccl_rpn_node_mkbool(node, node_sub, CCL_RPN_OR);
         }
         if (!node)
             node = search_terms(cclp, 0);
@@ -1173,7 +1225,7 @@ static struct ccl_rpn_node *search_elements(CCL_parser cclp,
  */
 static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
 {
-    struct ccl_rpn_node *p1, *p2, *pn;
+    struct ccl_rpn_node *p1, *p2;
     if (!(p1 = search_elements(cclp, qa)))
         return NULL;
     while (1)
@@ -1188,11 +1240,7 @@ static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
                 ccl_rpn_delete(p1);
                 return NULL;
             }
-            pn = ccl_rpn_node_create(CCL_RPN_AND);
-            pn->u.p[0] = p1;
-            pn->u.p[1] = p2;
-            pn->u.p[2] = 0;
-            p1 = pn;
+            p1 = ccl_rpn_node_mkbool(p1, p2, CCL_RPN_AND);
             continue;
         case CCL_TOK_OR:
             ADVANCE;
@@ -1202,11 +1250,7 @@ static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
                 ccl_rpn_delete(p1);
                 return NULL;
             }
-            pn = ccl_rpn_node_create(CCL_RPN_OR);
-            pn->u.p[0] = p1;
-            pn->u.p[1] = p2;
-            pn->u.p[2] = 0;
-            p1 = pn;
+            p1 = ccl_rpn_node_mkbool(p1, p2, CCL_RPN_OR);
             continue;
         case CCL_TOK_NOT:
             ADVANCE;
@@ -1216,11 +1260,7 @@ static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa)
                 ccl_rpn_delete(p1);
                 return NULL;
             }
-            pn = ccl_rpn_node_create(CCL_RPN_NOT);
-            pn->u.p[0] = p1;
-            pn->u.p[1] = p2;
-            pn->u.p[2] = 0;
-            p1 = pn;
+            p1 = ccl_rpn_node_mkbool(p1, p2, CCL_RPN_NOT);
             continue;
         }
         break;