X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcclfind.c;h=beec79328e2e8ce1069e09552776d8ab496f7eb6;hp=52ea57c7d2636ca1adb5f5ca62e951c13b47ddb0;hb=4d1450cc691292cef5bcfdd41cefc030e4dabbf6;hpb=b1df5f9013d82510f6250d93623a0126ec19265f

diff --git a/src/cclfind.c b/src/cclfind.c
index 52ea57c..beec793 100644
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -100,7 +100,7 @@ struct ccl_rpn_node *ccl_rpn_node_create(enum ccl_rpn_kind kind)
     ccl_assert(p);
     p->kind = kind;
 
-    switch(kind)
+    switch (kind)
     {
     case CCL_RPN_TERM:
         p->u.t.attr_list = 0;
@@ -113,6 +113,54 @@ struct ccl_rpn_node *ccl_rpn_node_create(enum ccl_rpn_kind kind)
     return p;
 }
 
+static struct ccl_rpn_node *ccl_rpn_dup(struct ccl_rpn_node *rpn)
+{
+    struct ccl_rpn_node *n;
+    struct ccl_rpn_attr *attr, **attrp;
+    if (!rpn)
+        return 0;
+    n = ccl_rpn_node_create(rpn->kind);
+    switch (rpn->kind)
+    {
+    case CCL_RPN_AND:
+    case CCL_RPN_OR:
+    case CCL_RPN_NOT:
+        n->u.p[0] = ccl_rpn_dup(rpn->u.p[0]);
+        n->u.p[1] = ccl_rpn_dup(rpn->u.p[1]);
+        break;
+    case CCL_RPN_TERM:
+        n->u.t.term = xstrdup(rpn->u.t.term);
+        n->u.t.qual = rpn->u.t.qual ? xstrdup(rpn->u.t.qual) : 0;
+        attrp = &n->u.t.attr_list;
+        for (attr = rpn->u.t.attr_list; attr; attr = attr->next)
+        {
+            *attrp = (struct ccl_rpn_attr *) xmalloc(sizeof(**attrp));
+            (*attrp)->kind = attr->kind;
+            (*attrp)->type = attr->type;
+            if (attr->kind == CCL_RPN_ATTR_STRING)
+                (*attrp)->value.str = xstrdup(attr->value.str);
+            else
+                (*attrp)->value.numeric = attr->value.numeric;
+            if (attr->set)
+                (*attrp)->set = xstrdup(attr->set);
+            else
+                (*attrp)->set = 0;
+            attrp = &(*attrp)->next;
+        }
+        *attrp = 0;
+        break;
+    case CCL_RPN_SET:
+        n->u.setname = xstrdup(rpn->u.setname);
+        break;
+    case CCL_RPN_PROX:
+        n->u.p[0] = ccl_rpn_dup(rpn->u.p[0]);
+        n->u.p[1] = ccl_rpn_dup(rpn->u.p[1]);
+        n->u.p[2] = ccl_rpn_dup(rpn->u.p[2]);
+        break;
+    }
+    return n;
+}
+
 /**
  * ccl_rpn_delete: Delete RPN tree.
  * rpn:   Pointer to tree.
@@ -159,7 +207,7 @@ static struct ccl_rpn_node *find_spec(CCL_parser cclp, ccl_qualifier_t *qa);
 
 static int is_term_ok(int look, int *list)
 {
-    for (;*list >= 0; list++)
+    for (; *list >= 0; list++)
         if (look == *list)
             return 1;
     return 0;
@@ -170,9 +218,7 @@ static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa);
 static struct ccl_rpn_attr *add_attr_node(struct ccl_rpn_node *p,
                                            const char *set, int type)
 {
-    struct ccl_rpn_attr *n;
-
-    n = (struct ccl_rpn_attr *)xmalloc(sizeof(*n));
+    struct ccl_rpn_attr *n = (struct ccl_rpn_attr *) xmalloc(sizeof(*n));
     ccl_assert(n);
     if (set)
         n->set = xstrdup(set);
@@ -181,7 +227,6 @@ static struct ccl_rpn_attr *add_attr_node(struct ccl_rpn_node *p,
     n->type = type;
     n->next = p->u.t.attr_list;
     p->u.t.attr_list = n;
-
     return n;
 }
 
@@ -195,9 +240,7 @@ static struct ccl_rpn_attr *add_attr_node(struct ccl_rpn_node *p,
 void ccl_add_attr_numeric(struct ccl_rpn_node *p, const char *set,
                           int type, int value)
 {
-    struct ccl_rpn_attr *n;
-
-    n = add_attr_node(p, set, type);
+    struct ccl_rpn_attr *n = add_attr_node(p, set, type);
     n->kind = CCL_RPN_ATTR_NUMERIC;
     n->value.numeric = value;
 }
@@ -205,9 +248,7 @@ void ccl_add_attr_numeric(struct ccl_rpn_node *p, const char *set,
 void ccl_add_attr_string(struct ccl_rpn_node *p, const char *set,
                          int type, char *value)
 {
-    struct ccl_rpn_attr *n;
-
-    n = add_attr_node(p, set, type);
+    struct ccl_rpn_attr *n = add_attr_node(p, set, type);
     n->kind = CCL_RPN_ATTR_STRING;
     n->value.str = xstrdup(value);
 }
@@ -231,7 +272,7 @@ static size_t cmp_operator(const char **aliases, const char *input)
 #define CCL_CHARS "#?\\"
 
 static int has_ccl_masking(const char *src_str,
-                           int src_len,
+                           size_t src_len,
                            const char **truncation_aliases,
                            const char **mask_aliases)
 {
@@ -327,15 +368,12 @@ static int append_term(CCL_parser cclp, const char *src_str, size_t src_len,
 }
 
 
-#if YAZ_781
 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,
-                                             int no, int term_len,
-                                             const char **truncation_aliases,
-                                             const char **mask_aliases,
+                                             size_t no,
                                              int is_phrase,
-                                             int is_ccl_masked,
                                              int auto_group)
 {
     struct ccl_rpn_node *p;
@@ -350,8 +388,42 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
     int right_trunc = 0;
     int regex_trunc = 0;
     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");
+    if (!truncation_aliases)
+    {
+        truncation_aliases = t_default;
+        t_default[0] = "?";
+        t_default[1] = 0;
+    }
+    mask_aliases =
+        ccl_qual_search_special(cclp->bibset, "mask");
+    if (!mask_aliases)
+    {
+        mask_aliases = m_default;
+        m_default[0] = "#";
+        m_default[1] = 0;
+    }
+    for (i = 0; i < no; i++)
+    {
+        if (has_ccl_masking(lookahead->name, lookahead->len,
+                            truncation_aliases,
+                            mask_aliases))
+            is_ccl_masked = 1;
+
+        term_len += 1 + lookahead->len + lookahead->ws_prefix_len;
+        lookahead = lookahead->next;
+    }
+    lookahead = lookahead0;
 
     p = ccl_rpn_node_create(CCL_RPN_TERM);
     p->u.t.attr_list = NULL;
@@ -509,7 +581,77 @@ static struct ccl_rpn_node *ccl_term_one_use(CCL_parser cclp,
     }
     return p;
 }
-#endif
+
+static struct ccl_rpn_node *split_recur(CCL_parser cclp, ccl_qualifier_t *qa,
+                                        struct ccl_rpn_node *parent,
+                                        struct ccl_token **ar, size_t sz)
+{
+    size_t l;
+    struct ccl_rpn_node *p_top = 0;
+    assert(sz > 0);
+    for (l = 1; l <= sz; l++)
+    {
+        struct ccl_rpn_node *p1;
+        struct ccl_rpn_node *p2 = ccl_term_one_use(cclp, ar[0],
+                                                   /* attr_use */0,
+                                                   qa, l,
+                                                   l > 1,
+                                                   /* auto_group */0);
+        if (!p2)
+            return 0;
+        if (parent)
+        {
+            struct ccl_rpn_node *tmp = ccl_rpn_node_create(CCL_RPN_AND);
+            tmp->u.p[0] = l > 1 ? ccl_rpn_dup(parent) : parent;
+            tmp->u.p[1] = p2;
+            p2 = tmp;
+        }
+        if (sz > l)
+            p1 = split_recur(cclp, qa, p2, ar + l, sz - l);
+        else
+            p1 = p2;
+        if (p_top)
+        {
+            struct ccl_rpn_node *tmp = ccl_rpn_node_create(CCL_RPN_OR);
+            tmp->u.p[0] = p_top;
+            tmp->u.p[1] = p1;
+            p_top = tmp;
+        }
+        else
+            p_top = p1;
+    }
+    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;
+    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;
+    }
+    p = split_recur(cclp, qa, 0, ar, sz);
+    xfree(ar);
+    for (i = 0; i < sz; i++)
+        ADVANCE;
+    return p;
+}
 
 /**
  * search_term: Parse CCL search term.
@@ -528,29 +670,6 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
     int and_list = 0;
     int auto_group = 0;
     int or_list = 0;
-    const char **truncation_aliases;
-    const char *t_default[2];
-    const char **mask_aliases;
-    const char *m_default[2];
-
-    truncation_aliases =
-        ccl_qual_search_special(cclp->bibset, "truncation");
-    if (!truncation_aliases)
-    {
-        truncation_aliases = t_default;
-        t_default[0] = "?";
-        t_default[1] = 0;
-    }
-
-    mask_aliases =
-        ccl_qual_search_special(cclp->bibset, "mask");
-    if (!mask_aliases)
-    {
-        mask_aliases = m_default;
-        m_default[0] = "#";
-        m_default[1] = 0;
-    }
-
 
     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
         and_list = 1;
@@ -558,26 +677,15 @@ 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;
-        int is_ccl_masked = 0;
-#if YAZ_781
-#else
-        char *attset;
-        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;
-        int regex_trunc = 0;
-        int z3958_trunc = 0;
-#endif
         size_t max = 200;
         if (and_list || or_list || !multi)
             max = 1;
@@ -595,12 +703,6 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
             for (i = 0; i<lookahead->len; i++)
                 if (lookahead->name[i] == ' ')
                     this_is_phrase = 1;
-
-            if (has_ccl_masking(lookahead->name, lookahead->len,
-                                truncation_aliases,
-                                mask_aliases))
-                is_ccl_masked = 1;
-
             if (auto_group)
             {
                 if (no > 0 && (is_phrase || is_phrase != this_is_phrase))
@@ -609,14 +711,12 @@ 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 . */
 
-#if YAZ_781
         /* go through all attributes and add them to the attribute list */
         for (i = 0; qa && qa[i]; i++)
         {
@@ -626,10 +726,9 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
                 if (attr->type == 1)
                 {
                     struct ccl_rpn_node *tmp2;
-                    tmp2 = ccl_term_one_use(cclp, attr, qa, no, len,
-                                            truncation_aliases, mask_aliases,
-                                            is_phrase, is_ccl_masked,
-                                            auto_group);
+                    tmp2 = ccl_term_one_use(cclp, cclp->look_token,
+                                            attr, qa, no,
+                                            is_phrase, auto_group);
                     if (!tmp2)
                     {
                         ccl_rpn_delete(p);
@@ -648,172 +747,13 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
                 }
         }
         if (!p)
-        {
-            p = ccl_term_one_use(cclp, 0 /* attr: no use */, qa, no, len,
-                                 truncation_aliases, mask_aliases,
-                                 is_phrase, is_ccl_masked, auto_group);
-            if (!p)
-                return 0;
-        }
+            p = ccl_term_one_use(cclp, cclp->look_token,
+                                 0 /* attr: no use */, qa, no,
+                                 is_phrase, auto_group);
         for (i = 0; i < no; i++)
             ADVANCE;
-#else
-        /* create the term node, but wait a moment before adding the term */
-        p = ccl_rpn_node_create(CCL_RPN_TERM);
-        p->u.t.attr_list = NULL;
-        p->u.t.term = NULL;
-        if (qa && qa[0])
-        {
-            const char *n = ccl_qual_get_name(qa[0]);
-            if (n)
-                p->u.t.qual = xstrdup(n);
-        }
-
-        /* 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)
-                switch(attr->kind)
-                {
-                case CCL_RPN_ATTR_STRING:
-                    ccl_add_attr_string(p, attr->set, attr->type,
-                                        attr->value.str);
-                    break;
-                case CCL_RPN_ATTR_NUMERIC:
-                    if (attr->value.numeric > 0)
-                    {   /* 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);
-                    }
-                }
-        }
-        attset = 0;
-        if (structure_value == -1 && (
-                auto_group ||
-                qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
-            )
-        {
-            if (!is_phrase)
-                ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 2);
-            else
-                ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 1);
-        }
-
-        if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_REGEX,
-                          &attset))
-        {
-            if (is_ccl_masked)
-                regex_trunc = 1; /* regex trunc (102) allowed */
-        }
-        else if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_Z3958,
-                          &attset))
-        {
-            if (is_ccl_masked)
-                z3958_trunc = 1; /* Z39.58 trunc (CCL) trunc allowed */
-        }
-
-        /* make the RPN token */
-        p->u.t.term = (char *)xmalloc(len * 2 + 2);
-        ccl_assert(p->u.t.term);
-        p->u.t.term[0] = '\0';
-        for (i = 0; i<no; i++)
-        {
-            const char *src_str = cclp->look_token->name;
-            size_t src_len = cclp->look_token->len;
-
-            if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
-            {
-                strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
-                        cclp->look_token->ws_prefix_len);
-            }
-            if (append_term(cclp, src_str, src_len, p->u.t.term, regex_trunc,
-                            z3958_trunc, truncation_aliases, mask_aliases,
-                            i == 0, i == no - 1,
-                            &left_trunc, &right_trunc))
-            {
-                ccl_rpn_delete(p);
-                return NULL;
-            }
-            ADVANCE;
-        }
-        if (left_trunc && right_trunc)
-        {
-            if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_BOTH,
-                                &attset))
-            {
-                cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
-                ccl_rpn_delete(p);
-                return NULL;
-            }
-            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 3);
-        }
-        else if (right_trunc)
-        {
-            if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_RIGHT,
-                                 &attset))
-            {
-                cclp->error_code = CCL_ERR_TRUNC_NOT_RIGHT;
-                ccl_rpn_delete(p);
-                return NULL;
-            }
-            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 1);
-        }
-        else if (left_trunc)
-        {
-            if (!qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_LEFT,
-                                &attset))
-            {
-                cclp->error_code = CCL_ERR_TRUNC_NOT_LEFT;
-                ccl_rpn_delete(p);
-                return NULL;
-            }
-            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 2);
-        }
-        else if (regex_trunc)
-        {
-            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 102);
-        }
-        else if (z3958_trunc)
-        {
-            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 104);
-        }
-        else
-        {
-            if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
-                               &attset))
-                ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 100);
-        }
-#endif
+        if (!p)
+            return 0;
         /* make the top node point to us.. */
         if (p_top)
         {
@@ -877,7 +817,6 @@ static struct ccl_rpn_node *search_terms2(CCL_parser cclp,
 }
 
 
-
 static
 struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
                                       ccl_qualifier_t *ap, char *attset)