isomorfeus-ferret 0.17.2 → 0.17.4

data/ext/isomorfeus_ferret_ext/frb_lazy_doc.c

@@ -1,8 +1,6 @@
 #include "frt_index.h"
 #include "isomorfeus_ferret.h"
 
-extern VALUE rb_hash_update(int argc, VALUE *argv, VALUE self);
-
 extern VALUE sym_each;
 extern ID id_eql;
 
@@ -97,7 +95,7 @@ static VALUE frb_ld_df_load(VALUE self, VALUE rkey, FrtLazyDocField *lazy_df) {
   rLazyDoc *rld = DATA_PTR(self);
   VALUE rdata;
   if (lazy_df->size == 1) {
-    char *data = frt_lazy_df_get_data(lazy_df, 0);
+    const char *data = frt_lazy_df_get_data(lazy_df, 0);
     rdata = rb_str_new(data, lazy_df->data[0].length);
     rb_enc_associate(rdata, lazy_df->data[0].encoding);
   } else {
@@ -105,7 +103,7 @@ static VALUE frb_ld_df_load(VALUE self, VALUE rkey, FrtLazyDocField *lazy_df) {
     VALUE rstr;
     rdata = rb_ary_new2(lazy_df->size);
     for (i = 0; i < lazy_df->size; i++) {
-      char *data = frt_lazy_df_get_data(lazy_df, i);
+      const char *data = frt_lazy_df_get_data(lazy_df, i);
       rstr = rb_str_new(data, lazy_df->data[i].length);
       rb_enc_associate(rstr, lazy_df->data[i].encoding);
       rb_ary_store(rdata, i, rstr);
@@ -127,7 +125,7 @@ static VALUE frb_ld_load(VALUE self) {
   if (ld->loaded) return self;
   int i;
   FrtLazyDocField *lazy_df;
-  for (i = 0; i < ld->size; i++) {
+  for (i = 0; i < ld->field_count; i++) {
     lazy_df = ld->fields[i];
     if (!(lazy_df->loaded)) frb_ld_df_load(self, ID2SYM(lazy_df->name), lazy_df);
   }
@@ -148,8 +146,8 @@ static VALUE frb_ld_fields(VALUE self) {
   VALUE rfields = rb_ivar_get(self, id_fields);
   if (rfields == Qnil) {
     int i;
-    rfields = rb_ary_new2(ld->size);
-    for (i = 0; i < ld->size; i++) {
+    rfields = rb_ary_new2(ld->field_count);
+    for (i = 0; i < ld->field_count; i++) {
       rb_ary_store(rfields, i, ID2SYM(ld->fields[i]->name));
     }
     rb_ivar_set(self, id_fields, rfields);
@@ -208,9 +206,9 @@ static VALUE frb_ld_equal(VALUE self, VALUE other) {
     rLazyDoc *other_rld;
     TypedData_Get_Struct(other, rLazyDoc, &frb_ld_t, other_rld);
     other_h = frb_ld_to_h(other);
-    other_size = other_rld->doc->size;
+    other_size = other_rld->doc->field_count;
   }
-  if (ld->size == other_size) {
+  if (ld->field_count == other_size) {
     VALUE self_h = frb_ld_to_h(self);
     return rb_funcall(self_h, id_equal, 1, other_h);
   }
@@ -278,7 +276,7 @@ static VALUE frb_ld_any(int argc, VALUE *argv, VALUE self) {
   FrtLazyDoc *ld = rld->doc;
   if (argc == 0) {
     if (!rb_block_given_p()) {
-      return (ld->size > 0) ? Qtrue : Qfalse;
+      return (ld->field_count > 0) ? Qtrue : Qfalse;
     } else {
       if (!ld->loaded) frb_ld_load(self);
       VALUE res = Qnil;
@@ -375,7 +373,7 @@ static VALUE frb_ld_each_value(VALUE self) {
 
 static VALUE frb_ld_empty(VALUE self) {
   FrtLazyDoc *ld = ((rLazyDoc *)DATA_PTR(self))->doc;
-  return (ld->size == 0) ? Qtrue : Qfalse;
+  return (ld->field_count == 0) ? Qtrue : Qfalse;
 }
 
 static VALUE frb_ld_eql(VALUE self, VALUE other) {
@@ -389,9 +387,9 @@ static VALUE frb_ld_eql(VALUE self, VALUE other) {
   } else {
     TypedData_Get_Struct(other, rLazyDoc, &frb_ld_t, other_rld);
     other_h = frb_ld_to_h(other);
-    other_size = other_rld->doc->size;
+    other_size = other_rld->doc->field_count;
   }
-  if (ld->size == other_size) {
+  if (ld->field_count == other_size) {
     VALUE self_h = frb_ld_to_h(self);
     return rb_funcall(self_h, id_eql, 1, other_h);
   }
@@ -475,7 +473,7 @@ static VALUE frb_ld_has_value(VALUE self, VALUE value) {
   if (!ld->loaded) frb_ld_load(self);
   int i;
   VALUE hvalue;
-  for (i=0; i<ld->size; i++) {
+  for (i=0; i<ld->field_count; i++) {
     hvalue = (VALUE)frt_h_get(rld->hash, (void *)ID2SYM(ld->fields[i]->name));
     hvalue = rb_funcall(hvalue, id_equal, 1, value);
     if (hvalue == Qtrue) return Qtrue;
@@ -499,7 +497,7 @@ static VALUE frb_ld_key(VALUE self, VALUE value) {
   if (!ld->loaded) frb_ld_load(self);
   int i;
   VALUE hvalue;
-  for (i=0; i<ld->size; i++) {
+  for (i=0; i<ld->field_count; i++) {
     hvalue = (VALUE)frt_h_get(rld->hash, (void *)ID2SYM(ld->fields[i]->name));
     hvalue = rb_funcall(hvalue, id_equal, 1, value);
     if (hvalue == Qtrue) return ID2SYM(ld->fields[i]->name);
@@ -509,7 +507,7 @@ static VALUE frb_ld_key(VALUE self, VALUE value) {
 
 static VALUE frb_ld_length(VALUE self) {
   FrtLazyDoc *ld = ((rLazyDoc *)DATA_PTR(self))->doc;
-  return INT2FIX(ld->size);
+  return INT2FIX(ld->field_count);
 }
 
 static VALUE frb_ld_merge(int argc, VALUE *argv, VALUE self) {

data/ext/isomorfeus_ferret_ext/frb_search.c

@@ -6,7 +6,8 @@
 
 // #undef close
 
-VALUE mSearch;
+static VALUE mSearch;
+static VALUE mSpans;
 
 static VALUE cHit;
 static VALUE cTopDocs;
@@ -92,7 +93,7 @@ static VALUE sym_integer;
 static VALUE sym_float;
 static VALUE sym_string;
 static VALUE sym_auto;
-static VALUE sym_doc_id;
+static VALUE sym_doc_num;
 static VALUE sym_score;
 static VALUE sym_byte;
 
@@ -145,7 +146,7 @@ extern VALUE frb_get_lazy_doc(FrtLazyDoc *lazy_doc);
  ****************************************************************************/
 
 static VALUE frb_get_hit(FrtHit *hit) {
-    return rb_struct_new(cHit, INT2FIX(hit->doc), rb_float_new((double)hit->score), NULL);
+    return rb_struct_new(cHit, INT2FIX(hit->doc_num), rb_float_new((double)hit->score), NULL);
 }
 
 /****************************************************************************
@@ -197,10 +198,10 @@ static VALUE frb_td_to_s(int argc, VALUE *argv, VALUE self) {
 
     for (i = 0; i < len; i++) {
         VALUE rhit = RARRAY_PTR(rhits)[i];
-        int doc_id = FIX2INT(rb_funcall(rhit, id_doc, 0));
+        int doc_num = FIX2INT(rb_funcall(rhit, id_doc, 0));
         const char *value = "";
         size_t value_len = 0;
-        FrtLazyDoc *lzd = sea->get_lazy_doc(sea, doc_id);
+        FrtLazyDoc *lzd = sea->get_lazy_doc(sea, doc_num);
         FrtLazyDocField *lzdf = frt_lazy_doc_get(lzd, field);
         if (NULL != lzdf) {
             value = frt_lazy_df_get_data(lzdf, 0);
@@ -211,7 +212,7 @@ static VALUE frb_td_to_s(int argc, VALUE *argv, VALUE self) {
             FRT_REALLOC_N(str, char, capa);
         }
 
-        sprintf(str + p, "\t%d \"%s\": %0.5f\n", doc_id, value,
+        sprintf(str + p, "\t%d \"%s\": %0.5f\n", doc_num, value,
                 NUM2DBL(rb_funcall(rhit, id_score, 0)));
         p += strlen(str + p);
         frt_lazy_doc_close(lzd);
@@ -229,7 +230,7 @@ static char *frb_lzd_load_to_json(FrtLazyDoc *lzd, char **str, char *s, int *sle
 	int len = diff, l;
 	FrtLazyDocField *f;
 
-	for (i = 0; i < lzd->size; i++) {
+	for (i = 0; i < lzd->field_count; i++) {
 		f = lzd->fields[i];
         /* 3 times length of field to make space for quoted quotes ('"') and
          * 4 times field elements to make space for '"' around fields and ','
@@ -244,7 +245,7 @@ static char *frb_lzd_load_to_json(FrtLazyDoc *lzd, char **str, char *s, int *sle
         s = *str + diff;
     }
 
-	for (i = 0; i < lzd->size; i++) {
+	for (i = 0; i < lzd->field_count; i++) {
         const char *field_name;
 		f = lzd->fields[i];
         field_name = rb_id2name(f->name);
@@ -278,7 +279,7 @@ static VALUE frb_td_to_json(VALUE self) {
 	FrtLazyDoc *lzd;
 	FrtSearcher *sea = (FrtSearcher *)DATA_PTR(rb_funcall(self, id_searcher, 0));
 	const int num_hits = RARRAY_LEN(rhits);
-	int doc_id;
+	int doc_num;
     int len = 32768;
 	char *str = FRT_ALLOC_N(char, len);
     char *s = str;
@@ -289,8 +290,8 @@ static VALUE frb_td_to_json(VALUE self) {
         if (i) *(s++) = ',';
         *(s++) = '{';
 		rhit = RARRAY_PTR(rhits)[i];
-		doc_id = FIX2INT(rb_funcall(rhit, id_doc, 0));
-		lzd = sea->get_lazy_doc(sea, doc_id);
+		doc_num = FIX2INT(rb_funcall(rhit, id_doc, 0));
+		lzd = sea->get_lazy_doc(sea, doc_num);
 		s = frb_lzd_load_to_json(lzd, &str, s, &len);
         frt_lazy_doc_close(lzd);
         *(s++) = '}';
@@ -1292,8 +1293,7 @@ static VALUE frb_phq_init(int argc, VALUE *argv, VALUE self) {
  *    # doesn't match => "big house"
  */
 static VALUE
-frb_phq_add(int argc, VALUE *argv, VALUE self)
-{
+frb_phq_add(int argc, VALUE *argv, VALUE self) {
     VALUE rterm, rpos_inc;
     int pos_inc = 1;
     FrtQuery *q = (FrtQuery *)DATA_PTR(self);
@@ -1688,8 +1688,7 @@ extern float frt_qp_default_fuzzy_min_sim;
  *  Set the default value for +:min_similarity+
  */
 static VALUE
-frb_fq_set_dms(VALUE self, VALUE val)
-{
+frb_fq_set_dms(VALUE self, VALUE val) {
     double min_sim = NUM2DBL(val);
     if (min_sim >= 1.0) {
         rb_raise(rb_eArgError,
@@ -1710,8 +1709,7 @@ frb_fq_set_dms(VALUE self, VALUE val)
  *  Get the default value for +:prefix_length+
  */
 static VALUE
-frb_fq_get_dpl(VALUE self)
-{
+frb_fq_get_dpl(VALUE self) {
     return rb_cvar_get(cFuzzyQuery, id_default_prefix_length);
 }
 
@@ -1723,8 +1721,7 @@ extern int frt_qp_default_fuzzy_pre_len;
  *  Set the default value for +:prefix_length+
  */
 static VALUE
-frb_fq_set_dpl(VALUE self, VALUE val)
-{
+frb_fq_set_dpl(VALUE self, VALUE val) {
     int pre_len = FIX2INT(val);
     if (pre_len < 0) {
         rb_raise(rb_eArgError,
@@ -2695,7 +2692,7 @@ static int get_sort_type(VALUE rtype) {
         return FRT_SORT_TYPE_STRING;
     } else if (rtype == sym_score) {
         return FRT_SORT_TYPE_SCORE;
-    } else if (rtype == sym_doc_id) {
+    } else if (rtype == sym_doc_num) {
         return FRT_SORT_TYPE_DOC;
     } else if (rtype == sym_float) {
         return FRT_SORT_TYPE_FLOAT;
@@ -2703,7 +2700,7 @@ static int get_sort_type(VALUE rtype) {
         return FRT_SORT_TYPE_AUTO;
     } else {
         rb_raise(rb_eArgError, ":%s is an unknown sort-type. Please choose "
-                 "from [:integer, :float, :string, :auto, :score, :doc_id]",
+                 "from [:integer, :float, :string, :auto, :score, :doc_num]",
                  rb_id2name(SYM2ID(rtype)));
     }
     return FRT_SORT_TYPE_DOC;
@@ -2720,7 +2717,7 @@ static int get_sort_type(VALUE rtype) {
  *
  *  :type::         Default: +:auto+. Specifies how a field should be sorted.
  *                  Choose from one of; +:auto+, +:integer+, +:float+,
- *                  +:string+, +:byte+, +:doc_id+ or +:score+. +:auto+ will
+ *                  +:string+, +:byte+, +:doc_num+ or +:score+. +:auto+ will
  *                  check the datatype of the field by trying to parse it into
  *                  either a number or a float before settling on a string
  *                  sort. String sort is locale dependent and works for
@@ -2789,7 +2786,7 @@ static VALUE frb_sf_get_name(VALUE self) {
  *     sort_field.type -> symbol
  *
  *  Return the type of sort. Should be one of; +:auto+, +:integer+, +:float+,
- *  +:string+, +:byte+, +:doc_id+ or +:score+.
+ *  +:string+, +:byte+, +:doc_num+ or +:score+.
  */
 static VALUE frb_sf_get_type(VALUE self) {
     GET_SF();
@@ -2799,7 +2796,7 @@ static VALUE frb_sf_get_type(VALUE self) {
         case FRT_SORT_TYPE_FLOAT:   return sym_float;
         case FRT_SORT_TYPE_STRING:  return sym_string;
         case FRT_SORT_TYPE_AUTO:    return sym_auto;
-        case FRT_SORT_TYPE_DOC:     return sym_doc_id;
+        case FRT_SORT_TYPE_DOC:     return sym_doc_num;
         case FRT_SORT_TYPE_SCORE:   return sym_score;
     }
     return Qnil;
@@ -2903,7 +2900,7 @@ static void frb_parse_sort_str(FrtSort *sort, char *xsort_str) {
 
         if (strcmp("SCORE", s) == 0) {
             sf = frt_sort_field_score_new(reverse);
-        } else if (strcmp("DOC_ID", s) == 0) {
+        } else if (strcmp("DOC_NUM", s) == 0) {
             sf = frt_sort_field_doc_new(reverse);
         } else {
             sf = frt_sort_field_auto_new(rb_intern(s), reverse);
@@ -2942,7 +2939,7 @@ static void frb_sort_add(FrtSort *sort, VALUE rsf, bool reverse) {
 #define GET_SORT() FrtSort *sort = (FrtSort *)DATA_PTR(self)
 /*
  *  call-seq:
- *     Sort.new(sort_fields = [SortField::SCORE, SortField::DOC_ID], reverse = false) -> Sort
+ *     Sort.new(sort_fields = [SortField::SCORE, SortField::DOC_NUM], reverse = false) -> Sort
  *
  *  Create a new Sort object. If +reverse+ is true, all sort_fields will be
  *  reversed so if any of them are already reversed the  will be turned back
@@ -3064,16 +3061,32 @@ static VALUE frb_sea_doc_freq(VALUE self, VALUE rfield, VALUE rterm) {
 
 /*
  *  call-seq:
- *     searcher.get_document(doc_id) -> LazyDoc
- *     searcher[doc_id] -> LazyDoc
+ *     searcher.get_document(doc_num) -> LazyDoc
+ *     searcher[doc_num] -> LazyDoc
  *
  *  Retrieve a document from the index. See LazyDoc for more details on the
  *  document returned. Documents are referenced internally by document ids
  *  which are returned by the Searchers search methods.
  */
-static VALUE frb_sea_doc(VALUE self, VALUE rdoc_id) {
+static VALUE frb_sea_doc(VALUE self, VALUE rdoc_num) {
+    int ex_code = 0;
+    const char *msg = NULL;
     GET_SEA();
-    return frb_get_lazy_doc(sea->get_lazy_doc(sea, FIX2INT(rdoc_id)));
+    VALUE ld = Qnil;
+
+    FRT_TRY
+        ld = frb_get_lazy_doc(sea->get_lazy_doc(sea, FIX2INT(rdoc_num)));
+    FRT_XCATCHALL
+        ex_code = xcontext.excode;
+        msg = xcontext.msg;
+        FRT_HANDLED();
+    FRT_XENDTRY
+
+    if (ex_code && msg) {
+        frb_raise(ex_code, msg);
+    }
+
+    return ld;
 }
 
 /*
@@ -3085,13 +3098,13 @@ static VALUE frb_sea_doc(VALUE self, VALUE rdoc_id) {
  *  there are no deletions, this number also refers to the number of documents
  *  in the index.
  */
-static VALUE frb_sea_max_doc(VALUE self) {
+static VALUE frb_sea_max_doc_num(VALUE self) {
     GET_SEA();
-    return INT2FIX(sea->max_doc(sea));
+    return INT2FIX(sea->max_doc_num(sea));
 }
 
-static float call_filter_proc(int doc_id, float score, FrtSearcher *sea, void *arg) {
-    VALUE val = rb_funcall((VALUE)arg, id_call, 3, INT2FIX(doc_id), rb_float_new((double)score), sea->rsea);
+static float call_filter_proc(int doc_num, float score, FrtSearcher *sea, void *arg) {
+    VALUE val = rb_funcall((VALUE)arg, id_call, 3, INT2FIX(doc_num), rb_float_new((double)score), sea->rsea);
     switch (TYPE(val)) {
         case T_NIL:
         case T_FALSE:
@@ -3192,8 +3205,7 @@ static FrtTopDocs *frb_sea_search_internal(FrtQuery *query, VALUE roptions, FrtS
                 post_filter_holder.filter_func = &call_filter_proc;
                 post_filter_holder.arg = (void *)rval;
                 post_filter = &post_filter_holder;
-            }
-            else {
+            } else {
                 post_filter = DATA_PTR(rval);
             }
         }
@@ -3249,7 +3261,7 @@ static FrtTopDocs *frb_sea_search_internal(FrtQuery *query, VALUE roptions, FrtS
  *                  to specify a fields type to sort it correctly. For more
  *                  on this, see the documentation for SortField
  *  :filter::       a Filter object to filter the search results with
- *  :filter_proc::  a filter Proc is a Proc which takes the doc_id, the score
+ *  :filter_proc::  a filter Proc is a Proc which takes the doc_num, the score
  *                  and the Searcher object as its parameters and returns
  *                  either a Boolean value specifying whether the result
  *                  should be included in the result set, or a Float between 0
@@ -3268,13 +3280,13 @@ static VALUE frb_sea_search(int argc, VALUE *argv, VALUE self) {
 
 /*
  *  call-seq:
- *     searcher.search_each(query, options = {}) {|doc_id, score| do_something}
+ *     searcher.search_each(query, options = {}) {|doc_num, score| do_something}
  *         -> total_hits
  *
  *  Run a query through the Searcher on the index. A TopDocs object is
  *  returned with the relevant results. The +query+ is a Query object. The
  *  Searcher#search_each method yields the internal document id (used to
- *  reference documents in the Searcher object like this; +searcher[doc_id]+)
+ *  reference documents in the Searcher object like this; +searcher[doc_num]+)
  *  and the search score for that document. It is possible for the score to be
  *  greater than 1.0 for some queries and taking boosts into account. This
  *  method will also normalize scores to the range 0.0..1.0 when the max-score
@@ -3302,7 +3314,7 @@ static VALUE frb_sea_search(int argc, VALUE *argv, VALUE self) {
  *                  to specify a fields type to sort it correctly. For more
  *                  on this, see the documentation for SortField
  *  :filter::       a Filter object to filter the search results with
- *  :filter_proc::  a filter Proc is a Proc which takes the doc_id, the score
+ *  :filter_proc::  a filter Proc is a Proc which takes the doc_num, the score
  *                  and the Searcher object as its parameters and returns a
  *                  Boolean value specifying whether the result should be
  *                  included in the result set.
@@ -3323,7 +3335,7 @@ static VALUE frb_sea_search_each(int argc, VALUE *argv, VALUE self) {
 
     /* yield normalized scores */
     for (i = 0; i < td->size; i++) {
-        rb_yield_values(2, INT2FIX(td->hits[i]->doc), rb_float_new((double)(td->hits[i]->score/max_score)));
+        rb_yield_values(2, INT2FIX(td->hits[i]->doc_num), rb_float_new((double)(td->hits[i]->score/max_score)));
     }
 
     rtotal_hits = INT2FIX(td->total_hits);
@@ -3396,11 +3408,9 @@ static VALUE frb_sea_scan(int argc, VALUE *argv, VALUE self) {
                 if (limit <= 0) {
                     rb_raise(rb_eArgError, ":limit must be > 0");
                 }
-            }
-            else if (rval == sym_all) {
+            } else if (rval == sym_all) {
                 limit = INT_MAX;
-            }
-            else {
+            } else {
                 rb_raise(rb_eArgError, "%s is not a sensible :limit value "
                          "Please use a positive integer or :all",
                          rs2s(rb_obj_as_string(rval)));
@@ -3420,14 +3430,14 @@ static VALUE frb_sea_scan(int argc, VALUE *argv, VALUE self) {
 
 /*
  *  call-seq:
- *     searcher.explain(query, doc_id) -> Explanation
+ *     searcher.explain(query, doc_num) -> Explanation
  *
  *  Create an explanation object to explain the score returned for a
- *  particular document at +doc_id+ in the index for the query +query+.
+ *  particular document at +doc_num+ in the index for the query +query+.
  *
  *  Usually used like this;
  *
- *    puts searcher.explain(query, doc_id).to_s
+ *    puts searcher.explain(query, doc_num).to_s
  */
 
 static size_t frb_explanation_size(const void *p) {
@@ -3458,17 +3468,17 @@ static VALUE frb_expl_alloc(VALUE rclass) {
     return TypedData_Wrap_Struct(rclass, &frb_explanation_t, e);
 }
 
-static VALUE frb_sea_explain(VALUE self, VALUE rquery, VALUE rdoc_id) {
+static VALUE frb_sea_explain(VALUE self, VALUE rquery, VALUE rdoc_num) {
     GET_SEA();
     FrtQuery *query = DATA_PTR(rquery);
     FrtExplanation *expl;
-    expl = sea->explain(sea, query, FIX2INT(rdoc_id));
+    expl = sea->explain(sea, query, FIX2INT(rdoc_num));
     return TypedData_Wrap_Struct(cExplanation, &frb_explanation_t, expl);
 }
 
 /*
  *  call-seq:
- *     searcher.highlight(query, doc_id, field, options = {}) -> Array
+ *     searcher.highlight(query, doc_num, field, options = {}) -> Array
  *
  *  Returns an array of strings with the matches highlighted.
  *
@@ -3490,7 +3500,7 @@ static VALUE frb_sea_explain(VALUE self, VALUE rquery, VALUE rdoc_id) {
  */
 static VALUE frb_sea_highlight(int argc, VALUE *argv, VALUE self) {
     GET_SEA();
-    VALUE rquery, rdoc_id, rfield, roptions, v;
+    VALUE rquery, rdoc_num, rfield, roptions, v;
     int excerpt_length = 150;
     int num_excerpts = 2;
     const char *pre_tag = "<b>";
@@ -3498,7 +3508,7 @@ static VALUE frb_sea_highlight(int argc, VALUE *argv, VALUE self) {
     const char *ellipsis = "...";
     char **excerpts;
 
-    rb_scan_args(argc, argv, "31", &rquery, &rdoc_id, &rfield, &roptions);
+    rb_scan_args(argc, argv, "31", &rquery, &rdoc_num, &rfield, &roptions);
     FrtQuery *query = DATA_PTR(rquery);
     if (argc > 3) {
         if (TYPE(roptions) != T_HASH) {
@@ -3511,8 +3521,7 @@ static VALUE frb_sea_highlight(int argc, VALUE *argv, VALUE self) {
             if (v == sym_all) {
                 num_excerpts = 1;
                 excerpt_length = INT_MAX/2;
-            }
-            else {
+            } else {
                 excerpt_length = FIX2INT(v);
             }
         }
@@ -3529,7 +3538,7 @@ static VALUE frb_sea_highlight(int argc, VALUE *argv, VALUE self) {
 
     if ((excerpts = frt_searcher_highlight(sea,
                                        query,
-                                       FIX2INT(rdoc_id),
+                                       FIX2INT(rdoc_num),
                                        frb_field(rfield),
                                        excerpt_length,
                                        num_excerpts,
@@ -3605,6 +3614,7 @@ static VALUE frb_sea_init(VALUE self, VALUE obj) {
     if (TYPE(obj) == T_STRING) {
         frb_create_dir(obj);
         store = frt_open_mdbx_store(rs2s(obj));
+        store->create_folder(store, segm_idx_name);
         ir = frt_ir_open(NULL, store);
         ir->rir = TypedData_Wrap_Struct(cIndexReader, &frb_index_reader_t, ir);
     } else {
@@ -3859,7 +3869,7 @@ static void Init_TopDocs(void) {
  *
  *  == Example
  *
- *    puts searcher.explain(query, doc_id).to_s
+ *    puts searcher.explain(query, doc_num).to_s
  */
 static void Init_Explanation(void) {
     cExplanation = rb_define_class_under(mSearch, "Explanation", rb_cObject);
@@ -4738,7 +4748,7 @@ static void Init_Filter(void) {
  *  * :float
  *  * :string
  *  * :byte
- *  * :doc_id
+ *  * :doc_num
  *  * :score
  *
  *  The type of the SortField is set by passing it as a parameter to the
@@ -4774,7 +4784,7 @@ static void Init_SortField(void) {
     sym_float = ID2SYM(rb_intern("float"));
     sym_string = ID2SYM(rb_intern("string"));
     sym_auto = ID2SYM(rb_intern("auto"));
-    sym_doc_id = ID2SYM(rb_intern("doc_id"));
+    sym_doc_num = ID2SYM(rb_intern("doc_num"));
     sym_score = ID2SYM(rb_intern("score"));
     sym_byte = ID2SYM(rb_intern("byte"));
 
@@ -4794,12 +4804,12 @@ static void Init_SortField(void) {
     rb_define_const(cSortField, "SCORE_REV", TypedData_Wrap_Struct(cSortField, &frb_sort_field_t, FRT_SORT_FIELD_SCORE_REV));
     FRT_SORT_FIELD_SCORE_REV->rfield = rb_const_get(cSortField, rb_intern("SCORE_REV"));
 
-    rb_define_const(cSortField, "DOC_ID", TypedData_Wrap_Struct(cSortField, &frb_sort_field_t, FRT_SORT_FIELD_DOC));
-    oSORT_FIELD_DOC = rb_const_get(cSortField, rb_intern("DOC_ID"));
+    rb_define_const(cSortField, "DOC_NUM", TypedData_Wrap_Struct(cSortField, &frb_sort_field_t, FRT_SORT_FIELD_DOC));
+    oSORT_FIELD_DOC = rb_const_get(cSortField, rb_intern("DOC_NUM"));
     FRT_SORT_FIELD_DOC->rfield = oSORT_FIELD_DOC;
 
-    rb_define_const(cSortField, "DOC_ID_REV", TypedData_Wrap_Struct(cSortField, &frb_sort_field_t, FRT_SORT_FIELD_DOC_REV));
-    FRT_SORT_FIELD_DOC_REV->rfield = rb_const_get(cSortField, rb_intern("DOC_ID_REV"));
+    rb_define_const(cSortField, "DOC_NUM_REV", TypedData_Wrap_Struct(cSortField, &frb_sort_field_t, FRT_SORT_FIELD_DOC_REV));
+    FRT_SORT_FIELD_DOC_REV->rfield = rb_const_get(cSortField, rb_intern("DOC_NUM_REV"));
 }
 
 /*
@@ -4861,8 +4871,8 @@ static void Init_Sort(void) {
  *
  *    searcher.search_each(TermQuery.new(:content, "ferret")
  *                         :filter => RangeFilter.new(:date, :< => "2006"),
- *                         :sort => "date DESC, title") do |doc_id, score|
- *        puts "#{searcher[doc_id][title] scored #{score}"
+ *                         :sort => "date DESC, title") do |doc_num, score|
+ *        puts "#{searcher[doc_num][title] scored #{score}"
  *    end
  */
 static void Init_Searcher(void) {
@@ -4892,7 +4902,7 @@ static void Init_Searcher(void) {
     rb_define_method(cSearcher, "doc_freq", frb_sea_doc_freq, 2);
     rb_define_method(cSearcher, "get_document", frb_sea_doc, 1);
     rb_define_method(cSearcher, "[]", frb_sea_doc, 1);
-    rb_define_method(cSearcher, "max_doc", frb_sea_max_doc, 0);
+    rb_define_method(cSearcher, "max_doc_num", frb_sea_max_doc_num, 0);
     rb_define_method(cSearcher, "search", frb_sea_search, -1);
     rb_define_method(cSearcher, "search_each", frb_sea_search_each, -1);
     rb_define_method(cSearcher, "scan", frb_sea_scan, -1);
@@ -4934,9 +4944,7 @@ static void Init_MultiSearcher(void) {
  *
  *  Happy Ferreting!!
  */
-void
-Init_Search(void)
-{
+void Init_Search(void) {
     mSearch = rb_define_module_under(mFerret, "Search");
 
     fsym_id = rb_intern("id");