isomorfeus-ferret 0.12.7 → 0.13.0

data/ext/isomorfeus_ferret_ext/frt_search.c

@@ -3,6 +3,8 @@
 #include "frt_search.h"
 #include "frt_array.h"
 
+#undef close
+
 /***************************************************************************
  *
  * FrtExplanation - Used to give details for query scores
@@ -305,47 +307,39 @@ static FrtQuery *q_rewrite(FrtQuery *self, FrtIndexReader *ir)
     return self;
 }
 
-static void q_extract_terms(FrtQuery *self, FrtHashSet *terms)
-{
+static void q_extract_terms(FrtQuery *self, FrtHashSet *terms) {
     /* do nothing by default */
     (void)self;
     (void)terms;
 }
 
-FrtSimilarity *frt_q_get_similarity_i(FrtQuery *self, FrtSearcher *searcher)
-{
+FrtSimilarity *frt_q_get_similarity_i(FrtQuery *self, FrtSearcher *searcher) {
     (void)self;
     return searcher->get_similarity(searcher);
 }
 
-void frt_q_destroy_i(FrtQuery *self)
-{
+void frt_q_destroy_i(FrtQuery *self) {
     free(self);
 }
 
-void frt_q_deref(FrtQuery *self)
-{
-    if (--(self->ref_cnt) == 0) {
+void frt_q_deref(FrtQuery *self) {
+    if (--(self->ref_cnt) == 0)
         self->destroy_i(self);
-    }
 }
 
-FrtWeight *frt_q_create_weight_unsup(FrtQuery *self, FrtSearcher *searcher)
-{
+FrtWeight *frt_q_create_weight_unsup(FrtQuery *self, FrtSearcher *searcher) {
     (void)self;
     (void)searcher;
-    FRT_RAISE(FRT_UNSUPPORTED_ERROR,
-          "Create weight is unsupported for this type of query");
+    FRT_RAISE(FRT_UNSUPPORTED_ERROR, "Create weight is unsupported for this type of query");
     return NULL;
 }
 
-FrtWeight *frt_q_weight(FrtQuery *self, FrtSearcher *searcher)
-{
-    FrtQuery      *query   = searcher->rewrite(searcher, self);
-    FrtWeight     *weight  = query->create_weight_i(query, searcher);
-    float       sum     = weight->sum_of_squared_weights(weight);
-    FrtSimilarity *sim     = query->get_similarity(query, searcher);
-    float       norm    = frt_sim_query_norm(sim, sum);
+FrtWeight *frt_q_weight(FrtQuery *self, FrtSearcher *searcher) {
+    FrtQuery    *query = searcher->rewrite(searcher, self);
+    FrtWeight  *weight = query->create_weight_i(query, searcher);
+    float          sum = weight->sum_of_squared_weights(weight);
+    FrtSimilarity *sim = query->get_similarity(query, searcher);
+    float         norm = frt_sim_query_norm(sim, sum);
     frt_q_deref(query);
 
     weight->normalize(weight, norm);
@@ -353,8 +347,8 @@ FrtWeight *frt_q_weight(FrtQuery *self, FrtSearcher *searcher)
 }
 
 #define BQ(query) ((FrtBooleanQuery *)(query))
-FrtQuery *frt_q_combine(FrtQuery **queries, int q_cnt)
-{
+
+FrtQuery *frt_q_combine(FrtQuery **queries, int q_cnt) {
     int i;
     FrtQuery *q, *ret_q;
     FrtHashSet *uniques = frt_hs_new((frt_hash_ft)&frt_q_hash, (frt_eq_ft)&frt_q_eq, NULL);
@@ -402,28 +396,24 @@ FrtQuery *frt_q_combine(FrtQuery **queries, int q_cnt)
     return ret_q;
 }
 
-unsigned long long frt_q_hash(FrtQuery *self)
-{
+unsigned long long frt_q_hash(FrtQuery *self) {
     return (self->hash(self) << 5) | self->type;
 }
 
-int frt_q_eq(FrtQuery *self, FrtQuery *o)
-{
+int frt_q_eq(FrtQuery *self, FrtQuery *o) {
     return (self == o)
         || ((self->type == o->type)
             && (self->boost == o->boost)
             && self->eq(self, o));
 }
 
-static FrtMatchVector *q_get_matchv_i(FrtQuery *self, FrtMatchVector *mv, FrtTermVector *tv)
-{
+static FrtMatchVector *q_get_matchv_i(FrtQuery *self, FrtMatchVector *mv, FrtTermVector *tv) {
     /* be default we don't add any matches */
     (void)self; (void)tv;
     return mv;
 }
 
-FrtQuery *frt_q_create(size_t size)
-{
+FrtQuery *frt_q_create(size_t size) {
     FrtQuery *self = (FrtQuery *)frt_ecalloc(size);
 #ifdef DEBUG
     if (size < sizeof(FrtQuery)) {
@@ -431,13 +421,14 @@ FrtQuery *frt_q_create(size_t size)
               "the size of a Query struct <%d>", (int)size, (int)sizeof(FrtQuery));
     }
 #endif
-    self->boost             = 1.0f;
-    self->rewrite           = &q_rewrite;
-    self->get_similarity    = &frt_q_get_similarity_i;
-    self->extract_terms     = &q_extract_terms;
-    self->get_matchv_i      = &q_get_matchv_i;
-    self->weight            = NULL;
-    self->ref_cnt           = 1;
+    self->boost          = 1.0f;
+    self->rewrite        = &q_rewrite;
+    self->get_similarity = &frt_q_get_similarity_i;
+    self->extract_terms  = &q_extract_terms;
+    self->get_matchv_i   = &q_get_matchv_i;
+    self->weight         = NULL;
+    self->ref_cnt        = 1;
+    self->rquery         = Qnil;
     return self;
 }
 
@@ -447,14 +438,12 @@ FrtQuery *frt_q_create(size_t size)
  *
  ***************************************************************************/
 
-void frt_scorer_destroy_i(FrtScorer *scorer)
-{
+void frt_scorer_destroy_i(FrtScorer *scorer) {
     free(scorer);
 }
 
-FrtScorer *frt_scorer_create(size_t size, FrtSimilarity *similarity)
-{
-    FrtScorer *self        = (FrtScorer *)frt_ecalloc(size);
+FrtScorer *frt_scorer_create(size_t size, FrtSimilarity *similarity) {
+    FrtScorer *self     = (FrtScorer *)frt_ecalloc(size);
 #ifdef DEBUG
     if (size < sizeof(FrtScorer)) {
         FRT_RAISE(FRT_ARG_ERROR, "size of scorer <%d> should be at least <%d>",
@@ -466,13 +455,11 @@ FrtScorer *frt_scorer_create(size_t size, FrtSimilarity *similarity)
     return self;
 }
 
-bool frt_scorer_doc_less_than(const FrtScorer *s1, const FrtScorer *s2)
-{
+bool frt_scorer_doc_less_than(const FrtScorer *s1, const FrtScorer *s2) {
     return s1->doc < s2->doc;
 }
 
-int frt_scorer_doc_cmp(const void *p1, const void *p2)
-{
+int frt_scorer_doc_cmp(const void *p1, const void *p2) {
     return (*(FrtScorer **)p1)->doc - (*(FrtScorer **)p2)->doc;
 }
 
@@ -497,8 +484,7 @@ static int match_range_cmp(const void *p1, const void *p2)
 
 
 /* ** FrtMatchVector ** */
-FrtMatchVector *frt_matchv_new()
-{
+FrtMatchVector *frt_matchv_new(void) {
     FrtMatchVector *matchv = FRT_ALLOC(FrtMatchVector);
 
     matchv->size = 0;
@@ -597,11 +583,7 @@ void frt_matchv_destroy(FrtMatchVector *self)
  *
  ***************************************************************************/
 
-FrtMatchVector *frt_searcher_get_match_vector(FrtSearcher *self,
-                                       FrtQuery *query,
-                                       const int doc_num,
-                                       FrtSymbol field)
-{
+FrtMatchVector *frt_searcher_get_match_vector(FrtSearcher *self, FrtQuery *query, const int doc_num, ID field) {
     FrtMatchVector *mv = frt_matchv_new();
     bool rewrite = query->get_matchv_i == q_get_matchv_i;
     FrtTermVector *tv = self->get_term_vector(self, doc_num, field);
@@ -618,8 +600,7 @@ FrtMatchVector *frt_searcher_get_match_vector(FrtSearcher *self,
     return mv;
 }
 
-typedef struct Excerpt
-{
+typedef struct Excerpt {
     int start;
     int end;
     int start_pos;
@@ -812,13 +793,12 @@ static char *highlight_field(FrtMatchVector *mv,
 char **frt_searcher_highlight(FrtSearcher *self,
                           FrtQuery *query,
                           const int doc_num,
-                          FrtSymbol field,
+                          ID field,
                           const int excerpt_len,
                           const int num_excerpts,
                           const char *pre_tag,
                           const char *post_tag,
-                          const char *ellipsis)
-{
+                          const char *ellipsis) {
     char **excerpt_strs = NULL;
     FrtTermVector *tv = self->get_term_vector(self, doc_num, field);
     FrtLazyDoc *lazy_doc = self->get_lazy_doc(self, doc_num);
@@ -970,8 +950,7 @@ static FrtSimilarity *sea_get_similarity(FrtSearcher *self)
 
 #define ISEA(searcher) ((FrtIndexSearcher *)(searcher))
 
-int frt_isea_doc_freq(FrtSearcher *self, FrtSymbol field, const char *term)
-{
+int frt_isea_doc_freq(FrtSearcher *self, ID field, const char *term) {
     return frt_ir_doc_freq(ISEA(self)->ir, field, term);
 }
 
@@ -1041,9 +1020,9 @@ static FrtTopDocs *isea_search_w(FrtSearcher *self,
         }
     } else {
         hq = frt_pq_new(max_size, (frt_lt_ft)&hit_lt, &free);
-        hq_pop = &hit_pq_pop;
         hq_insert = &hit_pq_insert;
         hq_destroy = &frt_pq_destroy;
+        hq_pop = &hit_pq_pop;
     }
 
     scorer = weight->scorer(weight, ISEA(self)->ir);
@@ -1080,7 +1059,6 @@ static FrtTopDocs *isea_search_w(FrtSearcher *self,
     } else {
         num_docs = 0;
     }
-    frt_pq_clear(hq);
     hq_destroy(hq);
     return frt_td_new(total_hits, num_docs, score_docs, max_score);
 }
@@ -1151,12 +1129,7 @@ static void isea_search_each(FrtSearcher *self, FrtQuery *query, FrtFilter *filt
  * Note: Unlike the offset_docnum in other search methods, this offset_docnum
  * refers to document number and not hit.
  */
-static int isea_search_unscored_w(FrtSearcher *self,
-                                  FrtWeight *weight,
-                                  int *buf,
-                                  int limit,
-                                  int offset_docnum)
-{
+static int isea_search_unscored_w(FrtSearcher *self, FrtWeight *weight, int *buf, int limit, int offset_docnum) {
     int count = 0;
     FrtScorer *scorer = weight->scorer(weight, ISEA(self)->ir);
     if (scorer) {
@@ -1170,12 +1143,7 @@ static int isea_search_unscored_w(FrtSearcher *self,
     return count;
 }
 
-static int isea_search_unscored(FrtSearcher *self,
-                                FrtQuery *query,
-                                int *buf,
-                                int limit,
-                                int offset_docnum)
-{
+static int isea_search_unscored(FrtSearcher *self, FrtQuery *query, int *buf, int limit, int offset_docnum) {
     int count;
     FrtWeight *weight = frt_q_weight(query, self);
     count = isea_search_unscored_w(self, weight, buf, limit, offset_docnum);
@@ -1183,8 +1151,7 @@ static int isea_search_unscored(FrtSearcher *self,
     return count;
 }
 
-static FrtQuery *isea_rewrite(FrtSearcher *self, FrtQuery *original)
-{
+static FrtQuery *isea_rewrite(FrtSearcher *self, FrtQuery *original) {
     int q_is_destroyed = false;
     FrtQuery *query = original;
     FrtQuery *rewritten_query = query->rewrite(query, ISEA(self)->ir);
@@ -1197,41 +1164,34 @@ static FrtQuery *isea_rewrite(FrtSearcher *self, FrtQuery *original)
     return query;
 }
 
-static FrtExplanation *isea_explain(FrtSearcher *self,
-                                 FrtQuery *query,
-                                 int doc_num)
-{
+static FrtExplanation *isea_explain(FrtSearcher *self, FrtQuery *query, int doc_num) {
     FrtWeight *weight = frt_q_weight(query, self);
     FrtExplanation *e = weight->explain(weight, ISEA(self)->ir, doc_num);
     weight->destroy(weight);
     return e;
 }
 
-static FrtExplanation *isea_explain_w(FrtSearcher *self, FrtWeight *w, int doc_num)
-{
+static FrtExplanation *isea_explain_w(FrtSearcher *self, FrtWeight *w, int doc_num) {
     return w->explain(w, ISEA(self)->ir, doc_num);
 }
 
-static FrtTermVector *isea_get_term_vector(FrtSearcher *self,
-                                          const int doc_num,
-                                          FrtSymbol field)
-{
+static FrtTermVector *isea_get_term_vector(FrtSearcher *self, const int doc_num, ID field) {
     FrtIndexReader *ir = ISEA(self)->ir;
     return ir->term_vector(ir, doc_num, field);
 }
 
-static void isea_close(FrtSearcher *self)
-{
+static void isea_close(FrtSearcher *self) {
     if (ISEA(self)->ir && ISEA(self)->close_ir) {
         frt_ir_close(ISEA(self)->ir);
     }
     free(self);
 }
 
-FrtSearcher *frt_isea_new(FrtIndexReader *ir)
-{
-    FrtSearcher *self          = (FrtSearcher *)FRT_ALLOC(FrtIndexSearcher);
+FrtSearcher *frt_isea_alloc(void) {
+    return (FrtSearcher *)FRT_ALLOC(FrtIndexSearcher);
+}
 
+FrtSearcher *frt_isea_init(FrtSearcher *self, FrtIndexReader *ir) {
     ISEA(self)->ir          = ir;
     ISEA(self)->close_ir    = true;
 
@@ -1253,10 +1213,15 @@ FrtSearcher *frt_isea_new(FrtIndexReader *ir)
     self->get_term_vector   = &isea_get_term_vector;
     self->get_similarity    = &sea_get_similarity;
     self->close             = &isea_close;
-
+    self->rsea              = Qnil;
     return self;
 }
 
+FrtSearcher *frt_isea_new(FrtIndexReader *ir) {
+    FrtSearcher *self = frt_isea_alloc();
+    return frt_isea_init(self, ir);
+}
+
 /***************************************************************************
  *
  * CachedDFSearcher
@@ -1271,8 +1236,7 @@ typedef struct CachedDFSearcher
     int      max_doc;
 } CachedDFSearcher;
 
-static int cdfsea_doc_freq(FrtSearcher *self, FrtSymbol field, const char *text)
-{
+static int cdfsea_doc_freq(FrtSearcher *self, ID field, const char *text) {
     FrtTerm term;
     int *df;
     term.field = field;
@@ -1358,9 +1322,7 @@ static FrtExplanation *cdfsea_explain_w(FrtSearcher *self, FrtWeight *w, int doc
     return NULL;
 }
 
-static FrtTermVector *cdfsea_get_term_vector(FrtSearcher *self, const int doc_num,
-                                          FrtSymbol field)
-{
+static FrtTermVector *cdfsea_get_term_vector(FrtSearcher *self, const int doc_num, ID field) {
     (void)self; (void)doc_num; (void)field;
     FRT_RAISE(FRT_UNSUPPORTED_ERROR, "%s", FRT_UNSUPPORTED_ERROR_MSG);
     return NULL;
@@ -1436,8 +1398,7 @@ static int msea_get_searcher_index(FrtSearcher *self, int n)
     return hi;
 }
 
-static int msea_doc_freq(FrtSearcher *self, FrtSymbol field, const char *term)
-{
+static int msea_doc_freq(FrtSearcher *self, ID field, const char *term) {
     int i;
     int doc_freq = 0;
     FrtMultiSearcher *msea = MSEA(self);
@@ -1599,8 +1560,7 @@ static int msea_search_unscored(FrtSearcher *self,
                                 FrtQuery *query,
                                 int *buf,
                                 int limit,
-                                int offset_docnum)
-{
+                                int offset_docnum) {
     int count;
     FrtWeight *weight = frt_q_weight(query, self);
     count = msea_search_unscored_w(self, weight, buf, limit, offset_docnum);
@@ -1636,8 +1596,7 @@ static FrtTopDocs *msea_search_w(FrtSearcher *self,
                               FrtFilter *filter,
                               FrtSort *sort,
                               FrtPostFilter *post_filter,
-                              bool load_fields)
-{
+                              bool load_fields) {
     int max_size = num_docs + (num_docs == INT_MAX ? 0 : first_doc);
     int i;
     int total_hits = 0;
@@ -1705,8 +1664,7 @@ static FrtTopDocs *msea_search(FrtSearcher *self,
                             FrtFilter *filter,
                             FrtSort *sort,
                             FrtPostFilter *post_filter,
-                            bool load_fields)
-{
+                            bool load_fields) {
     FrtTopDocs *td;
     FrtWeight *weight = frt_q_weight(query, self);
     td = msea_search_w(self, weight, first_doc, num_docs, filter,
@@ -1715,8 +1673,7 @@ static FrtTopDocs *msea_search(FrtSearcher *self,
     return td;
 }
 
-static FrtQuery *msea_rewrite(FrtSearcher *self, FrtQuery *original)
-{
+static FrtQuery *msea_rewrite(FrtSearcher *self, FrtQuery *original) {
     int i;
     FrtSearcher *s;
     FrtMultiSearcher *msea = MSEA(self);
@@ -1735,8 +1692,7 @@ static FrtQuery *msea_rewrite(FrtSearcher *self, FrtQuery *original)
     return rewritten;
 }
 
-static FrtExplanation *msea_explain(FrtSearcher *self, FrtQuery *query, int doc_num)
-{
+static FrtExplanation *msea_explain(FrtSearcher *self, FrtQuery *query, int doc_num) {
     FrtMultiSearcher *msea = MSEA(self);
     int i = msea_get_searcher_index(self, doc_num);
     FrtWeight *w = frt_q_weight(query, self);
@@ -1746,8 +1702,7 @@ static FrtExplanation *msea_explain(FrtSearcher *self, FrtQuery *query, int doc_
     return e;
 }
 
-static FrtExplanation *msea_explain_w(FrtSearcher *self, FrtWeight *w, int doc_num)
-{
+static FrtExplanation *msea_explain_w(FrtSearcher *self, FrtWeight *w, int doc_num) {
     FrtMultiSearcher *msea = MSEA(self);
     int i = msea_get_searcher_index(self, doc_num);
     FrtSearcher *s = msea->searchers[i];
@@ -1755,22 +1710,18 @@ static FrtExplanation *msea_explain_w(FrtSearcher *self, FrtWeight *w, int doc_n
     return e;
 }
 
-static FrtTermVector *msea_get_term_vector(FrtSearcher *self, const int doc_num,
-                                        FrtSymbol field)
-{
+static FrtTermVector *msea_get_term_vector(FrtSearcher *self, const int doc_num, ID field) {
     FrtMultiSearcher *msea = MSEA(self);
     int i = msea_get_searcher_index(self, doc_num);
     FrtSearcher *s = msea->searchers[i];
     return s->get_term_vector(s, doc_num - msea->starts[i], field);
 }
 
-static FrtSimilarity *msea_get_similarity(FrtSearcher *self)
-{
+static FrtSimilarity *msea_get_similarity(FrtSearcher *self) {
     return self->similarity;
 }
 
-static void msea_close(FrtSearcher *self)
-{
+static void msea_close(FrtSearcher *self) {
     int i;
     FrtSearcher *s;
     FrtMultiSearcher *msea = MSEA(self);
@@ -1785,10 +1736,12 @@ static void msea_close(FrtSearcher *self)
     free(self);
 }
 
-FrtSearcher *frt_msea_new(FrtSearcher **searchers, int s_cnt, bool close_subs)
-{
+FrtSearcher *frt_msea_alloc(void) {
+    return (FrtSearcher *)FRT_ALLOC(FrtMultiSearcher);
+}
+
+FrtSearcher *frt_msea_init(FrtSearcher *self, FrtSearcher **searchers, int s_cnt, bool close_subs) {
     int i, max_doc = 0;
-    FrtSearcher *self = (FrtSearcher *)FRT_ALLOC(FrtMultiSearcher);
     int *starts = FRT_ALLOC_N(int, s_cnt + 1);
     for (i = 0; i < s_cnt; i++) {
         starts[i] = max_doc;
@@ -1796,29 +1749,35 @@ FrtSearcher *frt_msea_new(FrtSearcher **searchers, int s_cnt, bool close_subs)
     }
     starts[i] = max_doc;
 
-    MSEA(self)->s_cnt           = s_cnt;
-    MSEA(self)->searchers       = searchers;
-    MSEA(self)->starts          = starts;
-    MSEA(self)->max_doc         = max_doc;
-    MSEA(self)->close_subs      = close_subs;
-
-    self->similarity            = frt_sim_create_default();
-    self->doc_freq              = &msea_doc_freq;
-    self->get_doc               = &msea_get_doc;
-    self->get_lazy_doc          = &msea_get_lazy_doc;
-    self->max_doc               = &msea_max_doc;
-    self->create_weight         = &msea_create_weight;
-    self->search                = &msea_search;
-    self->search_w              = &msea_search_w;
-    self->search_each           = &msea_search_each;
-    self->search_each_w         = &msea_search_each_w;
-    self->search_unscored       = &msea_search_unscored;
-    self->search_unscored_w     = &msea_search_unscored_w;
-    self->rewrite               = &msea_rewrite;
-    self->explain               = &msea_explain;
-    self->explain_w             = &msea_explain_w;
-    self->get_term_vector       = &msea_get_term_vector;
-    self->get_similarity        = &msea_get_similarity;
-    self->close                 = &msea_close;
+    MSEA(self)->s_cnt       = s_cnt;
+    MSEA(self)->searchers   = searchers;
+    MSEA(self)->starts      = starts;
+    MSEA(self)->max_doc     = max_doc;
+    MSEA(self)->close_subs  = close_subs;
+
+    self->similarity        = frt_sim_create_default();
+    self->doc_freq          = &msea_doc_freq;
+    self->get_doc           = &msea_get_doc;
+    self->get_lazy_doc      = &msea_get_lazy_doc;
+    self->max_doc           = &msea_max_doc;
+    self->create_weight     = &msea_create_weight;
+    self->search            = &msea_search;
+    self->search_w          = &msea_search_w;
+    self->search_each       = &msea_search_each;
+    self->search_each_w     = &msea_search_each_w;
+    self->search_unscored   = &msea_search_unscored;
+    self->search_unscored_w = &msea_search_unscored_w;
+    self->rewrite           = &msea_rewrite;
+    self->explain           = &msea_explain;
+    self->explain_w         = &msea_explain_w;
+    self->get_term_vector   = &msea_get_term_vector;
+    self->get_similarity    = &msea_get_similarity;
+    self->close             = &msea_close;
+    self->rsea              = Qnil;
     return self;
 }
+
+FrtSearcher *frt_msea_new(FrtSearcher **searchers, int s_cnt, bool close_subs) {
+    FrtSearcher *self = frt_msea_alloc();
+    return frt_msea_init(self, searchers, s_cnt, close_subs);
+}