Changeset 0a55ed69e9335dedb34800a9364559814f7b38ee

Timestamp:

04/24/08 21:02:02 (9 months ago)

Author:

dave <dave@…>

Parents:

ad7faee95a07fccb5ea03fcc0e0875729617ab0c

Children:

8eb44963185f8afa008b1abfe3f6fad0839b2efd

git-committer:

dave <dave@06fd6eb0-0002-0410-a719-e5602cce40bc> / 2008-04-24T11:02:02Z+0000

Message:

Updated the ruby bindings to use the new Symbols

git-svn-id: svn+ssh://davebalmain.com/home/dave/repos/ferret/trunk@1041 06fd6eb0-0002-0410-a719-e5602cce40bc

Location:

ruby

Files:

• .gitignore (modified) (2 diffs)
• ext/ferret.c (modified) (4 diffs)
• ext/ferret.h (modified) (3 diffs)
• ext/r_analysis.c (modified) (3 diffs)
• ext/r_index.c (modified) (18 diffs)
• ext/r_qparser.c (modified) (1 diff)
• ext/r_search.c (modified) (15 diffs)

ruby/.gitignore

@@ -72 +72 @@
 ext/index.c
 ext/index.h
-ext/intern.c
-ext/intern.h
 ext/internal.h
 ext/libstemmer.h
@@ -140 +138 @@
 ext/store.c
 ext/store.h
+ext/symbol.c
+ext/symbol.h
 ext/term_vectors.c
 ext/win32.h

ruby/ext/ferret.c

@@ -5 +5 @@
 #include "hashset.h"
 #include "threading.h"
-#include "intern.h"
+#include "symbol.h"
 #include "internal.h"
 
@@ -179 +179 @@
 }
 
-const char *
+Symbol
 rintern(VALUE rstr)
 {
@@ -186 +186 @@
 }
 
-char *
+Symbol
 frb_field(VALUE rfield)
 {
     switch (TYPE(rfield)) {
         case T_SYMBOL:
-            return rb_id2name(SYM2ID(rfield));
+            return I(rb_id2name(SYM2ID(rfield)));
         case T_STRING:
-            return rs2s(rfield);
+            return I(rs2s(rfield));
         default:
             rb_raise(rb_eArgError, "field name must be a symbol");
-    }
-    return NULL;
+            return NULL;
+    }
 }
 
@@ -290 +290 @@
 static ID id_text;
 
-VALUE frb_get_term(const char *field, const char *text)
+VALUE frb_get_term(Symbol field, const char *text)
 {
     return rb_struct_new(cTerm,
-                         ID2SYM(rb_intern(field)),
+                         SYM2RSYM(field),
                          rb_str_new2(text),
                          NULL);

ruby/ext/ferret.h

@@ -5 +5 @@
 #include "hashset.h"
 #include "document.h"
+#include "symbol.h"
 #include "internal.h"
 
@@ -65 +66 @@
 extern VALUE frb_hs_to_rb_ary(HashSet *hs);
 extern void *frb_rb_data_ptr(VALUE val);
-extern char * frb_field(VALUE rfield);
-extern VALUE frb_get_term(const char *field, const char *term);
+extern Symbol frb_field(VALUE rfield);
+extern VALUE frb_get_term(Symbol field, const char *term);
 extern char *json_concat_string(char *s, char *field);
 extern char *rs2s(VALUE rstr);
 extern char *rstrdup(VALUE rstr);
-extern const char *rintern(VALUE rstr);
+extern Symbol rintern(VALUE rstr);
 #define Frt_Make_Struct(klass)\
   rb_data_object_alloc(klass,NULL,(RUBY_DATA_FUNC)NULL,(RUBY_DATA_FUNC)NULL)
@@ -90 +91 @@
 #endif
 
+#define SYM2RSYM(_sym) ID2SYM(rb_intern(S(_sym)))
 #define frb_mark_cclass(klass) rb_ivar_set(klass, id_cclass, Qtrue)
 #define frb_is_cclass(obj) (rb_ivar_get(CLASS_OF(obj), id_cclass) == Qtrue)

ruby/ext/r_analysis.c

@@ -1149 +1149 @@
 
 static TokenStream *
-cwa_get_ts(Analyzer *a, const char *field, char *text)
+cwa_get_ts(Analyzer *a, Symbol field, char *text)
 {
     VALUE rts = rb_funcall(CWA(a)->ranalyzer, id_token_stream, 2,
-                           ID2SYM(rb_intern(field)), rb_str_new2(text));
+                           SYM2RSYM(field), rb_str_new2(text));
     return frb_get_cwrapped_rts(rts);
 } 
@@ -1481 +1481 @@
 {
     Analyzer *pfa, *a;
-    char *field = frb_field(rfield);
+    Symbol field = frb_field(rfield);
     GET_A(pfa, self);
 
@@ -1491 +1491 @@
     if (a->get_ts == cwa_get_ts) {
         return rb_funcall(CWA(a)->ranalyzer, id_token_stream, 2,
-                          ID2SYM(rb_intern(field)), rb_str_new2(rs2s(rstring)));
+                          SYM2RSYM(field), rb_str_new2(rs2s(rstring)));
     }
     else {

ruby/ext/r_index.c

@@ -52 +52 @@
 static VALUE sym_with_offsets;
 static VALUE sym_with_positions_offsets;
+
+static Symbol fsym_content;
 
 static ID id_term;
@@ -207 +209 @@
 {
     FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
-    return ID2SYM(rb_intern(fi->name));
+    return ID2SYM(rb_intern(S(fi->name)));
 }
 
@@ -486 +488 @@
                        }
         case T_SYMBOL:
+        case T_STRING:
             rfi = frb_get_field_info(fis_get_field(fis, frb_field(ridx)));
             break;
+            /*
         case T_STRING:
             rfi = frb_get_field_info(fis_get_field(fis, StringValuePtr(ridx)));
             break;
+            */
         default:
             rb_raise(rb_eArgError, "Can't index FieldInfos with %s",
@@ -634 +639 @@
     int i;
     for (i = 0; i < fis->size; i++) {
-        rb_ary_push(rfield_names, ID2SYM(rb_intern(fis->fields[i]->name)));
+        rb_ary_push(rfield_names, SYM2RSYM(fis->fields[i]->name));
     }
     return rfield_names;
@@ -654 +659 @@
     for (i = 0; i < fis->size; i++) {
         if (!fi_is_tokenized(fis->fields[i])) continue;
-        rb_ary_push(rfield_names, ID2SYM(rb_intern(fis->fields[i]->name)));
+        rb_ary_push(rfield_names, SYM2RSYM(fis->fields[i]->name));
     }
     return rfield_names;
@@ -749 +754 @@
 {
     TermEnum *te = (TermEnum *)DATA_PTR(self);
-    return frb_te_get_set_term(self, te->skip_to(te, frb_field(rterm)));
+    return frb_te_get_set_term(self, te->skip_to(te, rs2s(rterm)));
 }
 
@@ -1240 +1245 @@
     VALUE rfield, rterms, *rts;
     VALUE roffsets = Qnil;
-    rfield = ID2SYM(rb_intern(tv->field));
+    rfield = SYM2RSYM(tv->field);
 
     rterms = rb_ary_new2(t_cnt);
@@ -1432 +1437 @@
     } else {
         Document *doc = (Document *)arg;
-        char *field;
+        Symbol field = frb_field(key);
         VALUE val;
         DocField *df;
-        switch (TYPE(key)) {
-            case T_STRING:
-                field = rs2s(key);
-                break;
-            case T_SYMBOL:
-                field = rb_id2name(SYM2ID(key));
-                break;
-            default:
-                rb_raise(rb_eArgError,
-                         "%s cannot be a key to a field. Field keys must "
-                         " be symbols.", rs2s(rb_obj_as_string(key)));
-                break;
-        }
         if (NULL == (df = doc_get_field(doc, field))) {
             df = df_new(field);
@@ -1497 +1489 @@
             {
                 int i;
-                df = df_new("content");
+                df = df_new(fsym_content);
                 df->destroy_data = true;
                 for (i = 0; i < RARRAY(rdoc)->len; i++) {
@@ -1507 +1499 @@
             break;
         case T_SYMBOL:
-            df = df_add_data(df_new("content"), rb_id2name(SYM2ID(rdoc)));
+            df = df_add_data(df_new(fsym_content), rb_id2name(SYM2ID(rdoc)));
             doc_add_field(doc, df);
             break;
         case T_STRING:
-            df = df_add_data_len(df_new("content"), rs2s(rdoc),
+            df = df_add_data_len(df_new(fsym_content), rs2s(rdoc),
                                  RSTRING(rdoc)->len);
             doc_add_field(doc, df);
@@ -1517 +1509 @@
         default:
             val = rb_obj_as_string(rdoc);
-            df = df_add_data_len(df_new("content"), rstrdup(val),
+            df = df_add_data_len(df_new(fsym_content), rstrdup(val),
                                  RSTRING(val)->len);
             df->destroy_data = true;
@@ -1988 +1980 @@
 {
     LazyDoc *lazy_doc = (LazyDoc *)DATA_PTR(rb_ivar_get(self, id_data));
-    char *field = NULL;
-    switch (TYPE(rkey)) {
-        case T_STRING:
-            field = rs2s(rkey);
-            rkey = ID2SYM(rb_intern(field));
-            break;
-        case T_SYMBOL:
-            field = frb_field(rkey);
-            break;
-        default:
-            rb_raise(rb_eArgError,
-                     "%s cannot be a key to a field. Field keys must "
-                     " be symbols.", rs2s(rb_obj_as_string(rkey)));
-            break;
-    }
-    return frb_lazy_df_load(self, rkey, h_get(lazy_doc->field_dict, field));
+    Symbol field = frb_field(rkey);
+    VALUE rfield = SYM2RSYM(field);
+
+    return frb_lazy_df_load(self, rfield, h_get(lazy_doc->field_dict, field));
 }
 
@@ -2033 +2013 @@
     for (i = 0; i < lazy_doc->size; i++) {
         LazyDocField *lazy_df = lazy_doc->fields[i];
-        frb_lazy_df_load(self, ID2SYM(rb_intern(lazy_df->name)), lazy_df);
+        frb_lazy_df_load(self, SYM2RSYM(lazy_df->name), lazy_df);
     }
     return self;
@@ -2052 +2032 @@
 
     for (i = 0; i < lazy_doc->size; i++) {
-        RARRAY(rfields)->ptr[i] = ID2SYM(rb_intern(lazy_doc->fields[i]->name));
+        RARRAY(rfields)->ptr[i] = SYM2RSYM(lazy_doc->fields[i]->name);
         RARRAY(rfields)->len++;
     }
@@ -2191 +2171 @@
         FieldInfo *fi = fis->fields[i];
         rb_hash_aset(rfield_num_map,
-                     ID2SYM(rb_intern(fi->name)),
+                     SYM2RSYM(fi->name),
                      INT2FIX(fi->number));
     }
@@ -2660 +2640 @@
     int i;
     for (i = 0; i < fis->size; i++) {
-        rb_ary_push(rfield_names, ID2SYM(rb_intern(fis->fields[i]->name)));
+        rb_ary_push(rfield_names, SYM2RSYM(fis->fields[i]->name));
     }
     return rfield_names;
@@ -2696 +2676 @@
     for (i = 0; i < fis->size; i++) {
         if (!fi_is_tokenized(fis->fields[i])) continue;
-        rb_ary_push(rfield_names, ID2SYM(rb_intern(fis->fields[i]->name)));
+        rb_ary_push(rfield_names, SYM2RSYM(fis->fields[i]->name));
     }
     return rfield_names;
@@ -3486 +3466 @@
     sym_analyzer  = ID2SYM(rb_intern("analyzer"));
     sym_close_dir = ID2SYM(rb_intern("close_dir"));
+    fsym_content  = I("content");
 
     Init_TermVector();

ruby/ext/r_qparser.c

@@ -1 @@
-#include "intern.h"
+#include "symbol.h"
 #include "ferret.h"
 #include "search.h"

ruby/ext/r_search.c

@@ -127 +127 @@
 static VALUE sym_ellipsis;      
 
+static Symbol fsym_id;
+
 extern VALUE cIndexReader;
 extern void frb_ir_free(void *p);
@@ -196 +198 @@
     int p = 0;
     char *str = ALLOC_N(char, len * 64 + 100);
-    char *field = "id";
+    Symbol field = fsym_id;
     VALUE rstr;
 
@@ -250 +252 @@
          * between fields. Add 100 for '[', ']' and good safety.
          */
-        len += strlen(f->name) + f->len * 3 + 100 + 4 * f->size;
+        len += sym_len(f->name) + f->len * 3 + 100 + 4 * f->size;
     }
 
@@ -263 +265 @@
                 if (i)  *(s++) = ',';
         *(s++) = '"';
-        l = strlen(f->name);
+        l = sym_len(f->name);
         memcpy(s, f->name, l);
         s += l;
@@ -401 +403 @@
     GET_Q();
     VALUE rstr, rfield;
-    char *str, *field = "";
+    char *str;
+    Symbol field = NULL;
     if (rb_scan_args(argc, argv, "01", &rfield)) {
         field = frb_field(rfield);
@@ -597 +600 @@
 frb_tq_init(VALUE self, VALUE rfield, VALUE rterm)
 {
-    char *field = frb_field(rfield);
+    Symbol field = frb_field(rfield);
     char *term = rs2s(rb_obj_as_string(rterm));
     Query *q = tq_new(field, term);
@@ -714 +717 @@
 }
 
-typedef Query *(*mtq_maker_ft)(const char *field, const char *term);
+typedef Query *(*mtq_maker_ft)(Symbol field, const char *term);
 
 static int
@@ -913 +916 @@
     int len;
     GET_BC();
-    qstr = bc->query->to_s(bc->query, "");
+    qstr = bc->query->to_s(bc->query, NULL);
     switch (bc->occur) {
         case BC_SHOULD:
@@ -2175 +2178 @@
     int type = SORT_TYPE_AUTO;
     int is_reverse = false;
-    char *field;
+    Symbol field;
 
     if (rb_scan_args(argc, argv, "11", &rfield, &roptions) == 2) {
@@ -2192 +2195 @@
 
     sf = sort_field_new(field, type, is_reverse);
-    if (sf->field == NULL && field) {
-        sf->field = estrdup(field);
+    if (sf->field == NULL) {
+        sf->field = field;
     }
 
@@ -2227 +2230 @@
 {
     GET_SF();
-    return sf->field ? ID2SYM(rb_intern(sf->field)) : Qnil;
+    return sf->field ? SYM2RSYM(sf->field) : Qnil;
 }
 
@@ -2349 +2352 @@
             sf = sort_field_doc_new(reverse);
         } else {
-            sf = sort_field_auto_new(s, reverse);
+            sf = sort_field_auto_new(I(s), reverse);
         }
         frb_get_sf(sf);
@@ -2369 +2372 @@
             break;
         case T_SYMBOL:
-            rsf = rb_obj_as_string(rsf);
-            sf = sort_field_auto_new(rs2s(rsf), reverse);
+            sf = sort_field_auto_new(frb_field(rsf), reverse);
             /* need to give it a ruby object so it'll be freed when the
              * sort is garbage collected */
@@ -2616 +2618 @@
     }
     else {
-        filter = filt_create(sizeof(CWrappedFilter), "CWrappedFilter");
-        filter->hash     = &cwfilt_hash;
-        filter->eq       = &cwfilt_eq;
-        filter->get_bv_i = &cwfilt_get_bv_i;
+        filter               = filt_new(CWrappedFilter);
+        filter->hash         = &cwfilt_hash;
+        filter->eq           = &cwfilt_eq;
+        filter->get_bv_i     = &cwfilt_get_bv_i;
         CWF(filter)->rfilter = rval;
     }
@@ -4418 +4420 @@
     mSearch = rb_define_module_under(mFerret, "Search");
 
+    fsym_id = I("id");
+
     Init_Hit();
     Init_TopDocs();