isomorfeus-ferret 0.17.2 → 0.17.3

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;
@@ -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);
@@ -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,
@@ -3072,8 +3069,24 @@ static VALUE frb_sea_doc_freq(VALUE self, VALUE rfield, VALUE rterm) {
  *  which are returned by the Searchers search methods.
  */
 static VALUE frb_sea_doc(VALUE self, VALUE rdoc_id) {
+    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_id)));
+    FRT_XCATCHALL
+        ex_code = xcontext.excode;
+        msg = xcontext.msg;
+        FRT_HANDLED();
+    FRT_XENDTRY
+
+    if (ex_code && msg) {
+        frb_raise(ex_code, msg);
+    }
+
+    return ld;
 }
 
 /*
@@ -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);
             }
         }
@@ -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)));
@@ -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);
             }
         }
@@ -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 {
@@ -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");

data/ext/isomorfeus_ferret_ext/frb_store.c

@@ -8,8 +8,10 @@ static ID id_ref_cnt;
 VALUE cLock;
 VALUE cLockError;
 VALUE cDirectory;
-VALUE cRAMDirectory;
 VALUE cMDBXDirectory;
+VALUE cRAMDirectory;
+VALUE cObjectStore;
+VALUE mStore;
 
 /****************************************************************************
  * Lock Methods
@@ -307,7 +309,7 @@ static VALUE frb_dir_close(VALUE self) {
 
 /*
  *  call-seq:
- *     dir.exists?(file_name) -> nil
+ *     dir.exist?(file_name) -> nil
  *
  *  Return true if a file with the name +file_name+ exists in the directory.
  */
@@ -319,7 +321,7 @@ static VALUE frb_dir_exists(VALUE self, VALUE rfname) {
     bool res;
 
     FRT_TRY
-        res = store->exists(store, rs2s(rfname));
+        res = store->exists(store, segm_idx_name, rs2s(rfname));
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -346,7 +348,7 @@ static VALUE frb_dir_touch(VALUE self, VALUE rfname) {
     StringValue(rfname);
 
     FRT_TRY
-        store->touch(store, rs2s(rfname));
+        store->touch(store, segm_idx_name, rs2s(rfname));
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -373,7 +375,7 @@ static VALUE frb_dir_delete(VALUE self, VALUE rfname) {
     StringValue(rfname);
     bool res;
     FRT_TRY
-        res = (store->remove(store, rs2s(rfname)) == 0);
+        res = (store->remove(store, segm_idx_name, rs2s(rfname)) == 0);
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -399,7 +401,7 @@ static VALUE frb_dir_file_count(VALUE self) {
     FrtStore *store = DATA_PTR(self);
     int cnt = 0;
     FRT_TRY
-        cnt = INT2FIX(store->count(store));
+        cnt = INT2FIX(store->count(store, segm_idx_name));
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -425,7 +427,7 @@ static VALUE frb_dir_refresh(VALUE self) {
     FrtStore *store = DATA_PTR(self);
 
     FRT_TRY
-        store->clear_all(store);
+        store->clear_all(store, segm_idx_name);
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -454,7 +456,7 @@ static VALUE frb_dir_rename(VALUE self, VALUE rfrom, VALUE rto) {
     StringValue(rfrom);
     StringValue(rto);
     FRT_TRY
-        store->rename(store, rs2s(rfrom), rs2s(rto));
+        store->rename(store, segm_idx_name, rs2s(rfrom), rs2s(rto));
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -486,7 +488,7 @@ static VALUE frb_dir_make_lock(VALUE self, VALUE rlock_name) {
     FrtStore *store = DATA_PTR(self);
     StringValue(rlock_name);
     FRT_TRY
-        lock = frt_open_lock(store, rs2s(rlock_name));
+        lock = frt_open_lock(store, segm_idx_name, rs2s(rlock_name));
     FRT_XCATCHALL
         ex_code = xcontext.excode;
         msg = xcontext.msg;
@@ -529,10 +531,13 @@ static VALUE frb_ramdir_init(int argc, VALUE *argv, VALUE self) {
         case 1: {
                     FrtStore *ostore;
                     TypedData_Get_Struct(rdir, FrtStore, &frb_store_t, ostore);
-                    frt_open_ram_store_and_copy(store, ostore, false);
+                    frt_open_ram_store_and_copy(store, ostore, segm_idx_name, false);
                     break;
                 }
-        default: frt_open_ram_store(store);
+        default: {
+            frt_open_ram_store(store);
+            store->create_folder(store, segm_idx_name);
+        }
     }
     store->rstore = self;
     rb_ivar_set(self, id_ref_cnt, INT2FIX(0));
@@ -583,7 +588,8 @@ static VALUE frb_mdbxdir_new(int argc, VALUE *argv, VALUE klass) {
 
     FRT_TRY
         store = frt_open_mdbx_store(rs2s(rpath));
-        if (create) store->clear_all(store);
+        store->create_folder(store, segm_idx_name);
+        if (create) store->clear_all(store, segm_idx_name);
         self = store->rstore;
         if (self == Qnil || DATA_PTR(self) == NULL) {
             self = TypedData_Wrap_Struct(klass, &frb_store_t, store);
@@ -634,7 +640,6 @@ void Init_Directory(void) {
     cDirectory = rb_define_class_under(mStore, "Directory", rb_cObject);
     rb_define_const(cDirectory, "LOCK_PREFIX", rb_str_new2(FRT_LOCK_PREFIX));
     rb_define_method(cDirectory, "close", frb_dir_close, 0);
-    rb_define_method(cDirectory, "exists?", frb_dir_exists, 1);
     rb_define_method(cDirectory, "exist?", frb_dir_exists, 1);
     rb_define_method(cDirectory, "touch", frb_dir_touch, 1);
     rb_define_method(cDirectory, "delete", frb_dir_delete, 1);
@@ -707,6 +712,14 @@ void Init_MDBXDirectory(void) {
     rb_define_singleton_method(cMDBXDirectory, "new", frb_mdbxdir_new, -1);
 }
 
+void Init_ObjectStore(void) {
+    cObjectStore = rb_define_class_under(mStore, "ObjectStore", rb_cObject);
+    // rb_define_alloc_func(cObjectStore, frb_obst_alloc);
+    // rb_define_method(cObjectStore, "initialize", frb_obst_init, -1);
+    // rb_define_method(cObjectStore, "fetch", frb_obst_fetch, -1);
+    // rb_define_method(cObjectStore, "key?", frb_obst_key?, -1);
+    // rb_define_method(cObjectStore, "store", frb_obst_store, -1);
+}
 /*
  *  Document-module: Ferret::Store
  *
@@ -724,4 +737,5 @@ void Init_Store(void) {
     Init_Lock();
     Init_RAMDirectory();
     Init_MDBXDirectory();
+    Init_ObjectStore();
 }

data/ext/isomorfeus_ferret_ext/frb_utils.c

@@ -599,8 +599,7 @@ static int frb_mulmap_add_mappings_i(VALUE key, VALUE value, VALUE arg) {
             for (i = RARRAY_LEN(key) - 1; i >= 0; i--) {
                 frb_mulmap_add_mapping_i(mulmap, RARRAY_PTR(key)[i], to);
             }
-        }
-        else {
+        } else {
             frb_mulmap_add_mapping_i(mulmap, key, to);
         }
     }
@@ -904,8 +903,7 @@ static VALUE frb_pq_insert(VALUE self, VALUE elem) {
     GET_PQ(pq, self);
     if (pq->size < pq->capa) {
         frb_pq_push(pq, elem);
-    }
-    else if (pq->size > 0 && frb_pq_lt(pq->proc, pq->heap[1], elem)) {
+    } else if (pq->size > 0 && frb_pq_lt(pq->proc, pq->heap[1], elem)) {
         pq->heap[1] = elem;
         frb_pq_down(pq);
     }
@@ -957,8 +955,7 @@ static VALUE frb_pq_pop(VALUE self) {
         pq->size--;
         frb_pq_down(pq);                      /* adjust heap */
         return result;
-    }
-    else {
+    } else {
         return Qnil;
     }
 }

data/ext/isomorfeus_ferret_ext/frt_analysis.c

@@ -52,7 +52,7 @@ static bool cp_enc_istok(OnigCodePoint cp, rb_encoding *enc) {
     return false;
 }
 
-static inline int get_cp(char *start, char *end, int *cp_len, rb_encoding *enc) {
+static inline int get_cp(const char *start, const char *end, int *cp_len, rb_encoding *enc) {
     if (start >= end) {
         *cp_len = 0;
         return 0;
@@ -64,7 +64,7 @@ static inline int get_cp(char *start, char *end, int *cp_len, rb_encoding *enc)
 /*** FrtToken ****************************************************************/
 /*****************************************************************************/
 
-FrtToken *frt_tk_set(FrtToken *tk, char *text, int tlen, frt_off_t start, frt_off_t end, int pos_inc, rb_encoding *encoding) {
+FrtToken *frt_tk_set(FrtToken *tk, const char *text, int tlen, frt_off_t start, frt_off_t end, int pos_inc, rb_encoding *encoding) {
     if (tlen >= FRT_MAX_WORD_SIZE) {
         char *head_last = rb_enc_left_char_head(text, text + FRT_MAX_WORD_SIZE - 1, text + tlen, encoding);
         tlen = head_last - text;
@@ -89,11 +89,11 @@ FrtToken *frt_tk_set(FrtToken *tk, char *text, int tlen, frt_off_t start, frt_of
     return tk;
 }
 
-static FrtToken *frt_tk_set_ts(FrtToken *tk, char *start, char *end, char *text, int pos_inc, rb_encoding *encoding) {
+FrtToken *frt_tk_set_ts(FrtToken *tk, const char *start, const char *end, const char *text, int pos_inc, rb_encoding *encoding) {
     return frt_tk_set(tk, start, (int)(end - start), (off_t)(start - text), (off_t)(end - text), pos_inc, encoding);
 }
 
-FrtToken *frt_tk_set_no_len(FrtToken *tk, char *text, frt_off_t start, frt_off_t end, int pos_inc, rb_encoding *encoding) {
+FrtToken *frt_tk_set_no_len(FrtToken *tk, const char *text, frt_off_t start, frt_off_t end, int pos_inc, rb_encoding *encoding) {
     return frt_tk_set(tk, text, (int)strlen(text), start, end, pos_inc, encoding);
 }
 
@@ -138,7 +138,7 @@ void frt_ts_deref(FrtTokenStream *ts) {
         ts->destroy_i(ts);
 }
 
-FrtTokenStream *frt_ts_reset(FrtTokenStream *ts, char *text, rb_encoding *encoding) {
+FrtTokenStream *frt_ts_reset(FrtTokenStream *ts, const char *text, rb_encoding *encoding) {
     ts->t = ts->text = text;
     ts->length = strlen(text);
     ts->encoding = encoding;
@@ -224,14 +224,13 @@ FrtTokenStream *frt_non_tokenizer_new(void) {
 /*** FrtWhiteSpaceTokenizer **************************************************/
 /*****************************************************************************/
 
-static FrtToken *wst_next(FrtTokenStream *ts)
-{
+static FrtToken *wst_next(FrtTokenStream *ts) {
     int cp_len = 0;
     OnigCodePoint cp;
     rb_encoding *enc = ts->encoding;
-    char *end = ts->text + ts->length;
-    char *start;
-    char *t = ts->t;
+    const char *end = ts->text + ts->length;
+    const char *start;
+    const char *t = ts->t;
 
     cp = get_cp(t, end, &cp_len, enc);
     if (cp < 1)
@@ -278,9 +277,9 @@ static FrtToken *lt_next(FrtTokenStream *ts) {
     int cp_len = 0;
     OnigCodePoint cp;
     rb_encoding *enc = ts->encoding;
-    char *end = ts->text + ts->length;
-    char *start;
-    char *t = ts->t;
+    const char *end = ts->text + ts->length;
+    const char *start;
+    const char *t = ts->t;
 
     cp = get_cp(t, end, &cp_len, enc);
     if (cp < 1)
@@ -324,9 +323,9 @@ FrtTokenStream *frt_letter_tokenizer_new(void) {
 /*****************************************************************************/
 
 static int std_get_alnum(FrtTokenStream *ts, char *token, OnigCodePoint cp, int *cp_len_p, OnigCodePoint *cp_out_p, rb_encoding *enc) {
-    char *end = ts->text + ts->length;
-    char *t = ts->t;
-    char *tt = ts->t;
+    const char *end = ts->text + ts->length;
+    const char *t = ts->t;
+    const char *tt = ts->t;
     int cp_len = *cp_len_p;
 
     while (cp > 0 && rb_enc_isalnum(cp, enc)) {
@@ -349,10 +348,9 @@ static int std_get_alnum(FrtTokenStream *ts, char *token, OnigCodePoint cp, int
  * (alnum) = [a-zA-Z0-9]
  * (punc) = [_\/.,-]
  */
-static int std_get_number(FrtTokenStream *ts, char *start, char *end, OnigCodePoint cp, int cp_len_a, rb_encoding *enc) {
-
+static int std_get_number(FrtTokenStream *ts, const char *start, const char *end, OnigCodePoint cp, int cp_len_a, rb_encoding *enc) {
     OnigCodePoint cp_1 = 0;
-    char *t = start;
+    const char *t = start;
     int cp_len = cp_len_a;
     int cp_1_len = 0;
     int last_seen_digit = 2;
@@ -385,10 +383,10 @@ static int std_get_number(FrtTokenStream *ts, char *start, char *end, OnigCodePo
     }
 }
 
-static int std_get_apostrophe(FrtTokenStream *ts, char *input, OnigCodePoint cp, int *cp_len_p, rb_encoding *enc) {
+static int std_get_apostrophe(FrtTokenStream *ts, const char *input, OnigCodePoint cp, int *cp_len_p, rb_encoding *enc) {
     int cp_len = *cp_len_p;
-    char *end = ts->text + ts->length;
-    char *t = input;
+    const char *end = ts->text + ts->length;
+    const char *t = input;
 
     while (cp_len > 0 && (rb_enc_isalpha(cp, enc) || cp == cp_apostrophe)) {
         t += cp_len;
@@ -397,14 +395,14 @@ static int std_get_apostrophe(FrtTokenStream *ts, char *input, OnigCodePoint cp,
     return (int)(t - input);
 }
 
-static char *std_get_url(FrtTokenStream *ts, char *start, char *end, char *token, int *len, int bufred) {
+static const char *std_get_url(FrtTokenStream *ts, const char *start, const char *end, char *token, int *len, int bufred) {
     rb_encoding *enc = ts->encoding;
     OnigCodePoint cp;
     OnigCodePoint prev_cp = 0;
     int cp_len = 0;
     int prev_cp_len = 0;
-    char *t = start;
-    char *tt = start;
+    const char *t = start;
+    const char *tt = start;
 
     cp = get_cp(t, end, &cp_len, enc);
     while (cp > 0 && cp_enc_isurlc(cp, enc)) {
@@ -432,9 +430,9 @@ static char *std_get_url(FrtTokenStream *ts, char *start, char *end, char *token
 }
 
 /* Company names can contain '@' and '&' like AT&T and Excite@Home. */
-static int std_get_company_name(FrtTokenStream *ts, char *start, char* end) {
+static int std_get_company_name(FrtTokenStream *ts, const char *start, const char* end) {
     rb_encoding *enc = ts->encoding;
-    char * t = start;
+    const char * t = start;
     OnigCodePoint cp;
     int cp_len = 0;
 
@@ -452,8 +450,8 @@ static int std_advance_to_start(FrtTokenStream *ts, int *cp_len_p, OnigCodePoint
     int cp_next = 0;
     int cp_len_next = 0;
     OnigCodePoint cp;
-    char *end = ts->text + ts->length;
-    char *t = ts->t;
+    const char *end = ts->text + ts->length;
+    const char *t = ts->t;
 
     cp = get_cp(t, end, &cp_len, enc);
     while (cp > 0 && !rb_enc_isalnum(cp, enc)) {
@@ -472,11 +470,9 @@ static int std_advance_to_start(FrtTokenStream *ts, int *cp_len_p, OnigCodePoint
 }
 
 static FrtToken *std_next(FrtTokenStream *ts) {
-    char *s;
-    char *t;
-    char *start = NULL;
-    char *end;
-    char *num_end = NULL;
+    const char *s, *t, *end;
+    const char *start = NULL;
+    const char *num_end = NULL;
     char token[FRT_MAX_WORD_SIZE + 1];
     OnigCodePoint cp = 0;
     OnigCodePoint cp_1 = 0;
@@ -528,13 +524,11 @@ static FrtToken *std_next(FrtTokenStream *ts) {
             t -= 2;
             frt_tk_set_ts(&(ts->token), start, t, ts->text, 1, enc);
             ts->token.end += 2;
-        }
-        else if (t[-1] == '\'') {
+        } else if (t[-1] == '\'') {
             t -= 1;
             frt_tk_set_ts(&(ts->token), start, t, ts->text, 1, enc);
             ts->token.end += 1;
-        }
-        else {
+        } else {
             frt_tk_set_ts(&(ts->token), start, t, ts->text, 1, enc);
         }
         return &(ts->token);
@@ -606,8 +600,7 @@ static FrtToken *std_next(FrtTokenStream *ts) {
         if (cp == cp_at) {
             if (seen_at_symbol) {
                 break; /* we can only have one @ symbol */
-            }
-            else {
+            } else {
                 seen_at_symbol = true;
             }
         }
@@ -693,7 +686,7 @@ static FrtTokenStream *filter_clone_i(FrtTokenStream *ts) {
     return frt_filter_clone_size(ts, sizeof(FrtTokenFilter));
 }
 
-static FrtTokenStream *filter_reset(FrtTokenStream *ts, char *text, rb_encoding *encoding) {
+static FrtTokenStream *filter_reset(FrtTokenStream *ts, const char *text, rb_encoding *encoding) {
     TkFilt(ts)->sub_ts->reset(TkFilt(ts)->sub_ts, text, encoding);
     return ts;
 }
@@ -837,7 +830,7 @@ static FrtToken *mf_next(FrtTokenStream *ts) {
     return tk;
 }
 
-static FrtTokenStream *mf_reset(FrtTokenStream *ts, char *text, rb_encoding *encoding) {
+static FrtTokenStream *mf_reset(FrtTokenStream *ts, const char *text, rb_encoding *encoding) {
     FrtMultiMapper *mm = MFilt(ts)->mapper;
     if (mm->d_size == 0)
         frt_mulmap_compile(MFilt(ts)->mapper);
@@ -1100,7 +1093,7 @@ static void frt_a_standard_destroy_i(FrtAnalyzer *a) {
     free(a);
 }
 
-static FrtTokenStream *a_standard_get_ts(FrtAnalyzer *a, ID field, char *text, rb_encoding *encoding) {
+static FrtTokenStream *a_standard_get_ts(FrtAnalyzer *a, ID field, const char *text, rb_encoding *encoding) {
     FrtTokenStream *ts;
     (void)field;
     ts = frt_ts_clone(a->current_ts);
@@ -1112,7 +1105,7 @@ FrtAnalyzer *frt_analyzer_alloc(void) {
 }
 
 void frt_analyzer_init(FrtAnalyzer *a, FrtTokenStream *ts, void (*destroy_i)(FrtAnalyzer *a),
-                       FrtTokenStream *(*get_ts)(FrtAnalyzer *a, ID field, char *text, rb_encoding *encoding)) {
+                       FrtTokenStream *(*get_ts)(FrtAnalyzer *a, ID field, const char *text, rb_encoding *encoding)) {
     a->current_ts = ts;
     a->destroy_i = (destroy_i ? destroy_i : &frt_a_standard_destroy_i);
     a->get_ts = (get_ts ? get_ts : &a_standard_get_ts);
@@ -1121,7 +1114,7 @@ void frt_analyzer_init(FrtAnalyzer *a, FrtTokenStream *ts, void (*destroy_i)(Frt
 }
 
 FrtAnalyzer *frt_analyzer_new(FrtTokenStream *ts, void (*destroy_i)(FrtAnalyzer *a),
-                       FrtTokenStream *(*get_ts)(FrtAnalyzer *a, ID field, char *text, rb_encoding *encoding)) {
+                       FrtTokenStream *(*get_ts)(FrtAnalyzer *a, ID field, const char *text, rb_encoding *encoding)) {
     FrtAnalyzer *a = frt_analyzer_alloc();
     frt_analyzer_init(a, ts, destroy_i, get_ts);
     return a;
@@ -1214,7 +1207,7 @@ static void pfa_destroy_i(FrtAnalyzer *self) {
     free(self);
 }
 
-static FrtTokenStream *pfa_get_ts(FrtAnalyzer *self, ID field, char *text, rb_encoding *encoding) {
+static FrtTokenStream *pfa_get_ts(FrtAnalyzer *self, ID field, const char *text, rb_encoding *encoding) {
     FrtAnalyzer *a = (FrtAnalyzer *)frt_h_get(PFA(self)->dict, (void *)field);
     if (a == NULL)
         a = PFA(self)->default_a;