aurelian-ruby-ahocorasick 0.3.1 → 0.3.3

data/ext/ruby-ahocorasick.c

@@ -15,7 +15,7 @@
 //  * kwt.find_first("str")
 //  * kwt.find_all  ("str")
 //
-// TODO: rename search to find_all
+// TODO: change last_id and dictionary_size to long
 //
 
 #include <ruby.h>
@@ -37,6 +37,15 @@ struct kwt_struct_data {
   int is_frozen;
 };
 
+// int
+// rb_add_string(struct kwt_struct_data *kwt, char *word, int size, int id) {
+//   if(ac_add_string( kwt->tree, word, size, id ) == 0)
+//     return 0;
+//   kwt->dictionary_size++;
+//   kwt->last_id= id+1;
+//   return 1;
+// }
+
 /*
  * call-seq: initialize
  *
@@ -114,7 +123,7 @@ rb_kwt_make(VALUE self)
  *
  */
 static VALUE
-rb_kwt_search(int argc, VALUE *argv, VALUE self)
+rb_kwt_find_all(int argc, VALUE *argv, VALUE self)
 {
   char * result;        // itermediate result
   char * remain;        // returned by ac_search, the remaing text to search
@@ -137,7 +146,6 @@ rb_kwt_search(int argc, VALUE *argv, VALUE self)
     kwt_data->is_frozen = 1;
   }
   // prepare the return value
-  // v_results= rb_block_given_p()? Qnil : rb_ary_new();
   v_results= rb_ary_new();
   // fail quickly and return the empty array
   if(kwt_data->dictionary_size == 0) 
@@ -148,32 +156,19 @@ rb_kwt_search(int argc, VALUE *argv, VALUE self)
   while((remain= ac_search(kwt_data->tree, &lgt, &id, &ends_at)) != NULL) {
     // this is an individual result as a hash
     v_result= rb_hash_new();
-    rb_hash_aset( v_result, sym_id, INT2FIX(id) );
+    rb_hash_aset( v_result, sym_id,        INT2FIX(id) );
     rb_hash_aset( v_result, sym_starts_at, INT2FIX( ends_at - lgt - 1 ) );
-    rb_hash_aset( v_result, sym_ends_at, INT2FIX( ends_at - 1 ) );
+    rb_hash_aset( v_result, sym_ends_at,   INT2FIX( ends_at - 1 ) );
     result = (char*) malloc (sizeof(char)*lgt);
     sprintf( result, "%.*s", lgt, remain);
     rb_hash_aset( v_result, sym_value, rb_str_new(result, lgt) );
-
-    // yield this hash or, add it to the results
-    // if(rb_block_given_p())
-    //   rb_yield(v_result);
-    // else
     rb_ary_push( v_results, v_result );
     free(result);
   }
-  
   // TODO: maybe the Tree can be re-opened to add new items to dictionary
-
-  // return the results or nil if none
-  // if( v_results != Qnil && RARRAY(v_results)->len > 0 ) {
   return v_results;
-  // } else {
-  //   return Qnil;
-  // }
 }
 
-
 /*
  * Document-method: size
  * call-seq: size
@@ -194,7 +189,6 @@ rb_kwt_size(VALUE self)
   return INT2FIX(kwt_data->dictionary_size);
 }
 
-
 /*
  * Document-method: add_string
  * call-seq: add_string
@@ -228,7 +222,6 @@ rb_kwt_add_string(int argc, VALUE *argv, VALUE self)
  
   Check_Type(v_string, T_STRING);
   string= RSTRING(v_string)->ptr;
-
   KeywordTree(self, kwt_data);
 
   if(kwt_data->is_frozen == 1)
@@ -244,16 +237,12 @@ rb_kwt_add_string(int argc, VALUE *argv, VALUE self)
     id= NUM2INT(v_id);
   }
   
-  // printf("[internal]==> %d\n", id);
-  
-  if( ac_add_string(kwt_data->tree, string, strlen(string), id) == 0 ) {
+  if(ac_add_string(kwt_data->tree, string, strlen(string), id) == 0)
     rb_raise(rb_eRuntimeError, "Failed to add `%s\", duplicate id `%d\"?", string, id);
-  }
 
   kwt_data->last_id= id + 1;
   kwt_data->dictionary_size++;
-  // printf("[internal]==> %d\n", id);
-  return id;
+  return INT2FIX(id);
 }
 
 /*
@@ -293,14 +282,14 @@ rb_kwt_new_from_file(int argc, VALUE *argv, VALUE klass)
   KeywordTree( self, kwt_data );
 
   dictionary = fopen( RSTRING( f_string )->ptr, "r" );
-  if(dictionary == NULL) {
+  if(dictionary == NULL)
     rb_raise(rb_eRuntimeError, "Cannot open `%s\". No such file?", RSTRING(f_string)->ptr);
-  }
 
   while(fgets(word, 1024, dictionary) != NULL) {
     ac_add_string(kwt_data->tree, word, strlen(word)-1, id++);
     kwt_data->dictionary_size++;
   }
+
   kwt_data->last_id= id+1;
   fclose(dictionary);
   return self;
@@ -331,17 +320,16 @@ void Init_ahocorasick() {
   rb_define_method(rb_cKeywordTree, "size", rb_kwt_size, 0);
   rb_define_method(rb_cKeywordTree, "make", rb_kwt_make, 0);
   rb_define_method(rb_cKeywordTree, "add_string", rb_kwt_add_string, -1);
+  rb_define_alias(rb_cKeywordTree, "<<", "add_string");
 
-  rb_define_method(rb_cKeywordTree, "search", rb_kwt_search, -1);
-  rb_define_alias(rb_cKeywordTree, "find_all", "search");
+  rb_define_method(rb_cKeywordTree, "find_all", rb_kwt_find_all, -1);
+  rb_define_alias(rb_cKeywordTree, "search", "find_all");
 
-  rb_define_alias(rb_cKeywordTree, "<<", "add_string");
   rb_define_singleton_method(rb_cKeywordTree, "from_file", rb_kwt_new_from_file, -1);
 
-  sym_id= ID2SYM(rb_intern("id"));
-  sym_value= ID2SYM(rb_intern("value"));
-  sym_ends_at= ID2SYM( rb_intern("ends_at") );
+  sym_id       = ID2SYM(rb_intern("id"));
+  sym_value    = ID2SYM(rb_intern("value"));
+  sym_ends_at  = ID2SYM( rb_intern("ends_at") );
   sym_starts_at= ID2SYM( rb_intern("starts_at") );
-
 }
 

data/spec/ahocorasick_spec.rb

@@ -67,7 +67,7 @@ describe KeywordTree do
       #   01234567890123456789023
       #                 |       |
       q= "data moved to bucurești"
-      @kwt.search(q).each do | result |
+      @kwt.find_all(q).each do | result |
         result[:starts_at].should == 14
         result[:ends_at].should   == 24
       end
@@ -77,7 +77,7 @@ describe KeywordTree do
       @kwt << "expected"
       #    012345678901234578901234567890
       q = "moved to bucurești as expected"
-      @kwt.search(q).each do | r |
+      @kwt.find_all(q).each do | r |
         r[:starts_at].should == 23
         r[:ends_at].should   == q.size
         (r[:ends_at]-r[:starts_at]).should == r[:value].size
@@ -86,15 +86,15 @@ describe KeywordTree do
     
     it "even more unicode" do
       @kwt << "șșt"
-      #                    0124789
-      result= @kwt.search("mușștar").first
+      #                      0124789
+      result= @kwt.find_all("mușștar").first
       result[:starts_at].should == 2
       result[:ends_at].should == result[:starts_at] + "șșt".size
     end
 
     it "checks for result length" do
       @kwt << "foo"
-      result= @kwt.search("foo").first
+      result= @kwt.find_all("foo").first
       #          4                 0
       (result[:ends_at]-result[:starts_at]).should == result[:value].size
       "foo"[result[:ends_at]].should == nil
@@ -110,7 +110,7 @@ describe KeywordTree do
     end
 
     it "should match on context" do
-      @kwt.search("I've moved the data to a new database").size.should == 4
+      @kwt.find_all("I've moved the data to a new database").size.should == 4
     end
 
   end
@@ -146,8 +146,8 @@ describe KeywordTree do
     it "should work to add a random id" do
       kwt= KeywordTree.new
       kwt << "baz"
-      kwt.add_string "foo", 1990
-      kwt << "bar"
+      kwt.add_string("foo", 1990).should == 1990
+      kwt.add_string("bar").should == 1991
       kwt.size.should == 3
     end
 
@@ -185,7 +185,7 @@ describe KeywordTree do
       start= Time.now
       k= KeywordTree.from_file File.dirname(__FILE__) + "/data/en.words"
       load_time= Time.now
-      results= k.search( File.read( File.dirname(__FILE__) + "/data/melville-moby_dick.txt" ) )
+      results= k.find_all( File.read( File.dirname(__FILE__) + "/data/melville-moby_dick.txt" ) )
       puts "\n%d words re-loaded in %s seconds.\nGot %d results in %s seconds" % [k.size, (load_time - start), results.size, (Time.now-load_time)]
       (Time.now-load_time).should < 1.2
     end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: aurelian-ruby-ahocorasick
 version: !ruby/object:Gem::Version 
-  version: 0.3.1
+  version: 0.3.3
 platform: ruby
 authors: 
 - Aurelian Oancea
@@ -14,7 +14,7 @@ default_executable:
 dependencies: []
 
 description: Expose Aho-Corasick implementation from Strmat to Ruby.
-email: aurelian [ at ] locknet.ro
+email: oancea at gmail dot com
 executables: []
 
 extensions: