whistlepig 0.10 → 0.11

data/ext/whistlepig/whistlepig.c

@@ -447,6 +447,44 @@ static VALUE query_or(VALUE self, VALUE v_other) {
   return o_result;
 }
 
+/*
+ * call-seq: snippetize(field, string, max_num_results=10)
+ *
+ * Returns an array of [start, end] subarrays that mark the matching positions of
+ * the query within the string. 'field' determines which field the string is
+ * taken to be.
+ *
+ */
+static VALUE query_snippetize(int argc, VALUE* argv, VALUE self) {
+  VALUE v_field, v_string, v_max_num_results;
+  rb_scan_args(argc, argv, "21", &v_field, &v_string, &v_max_num_results);
+  Check_Type(v_string, T_STRING);
+
+  wp_query* query; Data_Get_Struct(self, wp_query, query);
+  uint32_t num_results, max_num_results;
+  if(NIL_P(v_max_num_results)) max_num_results = 10;
+  else max_num_results = NUM2INT(v_max_num_results);
+
+  uint32_t* start_offsets = malloc(sizeof(uint32_t) * max_num_results);
+  uint32_t* end_offsets = malloc(sizeof(uint32_t) * max_num_results);
+
+  wp_error* e = wp_snippetize_string(query, RSTRING_PTR(v_field), RSTRING_PTR(v_string), max_num_results, &num_results, start_offsets, end_offsets);
+  RAISE_IF_NECESSARY(e);
+
+  VALUE array = rb_ary_new2(num_results);
+  for(uint32_t i = 0; i < num_results; i++) {
+    VALUE subarray = rb_ary_new2(2);
+    rb_ary_store(subarray, 0, INT2NUM(start_offsets[i]));
+    rb_ary_store(subarray, 1, INT2NUM(end_offsets[i]));
+    rb_ary_store(array, i, subarray);
+  }
+
+  free(start_offsets);
+  free(end_offsets);
+
+  return array;
+}
+
 static VALUE query_init(VALUE self, VALUE query) {
   rb_iv_set(self, "@query", query);
   return self;
@@ -568,6 +606,7 @@ void Init_whistlepig() {
   rb_define_method(c_query, "to_s", query_to_s, 0);
   rb_define_method(c_query, "clone", query_clone, 0);
   rb_define_method(c_query, "term_map", query_map_terms, 0);
+  rb_define_method(c_query, "snippetize", query_snippetize, -1);
   rb_define_attr(c_query, "query", 1, 0);
 
   c_error = rb_define_class_under(m_whistlepig, "Error", rb_eStandardError);

data/ext/whistlepig/whistlepig.h

@@ -12,6 +12,8 @@
 #include "query-parser.h"
 #include "lock.h"
 #include "error.h"
+#include "rarray.h"
+#include "snippeter.h"
 
 // see comments in index.c
 char* strdup(const char* old);

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: whistlepig
 version: !ruby/object:Gem::Version
-  version: '0.10'
+  version: '0.11'
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,8 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-04-01 00:00:00.000000000 Z
+date: 2012-04-10 00:00:00.000000000 -07:00
+default_executable: 
 dependencies: []
 description: Whistlepig is a minimalist realtime full-text search index. Its goal
   is to be as small and minimally-featured as possible, while still remaining useful,
@@ -27,40 +28,45 @@ files:
 - COPYING
 - ext/whistlepig/extconf.rb
 - lib/whistlepig.rb
+- ext/whistlepig/query-parser.lex.h
+- ext/whistlepig/entry.h
+- ext/whistlepig/lock.c
+- ext/whistlepig/stringmap.c
+- ext/whistlepig/tokenizer.lex.h
+- ext/whistlepig/whistlepig.h
 - ext/whistlepig/error.c
+- ext/whistlepig/snippeter.h
 - ext/whistlepig/lock.h
-- ext/whistlepig/tokenizer.lex.c
-- ext/whistlepig/query-parser.tab.c
-- ext/whistlepig/defaults.h
+- ext/whistlepig/extconf.h
 - ext/whistlepig/stringmap.h
-- ext/whistlepig/query.c
-- ext/whistlepig/stringmap.c
-- ext/whistlepig/query.h
-- ext/whistlepig/segment.h
-- ext/whistlepig/entry.c
-- ext/whistlepig/search.c
-- ext/whistlepig/mmap-obj.c
-- ext/whistlepig/query-parser.tab.h
 - ext/whistlepig/query-parser.lex.c
-- ext/whistlepig/query-parser.c
+- ext/whistlepig/defaults.h
+- ext/whistlepig/tokenizer.lex.c
+- ext/whistlepig/termhash.c
+- ext/whistlepig/query-parser.h
+- ext/whistlepig/index.c
 - ext/whistlepig/stringpool.c
+- ext/whistlepig/query.h
+- ext/whistlepig/query-parser.c
 - ext/whistlepig/stringpool.h
-- ext/whistlepig/whistlepig.h
+- ext/whistlepig/mmap-obj.c
+- ext/whistlepig/whistlepig.c
+- ext/whistlepig/search.c
+- ext/whistlepig/snippeter.c
+- ext/whistlepig/termhash.h
+- ext/whistlepig/query.c
+- ext/whistlepig/query-parser.tab.h
 - ext/whistlepig/khash.h
-- ext/whistlepig/index.c
-- ext/whistlepig/search.h
-- ext/whistlepig/query-parser.lex.h
-- ext/whistlepig/mmap-obj.h
-- ext/whistlepig/tokenizer.lex.h
+- ext/whistlepig/query-parser.tab.c
+- ext/whistlepig/entry.c
 - ext/whistlepig/index.h
-- ext/whistlepig/entry.h
-- ext/whistlepig/termhash.h
-- ext/whistlepig/error.h
-- ext/whistlepig/whistlepig.c
-- ext/whistlepig/termhash.c
+- ext/whistlepig/segment.h
+- ext/whistlepig/mmap-obj.h
+- ext/whistlepig/rarray.h
 - ext/whistlepig/segment.c
-- ext/whistlepig/query-parser.h
-- ext/whistlepig/lock.c
+- ext/whistlepig/search.h
+- ext/whistlepig/error.h
+has_rdoc: true
 homepage: http://masanjin.net/whistlepig
 licenses: []
 post_install_message: 
@@ -87,7 +93,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 1.8.11
+rubygems_version: 1.6.0
 signing_key: 
 specification_version: 3
 summary: a minimalist realtime full-text search index