xapian-bindings/ruby/smoketest.rb

   1 #!/usr/bin/ruby -w
   2 #
   3 # smoketest.rb - test Xapian bindings for Ruby
   4 # Original version by Paul Legato (plegato@nks.net), 4/17/2006
   5 #
   6 # Originally based on smoketest.php from the PHP4 bindings.
   7 #
   8 # Copyright (C) 2006 Networked Knowledge Systems, Inc.
   9 # Copyright (C) 2008,2009,2010,2011,2016 Olly Betts
  10 # Copyright (C) 2010 Richard Boulton
  11 #
  12 # This program is free software; you can redistribute it and/or
  13 # modify it under the terms of the GNU General Public License as
  14 # published by the Free Software Foundation; either version 2 of the
  15 # License, or (at your option) any later version.
  16 #
  17 # This program is distributed in the hope that it will be useful,
  18 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20 # GNU General Public License for more details.
  21 #
  22 # You should have received a copy of the GNU General Public License
  23 # along with this program; if not, write to the Free Software
  24 # Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301
  25 # USA
  26
  27 require 'test/unit'
  28 require 'tmpdir'
  29 require 'xapian'
  30
  31 class TestMatchDecider < Xapian::MatchDecider
  32   def __call__(doc)
  33     return doc.value(0) == "yes"
  34   end
  35 end
  36
  37 class XapianSmoketest < Test::Unit::TestCase
  38
  39   def setup
  40     @stem = Xapian::Stem.new("english")
  41
  42     @doc = Xapian::Document.new()
  43     @doc.data = "is there anybody out there?"
  44     @doc.add_posting(@stem.call("is"), 1)
  45     @doc.add_posting(@stem.call("there"), 2)
  46     @doc.add_posting(@stem.call("anybody"), 3)
  47     @doc.add_posting(@stem.call("out"), 4)
  48     @doc.add_posting(@stem.call("there"), 5)
  49     @doc.add_term("XYzzy")
  50
  51     @db = Xapian::inmemory_open()
  52     @db.add_document(@doc)
  53
  54     @enq = Xapian::Enquire.new(@db)
  55   end # setup
  56
  57   def test_version
  58     # Test the version number reporting functions give plausible results.
  59     @v = sprintf("%d.%d.%d", Xapian::major_version(), Xapian::minor_version(),
  60                  Xapian::revision())
  61     @v2 = Xapian::version_string()
  62     assert_equal(@v2, @v)
  63   end # test_version
  64
  65   def test_stem
  66     assert_equal("Xapian::Stem(english)", @stem.description())
  67
  68     assert_equal("is", @stem.call("is"))
  69     assert_equal("go", @stem.call("going"))
  70     assert_equal("want", @stem.call("wanted"))
  71     assert_equal("refer", @stem.call("reference"))
  72   end # test_stem
  73
  74   # subtests are those on which some test_foo() method depends.
  75   def test_000_document
  76     assert_not_nil(@doc)
  77
  78     assert_equal("is there anybody out there?", @doc.data())
  79
  80     assert_equal(@doc.termlist_count(), 5)
  81     assert_equal("XYzzy", @doc.terms().first.term)
  82
  83     @doc.add_term("foo")
  84     assert_equal(6, @doc.termlist_count())
  85     assert_equal(@doc.terms.size(), @doc.termlist_count())
  86
  87   end # test_document
  88
  89   def test_001_database
  90     assert_not_nil(@db)
  91     assert_equal("WritableDatabase()", @db.description())
  92     assert_equal(1, @db.doccount())
  93   end # test_database
  94
  95   def test_002_queries
  96     assert_equal("Query((smoke OR test OR terms))",
  97                  Xapian::Query.new(Xapian::Query::OP_OR, ["smoke", "test", "terms"]).description())
  98
  99     phrase_query = Xapian::Query.new(Xapian::Query::OP_PHRASE, ["smoke", "test", "tuple"])
 100     xor_query = Xapian::Query.new(Xapian::Query::OP_XOR, [ Xapian::Query.new("smoke"), phrase_query, "string" ])
 101
 102     assert_equal("Query((smoke PHRASE 3 test PHRASE 3 tuple))", phrase_query.description())
 103     assert_equal("Query((smoke XOR (smoke PHRASE 3 test PHRASE 3 tuple) XOR string))", xor_query.description())
 104
 105     assert_equal([Xapian::Term.new("smoke", 1),
 106                   Xapian::Term.new("string", 1),
 107                   Xapian::Term.new("test", 1),
 108                   Xapian::Term.new("tuple", 1)], xor_query.terms())
 109
 110     assert_equal(Xapian::Query::OP_ELITE_SET, 10)
 111
 112     assert_equal("Query(<alldocuments>)", Xapian::Query::MatchAll.description())
 113     assert_equal("Query()", Xapian::Query::MatchNothing.description())
 114   end # test_queries
 115
 116   def test_003_enquire
 117     @enq = Xapian::Enquire.new(@db)
 118     assert_not_nil(@enq)
 119
 120     @enq.query = Xapian::Query.new(Xapian::Query::OP_OR, "there", "is")
 121     mset = @enq.mset(0, 10)
 122
 123     assert_equal(1, mset.size())
 124
 125     # Feature test for Enquire.matching_terms()
 126     assert_equal(2, @enq.matching_terms(mset.hit(0)).size())
 127     assert_equal([Xapian::Term.new("is", 1), Xapian::Term.new("there", 1)],
 128                  @enq.matching_terms(mset.hit(0)))
 129   end # test_enquire
 130
 131   def test_004_mset_iterator
 132     @enq = Xapian::Enquire.new(@db)
 133     assert_not_nil(@enq)
 134
 135     @enq.query = Xapian::Query.new(Xapian::Query::OP_OR, "there", "is")
 136     mset = @enq.mset(0, 10)
 137
 138     assert_equal(mset.matches().size(), mset.size())
 139   end
 140
 141
 142   def test_005_eset_iterator
 143     rset = Xapian::RSet.new
 144
 145     rset.add_document(1)
 146
 147     @enq = Xapian::Enquire.new(@db)
 148     @enq.query = Xapian::Query.new(Xapian::Query::OP_OR, "there", "is")
 149
 150     eset = @enq.eset(10, rset)
 151     assert_not_nil(eset)
 152
 153     assert_equal(3, eset.terms.size())
 154   end # test_eset_iter
 155
 156   # Feature test for Database.allterms
 157   def test_006_database_allterms
 158     assert_equal(5, @db.allterms.size())
 159     ou_terms = @db.allterms('ou')
 160     assert_equal(1, ou_terms.size())
 161     assert_equal('out', ou_terms[0].term)
 162   end
 163
 164   # Feature test for Database.postlist
 165   def test_007_database_postlist
 166     assert_equal(1, @db.postlist("there").size())
 167   end
 168
 169   # Feature test for Database.termlist
 170   def test_008_database_termlist
 171     assert_equal(5, @db.termlist(1).size())
 172   end
 173
 174   # Feature test for Database.positionlist
 175   def test_009_database_positionlist
 176     assert_equal(2, @db.positionlist(1, "there").size())
 177   end
 178
 179   # Feature test for Document.values
 180   def test_010_document_values
 181     assert_equal(0, @doc.values().size())
 182   end
 183
 184   def test_011_matchdecider
 185     @doc = Xapian::Document.new()
 186     @doc.data = "Two"
 187     @doc.add_posting(@stem.call("out"), 1)
 188     @doc.add_posting(@stem.call("source"), 2)
 189     @doc.add_value(0, "yes")
 190     @db.add_document(@doc)
 191
 192     @query = Xapian::Query.new(@stem.call("out"))
 193     enquire = Xapian::Enquire.new(@db)
 194     enquire.query = @query
 195     mset = enquire.mset(0, 10, nil, TestMatchDecider.new)
 196     assert_equal(mset.size(), 1)
 197     assert_equal(mset.docid(0), 2)
 198   end
 199
 200   def test_012_metadata
 201     assert_equal(@db.get_metadata('Foo'), '')
 202     @db.set_metadata('Foo', 'Foo')
 203     assert_equal(@db.get_metadata('Foo'), 'Foo')
 204   end
 205
 206   def test_013_scaleweight
 207     query = Xapian::Query.new("foo")
 208     query2 = Xapian::Query.new(Xapian::Query::OP_SCALE_WEIGHT, query, 5);
 209     assert_equal(query2.description(), "Query(5 * foo)")
 210   end
 211
 212   def test_014_sortable_serialise
 213     # In Xapian 1.0.13/1.1.1 and earlier, the SWIG generated wrapper code
 214     # didn't handle integer values > MAXINT for double parameters.
 215     v = 51767811298
 216     assert_equal(v, Xapian::sortable_unserialise(Xapian::sortable_serialise(v)))
 217   end
 218
 219   def test_015_valuecount_matchspy
 220     spy = Xapian::ValueCountMatchSpy.new(0)
 221     doc = Xapian::Document.new()
 222     doc.add_posting("term", 1)
 223     doc.add_value(0, "yes")
 224     @db.add_document(doc)
 225     @db.add_document(doc)
 226     doc.add_value(0, "maybe")
 227     @db.add_document(doc)
 228     doc.add_value(0, "no")
 229     @db.add_document(doc)
 230     query = Xapian::Query.new("term")
 231     enquire = Xapian::Enquire.new(@db)
 232     enquire.query = query
 233     enquire.add_matchspy(spy)
 234     mset = enquire.mset(0, 10)
 235     assert_equal(mset.size(), 4)
 236     assert_equal(spy.values.map{|i| "%s:%d"%[i.term, i.termfreq]} * ",",
 237                  "maybe:1,no:1,yes:2")
 238     assert_equal(spy.top_values(1).map{|i| "%s:%d"%[i.term, i.termfreq]} * ",",
 239                  "yes:2")
 240     assert_equal(spy.top_values(2).map{|i| "%s:%d"%[i.term, i.termfreq]} * ",",
 241                  "yes:2,maybe:1")
 242     assert_equal(spy.top_values(3).map{|i| "%s:%d"%[i.term, i.termfreq]} * ",",
 243                  "yes:2,maybe:1,no:1")
 244
 245     # Test the valuestream iterator, while we've got some data
 246     assert_equal(@db.valuestream(1).size(), 0)
 247     assert_equal(@db.valuestream(0).map{|i| "%d:%s"%[i.docid, i.value]}*",",
 248                  "2:yes,3:yes,4:maybe,5:no")
 249   end
 250
 251   def test_016_compactor
 252     Dir.mktmpdir("smokerb") {|tmpdir|
 253         db1path = "#{tmpdir}db1"
 254         db2path = "#{tmpdir}db2"
 255         db3path = "#{tmpdir}db3"
 256
 257         # Set up a couple of sample input databases
 258         db1 = Xapian::WritableDatabase.new(db1path, Xapian::DB_CREATE_OR_OVERWRITE)
 259         doc1 = Xapian::Document.new()
 260         doc1.add_term('Hello')
 261         doc1.add_term('Hello1')
 262         doc1.add_value(0, 'Val1')
 263         db1.set_metadata('key', '1')
 264         db1.set_metadata('key1', '1')
 265         db1.add_document(doc1)
 266         db1.commit()
 267
 268         db2 = Xapian::WritableDatabase.new(db2path, Xapian::DB_CREATE_OR_OVERWRITE)
 269         doc2 = Xapian::Document.new()
 270         doc2.add_term('Hello')
 271         doc2.add_term('Hello2')
 272         doc2.add_value(0, 'Val2')
 273         db2.set_metadata('key', '2')
 274         db2.set_metadata('key2', '2')
 275         db2.add_document(doc2)
 276         db2.commit()
 277
 278         # Compact with the default compactor
 279         # Metadata conflicts are resolved by picking the first value
 280         c = Xapian::Compactor.new()
 281         c.add_source(db1path)
 282         c.add_source(db2path)
 283         c.set_destdir(db3path)
 284         c.compact()
 285
 286         db3 = Xapian::Database.new(db3path)
 287         #assert_equal([(item.term, item.termfreq) for item in db3.allterms()],
 288         #       [('Hello', 2), ('Hello1', 1), ('Hello2', 1)])
 289         assert_equal(db3.document(1).value(0), 'Val1')
 290         assert_equal(db3.document(2).value(0), 'Val2')
 291         assert_equal(db3.get_metadata('key'), '1')
 292         assert_equal(db3.get_metadata('key1'), '1')
 293         assert_equal(db3.get_metadata('key2'), '2')
 294     }
 295   end
 296
 297   def test_016_latlongcoords_iterator
 298     coords = Xapian::LatLongCoords.new()
 299     coords.append(Xapian::LatLongCoord.new(0, 0))
 300     assert_equal(coords.size(), 1)
 301     assert_equal(coords.all.map{|i| "%s"%i.description}*",",
 302                  "Xapian::LatLongCoord(0, 0)")
 303   end
 304
 305 end # class XapianSmoketest