X-Git-Url: https://git.njae.me.uk/?a=blobdiff_plain;f=test%2Ftc_porter2_parts.rb;fp=test%2Ftc_porter2_parts.rb;h=824e2dd9de2ea3028d1fcb1d5456247e9aeabbf2;hb=a222cc928502d3bb6ecd3d03c532ce7f9804acb4;hp=b38016a78ee011d1f76712bc62332fe43cbfeb2b;hpb=9dd2505747fbbd6040e9516dd0ae2e8ec5c6a4d2;p=porter2stemmer.git diff --git a/test/tc_porter2_parts.rb b/test/tc_porter2_parts.rb index b38016a..824e2dd 100644 --- a/test/tc_porter2_parts.rb +++ b/test/tc_porter2_parts.rb @@ -3,16 +3,11 @@ # # This file tests each stage of the stemmer individually. - $:.unshift File.join(File.dirname(__FILE__), "..", "lib") require 'test/unit' require 'porter2' -#class String -# public :porter2_preprocess, :porter2_r1, :porter2_r2 -#end - class TestPorter2 < Test::Unit::TestCase def test_tidy @@ -36,10 +31,11 @@ class TestPorter2 < Test::Unit::TestCase assert_equal "abacdeY", "abacdey".porter2_preprocess assert_equal "abaYde", "abayde".porter2_preprocess assert_equal "kabaYde", "kabayde".porter2_preprocess + assert_equal "kabyaYde", "kabyayde".porter2_preprocess assert_equal "'", "'''".porter2_preprocess end - def test_find_R1 + def test_R1 assert_equal "iful", "beautiful".porter2_r1 assert_equal "y", "beauty".porter2_r1 assert_equal "", "beau".porter2_r1 @@ -66,6 +62,15 @@ class TestPorter2 < Test::Unit::TestCase assert_equal "al", "arsenal".porter2_r1 end + def test_R2 + assert_equal "ul", "beautiful".porter2_r2 + assert_equal "", "beauty".porter2_r2 + assert_equal "", "beau".porter2_r2 + assert_equal "adversion", "animadversion".porter2_r2 + assert_equal "", "sprinkled".porter2_r2 + assert_equal "ist", "eucharist".porter2_r2 + end + def test_ends_with_short_syllable? assert_equal true, "rap".porter2_ends_with_short_syllable? assert_equal true, "trap".porter2_ends_with_short_syllable? @@ -93,54 +98,45 @@ class TestPorter2 < Test::Unit::TestCase end end - def test_find_R2 - assert_equal "ul", "beautiful".porter2_r2 - assert_equal "", "beauty".porter2_r2 - assert_equal "", "beau".porter2_r2 - assert_equal "adversion", "animadversion".porter2_r2 - assert_equal "", "sprinkled".porter2_r2 - assert_equal "ist", "eucharist".porter2_r2 - end - def test_step_0 - assert_equal "abac", "abac".step_0 - assert_equal "abac", "abac'".step_0 - assert_equal "abac", "abac's".step_0 - assert_equal "abac", "abac's'".step_0 - assert_equal "ab'c", "ab'c".step_0 - assert_equal "ab'sc", "ab'sc".step_0 - assert_equal "ab's'c", "ab's'c".step_0 - assert_equal "ab'sc", "ab'sc's".step_0 - assert_equal "'", "'".step_0 - assert_equal "'s", "'s".step_0 - assert_equal "'s", "'s'".step_0 + assert_equal "abac", "abac".porter2_step0 + assert_equal "abac", "abac'".porter2_step0 + assert_equal "abac", "abac's".porter2_step0 + assert_equal "abac", "abac's'".porter2_step0 + assert_equal "ab'c", "ab'c".porter2_step0 + assert_equal "ab'sc", "ab'sc".porter2_step0 + assert_equal "ab's'c", "ab's'c".porter2_step0 + assert_equal "ab'sc", "ab'sc's".porter2_step0 + assert_equal "'", "'".porter2_step0 + assert_equal "'s", "'s".porter2_step0 + assert_equal "'s", "'s'".porter2_step0 end def test_step_1a - assert_equal "abacde", "abacde".step_1a - assert_equal "abacess", "abacesses".step_1a - assert_equal "tie", "ties".step_1a - assert_equal "tie", "tied".step_1a - assert_equal "cri", "cries".step_1a - assert_equal "cri", "cried".step_1a - assert_equal "gas", "gas".step_1a - assert_equal "this", "this".step_1a - assert_equal "gap", "gaps".step_1a - assert_equal "kiwi", "kiwis".step_1a - assert_equal "abacus", "abacus".step_1a - assert_equal "abacess", "abacess".step_1a + assert_equal "abacde", "abacde".porter2_step1a + assert_equal "abacess", "abacesses".porter2_step1a + assert_equal "tie", "ties".porter2_step1a + assert_equal "tie", "tied".porter2_step1a + assert_equal "cri", "cries".porter2_step1a + assert_equal "cri", "cried".porter2_step1a + assert_equal "gas", "gas".porter2_step1a + assert_equal "this", "this".porter2_step1a + assert_equal "gap", "gaps".porter2_step1a + assert_equal "kiwi", "kiwis".porter2_step1a + assert_equal "abacus", "abacus".porter2_step1a + assert_equal "abacess", "abacess".porter2_step1a end def test_step_1b - assert_equal "abacde", "abacde".step_1b + assert_equal "abacde", "abacde".porter2_step1b words_non_gb = {"luxuriated" => "luxuriate", "luxuriating" => "luxuriate", "hopping" => "hop", "hopped" => "hop", "hoped" => "hope", "hoping" => "hope", "atomized" => "atomize", "atomised" => "atomis", "addicted" => "addict", "bleed" => "bleed" } words_non_gb.each do |original, stemmed| - assert_equal stemmed, original.step_1b, - "#{original} should have stemmed to #{stemmed} but got #{original.step_1b(original.porter2_r1)} instead" + assert_equal stemmed, original.porter2_step1b, + "#{original} should have stemmed to #{stemmed} but got #{original.porter2_step1b(original.porter2_r1)} instead" end words_gb = {"luxuriated" => "luxuriate", "luxuriating" => "luxuriate", "hopping" => "hop", "hopped" => "hop", @@ -148,156 +144,156 @@ class TestPorter2 < Test::Unit::TestCase "atomized" => "atomize", "atomised" => "atomise", "addicted" => "addict", "bleed" => "bleed" } words_gb.each do |original, stemmed| - assert_equal stemmed, original.step_1b(true), - "#{original} should have stemmed to #{stemmed} but got #{original.step_1b(original.porter2_r1)} instead" + assert_equal stemmed, original.porter2_step1b(true), + "#{original} should have stemmed to #{stemmed} but got #{original.porter2_step1b(original.porter2_r1)} instead" end end def test_step_1c - assert_equal "cri", "cry".step_1c - assert_equal "by", "by".step_1c - assert_equal "saY", "saY".step_1c - assert_equal "abbeY", "abbeY".step_1c + assert_equal "cri", "cry".porter2_step1c + assert_equal "by", "by".porter2_step1c + assert_equal "saY", "saY".porter2_step1c + assert_equal "abbeY", "abbeY".porter2_step1c end def test_step_2 - assert_equal "abac", "abac".step_2 + assert_equal "abac", "abac".porter2_step2 - assert_equal "nationalize", "nationalization".step_2 - assert_equal "nationalisate", "nationalisation".step_2 - assert_equal "nationalize", "nationalization".step_2(true) - assert_equal "nationalise", "nationalisation".step_2(true) + assert_equal "nationalize", "nationalization".porter2_step2 + assert_equal "nationalisate", "nationalisation".porter2_step2 + assert_equal "nationalize", "nationalization".porter2_step2(true) + assert_equal "nationalise", "nationalisation".porter2_step2(true) # Repeat the steps to ensure that the english-gb behaviour isn't sticky - assert_equal "nationalize", "nationalization".step_2(false) - assert_equal "nationalisate", "nationalisation".step_2(false) - assert_equal "nationalize", "nationalization".step_2 - assert_equal "nationalisate", "nationalisation".step_2 + assert_equal "nationalize", "nationalization".porter2_step2(false) + assert_equal "nationalisate", "nationalisation".porter2_step2(false) + assert_equal "nationalize", "nationalization".porter2_step2 + assert_equal "nationalisate", "nationalisation".porter2_step2 - assert_equal "nationalize", "nationalizer".step_2 - assert_equal "nationaliser", "nationaliser".step_2 - assert_equal "nationalize", "nationalizer".step_2(true) - assert_equal "nationalise", "nationaliser".step_2(true) + assert_equal "nationalize", "nationalizer".porter2_step2 + assert_equal "nationaliser", "nationaliser".porter2_step2 + assert_equal "nationalize", "nationalizer".porter2_step2(true) + assert_equal "nationalise", "nationaliser".porter2_step2(true) - assert_equal "abaction", "abactional".step_2 - assert_equal "abacence", "abacenci".step_2 - assert_equal "abacance", "abacanci".step_2 - assert_equal "abacable", "abacabli".step_2 - assert_equal "abacent", "abacentli".step_2 - assert_equal "abacize", "abacizer".step_2 - assert_equal "abacize", "abacization".step_2 - assert_equal "abacate", "abacational".step_2 - assert_equal "abacate", "abacation".step_2 - assert_equal "abacate", "abacator".step_2 - assert_equal "abacal", "abacalism".step_2 - assert_equal "abacal", "abacaliti".step_2 - assert_equal "abacal", "abacalli".step_2 - assert_equal "abacful", "abacfulness".step_2 - assert_equal "abacous", "abacousli".step_2 - assert_equal "abacous", "abacousness".step_2 - assert_equal "abacive", "abaciveness".step_2 - assert_equal "abacive", "abaciviti".step_2 - assert_equal "abiliti", "abiliti".step_2 - assert_equal "abacble", "abacbiliti".step_2 - assert_equal "abacble", "abacbli".step_2 - assert_equal "abacful", "abacfulli".step_2 - assert_equal "abacless", "abaclessli".step_2 - assert_equal "abaclog", "abaclogi".step_2 + assert_equal "abaction", "abactional".porter2_step2 + assert_equal "abacence", "abacenci".porter2_step2 + assert_equal "abacance", "abacanci".porter2_step2 + assert_equal "abacable", "abacabli".porter2_step2 + assert_equal "abacent", "abacentli".porter2_step2 + assert_equal "abacize", "abacizer".porter2_step2 + assert_equal "abacize", "abacization".porter2_step2 + assert_equal "abacate", "abacational".porter2_step2 + assert_equal "abacate", "abacation".porter2_step2 + assert_equal "abacate", "abacator".porter2_step2 + assert_equal "abacal", "abacalism".porter2_step2 + assert_equal "abacal", "abacaliti".porter2_step2 + assert_equal "abacal", "abacalli".porter2_step2 + assert_equal "abacful", "abacfulness".porter2_step2 + assert_equal "abacous", "abacousli".porter2_step2 + assert_equal "abacous", "abacousness".porter2_step2 + assert_equal "abacive", "abaciveness".porter2_step2 + assert_equal "abacive", "abaciviti".porter2_step2 + assert_equal "abiliti", "abiliti".porter2_step2 + assert_equal "abacble", "abacbiliti".porter2_step2 + assert_equal "abacble", "abacbli".porter2_step2 + assert_equal "abacful", "abacfulli".porter2_step2 + assert_equal "abacless", "abaclessli".porter2_step2 + assert_equal "abaclog", "abaclogi".porter2_step2 - assert_equal "abac", "abacli".step_2 - assert_equal "abd", "abdli".step_2 - assert_equal "abe", "abeli".step_2 - assert_equal "abg", "abgli".step_2 - assert_equal "abh", "abhli".step_2 - assert_equal "abk", "abkli".step_2 - assert_equal "abm", "abmli".step_2 - assert_equal "abn", "abnli".step_2 - assert_equal "abr", "abrli".step_2 - assert_equal "abt", "abtli".step_2 - assert_equal "abali", "abali".step_2 + assert_equal "abac", "abacli".porter2_step2 + assert_equal "abd", "abdli".porter2_step2 + assert_equal "abe", "abeli".porter2_step2 + assert_equal "abg", "abgli".porter2_step2 + assert_equal "abh", "abhli".porter2_step2 + assert_equal "abk", "abkli".porter2_step2 + assert_equal "abm", "abmli".porter2_step2 + assert_equal "abn", "abnli".porter2_step2 + assert_equal "abr", "abrli".porter2_step2 + assert_equal "abt", "abtli".porter2_step2 + assert_equal "abali", "abali".porter2_step2 - assert_equal "bad", "badli".step_2 - assert_equal "fluentli", "fluentli".step_2 - assert_equal "geolog", "geologi".step_2 + assert_equal "bad", "badli".porter2_step2 + assert_equal "fluentli", "fluentli".porter2_step2 + assert_equal "geolog", "geologi".porter2_step2 end def test_step_3 - assert_equal "abac", "abac".step_3("") + assert_equal "abac", "abac".porter2_step3("") - assert_equal "national", "nationalize".step_3 - assert_equal "nationalise", "nationalise".step_3 - assert_equal "national", "nationalise".step_3(true) + assert_equal "national", "nationalize".porter2_step3 + assert_equal "nationalise", "nationalise".porter2_step3 + assert_equal "national", "nationalise".porter2_step3(true) # Repeat the steps to ensure that the english-gb behaviour isn't sticky - assert_equal "national", "nationalize".step_3(false) - assert_equal "nationalise", "nationalise".step_3(false) - assert_equal "national", "nationalize".step_3 - assert_equal "nationalise", "nationalise".step_3 + assert_equal "national", "nationalize".porter2_step3(false) + assert_equal "nationalise", "nationalise".porter2_step3(false) + assert_equal "national", "nationalize".porter2_step3 + assert_equal "nationalise", "nationalise".porter2_step3 - assert_equal "abaction", "abactional".step_3 - assert_equal "abacate", "abacational".step_3 - assert_equal "abacic", "abacicate".step_3 - assert_equal "abacic", "abaciciti".step_3 - assert_equal "abacic", "abacical".step_3 - assert_equal "abac", "abacful".step_3 - assert_equal "abac", "abacness".step_3 + assert_equal "abaction", "abactional".porter2_step3 + assert_equal "abacate", "abacational".porter2_step3 + assert_equal "abacic", "abacicate".porter2_step3 + assert_equal "abacic", "abaciciti".porter2_step3 + assert_equal "abacic", "abacical".porter2_step3 + assert_equal "abac", "abacful".porter2_step3 + assert_equal "abac", "abacness".porter2_step3 - assert_equal "abacabac", "abacabacative".step_3 - assert_equal "abacabac", "abacabacative".step_3 + assert_equal "abacabac", "abacabacative".porter2_step3 + assert_equal "abacabac", "abacabacative".porter2_step3 - assert_equal "dryness", "dryness".step_3 + assert_equal "dryness", "dryness".porter2_step3 end def test_step_4 - assert_equal "abac", "abac".step_4("") + assert_equal "abac", "abac".porter2_step4("") - assert_equal "nation", "nationize".step_4 - assert_equal "nationise", "nationise".step_4 - assert_equal "nation", "nationize".step_4(true) - assert_equal "nation", "nationise".step_4(true) - assert_equal "nation", "nationize".step_4(false) - assert_equal "nationise", "nationise".step_4(false) - assert_equal "nation", "nationize".step_4() - assert_equal "nationise", "nationise".step_4() + assert_equal "nation", "nationize".porter2_step4 + assert_equal "nationise", "nationise".porter2_step4 + assert_equal "nation", "nationize".porter2_step4(true) + assert_equal "nation", "nationise".porter2_step4(true) + assert_equal "nation", "nationize".porter2_step4(false) + assert_equal "nationise", "nationise".porter2_step4(false) + assert_equal "nation", "nationize".porter2_step4() + assert_equal "nationise", "nationise".porter2_step4() - assert_equal "abac", "abacal".step_4 - assert_equal "abac", "abacance".step_4 - assert_equal "abac", "abacence".step_4 - assert_equal "abac", "abacer".step_4 - assert_equal "abac", "abacic".step_4 - assert_equal "abacer", "abacerable".step_4 - assert_equal "abac", "abacible".step_4 - assert_equal "abac", "abacant".step_4 - assert_equal "abac", "abacement".step_4 # Check we handle overlapping suffixes properly - assert_equal "abacac", "abacacement".step_4 - assert_equal "abacac", "abacacment".step_4 - assert_equal "abac", "abacment".step_4 - assert_equal "abac", "abacent".step_4 - assert_equal "abac", "abacism".step_4 - assert_equal "abac", "abacate".step_4 - assert_equal "abac", "abaciti".step_4 - assert_equal "abac", "abacous".step_4 - assert_equal "abac", "abacive".step_4 - assert_equal "abac", "abacize".step_4 - assert_equal "abacion", "abacion".step_4 - assert_equal "abacs", "abacsion".step_4 - assert_equal "abact", "abaction".step_4 - assert_equal "abction", "abction".step_4 - assert_equal "ablut", "ablution".step_4 - assert_equal "agreement", "agreement".step_4 + assert_equal "abac", "abacal".porter2_step4 + assert_equal "abac", "abacance".porter2_step4 + assert_equal "abac", "abacence".porter2_step4 + assert_equal "abac", "abacer".porter2_step4 + assert_equal "abac", "abacic".porter2_step4 + assert_equal "abacer", "abacerable".porter2_step4 + assert_equal "abac", "abacible".porter2_step4 + assert_equal "abac", "abacant".porter2_step4 + assert_equal "abac", "abacement".porter2_step4 # Check we handle overlapping suffixes properly + assert_equal "abacac", "abacacement".porter2_step4 + assert_equal "abacac", "abacacment".porter2_step4 + assert_equal "abac", "abacment".porter2_step4 + assert_equal "abac", "abacent".porter2_step4 + assert_equal "abac", "abacism".porter2_step4 + assert_equal "abac", "abacate".porter2_step4 + assert_equal "abac", "abaciti".porter2_step4 + assert_equal "abac", "abacous".porter2_step4 + assert_equal "abac", "abacive".porter2_step4 + assert_equal "abac", "abacize".porter2_step4 + assert_equal "abacion", "abacion".porter2_step4 + assert_equal "abacs", "abacsion".porter2_step4 + assert_equal "abact", "abaction".porter2_step4 + assert_equal "abction", "abction".porter2_step4 + assert_equal "ablut", "ablution".porter2_step4 + assert_equal "agreement", "agreement".porter2_step4 - assert_equal "abcal", "abcal".step_4 # No removal if suffix isn't in R2 + assert_equal "abcal", "abcal".porter2_step4 # No removal if suffix isn't in R2 end def test_step_5 - assert_equal "abac", "abac".step_5 + assert_equal "abac", "abac".porter2_step5 - assert_equal "abacl", "abacll".step_5 - assert_equal "abcll", "abcll".step_5 + assert_equal "abacl", "abacll".porter2_step5 + assert_equal "abcll", "abcll".porter2_step5 - assert_equal "abc", "abc".step_5 - assert_equal "abl", "able".step_5 - assert_equal "abe", "abe".step_5 - assert_equal "abac", "abace".step_5 - assert_equal "bawac", "bawace".step_5 + assert_equal "abc", "abc".porter2_step5 + assert_equal "abl", "able".porter2_step5 + assert_equal "abe", "abe".porter2_step5 + assert_equal "abac", "abace".porter2_step5 + assert_equal "bawac", "bawace".porter2_step5 end def test_porter2_postprocess