basis/porter-stemmer/porter-stemmer-tests.factor

   1 USING: arrays assocs io kernel porter-stemmer sequences
   2 tools.test io.files io.encodings.utf8 ;
   3
   4 { 0 } [ "xa" consonant-seq ] unit-test
   5 { 0 } [ "xxaa" consonant-seq ] unit-test
   6 { 1 } [ "xaxa" consonant-seq ] unit-test
   7 { 2 } [ "xaxaxa" consonant-seq ] unit-test
   8 { 3 } [ "xaxaxaxa" consonant-seq ] unit-test
   9 { 3 } [ "zzzzxaxaxaxaeee" consonant-seq ] unit-test
  10
  11 { t } [ 0 "fish" consonant? ] unit-test
  12 { f } [ 0 "and" consonant? ] unit-test
  13 { t } [ 0 "yes" consonant? ] unit-test
  14 { f } [ 1 "gym" consonant? ] unit-test
  15
  16 { t } [ 5 "splitting" double-consonant? ] unit-test
  17 { f } [ 2 "feel" double-consonant? ] unit-test
  18
  19 { f } [ "xxxz" stem-vowel? ] unit-test
  20 { t } [ "baobab" stem-vowel? ] unit-test
  21
  22 { t } [ "hop" cvc? ] unit-test
  23 { t } [ "cav" cvc? ] unit-test
  24 { t } [ "lov" cvc? ] unit-test
  25 { t } [ "crim" cvc? ] unit-test
  26 { f } [ "show" cvc? ] unit-test
  27 { f } [ "box" cvc? ] unit-test
  28 { f } [ "tray" cvc? ] unit-test
  29 { f } [ "meet" cvc? ] unit-test
  30
  31 { "caress" } [ "caresses" step1a step1b "" like ] unit-test
  32 { "poni" } [ "ponies" step1a step1b "" like ] unit-test
  33 { "ti" } [ "ties" step1a step1b "" like ] unit-test
  34 { "caress" } [ "caress" step1a step1b "" like ] unit-test
  35 { "cat" } [ "cats" step1a step1b "" like ] unit-test
  36 { "feed" } [ "feed" step1a step1b "" like ] unit-test
  37 { "agree" } [ "agreed" step1a step1b "" like ] unit-test
  38 { "disable" } [ "disabled" step1a step1b "" like ] unit-test
  39 { "mat" } [ "matting" step1a step1b "" like ] unit-test
  40 { "mate" } [ "mating" step1a step1b "" like ] unit-test
  41 { "meet" } [ "meeting" step1a step1b "" like ] unit-test
  42 { "mill" } [ "milling" step1a step1b "" like ] unit-test
  43 { "mess" } [ "messing" step1a step1b "" like ] unit-test
  44 { "meet" } [ "meetings" step1a step1b "" like ] unit-test
  45
  46 { "fishi" } [ "fishy" step1c ] unit-test
  47 { "by" } [ "by" step1c ] unit-test
  48
  49 { "realizat" } [ "realization" step4 ] unit-test
  50 { "ion" } [ "ion" step4 ] unit-test
  51 { "able" } [ "able" step4 ] unit-test
  52
  53 { "fear" } [ "feare" step5 "" like ] unit-test
  54 { "mate" } [ "mate" step5 "" like ] unit-test
  55 { "hell" } [ "hell" step5 "" like ] unit-test
  56 { "mate" } [ "mate" step5 "" like ] unit-test
  57
  58 { { } } [
  59     "vocab:porter-stemmer/test/voc.txt" utf8 file-lines
  60     [ stem ] map
  61     "vocab:porter-stemmer/test/output.txt" utf8 file-lines
  62     zip [ = ] assoc-reject
  63 ] unit-test