},
      "italian_elision": {
        "type":         "elision",
        "articles": [
          "c", "l", "all", "dall", "dell",
          "nell", "sull", "coll", "pell",
          "gl", "agl", "dagl", "degl", "negl",
          "sugl", "un", "m", "t", "s", "v", "d"
        ]
      },
      "italian_stop": {
        "type":       "stop",
        "stopwords":  "_italian_"
      },

ca=ca\n\
ckb-iq=ckb-iq\n\
ckb_IQ=ckb-iq\n\
cs=cs\n\
da=da\n\
de=de\n\
el=el\n\
en=en\n\
en-ie=en-ie\n\
en_IE=en-ie\n\
es=es\n\
et=et\n\
eu=eu\n\
fa=fa\n\
fi=fi\n\
fr=fr\n\
gl=gl\n\
gu=gu\n\
he=he\n\
hi=hi\n\
hr=hr\n\
hu=hu\n\
hy=hy\n\
id=id\n\
it=it\n\
ja=ja\n\
ko=ko\n\
lt=lt\n\
lv=lv\n\
mk=mk\n\
ml=ml\n\
nl=nl\n\
no=no\n\
pa=pa\n\

      },
      "italian_elision": {
        "type":         "elision",
        "articles": [
          "c", "l", "all", "dall", "dell",
          "nell", "sull", "coll", "pell",
          "gl", "agl", "dagl", "degl", "negl",
          "sugl", "un", "m", "t", "s", "v", "d"
        ]
      },
      "italian_stop": {
        "type":       "stop",
        "stopwords":  "_italian_"
      },

      },
      "italian_elision": {
        "type":         "elision",
        "articles": [
          "c", "l", "all", "dall", "dell",
          "nell", "sull", "coll", "pell",
          "gl", "agl", "dagl", "degl", "negl",
          "sugl", "un", "m", "t", "s", "v", "d"
        ]
      },
      "italian_stop": {
        "type":       "stop",
        "stopwords":  "_italian_"
      },

ue?ofni?t&en?rops??ude?v&og?t???a??g!.&gro?hsadtob,lenap:.nomead,,oc?saak,t&en?ikcats,???i&a?v??k!.&gro?lim?moc?su,ten?ude?vog?xx,??m!.&drp?gro?lim?mo&c?n??oc?ude?vog??pk??n!.&clp,dtl,eman?gro?hcs?i!bom??l&im?oc,?m&oc?rif,?neg,ogn,ten?ude?vog?zib:.&gl,ld,no,o&c,g,??,?aw?i!b!mulp??car?d&art?dew??h&sif?tolc??k&iv?oo&b?c???ls?n&aelc?iart??p!pohs??re&enigne?tac??t&ad?ekram?hgil?lusnoc?neg?ov?soh!.tfarcnepo,??vi&g?l???o!.lbo,s??u&rehcisrev?smas?tarebsnegömrev???o&d?lb?og!.&duolc,etalsnart,???r&2n084q...

    // TODO make exception message escape non-printable characters
  }

  @Test
  fun parseDoesNotTrimOtherWhitespaceCharacters() {
    // Whitespace characters list from Google's Guava team: http://goo.gl/IcR9RD
    // line tabulation
    assertThat(parse("http://h/\u000b").encodedPath).isEqualTo("/%0B")
    // information separator 4
    assertThat(parse("http://h/\u001c").encodedPath).isEqualTo("/%1C")