src/dom/test/test-dom-configuration-basic

   1 #!/bin/sh
   2 #
   3 # Copyright (c) 2005 Jonas Fonseca
   4 #
   5
   6 test_description='Test the DOM configuration module
   7
   8 This test checks that the normalization performed by the DOM configuration
   9 is done correctly.
  10 '
  11
  12 . "$TEST_LIB"
  13
  14 test_normalize_output_equals () {
  15         desc="$1"; shift
  16         config="$1"; shift
  17         src="$1"; shift
  18         out="$1"; shift
  19
  20         URI="test:$(normalize "$desc")"
  21
  22         sgml-parser --src "$src" --normalize "$config" --uri "$URI" --src "$src" > output
  23         echo "#document: $URI" > expected
  24         echo "$out" | sed -n '2,$p' | sed 's/^/  /' >> expected
  25
  26         test_expect_success "$desc" 'cmp output expected'
  27 }
  28
  29
  30 ## Config strings ###########################################################
  31
  32 NOOP='cdata-sections,comments,element-content-whitespace,entities'
  33 NOCOMMENTS='cdata-sections,element-content-whitespace,entities'
  34 CDATA2TEXT='comments,element-content-whitespace,entities'
  35 ENTITIES='cdata-section,comments,element-content-whitespace'
  36 NOWSTEXT='cdata-section,comments,entities'
  37 NORM1=''
  38
  39
  40 ## No-ops ###################################################################
  41
  42 test_normalize_output_equals \
  43 'Normalization no-op.' \
  44 "$NOOP" \
  45 '<roswell>   <![CDATA[|  that  |]]><!-- ends -->  well</roswell>' \
  46 '
  47 element: roswell
  48   #text:
  49   #cdata-section: |  that  |
  50   #comment:  ends
  51   #text:   well'
  52
  53 test_normalize_output_equals \
  54 'Keep comments.' \
  55 "$NOOP" \
  56 'and<!-- comment:1 -->or<!-- comment:2 -->' \
  57 '
  58 #text: and
  59 #comment:  comment:1
  60 #text: or
  61 #comment:  comment:2 '
  62
  63 test_normalize_output_equals \
  64 'Keep CDATA sections ' \
  65 "$NOOP" \
  66 '<![CDATA[and]]>or<![CDATA[maybe]]>' \
  67 '
  68 #cdata-section: and
  69 #text: or
  70 #cdata-section: maybe'
  71
  72
  73 ## Comments #################################################################
  74
  75 test_normalize_output_equals \
  76 'Remove comments. (I)' \
  77 "$NOCOMMENTS" \
  78 "<no><!-- comment -->?</no>" \
  79 '
  80 element: no
  81   #text: ?'
  82
  83 test_normalize_output_equals \
  84 'Remove comments. (II)' \
  85 "$NOCOMMENTS" \
  86 '<!-- comment:1 -->and<!-- comment:2 -->' \
  87 '
  88 #text: and'
  89
  90 test_normalize_output_equals \
  91 'Remove comments. (III)' \
  92 "$NOCOMMENTS" \
  93 'nothing to see <!-- comment -->here' \
  94 '
  95 #text: nothing to see here'
  96
  97
  98 ## Entities #################################################################
  99
 100 # Entities should be shown 'verbatim' here after expansion.
 101
 102 test_normalize_output_equals \
 103 'Expand entities. (I)' \
 104 "$ENTITIES" \
 105 'a&lt;b&gt;c' \
 106 '
 107 #text: a&lt;b&gt;c'
 108
 109 test_normalize_output_equals \
 110 'Expand entities. (II)' \
 111 "$ENTITIES" \
 112 '&bad-entity&good-entity;' \
 113 '
 114 #text: &bad-entity;&good-entity;'
 115
 116 test_normalize_output_equals \
 117 'Expand entities. (III)' \
 118 "$ENTITIES" \
 119 '<a>&b;</a>' \
 120 '
 121 element: a
 122   #text: &b;'
 123
 124
 125 ## CDATA Sections ###########################################################
 126
 127 test_normalize_output_equals \
 128 'Replace CDATA section with text. (I)' \
 129 "$CDATA2TEXT" \
 130 '<![CDATA[a small text snippet]]>' \
 131 '
 132 #text: a small text snippet'
 133
 134 test_normalize_output_equals \
 135 'Replace CDATA section with text. (II)' \
 136 "$CDATA2TEXT" \
 137 '<![CDATA[a small]]> <![CDATA[text snippet]]>' \
 138 '
 139 #text: a small text snippet'
 140
 141 test_normalize_output_equals \
 142 'Replace CDATA section with text. (III)' \
 143 "$CDATA2TEXT" \
 144 'before <![CDATA[and]]> after' \
 145 '
 146 #text: before and after'
 147
 148
 149 ## Element Content Whitespace ###############################################
 150
 151 test_normalize_output_equals \
 152 'Remove element content whitespace. (I)' \
 153 "$NOWSTEXT" \
 154 '<a>
 155         <b>some text</b>
 156 </a>' \
 157 '
 158 element: a
 159   element: b
 160     #text: some text' \
 161
 162 # I haven't read the specs about this thing, for now it just blasts all
 163 # space-only text nodes. Probably not the wanted behaviour all of the time.
 164 # --jonas
 165 test_normalize_output_equals \
 166 'Remove element content whitespace. (II)' \
 167 "$NOWSTEXT" \
 168 '<e>space between &this; &that; gets removed</e>' \
 169 '
 170 element: e
 171   #text: space between
 172   entity-reference: this
 173   entity-reference: that
 174   #text:  gets removed'
 175
 176
 177 ## Mixes ####################################################################
 178
 179 test_normalize_output_equals \
 180 'Normalization mix #1. (I)' \
 181 "$NORM1" \
 182 'before <![CDATA[and]]> after &some;<!--comments--> remain' \
 183 '
 184 #text: before and after &some; remain'
 185
 186 test_normalize_output_equals \
 187 'Normalization mix #1. (II)' \
 188 "$NORM1" \
 189 '<!--a-->b&c;  <![CDATA[d]]>' \
 190 '
 191 #text: b&c;d'
 192
 193
 194 ## Special ELinks Extensions ################################################
 195
 196 test_normalize_output_equals \
 197 'Remove unknown (HTML) elements and attributes. (I)' \
 198 "$NOOP,unknown" \
 199 '<html wack="..."><title w00t="...">where?<doit>here!</doit></title></html>' \
 200 '
 201 element: html
 202   element: title
 203     #text: where?' \
 204
 205 test_normalize_output_equals \
 206 'Remove unknown (HTML) elements and attributes. (II)' \
 207 "$NOOP,unknown" \
 208 '<x y=""><z></z></x> aint no HTML' \
 209 '
 210 #text:  aint no HTML' \
 211
 212 test_normalize_output_equals \
 213 'Normalize whitespace. (I)' \
 214 "$NOOP,normalize-whitespace" \
 215 'Here    is      a
 216
 217
 218
 219 lot of     useless      space.' \
 220 '
 221 #text: Here is a lot of useless space.' \
 222
 223 test_normalize_output_equals \
 224 'Normalize whitespace. (II)' \
 225 "$CDATA2TEXT,normalize-whitespace" \
 226 'Could we <![CDATA[    please     ]]> read that again?' \
 227 '
 228 #text: Could we please read that again?' \
 229
 230 test_done