slaw 5.0.0 → 6.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +6 -0
- data/lib/slaw/grammars/za/act.treetop +2 -0
- data/lib/slaw/grammars/za/postprocess.rb +28 -0
- data/lib/slaw/parse/blocklists.rb +5 -0
- data/lib/slaw/parse/builder.rb +1 -16
- data/lib/slaw/parse/grammar_helpers.rb +8 -0
- data/lib/slaw/version.rb +1 -1
- data/spec/parse/blocklists_spec.rb +714 -0
- data/spec/parse/builder_spec.rb +0 -708
- data/spec/za/postprocess_spec.rb +117 -0
- metadata +7 -2
data/spec/parse/builder_spec.rb
CHANGED
@@ -7,714 +7,6 @@ describe Slaw::Parse::Builder do
|
|
7
7
|
let(:parser) { double("parser") }
|
8
8
|
subject { Slaw::Parse::Builder.new(parser: parser) }
|
9
9
|
|
10
|
-
describe '#adjust_blocklists' do
|
11
|
-
it 'should nest simple blocks' do
|
12
|
-
doc = xml2doc(subsection(<<XML
|
13
|
-
<blockList id="section-10.1.lst0" renest="true">
|
14
|
-
<item id="section-10.1.lst0.a">
|
15
|
-
<num>(a)</num>
|
16
|
-
<p>foo</p>
|
17
|
-
</item>
|
18
|
-
<item id="section-10.1.lst0.i">
|
19
|
-
<num>(i)</num>
|
20
|
-
<p>item-i</p>
|
21
|
-
</item>
|
22
|
-
<item id="section-10.1.lst0.ii">
|
23
|
-
<num>(ii)</num>
|
24
|
-
<p>item-ii</p>
|
25
|
-
</item>
|
26
|
-
<item id="section-10.1.lst0.iii">
|
27
|
-
<num>(iii)</num>
|
28
|
-
<p>item-iii</p>
|
29
|
-
</item>
|
30
|
-
<item id="section-10.1.lst0.aa">
|
31
|
-
<num>(aa)</num>
|
32
|
-
<p>item-aa</p>
|
33
|
-
</item>
|
34
|
-
<item id="section-10.1.lst0.bb">
|
35
|
-
<num>(bb)</num>
|
36
|
-
<p>item-bb</p>
|
37
|
-
</item>
|
38
|
-
</blockList>
|
39
|
-
XML
|
40
|
-
))
|
41
|
-
|
42
|
-
subject.adjust_blocklists(doc)
|
43
|
-
doc.to_s.should == subsection(<<XML
|
44
|
-
<blockList id="section-10.1.lst0">
|
45
|
-
<item id="section-10.1.lst0.a">
|
46
|
-
<num>(a)</num>
|
47
|
-
<blockList id="section-10.1.lst0.a.list0">
|
48
|
-
<listIntroduction>foo</listIntroduction>
|
49
|
-
<item id="section-10.1.lst0.a.list0.i">
|
50
|
-
<num>(i)</num>
|
51
|
-
<p>item-i</p>
|
52
|
-
</item>
|
53
|
-
<item id="section-10.1.lst0.a.list0.ii">
|
54
|
-
<num>(ii)</num>
|
55
|
-
<p>item-ii</p>
|
56
|
-
</item>
|
57
|
-
<item id="section-10.1.lst0.a.list0.iii">
|
58
|
-
<num>(iii)</num>
|
59
|
-
<blockList id="section-10.1.lst0.a.list0.iii.list0">
|
60
|
-
<listIntroduction>item-iii</listIntroduction>
|
61
|
-
<item id="section-10.1.lst0.a.list0.iii.list0.aa">
|
62
|
-
<num>(aa)</num>
|
63
|
-
<p>item-aa</p>
|
64
|
-
</item>
|
65
|
-
<item id="section-10.1.lst0.a.list0.iii.list0.bb">
|
66
|
-
<num>(bb)</num>
|
67
|
-
<p>item-bb</p>
|
68
|
-
</item>
|
69
|
-
</blockList>
|
70
|
-
</item>
|
71
|
-
</blockList>
|
72
|
-
</item>
|
73
|
-
</blockList>
|
74
|
-
XML
|
75
|
-
)
|
76
|
-
end
|
77
|
-
|
78
|
-
# -------------------------------------------------------------------------
|
79
|
-
|
80
|
-
it 'should jump back up a level' do
|
81
|
-
doc = xml2doc(subsection(<<XML
|
82
|
-
<blockList id="section-10.1.lst0" renest="true">
|
83
|
-
<item id="section-10.1.lst0.a">
|
84
|
-
<num>(a)</num>
|
85
|
-
<p>foo</p>
|
86
|
-
</item>
|
87
|
-
<item id="section-10.1.lst0.i">
|
88
|
-
<num>(i)</num>
|
89
|
-
<p>item-i</p>
|
90
|
-
</item>
|
91
|
-
<item id="section-10.1.lst0.ii">
|
92
|
-
<num>(ii)</num>
|
93
|
-
<p>item-ii</p>
|
94
|
-
</item>
|
95
|
-
<item id="section-10.1.lst0.c">
|
96
|
-
<num>(c)</num>
|
97
|
-
<p>item-c</p>
|
98
|
-
</item>
|
99
|
-
</blockList>
|
100
|
-
XML
|
101
|
-
))
|
102
|
-
|
103
|
-
subject.adjust_blocklists(doc)
|
104
|
-
doc.to_s.should == subsection(<<XML
|
105
|
-
<blockList id="section-10.1.lst0">
|
106
|
-
<item id="section-10.1.lst0.a">
|
107
|
-
<num>(a)</num>
|
108
|
-
<blockList id="section-10.1.lst0.a.list0">
|
109
|
-
<listIntroduction>foo</listIntroduction>
|
110
|
-
<item id="section-10.1.lst0.a.list0.i">
|
111
|
-
<num>(i)</num>
|
112
|
-
<p>item-i</p>
|
113
|
-
</item>
|
114
|
-
<item id="section-10.1.lst0.a.list0.ii">
|
115
|
-
<num>(ii)</num>
|
116
|
-
<p>item-ii</p>
|
117
|
-
</item>
|
118
|
-
</blockList>
|
119
|
-
</item>
|
120
|
-
<item id="section-10.1.lst0.c">
|
121
|
-
<num>(c)</num>
|
122
|
-
<p>item-c</p>
|
123
|
-
</item>
|
124
|
-
</blockList>
|
125
|
-
XML
|
126
|
-
)
|
127
|
-
end
|
128
|
-
|
129
|
-
# -------------------------------------------------------------------------
|
130
|
-
|
131
|
-
it 'should handle (i) correctly' do
|
132
|
-
doc = xml2doc(subsection(<<XML
|
133
|
-
<blockList id="section-10.1.lst0" renest="true">
|
134
|
-
<item id="section-10.1.lst0.h">
|
135
|
-
<num>(h)</num>
|
136
|
-
<p>foo</p>
|
137
|
-
</item>
|
138
|
-
<item id="section-10.1.lst0.i">
|
139
|
-
<num>(i)</num>
|
140
|
-
<p>item-i</p>
|
141
|
-
</item>
|
142
|
-
<item id="section-10.1.lst0.j">
|
143
|
-
<num>(j)</num>
|
144
|
-
<p>item-ii</p>
|
145
|
-
</item>
|
146
|
-
</blockList>
|
147
|
-
XML
|
148
|
-
))
|
149
|
-
|
150
|
-
subject.adjust_blocklists(doc)
|
151
|
-
doc.to_s.should == subsection(<<XML
|
152
|
-
<blockList id="section-10.1.lst0">
|
153
|
-
<item id="section-10.1.lst0.h">
|
154
|
-
<num>(h)</num>
|
155
|
-
<p>foo</p>
|
156
|
-
</item>
|
157
|
-
<item id="section-10.1.lst0.i">
|
158
|
-
<num>(i)</num>
|
159
|
-
<p>item-i</p>
|
160
|
-
</item>
|
161
|
-
<item id="section-10.1.lst0.j">
|
162
|
-
<num>(j)</num>
|
163
|
-
<p>item-ii</p>
|
164
|
-
</item>
|
165
|
-
</blockList>
|
166
|
-
XML
|
167
|
-
)
|
168
|
-
end
|
169
|
-
|
170
|
-
# -------------------------------------------------------------------------
|
171
|
-
|
172
|
-
it 'should handle (u) (v) and (x) correctly' do
|
173
|
-
doc = xml2doc(subsection(<<XML
|
174
|
-
<blockList id="section-10.1.lst0" renest="true">
|
175
|
-
<item id="section-10.1.lst0.t">
|
176
|
-
<num>(t)</num>
|
177
|
-
<p>foo</p>
|
178
|
-
</item>
|
179
|
-
<item id="section-10.1.lst0.u">
|
180
|
-
<num>(u)</num>
|
181
|
-
<p>item-i</p>
|
182
|
-
</item>
|
183
|
-
<item id="section-10.1.lst0.v">
|
184
|
-
<num>(v)</num>
|
185
|
-
<p>item-ii</p>
|
186
|
-
</item>
|
187
|
-
<item id="section-10.1.lst0.w">
|
188
|
-
<num>(w)</num>
|
189
|
-
<p>item-ii</p>
|
190
|
-
</item>
|
191
|
-
<item id="section-10.1.lst0.x">
|
192
|
-
<num>(x)</num>
|
193
|
-
<p>item-ii</p>
|
194
|
-
</item>
|
195
|
-
</blockList>
|
196
|
-
XML
|
197
|
-
))
|
198
|
-
|
199
|
-
subject.adjust_blocklists(doc)
|
200
|
-
doc.to_s.should == subsection(<<XML
|
201
|
-
<blockList id="section-10.1.lst0">
|
202
|
-
<item id="section-10.1.lst0.t">
|
203
|
-
<num>(t)</num>
|
204
|
-
<p>foo</p>
|
205
|
-
</item>
|
206
|
-
<item id="section-10.1.lst0.u">
|
207
|
-
<num>(u)</num>
|
208
|
-
<p>item-i</p>
|
209
|
-
</item>
|
210
|
-
<item id="section-10.1.lst0.v">
|
211
|
-
<num>(v)</num>
|
212
|
-
<p>item-ii</p>
|
213
|
-
</item>
|
214
|
-
<item id="section-10.1.lst0.w">
|
215
|
-
<num>(w)</num>
|
216
|
-
<p>item-ii</p>
|
217
|
-
</item>
|
218
|
-
<item id="section-10.1.lst0.x">
|
219
|
-
<num>(x)</num>
|
220
|
-
<p>item-ii</p>
|
221
|
-
</item>
|
222
|
-
</blockList>
|
223
|
-
XML
|
224
|
-
)
|
225
|
-
end
|
226
|
-
|
227
|
-
|
228
|
-
# -------------------------------------------------------------------------
|
229
|
-
|
230
|
-
it 'should handle (j) correctly' do
|
231
|
-
doc = xml2doc(subsection(<<XML
|
232
|
-
<blockList id="section-28.3.list2" renest="true">
|
233
|
-
<item id="section-28.3.list2.g">
|
234
|
-
<num>(g)</num>
|
235
|
-
<p>all <term refersTo="#term-memorial_work" id="trm381">memorial work</term> up to 150 mm in thickness must be securely attached to the base;</p>
|
236
|
-
</item>
|
237
|
-
<item id="section-28.3.list2.h">
|
238
|
-
<num>(h)</num>
|
239
|
-
<p>all the components of <term refersTo="#term-memorial_work" id="trm382">memorial work</term> must be completed before being brought into a <term refersTo="#term-cemetery" id="trm383">cemetery</term>;</p>
|
240
|
-
</item>
|
241
|
-
<item id="section-28.3.list2.i">
|
242
|
-
<num>(i)</num>
|
243
|
-
<p>footstones must consist of one solid piece;</p>
|
244
|
-
</item>
|
245
|
-
<item id="section-28.3.list2.j">
|
246
|
-
<num>(j)</num>
|
247
|
-
<p>in all cases where <term refersTo="#term-memorial_work" id="trm384">memorial work</term> rests on a base -</p>
|
248
|
-
</item>
|
249
|
-
<item id="section-28.3.list2.i">
|
250
|
-
<num>(i)</num>
|
251
|
-
<p>such <term refersTo="#term-memorial_work" id="trm385">memorial work</term> must have a foundation;</p>
|
252
|
-
</item>
|
253
|
-
<item id="section-28.3.list2.ii">
|
254
|
-
<num>(ii)</num>
|
255
|
-
<p>such <term refersTo="#term-memorial_work" id="trm386">memorial work</term> must be set with cement mortar;</p>
|
256
|
-
</item>
|
257
|
-
<item id="section-28.3.list2.iii">
|
258
|
-
<num>(iii)</num>
|
259
|
-
<p>the bottom base of a single <term refersTo="#term-memorial_work" id="trm387">memorial work</term> must not be less than 900mm long 220 mm wide x 250 mm thick and that of a double <term refersTo="#term-memorial_work" id="trm388">memorial work</term> not less than 2 286 mm long x 200 mm wide x 250 mm thick; and</p>
|
260
|
-
</item>
|
261
|
-
</blockList>
|
262
|
-
XML
|
263
|
-
))
|
264
|
-
|
265
|
-
subject.adjust_blocklists(doc)
|
266
|
-
doc.to_s.should == subsection(<<XML
|
267
|
-
<blockList id="section-28.3.list2">
|
268
|
-
<item id="section-28.3.list2.g">
|
269
|
-
<num>(g)</num>
|
270
|
-
<p>all <term refersTo="#term-memorial_work" id="trm381">memorial work</term> up to 150 mm in thickness must be securely attached to the base;</p>
|
271
|
-
</item>
|
272
|
-
<item id="section-28.3.list2.h">
|
273
|
-
<num>(h)</num>
|
274
|
-
<p>all the components of <term refersTo="#term-memorial_work" id="trm382">memorial work</term> must be completed before being brought into a <term refersTo="#term-cemetery" id="trm383">cemetery</term>;</p>
|
275
|
-
</item>
|
276
|
-
<item id="section-28.3.list2.i">
|
277
|
-
<num>(i)</num>
|
278
|
-
<p>footstones must consist of one solid piece;</p>
|
279
|
-
</item>
|
280
|
-
<item id="section-28.3.list2.j">
|
281
|
-
<num>(j)</num>
|
282
|
-
<blockList id="section-28.3.list2.j.list0">
|
283
|
-
<listIntroduction>in all cases where <term refersTo="#term-memorial_work" id="trm384">memorial work</term> rests on a base -</listIntroduction>
|
284
|
-
<item id="section-28.3.list2.j.list0.i">
|
285
|
-
<num>(i)</num>
|
286
|
-
<p>such <term refersTo="#term-memorial_work" id="trm385">memorial work</term> must have a foundation;</p>
|
287
|
-
</item>
|
288
|
-
<item id="section-28.3.list2.j.list0.ii">
|
289
|
-
<num>(ii)</num>
|
290
|
-
<p>such <term refersTo="#term-memorial_work" id="trm386">memorial work</term> must be set with cement mortar;</p>
|
291
|
-
</item>
|
292
|
-
<item id="section-28.3.list2.j.list0.iii">
|
293
|
-
<num>(iii)</num>
|
294
|
-
<p>the bottom base of a single <term refersTo="#term-memorial_work" id="trm387">memorial work</term> must not be less than 900mm long 220 mm wide x 250 mm thick and that of a double <term refersTo="#term-memorial_work" id="trm388">memorial work</term> not less than 2 286 mm long x 200 mm wide x 250 mm thick; and</p>
|
295
|
-
</item>
|
296
|
-
</blockList>
|
297
|
-
</item>
|
298
|
-
</blockList>
|
299
|
-
XML
|
300
|
-
)
|
301
|
-
end
|
302
|
-
|
303
|
-
# -------------------------------------------------------------------------
|
304
|
-
it 'should handle (I) correctly' do
|
305
|
-
doc = xml2doc(subsection(<<XML
|
306
|
-
<blockList id="section-28.3.list2" renest="true">
|
307
|
-
<item id="section-28.3.list2.g">
|
308
|
-
<num>(g)</num>
|
309
|
-
<p>all memorial work up to 150 mm in thickness must be securely attached to the base;</p>
|
310
|
-
</item>
|
311
|
-
<item id="section-28.3.list2.h">
|
312
|
-
<num>(h)</num>
|
313
|
-
<p>all the components of memorial work must be completed before being brought into a cemetery;</p>
|
314
|
-
</item>
|
315
|
-
<item id="section-28.3.list2.i">
|
316
|
-
<num>(i)</num>
|
317
|
-
<p>item i</p>
|
318
|
-
</item>
|
319
|
-
<item id="section-28.3.list2.I">
|
320
|
-
<num>(I)</num>
|
321
|
-
<p>a subitem</p>
|
322
|
-
</item>
|
323
|
-
<item id="section-28.3.list2.II">
|
324
|
-
<num>(II)</num>
|
325
|
-
<p>another subitem</p>
|
326
|
-
</item>
|
327
|
-
<item id="section-28.3.list2.j">
|
328
|
-
<num>(j)</num>
|
329
|
-
<p>in all cases where memorial work rests on a base -</p>
|
330
|
-
</item>
|
331
|
-
<item id="section-28.3.list2.i">
|
332
|
-
<num>(i)</num>
|
333
|
-
<p>such memorial work must have a foundation;</p>
|
334
|
-
</item>
|
335
|
-
<item id="section-28.3.list2.ii">
|
336
|
-
<num>(ii)</num>
|
337
|
-
<p>such memorial work must be set with cement mortar;</p>
|
338
|
-
</item>
|
339
|
-
<item id="section-28.3.list2.iii">
|
340
|
-
<num>(iii)</num>
|
341
|
-
<p>the bottom base of a single memorial work must not be less than 900mm long 220 mm wide x 250 mm thick and that of a double memorial work not less than 2 286 mm long x 200 mm wide x 250 mm thick; and</p>
|
342
|
-
</item>
|
343
|
-
</blockList>
|
344
|
-
XML
|
345
|
-
))
|
346
|
-
|
347
|
-
subject.adjust_blocklists(doc)
|
348
|
-
doc.to_s.should == subsection(<<XML
|
349
|
-
<blockList id="section-28.3.list2">
|
350
|
-
<item id="section-28.3.list2.g">
|
351
|
-
<num>(g)</num>
|
352
|
-
<p>all memorial work up to 150 mm in thickness must be securely attached to the base;</p>
|
353
|
-
</item>
|
354
|
-
<item id="section-28.3.list2.h">
|
355
|
-
<num>(h)</num>
|
356
|
-
<p>all the components of memorial work must be completed before being brought into a cemetery;</p>
|
357
|
-
</item>
|
358
|
-
<item id="section-28.3.list2.i">
|
359
|
-
<num>(i)</num>
|
360
|
-
<blockList id="section-28.3.list2.i.list0">
|
361
|
-
<listIntroduction>item i</listIntroduction>
|
362
|
-
<item id="section-28.3.list2.i.list0.I">
|
363
|
-
<num>(I)</num>
|
364
|
-
<p>a subitem</p>
|
365
|
-
</item>
|
366
|
-
<item id="section-28.3.list2.i.list0.II">
|
367
|
-
<num>(II)</num>
|
368
|
-
<p>another subitem</p>
|
369
|
-
</item>
|
370
|
-
</blockList>
|
371
|
-
</item>
|
372
|
-
<item id="section-28.3.list2.j">
|
373
|
-
<num>(j)</num>
|
374
|
-
<blockList id="section-28.3.list2.j.list1">
|
375
|
-
<listIntroduction>in all cases where memorial work rests on a base -</listIntroduction>
|
376
|
-
<item id="section-28.3.list2.j.list1.i">
|
377
|
-
<num>(i)</num>
|
378
|
-
<p>such memorial work must have a foundation;</p>
|
379
|
-
</item>
|
380
|
-
<item id="section-28.3.list2.j.list1.ii">
|
381
|
-
<num>(ii)</num>
|
382
|
-
<p>such memorial work must be set with cement mortar;</p>
|
383
|
-
</item>
|
384
|
-
<item id="section-28.3.list2.j.list1.iii">
|
385
|
-
<num>(iii)</num>
|
386
|
-
<p>the bottom base of a single memorial work must not be less than 900mm long 220 mm wide x 250 mm thick and that of a double memorial work not less than 2 286 mm long x 200 mm wide x 250 mm thick; and</p>
|
387
|
-
</item>
|
388
|
-
</blockList>
|
389
|
-
</item>
|
390
|
-
</blockList>
|
391
|
-
XML
|
392
|
-
)
|
393
|
-
end
|
394
|
-
|
395
|
-
# -------------------------------------------------------------------------
|
396
|
-
|
397
|
-
it 'should treat (aa) after (z) as siblings' do
|
398
|
-
doc = xml2doc(subsection(<<XML
|
399
|
-
<blockList id="list0">
|
400
|
-
<item id="list0.y">
|
401
|
-
<num>(y)</num>
|
402
|
-
<p>foo</p>
|
403
|
-
</item>
|
404
|
-
<item id="list0.z">
|
405
|
-
<num>(z)</num>
|
406
|
-
<p>item-z</p>
|
407
|
-
</item>
|
408
|
-
<item id="list0.aa">
|
409
|
-
<num>(aa)</num>
|
410
|
-
<p>item-aa</p>
|
411
|
-
</item>
|
412
|
-
<item id="list0.bb">
|
413
|
-
<num>(bb)</num>
|
414
|
-
<p>item-bb</p>
|
415
|
-
</item>
|
416
|
-
</blockList>
|
417
|
-
XML
|
418
|
-
))
|
419
|
-
|
420
|
-
subject.adjust_blocklists(doc)
|
421
|
-
doc.to_s.should == subsection(<<XML
|
422
|
-
<blockList id="list0">
|
423
|
-
<item id="list0.y">
|
424
|
-
<num>(y)</num>
|
425
|
-
<p>foo</p>
|
426
|
-
</item>
|
427
|
-
<item id="list0.z">
|
428
|
-
<num>(z)</num>
|
429
|
-
<p>item-z</p>
|
430
|
-
</item>
|
431
|
-
<item id="list0.aa">
|
432
|
-
<num>(aa)</num>
|
433
|
-
<p>item-aa</p>
|
434
|
-
</item>
|
435
|
-
<item id="list0.bb">
|
436
|
-
<num>(bb)</num>
|
437
|
-
<p>item-bb</p>
|
438
|
-
</item>
|
439
|
-
</blockList>
|
440
|
-
XML
|
441
|
-
)
|
442
|
-
end
|
443
|
-
|
444
|
-
# -------------------------------------------------------------------------
|
445
|
-
|
446
|
-
it 'should treat (AA) after (z) a sublist' do
|
447
|
-
doc = xml2doc(subsection(<<XML
|
448
|
-
<blockList id="list0" renest="true">
|
449
|
-
<item id="list0.y">
|
450
|
-
<num>(y)</num>
|
451
|
-
<p>foo</p>
|
452
|
-
</item>
|
453
|
-
<item id="list0.z">
|
454
|
-
<num>(z)</num>
|
455
|
-
<p>item-z</p>
|
456
|
-
</item>
|
457
|
-
<item id="list0.AA">
|
458
|
-
<num>(AA)</num>
|
459
|
-
<p>item-AA</p>
|
460
|
-
</item>
|
461
|
-
<item id="list0.BB">
|
462
|
-
<num>(BB)</num>
|
463
|
-
<p>item-BB</p>
|
464
|
-
</item>
|
465
|
-
</blockList>
|
466
|
-
XML
|
467
|
-
))
|
468
|
-
|
469
|
-
subject.adjust_blocklists(doc)
|
470
|
-
doc.to_s.should == subsection(<<XML
|
471
|
-
<blockList id="list0">
|
472
|
-
<item id="list0.y">
|
473
|
-
<num>(y)</num>
|
474
|
-
<p>foo</p>
|
475
|
-
</item>
|
476
|
-
<item id="list0.z">
|
477
|
-
<num>(z)</num>
|
478
|
-
<blockList id="list0.z.list0">
|
479
|
-
<listIntroduction>item-z</listIntroduction>
|
480
|
-
<item id="list0.z.list0.AA">
|
481
|
-
<num>(AA)</num>
|
482
|
-
<p>item-AA</p>
|
483
|
-
</item>
|
484
|
-
<item id="list0.z.list0.BB">
|
485
|
-
<num>(BB)</num>
|
486
|
-
<p>item-BB</p>
|
487
|
-
</item>
|
488
|
-
</blockList>
|
489
|
-
</item>
|
490
|
-
</blockList>
|
491
|
-
XML
|
492
|
-
)
|
493
|
-
end
|
494
|
-
|
495
|
-
# -------------------------------------------------------------------------
|
496
|
-
|
497
|
-
it 'should handle deeply nested lists' do
|
498
|
-
doc = xml2doc(subsection(<<XML
|
499
|
-
<blockList id="list0" renest="true">
|
500
|
-
<item id="list0.a">
|
501
|
-
<num>(a)</num>
|
502
|
-
<p>foo</p>
|
503
|
-
</item>
|
504
|
-
<item id="list0.b">
|
505
|
-
<num>(b)</num>
|
506
|
-
<p>item-b</p>
|
507
|
-
</item>
|
508
|
-
<item id="list0.i">
|
509
|
-
<num>(i)</num>
|
510
|
-
<p>item-b-i</p>
|
511
|
-
</item>
|
512
|
-
<item id="list0.aa">
|
513
|
-
<num>(aa)</num>
|
514
|
-
<p>item-i-aa</p>
|
515
|
-
</item>
|
516
|
-
<item id="list0.bb">
|
517
|
-
<num>(bb)</num>
|
518
|
-
<p>item-i-bb</p>
|
519
|
-
</item>
|
520
|
-
<item id="list0.ii">
|
521
|
-
<num>(ii)</num>
|
522
|
-
<p>item-b-ii</p>
|
523
|
-
</item>
|
524
|
-
<item id="list0.c">
|
525
|
-
<num>(c)</num>
|
526
|
-
<p>item-c</p>
|
527
|
-
</item>
|
528
|
-
<item id="list0.i">
|
529
|
-
<num>(i)</num>
|
530
|
-
<p>item-c-i</p>
|
531
|
-
</item>
|
532
|
-
<item id="list0.ii">
|
533
|
-
<num>(ii)</num>
|
534
|
-
<p>item-c-ii</p>
|
535
|
-
</item>
|
536
|
-
<item id="list0.iii">
|
537
|
-
<num>(iii)</num>
|
538
|
-
<p>item-c-iii</p>
|
539
|
-
</item>
|
540
|
-
</blockList>
|
541
|
-
XML
|
542
|
-
))
|
543
|
-
|
544
|
-
subject.adjust_blocklists(doc)
|
545
|
-
doc.to_s.should == subsection(<<XML
|
546
|
-
<blockList id="list0">
|
547
|
-
<item id="list0.a">
|
548
|
-
<num>(a)</num>
|
549
|
-
<p>foo</p>
|
550
|
-
</item>
|
551
|
-
<item id="list0.b">
|
552
|
-
<num>(b)</num>
|
553
|
-
<blockList id="list0.b.list0">
|
554
|
-
<listIntroduction>item-b</listIntroduction>
|
555
|
-
<item id="list0.b.list0.i">
|
556
|
-
<num>(i)</num>
|
557
|
-
<blockList id="list0.b.list0.i.list0">
|
558
|
-
<listIntroduction>item-b-i</listIntroduction>
|
559
|
-
<item id="list0.b.list0.i.list0.aa">
|
560
|
-
<num>(aa)</num>
|
561
|
-
<p>item-i-aa</p>
|
562
|
-
</item>
|
563
|
-
<item id="list0.b.list0.i.list0.bb">
|
564
|
-
<num>(bb)</num>
|
565
|
-
<p>item-i-bb</p>
|
566
|
-
</item>
|
567
|
-
</blockList>
|
568
|
-
</item>
|
569
|
-
<item id="list0.b.list0.ii">
|
570
|
-
<num>(ii)</num>
|
571
|
-
<p>item-b-ii</p>
|
572
|
-
</item>
|
573
|
-
</blockList>
|
574
|
-
</item>
|
575
|
-
<item id="list0.c">
|
576
|
-
<num>(c)</num>
|
577
|
-
<blockList id="list0.c.list1">
|
578
|
-
<listIntroduction>item-c</listIntroduction>
|
579
|
-
<item id="list0.c.list1.i">
|
580
|
-
<num>(i)</num>
|
581
|
-
<p>item-c-i</p>
|
582
|
-
</item>
|
583
|
-
<item id="list0.c.list1.ii">
|
584
|
-
<num>(ii)</num>
|
585
|
-
<p>item-c-ii</p>
|
586
|
-
</item>
|
587
|
-
<item id="list0.c.list1.iii">
|
588
|
-
<num>(iii)</num>
|
589
|
-
<p>item-c-iii</p>
|
590
|
-
</item>
|
591
|
-
</blockList>
|
592
|
-
</item>
|
593
|
-
</blockList>
|
594
|
-
XML
|
595
|
-
)
|
596
|
-
end
|
597
|
-
|
598
|
-
# -------------------------------------------------------------------------
|
599
|
-
|
600
|
-
it 'should jump back up a level when finding (i) near (h)' do
|
601
|
-
doc = xml2doc(subsection(<<XML
|
602
|
-
<blockList id="section-10.1.lst0" renest="true">
|
603
|
-
<item id="section-10.1.lst0.h">
|
604
|
-
<num>(h)</num>
|
605
|
-
<p>foo</p>
|
606
|
-
</item>
|
607
|
-
<item id="section-10.1.lst0.i">
|
608
|
-
<num>(i)</num>
|
609
|
-
<p>item-i</p>
|
610
|
-
</item>
|
611
|
-
<item id="section-10.1.lst0.ii">
|
612
|
-
<num>(ii)</num>
|
613
|
-
<p>item-ii</p>
|
614
|
-
</item>
|
615
|
-
<item id="section-10.1.lst0.i">
|
616
|
-
<num>(i)</num>
|
617
|
-
<p>item-i</p>
|
618
|
-
</item>
|
619
|
-
</blockList>
|
620
|
-
XML
|
621
|
-
))
|
622
|
-
|
623
|
-
subject.adjust_blocklists(doc)
|
624
|
-
doc.to_s.should == subsection(<<XML
|
625
|
-
<blockList id="section-10.1.lst0">
|
626
|
-
<item id="section-10.1.lst0.h">
|
627
|
-
<num>(h)</num>
|
628
|
-
<blockList id="section-10.1.lst0.h.list0">
|
629
|
-
<listIntroduction>foo</listIntroduction>
|
630
|
-
<item id="section-10.1.lst0.h.list0.i">
|
631
|
-
<num>(i)</num>
|
632
|
-
<p>item-i</p>
|
633
|
-
</item>
|
634
|
-
<item id="section-10.1.lst0.h.list0.ii">
|
635
|
-
<num>(ii)</num>
|
636
|
-
<p>item-ii</p>
|
637
|
-
</item>
|
638
|
-
</blockList>
|
639
|
-
</item>
|
640
|
-
<item id="section-10.1.lst0.i">
|
641
|
-
<num>(i)</num>
|
642
|
-
<p>item-i</p>
|
643
|
-
</item>
|
644
|
-
</blockList>
|
645
|
-
XML
|
646
|
-
)
|
647
|
-
end
|
648
|
-
|
649
|
-
# -------------------------------------------------------------------------
|
650
|
-
|
651
|
-
it 'should handle dotted numbers correctly' do
|
652
|
-
doc = xml2doc(subsection(<<XML
|
653
|
-
<blockList id="section-9.subsection-2.list2" renest="true">
|
654
|
-
<item id="section-9.subsection-2.list2.9.2.1">
|
655
|
-
<num>9.2.1</num>
|
656
|
-
<p>is incapable of trading because of an illness, provided that:</p>
|
657
|
-
</item>
|
658
|
-
<item id="section-9.subsection-2.list2.9.2.1.1">
|
659
|
-
<num>9.2.1.1</num>
|
660
|
-
<p>proof from a medical practitioner is provided to the City which certifies that the permit-holder is unable to trade; and</p>
|
661
|
-
</item>
|
662
|
-
<item id="section-9.subsection-2.list2.9.2.1.2">
|
663
|
-
<num>9.2.1.2</num>
|
664
|
-
<p>the dependent or assistant is only permitted to replace the permit-</p>
|
665
|
-
</item>
|
666
|
-
</blockList>
|
667
|
-
XML
|
668
|
-
))
|
669
|
-
|
670
|
-
subject.adjust_blocklists(doc)
|
671
|
-
doc.to_s.should == subsection(<<XML
|
672
|
-
<blockList id="section-9.subsection-2.list2">
|
673
|
-
<item id="section-9.subsection-2.list2.9.2.1">
|
674
|
-
<num>9.2.1</num>
|
675
|
-
<blockList id="section-9.subsection-2.list2.9.2.1.list0">
|
676
|
-
<listIntroduction>is incapable of trading because of an illness, provided that:</listIntroduction>
|
677
|
-
<item id="section-9.subsection-2.list2.9.2.1.list0.9.2.1.1">
|
678
|
-
<num>9.2.1.1</num>
|
679
|
-
<p>proof from a medical practitioner is provided to the City which certifies that the permit-holder is unable to trade; and</p>
|
680
|
-
</item>
|
681
|
-
<item id="section-9.subsection-2.list2.9.2.1.list0.9.2.1.2">
|
682
|
-
<num>9.2.1.2</num>
|
683
|
-
<p>the dependent or assistant is only permitted to replace the permit-</p>
|
684
|
-
</item>
|
685
|
-
</blockList>
|
686
|
-
</item>
|
687
|
-
</blockList>
|
688
|
-
XML
|
689
|
-
)
|
690
|
-
end
|
691
|
-
|
692
|
-
it 'should handle p tags just before' do
|
693
|
-
doc = xml2doc(subsection(<<XML
|
694
|
-
<p>intro</p>
|
695
|
-
<blockList id="section-10.1.lst0">
|
696
|
-
<item id="section-10.1.lst0.a">
|
697
|
-
<num>(a)</num>
|
698
|
-
<p>foo</p>
|
699
|
-
</item>
|
700
|
-
</blockList>
|
701
|
-
XML
|
702
|
-
))
|
703
|
-
|
704
|
-
subject.adjust_blocklists(doc)
|
705
|
-
doc.to_s.should == subsection(<<XML
|
706
|
-
<blockList id="section-10.1.lst0">
|
707
|
-
<listIntroduction>intro</listIntroduction>
|
708
|
-
<item id="section-10.1.lst0.a">
|
709
|
-
<num>(a)</num>
|
710
|
-
<p>foo</p>
|
711
|
-
</item>
|
712
|
-
</blockList>
|
713
|
-
XML
|
714
|
-
)
|
715
|
-
end
|
716
|
-
end
|
717
|
-
|
718
10
|
describe '#preprocess' do
|
719
11
|
it 'should split inline table cells into block table cells' do
|
720
12
|
text = <<EOS
|