Added support for pagebreaks (i.e. empty elements)
Change-Id: Ifa9114601f3d9fd954c5ba493a268e75ed9b834e
diff --git a/t/index/corpus/doc/0001/struct/structure.xml b/t/index/corpus/doc/0001/struct/structure.xml
index 68d4dc2..30edc1e 100644
--- a/t/index/corpus/doc/0001/struct/structure.xml
+++ b/t/index/corpus/doc/0001/struct/structure.xml
@@ -1,7 +1,7 @@
<?xml version="1.0" encoding="UTF-8"?>
<?xml-model href="span.rng" type="application/xml" schematypens="http://relaxng.org/ns/structure/1.0"?>
-<layer docid="A01_APR.13047" xmlns="http://ids-mannheim.de/ns/KorAP" version="KorAP-0.4">
+<layer docid="Corpus_Doc.0001" xmlns="http://ids-mannheim.de/ns/KorAP" version="KorAP-0.4">
<spanList>
<span id="s0" from="0" to="128" l="1">
<fs type="struct" xmlns="http://www.tei-c.org/ns/1.0">
@@ -89,6 +89,18 @@
</f>
</fs>
</span>
+ <span id="sx" from="42" to="42" l="7">
+ <fs type="struct" xmlns="http://www.tei-c.org/ns/1.0">
+ <f name="name">pb</f>
+ <f name="attr">
+ <fs type="attr">
+ <f name="n">2</f>
+ <f name="id">Corpus.Doc-0001-pb2</f>
+ <f name="TEIform">pb</f>
+ </fs>
+ </f>
+ </fs>
+ </span>
<span id="s10" from="43" to="128" l="6">
<fs type="struct" xmlns="http://www.tei-c.org/ns/1.0">
<f name="name">div</f>
diff --git a/t/index/dereko_struct.t b/t/index/dereko_struct.t
index 30c41f0..beaf248 100644
--- a/t/index/dereko_struct.t
+++ b/t/index/dereko_struct.t
@@ -21,7 +21,7 @@
# Empty element (from 0 to 0) on level 1, with TUI 2
is($data->{stream}->[0]->[1],
- '<>:dereko/s:idsHeader$<b>65<i>0<i>0<i>1<b>1<s>2',
+ '<>:dereko/s:idsHeader$<b>65<i>0<i>0<i>0<b>1<s>2',
'Empty element');
# Attributes:
@@ -46,7 +46,17 @@
'Attribute of idsHeader');
-diag 'TODO: Test for element spans';
+is($data->{stream}->[4]->[1],
+ '<>:dereko/s:s$<b>64<i>32<i>42<i>6<b>6<s>1',
+ 'Sentence span');
+
+is($data->{stream}->[4]->[2],
+ '@:dereko/s:broken:no$<b>17<s>1<i>6',
+ 'Attribute of sentence span');
+
+is($data->{stream}->[6]->[0],
+ '<>:dereko/s:pb$<b>65<i>42<i>42<i>6<b>6<s>1',
+ 'Pagebreak element');
done_testing;