Test for stemming
authorDennis Schafroth <dennis@indexdata.com>
Thu, 11 Nov 2010 12:13:04 +0000 (13:13 +0100)
committerDennis Schafroth <dennis@indexdata.com>
Thu, 11 Nov 2010 12:13:04 +0000 (13:13 +0100)
test/tsticu-3.input [new file with mode: 0644]
test/tsticu-3.output [new file with mode: 0644]
test/tsticu-3.xml [new file with mode: 0644]

diff --git a/test/tsticu-3.input b/test/tsticu-3.input
new file mode 100644 (file)
index 0000000..989f89d
--- /dev/null
@@ -0,0 +1,2 @@
+men running
+he runs 
diff --git a/test/tsticu-3.output b/test/tsticu-3.output
new file mode 100644 (file)
index 0000000..3139c0b
--- /dev/null
@@ -0,0 +1,4 @@
+1 1 'men' ''
+2 1 'run' ''
+3 2 'he' ''
+4 2 'run' ''
diff --git a/test/tsticu-3.xml b/test/tsticu-3.xml
new file mode 100644 (file)
index 0000000..f7db5c7
--- /dev/null
@@ -0,0 +1,7 @@
+<icu_chain locale="en">
+  <transform rule="[:Control:] Any-Remove"/>
+  <tokenize rule="l"/>
+  <transform rule="[[:WhiteSpace:][:Punctuation:]] Remove"/>
+  <casemap rule="l"/>
+  <stemming rule="english" />
+</icu_chain>