From aa5f96417ff81408b15b54aab35a3c16b845adf8 Mon Sep 17 00:00:00 2001 From: Patrick Simianer
Date: Mon, 29 Aug 2011 22:02:45 +0200
Subject: big update: working iterating, pretty output, test scripts and more
---
dtrain/test/wc_pipes/wordcount.cc | 38 ++++++++++++++++++++++++++++++++++++++
1 file changed, 38 insertions(+)
create mode 100644 dtrain/test/wc_pipes/wordcount.cc
(limited to 'dtrain/test/wc_pipes/wordcount.cc')
diff --git a/dtrain/test/wc_pipes/wordcount.cc b/dtrain/test/wc_pipes/wordcount.cc
new file mode 100644
index 00000000..39560a31
--- /dev/null
+++ b/dtrain/test/wc_pipes/wordcount.cc
@@ -0,0 +1,38 @@
+#include "wordcount.hh"
+
+
+void
+WordcountMapper::map(HadoopPipes::MapContext & context)
+{
+ typedef boost::tokenizer<> tokenizer_t;
+ tokenizer_t tokenizer(context.getInputValue());
+
+ for( tokenizer_t::const_iterator i = tokenizer.begin();
+ tokenizer.end() != i; ++i ) {
+ context.emit(boost::to_lower_copy(*i), "1");
+ }
+}
+
+void
+WordcountReducer::reduce(HadoopPipes::ReduceContext & context)
+{
+ uint32_t count( 0 );
+
+ do {
+ ++count;
+ } while( context.nextValue() );
+
+ std::cout << context.getInputKey() << endl;
+ context.emit( context.getInputKey(),
+ boost::lexical_cast