From 8d28a8e13dbe602783a505adb1df375b0d65efe0 Mon Sep 17 00:00:00 2001
From: Feffernoose <fefferburbia@gmail.com>
Date: Sun, 6 Oct 2013 19:51:45 -0400
Subject: Split rawr-ebooks and rawr-gen

Also wrote README
---
 gen.cpp | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 gen.cpp

(limited to 'gen.cpp')

diff --git a/gen.cpp b/gen.cpp
new file mode 100644
index 0000000..dc73e0f
--- /dev/null
+++ b/gen.cpp
@@ -0,0 +1,48 @@
+#include <cstdio>
+#include <list>
+#include <map>
+#include "kgramstats.h"
+#include <ctime>
+#include <vector>
+#include <cstdlib>
+#include <fstream>
+#include <iostream>
+#include <unistd.h>
+#include <yaml-cpp/yaml.h>
+
+using namespace::std;
+
+int main(int argc, char** args)
+{
+	srand(time(NULL));
+	
+	YAML::Node config = YAML::LoadFile("config.yml");
+	
+	ifstream infile(config["corpus"].as<std::string>().c_str());
+	string corpus;
+	string line;
+	while (getline(infile, line))
+	{
+		corpus += " " + line;
+	}
+	
+    cout << "Preprocessing corpus..." << endl;
+	kgramstats* stats = new kgramstats(corpus, 5);
+    
+    cout << "Generating..." << endl;
+	for (;;)
+	{
+		vector<string> doc = stats->randomSentence(rand() % 35 + 15);
+		string hi;
+		for (vector<string>::iterator it = doc.begin(); it != doc.end(); ++it)
+		{
+			hi += *it + " ";
+		}
+
+		cout << hi << endl;
+		
+        getc(stdin);
+	}
+	
+	return 0;
+}
\ No newline at end of file
-- 
cgit 1.4.1