From c947d87ab9e9510e322d8cb030f42b64f6f07dc4 Mon Sep 17 00:00:00 2001
From: Emanuel Guevel <guevel.emanuel@gmail.com>
Date: Fri, 4 Jan 2013 15:10:30 +0100
Subject: [PATCH] Add a test for to_utf8 component

---
 components/to_utf8/tests/.gitignore           |  1 +
 .../to_utf8/tests/output/to_utf8_test.out     |  4 ++
 components/to_utf8/tests/test.sh              | 18 ++++++
 components/to_utf8/tests/to_utf8_test.cpp     | 61 +++++++++++++++++++
 4 files changed, 84 insertions(+)
 create mode 100644 components/to_utf8/tests/.gitignore
 create mode 100644 components/to_utf8/tests/output/to_utf8_test.out
 create mode 100755 components/to_utf8/tests/test.sh
 create mode 100644 components/to_utf8/tests/to_utf8_test.cpp

diff --git a/components/to_utf8/tests/.gitignore b/components/to_utf8/tests/.gitignore
new file mode 100644
index 000000000..814490404
--- /dev/null
+++ b/components/to_utf8/tests/.gitignore
@@ -0,0 +1 @@
+*_test
diff --git a/components/to_utf8/tests/output/to_utf8_test.out b/components/to_utf8/tests/output/to_utf8_test.out
new file mode 100644
index 000000000..dcb32359a
--- /dev/null
+++ b/components/to_utf8/tests/output/to_utf8_test.out
@@ -0,0 +1,4 @@
+original:  Без вопросов отдаете ему рулет, зная, что позже вы сможете привести с собой своих друзей и тогда он получит по заслугам?
+converted: Без вопросов отдаете ему рулет, зная, что позже вы сможете привести с собой своих друзей и тогда он получит по заслугам?
+original:  Vous lui donnez le gâteau sans protester avant d’aller chercher tous vos amis et de revenir vous venger.
+converted: Vous lui donnez le gâteau sans protester avant d’aller chercher tous vos amis et de revenir vous venger.
diff --git a/components/to_utf8/tests/test.sh b/components/to_utf8/tests/test.sh
new file mode 100755
index 000000000..2d07708ad
--- /dev/null
+++ b/components/to_utf8/tests/test.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+
+make || exit
+
+mkdir -p output
+
+PROGS=*_test
+
+for a in $PROGS; do
+    if [ -f "output/$a.out" ]; then
+        echo "Running $a:"
+        ./$a | diff output/$a.out -
+    else
+        echo "Creating $a.out"
+        ./$a > "output/$a.out"
+        git add "output/$a.out"
+    fi
+done
diff --git a/components/to_utf8/tests/to_utf8_test.cpp b/components/to_utf8/tests/to_utf8_test.cpp
new file mode 100644
index 000000000..8c25c483e
--- /dev/null
+++ b/components/to_utf8/tests/to_utf8_test.cpp
@@ -0,0 +1,61 @@
+#include <iostream>
+#include <fstream>
+#include <cassert>
+#include <stdexcept>
+#include <typeinfo>
+
+#include "../to_utf8.hpp"
+
+std::string getFirstLine(const std::string &filename);
+void testEncoder(ToUTF8::FromType encoding, const std::string &legacyEncFile,
+                 const std::string &utf8File);
+
+/// Test character encoding conversion to and from UTF-8
+void testEncoder(ToUTF8::FromType encoding, const std::string &legacyEncFile,
+                 const std::string &utf8File)
+{
+    // get some test data
+    std::string legacyEncLine = getFirstLine(legacyEncFile);
+    std::string utf8Line = getFirstLine(utf8File);
+
+    // create an encoder for specified character encoding
+    ToUTF8::Utf8Encoder encoder;
+    encoder.setEncoding(encoding);
+
+    // convert text to UTF-8
+    std::string convertedUtf8Line = encoder.getUtf8(legacyEncLine);
+
+    std::cout << "original:  " << utf8Line          << std::endl;
+    std::cout << "converted: " << convertedUtf8Line << std::endl;
+
+    // check correctness
+    assert(convertedUtf8Line == utf8Line);
+
+    // convert UTF-8 text to legacy encoding
+    std::string convertedLegacyEncLine = encoder.getLegacyEnc(utf8Line);
+    // check correctness
+    assert(convertedLegacyEncLine == legacyEncLine);
+}
+
+std::string getFirstLine(const std::string &filename)
+{
+    std::string line;
+    std::ifstream text (filename.c_str());
+
+    if (!text.is_open())
+    {
+        throw std::runtime_error("Unable to open file " + filename);
+    }
+
+    std::getline(text, line);
+    text.close();
+
+    return line;
+}
+
+int main()
+{
+    testEncoder(ToUTF8::WINDOWS_1251, "data/russian-win1251.txt", "data/russian-utf8.txt");
+    testEncoder(ToUTF8::WINDOWS_1252, "data/french-win1252.txt", "data/french-utf8.txt");
+    return 0;
+}