create baseline tests for xml dump processing

[mediawiki_dump_tools.git] / test / Wikiq_Unit_Test.py

diff --git a/test/Wikiq_Unit_Test.py b/test/Wikiq_Unit_Test.py
new file mode 100644 (file)
index 0000000..3a1e24b
--- /dev/null
+++ b/test/Wikiq_Unit_Test.py
@@ -0,0 +1,108 @@
+import unittest
+import os
+import sys
+from shutil import copyfile
+
+# with / without pwr DONE
+# with / without url encode DONE
+# with / without collapse user DONE
+# with output to sdtout
+# note that the persistence radius is 7 by default
+# reading various file formats including
+#        7z, gz, bz2, xml 
+# wikia and wikipedia data
+# malformed xmls
+
+class Test_Basic(unittest.TestCase):
+
+    def setUp(self):
+        if not os.path.exists("test_output"):
+            os.mkdir("test_output")
+
+        self.wiki = 'sailormoon'
+        self.wikiq_out_name =  self.wiki + ".tsv"
+        self.test_output_dir = os.path.join(".", "test_output")
+        self.call_output = os.path.join(self.test_output_dir, self.wikiq_out_name)
+
+        self.infile = "{0}.xml.7z".format(self.wiki)
+        self.base_call = "../wikiq {0} -o {1}"
+        self.input_dir = "dumps"
+        self.input_file = os.path.join(".", self.input_dir,self.infile)
+        self.baseline_output_dir = "baseline_output"
+
+    def test_noargs(self):
+
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        os.system(call)
+
+        test_file =  "noargs_" + self.wikiq_out_name
+        copyfile(self.call_output, os.path.join(self.test_output_dir, test_file))
+
+        baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
+
+        test_lines = open(os.path.join(self.test_output_dir,test_file))
+        baseline_lines = open(baseline_file)
+        for test, baseline in zip(test_lines, baseline_lines):
+            self.assertEqual(test,baseline)
+
+        test_lines.close()
+        baseline_lines.close()
+
+    def test_collapse_user(self):
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        os.system(call)
+        call = call + " --collapse-user"
+
+        os.system(call)
+
+        test_file =  "collapse-user_" + self.wikiq_out_name
+        copyfile(self.call_output, os.path.join(self.test_output_dir, test_file))
+
+        baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
+
+        test_lines = open(os.path.join(self.test_output_dir,test_file))
+        baseline_lines = open(baseline_file)
+        for test, baseline in zip(test_lines, baseline_lines):
+            self.assertEqual(test,baseline)
+
+        test_lines.close()
+        baseline_lines.close()
+
+    def test_pwr(self):
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        call = call + " --persistence"
+        os.system(call)
+
+        test_file =  "persistence_" + self.wikiq_out_name
+        copyfile(self.call_output, os.path.join(self.test_output_dir, test_file))
+
+        baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
+
+        test_lines = open(os.path.join(self.test_output_dir,test_file))
+        baseline_lines = open(baseline_file)
+        for test, baseline in zip(test_lines, baseline_lines):
+            self.assertEqual(test,baseline)
+
+        test_lines.close()
+        baseline_lines.close()
+
+    def test_url_encode(self):
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        call = call + " --url-encode"
+        os.system(call)
+        test_file =  "url-encode_" + self.wikiq_out_name
+        copyfile(self.call_output, os.path.join(self.test_output_dir, test_file))
+        baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
+
+        test_lines = open(os.path.join(self.test_output_dir,test_file))
+        baseline_lines = open(baseline_file)
+        for test, baseline in zip(test_lines, baseline_lines):
+            self.assertEqual(test,baseline)
+
+        test_lines.close()
+        baseline_lines.close()
+
+
+        
+if __name__ == '__main__':
+    unittest.main()