add namespace filter parameter

[mediawiki_dump_tools.git] / test / Wikiq_Unit_Test.py

diff --git a/test/Wikiq_Unit_Test.py b/test/Wikiq_Unit_Test.py
index ab0a53255ae553034a22c354ec69f86c6466ca45..91ac928c721f1fb788326069b52829b0823613f6 100644 (file)
--- a/test/Wikiq_Unit_Test.py
+++ b/test/Wikiq_Unit_Test.py
@@ -1,9 +1,10 @@
 import unittest
 import os
-import sys
 import subprocess
 from shutil import copyfile
-import pdb
+import pandas as pd
+from pandas.util.testing import assert_frame_equal
+from io import StringIO
 
 # with / without pwr DONE
 # with / without url encode DONE
@@ -45,13 +46,32 @@ class Test_Wikipedia(unittest.TestCase):
         copyfile(self.call_output, test_file)
         baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
 
-        test_lines = open(test_file)
-        baseline_lines = open(baseline_file)
-        for test, baseline in zip(test_lines, baseline_lines):
-            self.assertEqual(test,baseline)
+        # as a test let's make sure that we get equal data frames
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
 
-        test_lines.close()
-        baseline_lines.close()
+    def test_WP_namespaces(self):
+        print(os.path.abspath('.'))
+        test_filename =  "namespaces_" + self.wikiq_out_name
+        test_file = os.path.join(self.test_output_dir, test_filename)
+        if os.path.exists(test_file):
+            os.remove(test_file)
+        
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        call = call + " -n 0 -n 1"
+        print(call)
+        proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+        proc.wait()
+        copyfile(self.call_output, test_file)
+        baseline_file = os.path.join(os.path.abspath("."), self.baseline_output_dir, test_filename)
+
+        # as a test let's make sure that we get equal data frames
+        test = pd.read_table(test_file)
+        num_wrong_ns = sum(~ test.namespace.isin({0,1}))
+        self.assertEquals(num_wrong_ns, 0)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
 
 
 class Test_Basic(unittest.TestCase):
@@ -86,13 +106,10 @@ class Test_Basic(unittest.TestCase):
 
         baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
 
-        test_lines = open(test_file)
-        baseline_lines = open(baseline_file)
-        for test, baseline in zip(test_lines, baseline_lines):
-            self.assertEqual(test, baseline)
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
 
-        test_lines.close()
-        baseline_lines.close()
 
     def test_collapse_user(self):
         test_filename =  "collapse-user_" + self.wikiq_out_name
@@ -109,14 +126,29 @@ class Test_Basic(unittest.TestCase):
         copyfile(self.call_output, test_file)
 
         baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
+
+    def test_pwr_segment(self):
+        test_filename =  "persistence_segment_" + self.wikiq_out_name
+        test_file = os.path.join(self.test_output_dir, test_filename)
+        if os.path.exists(test_file):
+            os.remove(test_file)
+        
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        call = call + " --persistence segment"
+        proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+        proc.wait()
+
 
-        test_lines = open(test_file)
-        baseline_lines = open(baseline_file)
-        for test, baseline in zip(test_lines, baseline_lines):
-            self.assertEqual(test,baseline)
+        copyfile(self.call_output, test_file)
+
+        baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
 
-        test_lines.close()
-        baseline_lines.close()
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
 
     def test_pwr_legacy(self):
         test_filename =  "persistence_legacy_" + self.wikiq_out_name
@@ -125,7 +157,27 @@ class Test_Basic(unittest.TestCase):
             os.remove(test_file)
         
         call = self.base_call.format(self.input_file, self.test_output_dir)
-        call = call + " --persistence-legacy"
+        call = call + " --persistence legacy"
+        proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
+        proc.wait()
+
+
+        copyfile(self.call_output, test_file)
+
+        baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
+
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
+
+    def test_pwr(self):
+        test_filename =  "persistence_" + self.wikiq_out_name
+        test_file = os.path.join(self.test_output_dir, test_filename)
+        if os.path.exists(test_file): 
+           os.remove(test_file)
+        
+        call = self.base_call.format(self.input_file, self.test_output_dir)
+        call = call + " --persistence"
         proc = subprocess.Popen(call,stdout=subprocess.PIPE,shell=True)
         proc.wait()
 
@@ -134,13 +186,10 @@ class Test_Basic(unittest.TestCase):
 
         baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
 
-        test_lines = open(test_file)
-        baseline_lines = open(baseline_file)
-        for test, baseline in zip(test_lines, baseline_lines):
-            self.assertEqual(test,baseline)
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
 
-        test_lines.close()
-        baseline_lines.close()
 
     def test_url_encode(self):
         test_filename =  "url-encode_" + self.wikiq_out_name
@@ -156,17 +205,12 @@ class Test_Basic(unittest.TestCase):
 
         copyfile(self.call_output, test_file)
         baseline_file = os.path.join(".", self.baseline_output_dir, test_filename)
+        test = pd.read_table(test_file)
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
 
-        test_lines = open(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()
 
 class Test_Malformed(unittest.TestCase):
-
     def setUp(self):
         if not os.path.exists("test_output"):
             os.mkdir("test_output")
@@ -207,42 +251,14 @@ class Test_Stdout(unittest.TestCase):
 
         call = self.base_call.format(self.input_file)
         proc = subprocess.run(call,stdout=subprocess.PIPE,shell=True)
-        outs = proc.stdout.decode('utf-8')
-        
-        test_file = "noargs_" + self.wikiq_out_name
-        baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
-
-        test_lines = outs.splitlines(True)
-        baseline_lines = open(baseline_file)
-        for test, baseline in zip(test_lines, baseline_lines):
-            self.assertEqual(test,baseline)
+        outs = proc.stdout.decode("utf8")
 
-    def test_persistence(self):
-
-        call = self.base_call.format(self.input_file) + " --persistence"
-        proc = subprocess.run(call,stdout=subprocess.PIPE,shell=True)
-        outs = proc.stdout.decode('utf-8')
-        
-        test_file = "persistence_" + self.wikiq_out_name
+        test_file = "noargs_" + self.wikiq_out_name
         baseline_file = os.path.join(".", self.baseline_output_dir, test_file)
-
-        test_lines = outs.splitlines(True)
-        baseline_lines = open(baseline_file)
-        for test, baseline in zip(test_lines, baseline_lines):
-            self.assertEqual(test,baseline)
-
-        # 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()
-
+        print(baseline_file)
+        test = pd.read_table(StringIO(outs))
+        baseline = pd.read_table(baseline_file)
+        assert_frame_equal(test,baseline)
         
 if __name__ == '__main__':
     unittest.main()