DuplicateFileBear: Add DuplicateFileBear

This bear checks for files with same content .

Closes #1479

bears/general/DuplicateFileBear.py

@@ -0,0 +1,38 @@
+import itertools
+
+from coalib.bears.GlobalBear import GlobalBear
+from coalib.results.Result import Result
+from coalib.results.RESULT_SEVERITY import RESULT_SEVERITY
+
+
+class DuplicateFileBear(GlobalBear):
+    LANGUAGES = {'All'}
+    AUTHORS = {'The coala developers'}
+    AUTHORS_EMAILS = {'[email protected]'}
+    LICENSE = 'AGPL-3.0'
+    CAN_DETECT = {'Duplication'}
+
+    def run(self):
+        """
+        Checks for Duplicate Files
+        """
+        if not self.file_dict:
+            yield Result(self, 'You did not add any file to compare',
+                         severity=RESULT_SEVERITY.MAJOR)
+        elif len(self.file_dict) == 1:
+            yield Result(self, 'You included only one file',
+                         severity=RESULT_SEVERITY.MAJOR)
+        else:
+            unique_tuples = [file_tuple for file_tuple in
+                             itertools.combinations(self.file_dict, 2)]
+
+            for file_pair in unique_tuples:
+                if (self.file_dict[file_pair[0]] ==
+                        self.file_dict[file_pair[1]]):
+                    first_file_name = file_pair[0]
+                    second_file_name = file_pair[1]
+                    message = ('File ' + first_file_name + ' is identical'
+                               ' to File ' + second_file_name)
+                    yield Result.from_values(origin=self, message=message,
+                                             severity=RESULT_SEVERITY.INFO,
+                                             file=first_file_name)

tests/general/DuplicatefileBearTest.py

@@ -0,0 +1,65 @@
+import unittest
+import os
+
+from coalib.settings.Section import Section
+from coalib.results.RESULT_SEVERITY import RESULT_SEVERITY
+from bears.general.DuplicateFileBear import DuplicateFileBear
+from queue import Queue
+
+
+def get_absolute_test_path(file):
+    return os.path.join(os.path.dirname(__file__),
+                        'duplicate_test_files', file)
+
+
+class DuplicateFileBearTest(unittest.TestCase):
+
+    def setUp(self):
+        self.section = Section('name')
+        self.queue = Queue()
+        self.file_dict = {}
+        self.test_files = ['complexFirst.txt', 'complexSecond.txt',
+                           'noMatch.txt', 'smallFirst.txt',
+                           'smallSecond.txt']
+
+    def get_results(self, files_to_check):
+        self.files = [get_absolute_test_path(file) for file in files_to_check]
+        for filename in self.files:
+            with open(filename, 'r', encoding='utf-8') as _file:
+                self.file_dict[filename] = tuple(_file.readlines())
+        self.maxDiff = None
+        self.uut = DuplicateFileBear(self.file_dict, self.section,
+                                     self.queue)
+        return list(self.uut.run())
+
+    def test_results_complete(self):
+        results = self.get_results(self.test_files)
+        messages = [result.message for result in results]
+        combined = '\t'.join(messages)
+        self.assertIn(get_absolute_test_path(
+            'complexSecond.txt'), combined.split())
+        self.assertIn(get_absolute_test_path(
+            'complexFirst.txt'), combined.split())
+        self.assertIn(get_absolute_test_path(
+            'smallFirst.txt'), combined.split())
+        self.assertIn(get_absolute_test_path(
+            'smallSecond.txt'), combined.split())
+        self.assertEquals(results[0].severity, RESULT_SEVERITY.INFO)
+
+    def test_results_no_duplicates(self):
+        results = self.get_results([self.test_files[2],
+                                    self.test_files[3]])
+        messages = [result.message for result in results]
+        self.assertEquals(messages, [])
+
+    def test_results_empty(self):
+        results = self.get_results([])
+        messages = [result.message for result in results]
+        self.assertEquals(messages, ['You did not add any file to compare'])
+        self.assertEquals(results[0].severity, RESULT_SEVERITY.MAJOR)
+
+    def test_result_single(self):
+        results = self.get_results([self.test_files[0]])
+        messages = [result.message for result in results]
+        self.assertEquals(messages, ['You included only one file'])
+        self.assertEquals(results[0].severity, RESULT_SEVERITY.MAJOR)

tests/general/duplicate_test_files/complexFirst.txt

@@ -0,0 +1 @@
+4 r34l ch4ll3n63: 123 ÄÖü ABc @€¥ §&% {[( ←↓→↑ ĦŊħ ß°^ \\\n\u2192

tests/general/duplicate_test_files/complexSecond.txt

@@ -0,0 +1 @@
+4 r34l ch4ll3n63: 123 ÄÖü ABc @€¥ §&% {[( ←↓→↑ ĦŊħ ß°^ \\\n\u2192

tests/general/duplicate_test_files/noMatch.txt

@@ -0,0 +1 @@
+I dont match with any other file!

tests/general/duplicate_test_files/smallFirst.txt

@@ -0,0 +1 @@
+This is a small file!

tests/general/duplicate_test_files/smallSecond.txt

@@ -0,0 +1 @@
+This is a small file!