CLOCBear: Add CLOCBear

This bear gives statistics about number of lines,
number of comment lines, and total number of lines.

Closes coala#1577

.ci/deps.sh

@@ -13,7 +13,7 @@ esac
 # apt-get commands
 export DEBIAN_FRONTEND=noninteractive
 
-deps="libclang1-3.4 indent mono-mcs chktex r-base julia golang-go luarocks verilator cppcheck flawfinder devscripts"
+deps="libclang1-3.4 indent mono-mcs chktex r-base julia golang-go luarocks verilator cppcheck flawfinder devscripts cloc"
 deps_infer="m4 opam"
 
 case $CIRCLE_BUILD_IMAGE in

.travis.yml

@@ -43,6 +43,7 @@ addons:
       - opam
       - php-codesniffer
       - verilator
+      - cloc
 
 cache:
   pip: true

bears/general/CLOCBear.py

@@ -0,0 +1,65 @@
+import re
+
+from coalib.bearlib.abstractions.Linter import linter
+from coalib.results.Result import Result
+from coalib.results.RESULT_SEVERITY import RESULT_SEVERITY
+from dependency_management.requirements.DistributionRequirement import (
+    DistributionRequirement)
+
+
+@linter(executable='cloc')
+class CLOCBear:
+    """
+    Summarises a file or number of files in directory structure
+    with total lines, number of comment lines, number of actual code lines
+    using CLI tool cloc
+    """
+    LANGUAGES = {'All'}
+    REQUIREMENTS = {DistributionRequirement(apt_get='cloc')}
+    AUTHORS = {'The coala developers'}
+    AUTHORS_EMAILS = {'[email protected]'}
+    LICENSE = 'AGPL-3.0'
+    CAN_DETECT = {'Number of comment lines',
+                  'Number of blank lines',
+                  'Number of code lines'}
+
+    @staticmethod
+    def create_arguments(filename, file, config_file):
+        return ('--csv', filename)
+
+    def process_output(self, output, filename, file):
+        lines = output.split('\n')
+        files, ignored_files = re.search(
+            r'(\d+).*?\n\s+(\d+)', '\n'.join(lines[1:3]), re.MULTILINE)\
+            .group(1, 2)
+        if(files == ignored_files):
+            msg = 'File does not belong to valid programming language.'
+            yield Result.from_values(origin=self,
+                                     message=msg,
+                                     file=filename,
+                                     severity=RESULT_SEVERITY.INFO)
+        else:
+            regex = re.compile(r'(\d+),(\S+),(\d+),(\d+),(\d+)')
+            # cloc tool gives summary of files
+            # starting from 5th(0-index) line up to second last line
+            for row in lines[5:-1]:
+                match = regex.match(row)
+
+                nfiles, lang = match.group(1, 2)
+                blank, comment, code = map(int, match.group(3, 4, 5))
+                total = blank + comment + code
+
+                report = '\n'.join(['Language: {0}'.format(lang),
+                                    'Total files: {0}'.format(nfiles),
+                                    'Total lines: {0}'.format(total),
+                                    'Code lines: {0} ({1:.2f}%)'.format(
+                                        code, code * 100.0 / total),
+                                    'Comment lines: {0} ({1:.2f}%)'.format(
+                                        comment, comment * 100.0 / total),
+                                    'Blank lines: {0} ({1:.2f}%)'.format(
+                                        blank, blank * 100.0 / total)
+                                    ])
+                yield Result.from_values(origin=self,
+                                         message=report,
+                                         file=filename,
+                                         severity=RESULT_SEVERITY.INFO)

tests/general/CLOCBearTest.py

@@ -0,0 +1,94 @@
+import os
+from queue import Queue
+
+from coalib.settings.Section import Section
+from coalib.testing.LocalBearTestHelper import LocalBearTestHelper
+from coalib.testing.BearTestHelper import generate_skip_decorator
+from coalib.results.RESULT_SEVERITY import RESULT_SEVERITY
+from coalib.results.Result import Result
+
+from bears.general.CLOCBear import CLOCBear
+
+
+def get_absolute_test_path(file):
+    return os.path.join(os.path.dirname(__file__),
+                        'cloc_test_files', file)
+
+
+def load_testfile(name):
+    with open(get_absolute_test_path(name)) as file:
+        return file.readlines()
+
+
+@generate_skip_decorator(CLOCBear)
+class CLOCBearTest(LocalBearTestHelper):
+
+    def setUp(self):
+        self.uut = CLOCBear(Section('name'), Queue())
+        self.test_files = ['example1.cpp', 'example2.py',
+                           'example3.cpp', 'example4.txt']
+        self.expected_results = {self.test_files[0]: {'LANGUAGE': 'C++',
+                                                      'FILES': 1,
+                                                      'CODE': 6,
+                                                      'COMMENTS': 5,
+                                                      'BLANK': 3},
+                                 self.test_files[1]: {'LANGUAGE': 'Python',
+                                                      'FILES': 1,
+                                                      'CODE': 3,
+                                                      'COMMENTS': 2,
+                                                      'BLANK': 2},
+                                 self.test_files[2]: {'LANGUAGE': 'C++',
+                                                      'FILES': 1,
+                                                      'CODE': 11,
+                                                      'COMMENTS': 0,
+                                                      'BLANK': 3},
+                                 self.test_files[3]: 'File does not belong '
+                                                     'to valid programming '
+                                                     'language.'
+                                 }
+
+    def build_message(self, filename):
+        result = self.expected_results[filename]
+
+        lang = result['LANGUAGE']
+        nfiles = result['FILES']
+        code = result['CODE']
+        comment = result['COMMENTS']
+        blank = result['BLANK']
+        total = code + comment + blank
+
+        message = '\n'.join(['Language: {0}'.format(lang),
+                             'Total files: {0}'.format(nfiles),
+                             'Total lines: {0}'.format(total),
+                             'Code lines: {0} ({1:.2f}%)'.format(
+                               code, code * 100.0 / total),
+                             'Comment lines: {0} ({1:.2f}%)'.format(
+                               comment, comment * 100.0 / total),
+                             'Blank lines: {0} ({1:.2f}%)'.format(
+                               blank, blank * 100.0 / total)
+                             ])
+        return message
+
+    def test_valid(self):
+        for filename in self.test_files[:-1]:
+            file_contents = load_testfile(filename)
+            self.check_results(
+                self.uut,
+                file_contents,
+                [Result.from_values('CLOCBear',
+                                    self.build_message(filename),
+                                    file=get_absolute_test_path(filename),
+                                    severity=RESULT_SEVERITY.INFO)],
+                filename=get_absolute_test_path(filename))
+
+    def test_invalid(self):
+        filename = self.test_files[3]
+        file_contents = load_testfile(filename)
+        self.check_results(
+            self.uut,
+            file_contents,
+            [Result.from_values('CLOCBear',
+                                str(self.expected_results[filename]),
+                                file=get_absolute_test_path(filename),
+                                severity=RESULT_SEVERITY.INFO)],
+            filename=get_absolute_test_path(filename))

tests/general/cloc_test_files/example1.cpp

@@ -0,0 +1,14 @@
+// This comments are for testing purpose.
+
+/*
+Just like any other programmer,
+This example file will also try to print only two words
+*/
+
+#include <cstdio>
+
+int main()
+{
+    std::cout << 'Hello World!' << std::endl;
+    return 0;
+}

tests/general/cloc_test_files/example2.py

@@ -0,0 +1,7 @@
+# This comments are for testing purpose
+# This program will just try to add two static numbers
+
+first = 5  # First number
+second = 10  # Second number
+
+sum = first + second

tests/general/cloc_test_files/example3.cpp

@@ -0,0 +1,14 @@
+#include <cstdio>
+
+#define PI 3.1415926
+using namespace std;
+
+
+int main()
+{
+    const double radii = 1.0;
+    const double area = PI * radii * radii;
+    cout << 'Area for the circle with radii ' << radii <<
+            ' is ' << area << endl;
+    return 0;
+}

tests/general/cloc_test_files/example4.txt

@@ -0,0 +1,2 @@
+This is normal text file. And it is not specific to any programming language.
+So, this has to detected as invalid file by cloc.