Improve the implementation of the AboutCollector, output_path is a params for the write_to_csv rather than on the class, #23

Author: tdruez

about_code_tool/about.py

@@ -893,12 +893,11 @@ def notice_text(self):
 
 
 class AboutCollector(object):
-    def __init__(self, input_path, output_path, verbosity):
+    def __init__(self, input_path, verbosity):
         # Setup the input and output paths
         self.original_input_path = input_path
         self.input_path = abspath(input_path)
         assert exists(self.input_path)
-        self.output_path = output_path
 
         # Setup the verbosity
         self.display_error = self.display_warning = False
@@ -909,10 +908,12 @@ def __init__(self, input_path, output_path, verbosity):
 
         self.about_files = []
         self.about_objects = []
+        self.about_data_list = []
 
         # Running the files collection and objects creation on instantiation
         self.collect_about_files()
         self.create_about_objects_from_files()
+        self.extract_about_data_from_objects()
 
     def collect_about_files(self):
         """
@@ -940,10 +941,13 @@ def create_about_objects_from_files(self):
 
         self.about_objects = about_objects
 
-    def extract_about_info(self):
+    def extract_about_data_from_objects(self):
         """
         Builds rows for each stored about objects.
         """
+        if self.about_data_list:  # Process only once.
+            return
+
         about_data_list = []
         warnings_count = errors_count = 0
 
@@ -982,21 +986,24 @@ def extract_about_info(self):
                     if about_object.warnings:
                         print("WARNING: %s\n" % about_object.warnings)
 
-        self.write_to_csv(about_data_list)
+        self.about_data_list = about_data_list
+
         if errors_count:
             print("%d errors detected." % errors_count)
         if warnings_count:
             print("%d warnings detected.\n" % warnings_count)
 
-    def write_to_csv(self, about_data_list):
+    def write_to_csv(self, output_path):
         """
         Write results in CSV file at output_path.
         """
-        with open(self.output_path, 'wb') as output_file:
+        header_row = ['about_file'] + MANDATORY_FIELDS + OPTIONAL_FIELDS + \
+                     ['warnings', 'errors']
+
+        with open(output_path, 'wb') as output_file:
             about_spec_writer = csv.writer(output_file)
-            about_spec_writer.writerow(['about_file'] + MANDATORY_FIELDS +
-                                       OPTIONAL_FIELDS + ['warnings', 'errors'])
-            for row in about_data_list:
+            about_spec_writer.writerow(header_row)
+            for row in self.about_data_list:
                 about_spec_writer.writerow(row)
 
     def generate_attribution(self, template_path='templates/default.html',
@@ -1105,8 +1112,8 @@ def main(parser, options, args):
         sys.exit(errno.EEXIST)
 
     if not exists(output_path) or (exists(output_path) and overwrite):
-        collector = AboutCollector(input_path, output_path, verbosity)
-        collector.extract_about_info()
+        collector = AboutCollector(input_path, verbosity)
+        collector.write_to_csv(output_path)
     else:
         # we should never reach this
         assert False, "Unsupported option(s)."

about_code_tool/genabout.py

@@ -31,7 +31,6 @@
 import os
 import shutil
 import sys
-import urllib2
 
 __version__ = '0.9.0'
 

about_code_tool/genattrib.py

@@ -16,7 +16,7 @@
 
 """
 This is a tool to generate component attribution based on a set of .ABOUT files.
-Optionally, one could pass a subset list of specific components for set of 
+Optionally, one could pass a subset list of specific components for set of
 .ABOUT files to generate attribution.
 """
 
@@ -135,7 +135,7 @@ def main(args, opts):
     input_path = args[0]
     output_path = args[1]
     component_subset_path = None if len(args) < 3 else args[2]
-    
+
     # TODO: need more path normalization (normpath, expanduser)
     # input_path = abspath(input_path)
     output_path = abspath(output_path)

about_code_tool/tests/test_about.py

@@ -51,8 +51,8 @@ def test_return_path_is_not_abspath_and_contains_subdirs_on_file(self):
             os.remove(test_output)
         except OSError:
             pass
-        collector = about.AboutCollector(test_input, test_output, 0)
-        collector.extract_about_info()
+        collector = about.AboutCollector(test_input, 0)
+        collector.write_to_csv(test_output)
         self.assertTrue(open(test_output).read().partition('\n')[2].startswith('about_code_tool/tests/testdata/thirdparty/django_snippets_2413.ABOUT'))
         os.remove(test_output)
 
@@ -64,8 +64,8 @@ def test_return_path_is_not_abspath_and_contains_subdirs_on_dir(self):
             os.remove(test_output)
         except OSError:
             pass
-        collector = about.AboutCollector(test_input, test_output, 0)
-        collector.extract_about_info()
+        collector = about.AboutCollector(test_input, 0)
+        collector.write_to_csv(test_output)
         self.assertTrue(open(test_output).read().partition('\n')[2].startswith('about_code_tool/tests/testdata/basic'))
         os.remove(test_output)
 
@@ -475,7 +475,7 @@ def test_remove_blank_lines_and_no_colon_fields(self):
 
     def test_generate_attribution(self):
         expected = u'version:2.4.3about_resource:httpd-2.4.3.tar.gzname:Apache HTTP Server'
-        about_collector = about.AboutCollector(join(TESTDATA_PATH, 'attrib/attrib.ABOUT'), None, 0)
+        about_collector = about.AboutCollector(join(TESTDATA_PATH, 'attrib/attrib.ABOUT'), 0)
         result = about_collector.generate_attribution(join(TESTDATA_PATH, 'attrib/test.template'))
         self.assertEqual(result, expected)