A few fixes

gitstats/data_generators/gen_loc_data.py

@@ -17,7 +17,7 @@ def gen_loc_data(conf, row_processor):
 
     :param conf: configuration (mostly used for date limits)
     :param row_processor: function to receive the callback
-    :return: total lines in repo
+    :return: (total_files, total_lines) in repo
     '''
 
     # line statistics
@@ -64,7 +64,20 @@ def gen_loc_data(conf, row_processor):
             else:
                 logging.warning(f'Failed to handle line "{line}"')
                 (files, inserted, deleted) = (0, 0, 0)
-    return total_lines
+
+    totals = getpipeoutput(
+        ['git diff --shortstat `git hash-object -t tree /dev/null`']
+    )
+    total_files, remainder = totals.strip().split(' file')
+    total_lines = remainder.split('hanged, ')[1].split(' insert')[0]
+    try:
+        total_files = int(total_files)
+        total_lines = int(total_lines)
+    except ValueError:
+        total_lines = 0
+        total_files = 0
+        logging.warning(f"Unable to parse results: {totals}")
+    return total_files, total_lines
 
 
 if __name__ == "__main__":

gitstats/git_csv_generator.py

@@ -43,6 +43,10 @@ class _FileHandles:
         self.loc_info_writer.writerow(['Repo', 'CommitHash', 'TimeStamp', 'FileCount', 'LinesInserted', 'LinesDeleted',
                                        'TotalLines'])
 
+        self.repo_info = open(os.path.join(output_dir, 'repo.csv'), 'w', encoding='utf8')
+        self.repo_info_writer = csv.writer(self.repo_info)
+        self.repo_info_writer.writerow(['Repo', 'TotalFiles', 'TotalLines'])
+
     def close(self):
         self.author_info.close()
         self.author_totals_info.close()
@@ -50,6 +54,7 @@ class _FileHandles:
         self.revision_info.close()
         self.files_info.close()
         self.loc_info.close()
+        self.repo_info.close()
 
 
 class GitCsvGenerator():
@@ -108,7 +113,8 @@ class GitCsvGenerator():
         def row_processor(row: LocByDate):
             self.files.loc_info_writer.writerow([self.projectname, row.hash, row.stamp, row.file_count,
                                                  row.lines_inserted, row.lines_deleted, row.total_lines])
-        gen_loc_data(self.conf, row_processor)
+        total_files, total_lines = gen_loc_data(self.conf, row_processor)
+        self.files.repo_info_writer.writerow([self.projectname, total_files, total_lines])
 
     def get_author_info(self):
         logging.info(f"Getting author info for {self.projectname}")

gitstats/gitdatacollector.py

@@ -62,7 +62,7 @@ class GitDataCollector(DataCollector):
             self.total_lines_added += row.lines_inserted
             self.total_lines_removed += row.lines_deleted
 
-        self.total_lines += gen_loc_data(self.conf, row_processor)
+        self.total_lines += gen_loc_data(self.conf, row_processor)[1]
 
     def get_file_info(self):
         # extensions and size of files