reworks backup strategy (#67)

phylofisher/tools.py

@@ -192,7 +192,6 @@ def first_backup(dfo, date_time):
     os.mkdir(f'{dfo}/backups/{date_time}')
     shutil.copytree(f'{dfo}/orthologs', f'{dfo}/backups/{date_time}/orthologs')
     shutil.copytree(f'{dfo}/paralogs', f'{dfo}/backups/{date_time}/paralogs')
-    shutil.copytree(f'{dfo}/proteomes', f'{dfo}/backups/{date_time}/proteomes')
     shutil.copy(f'{dfo}/metadata.tsv', f'{dfo}/backups/{date_time}/metadata.tsv')
     shutil.copy(f'{dfo}/tree_colors.tsv', f'{dfo}/backups/{date_time}/tree_colors.tsv')
 
@@ -210,13 +209,12 @@ def backup(dfo):
         os.mkdir(f'{dfo}/backups/{now}')
         os.mkdir(f'{dfo}/backups/{now}/orthologs')
         os.mkdir(f'{dfo}/backups/{now}/paralogs')
-        os.mkdir(f'{dfo}/backups/{now}/proteomes')
 
         for root, dirs, files in os.walk(f'{dfo}/backups'):
             for file in files:
                 back_files[file] = (os.path.join(root, file))
 
-        for folder in ['orthologs', 'paralogs', 'proteomes']:
+        for folder in ['orthologs', 'paralogs']:
             for root, dirs, files in os.walk(f'{dfo}/{folder}'):
                 for file in files:
                     db_files[file] = (os.path.join(root, file))
@@ -230,10 +228,8 @@ def backup(dfo):
             else:
                 dest = '/'.join(v.split('/')[-2:])
 
-            if k in back_files.keys() and get_md5(db_files[k]) == get_md5(back_files[k]):
-                os.symlink(f'{dfo}/backups/{latest_backup}/{dest}', f'{dfo}/backups/{now}/{dest}')
-            else:
-                shutil.copy(v, f'{dfo}/backups/{now}/{dest}')
+
+            shutil.copy(v, f'{dfo}/backups/{now}/{dest}')
 
     else:
         first_backup(dfo, now)

phylofisher/utilities/backup_restoration.py

@@ -45,9 +45,6 @@ def restore():
         os.remove(f'{args.database}/tree_colors.tsv')
     shutil.copy(f'{backup_dir}/tree_colors.tsv', f'{args.database}/tree_colors.tsv')
 
-    if os.path.isdir(f'{args.database}/proteomes'):
-        shutil.rmtree(f'{args.database}/proteomes')
-    shutil.copytree(f'{backup_dir}/proteomes', f'{args.database}/proteomes')
 
 
 if __name__ == '__main__':

phylofisher/utilities/purge.py

@@ -90,21 +90,6 @@ def delete_homologs(org_set):
             fasta_cleaner(file, org_set)
 
 
-def delete_proteomes(org_set, collapsed_taxa):
-    '''
-    Purges input proteomes
-    Does not purge collapsed taxa, because they do not have a input proteomes
-
-    :param org_set: organisms to remove
-    :type org_set: set
-    :param collapsed_taxa: collapsed taxa
-    :type collapsed_taxa: list
-    '''
-    for org in org_set:
-        if org not in collapsed_taxa:
-            os.remove(os.path.join(dfo, 'proteomes', f'{org}.faa.tar.gz'))
-
-
 def purge(collapsed_taxa):
     '''
     Purges taxa from database
@@ -131,7 +116,6 @@ def purge(collapsed_taxa):
                 res.writerow(line)
 
     delete_homologs(orgs_to_del)
-    delete_proteomes(orgs_to_del, collapsed_taxa)
 
 
 if __name__ == '__main__':