updated readme

README.md

@@ -5,6 +5,10 @@
 
 -----
 
+# Note
+
+## This repo is underdevelopment and not ready for routine usage
+
 **Table of Contents**
 
 - [Installation](#installation)

pyproject.toml

@@ -7,7 +7,7 @@ name = "mikrokondo-tools"
 dynamic = ["version"]
 description = ''
 readme = "README.md"
-requires-python = ">=3.7"
+requires-python = ">=3.8"
 license = "MIT"
 keywords = []
 authors = [

src/mikrokondo_tools/cli/__init__.py

@@ -4,8 +4,8 @@
 import click
 
 from mikrokondo_tools.__about__ import __version__
-from mikrokondo_tools.cli.format import fmt
 from mikrokondo_tools.cli.download import download
+from mikrokondo_tools.cli.format import fmt
 
 
 @click.group(context_settings={"help_option_names": ["-h", "--help"]}, invoke_without_command=True, no_args_is_help=True)

src/mikrokondo_tools/cli/download/__init__.py

@@ -1,21 +1,21 @@
 import os
 
 import click
-from mikrokondo_tools.download import download as dwnld
 
+from mikrokondo_tools.download import download as dwnld
 
-options = {"gtdb-sketch": dwnld.Constants.sketch_url, 
+options = {"gtdb-sketch": dwnld.Constants.sketch_url,
         "dehost": dwnld.Constants.dehost_idx,
         "kraken-std": dwnld.Constants.kraken2_std,
         "bakta-light": dwnld.Constants.bakta_light,
         "bakta-full": dwnld.Constants.bakta_full}
 
 @click.command(short_help="Download a database for mikrokondo", no_args_is_help=True)
-@click.option("-f", "--file", "file", type=click.Choice(options.keys()), required = True, 
+@click.option("-f", "--file", "file", type=click.Choice(options.keys()), required = True,
             help="Pick an option from the list to download")
 @click.option("-o", "--output", "output", default=os.getcwd(), type=click.Path(), help="An existing directory to download files too.")
 def download(file, output):
     """Download a external file for use in mikrokondo. This script only downloads the file and will not untar or unzip them.
     """
     option = options[file]
-    dwnld.download_file(output_dir=output, url=option)
+    dwnld.download_file(output_dir=output, url=option)

src/mikrokondo_tools/download/download.py

@@ -3,14 +3,13 @@
 
 Matthew Wells: 2024-04-18
 """
-import requests
-
-import logging
 import itertools
-import sys
+import logging
 import os
+import sys
 from dataclasses import dataclass
 
+import requests
 
 logging.basicConfig(stream=sys.stderr, level=logging.INFO) # TODO this should probably go somewhere else
 logger = logging.getLogger(__name__)
@@ -34,7 +33,7 @@ def get_file_name(input_uri: str) -> str:
 
 def download_file(url: str, output_dir: str = None) -> str:
     """Download a file from zenodo and put it in the correct output director
-    
+
     output_dir str|os.Path: Output location different then local directory
     """
     output_file = get_file_name(url)

src/mikrokondo_tools/format/format.py

@@ -64,7 +64,7 @@ def denested_information(keys: list[str], last_value: dict) -> dict:
 
     temp[Constants.type_field] = Constants.properties_type
     temp[Constants.extraction_field] = {keys[-1]: last_value}
-    
+
     return new_chain
 
 
@@ -93,25 +93,25 @@ def nest_schema(properties: dict) -> dict:
         if Constants.delimiter not in key:
             continue
         split_key = key.split(Constants.delimiter)
-        
+
         if new_dict.get(split_key[0]) is None:
             new_dict[split_key[0]] = {}
             new_dict[split_key[0]][Constants.type_field] = Constants.properties_type
             new_dict[split_key[0]][Constants.extraction_field] = {}
-    
+
         denested_data = denested_information(split_key[1:], values)
         if denested_data != values:
             # multiple fields to be set, update the properties instead of overwriting it
-            
+
             temp = denested_data
-            
+
             nd_temp = new_dict[split_key[0]][Constants.extraction_field]
             for i in split_key[1:-1]:
                 if nd_temp.get(i) is None:
                     nd_temp[i] = {}
                     nd_temp[i][Constants.type_field] = temp[i][Constants.type_field]
                     nd_temp[i][Constants.extraction_field] = {}
-                
+
                 nd_temp = nd_temp[i][Constants.extraction_field]
                 temp = temp[i][Constants.extraction_field]
 
@@ -160,7 +160,7 @@ def nest_properties(schema: dict) -> dict:
         new_properties = nest_schema(properties=props[Constants.extraction_field])
         del schema[Constants.nesting_field][k][Constants.extraction_field]
         schema[Constants.nesting_field][k][Constants.extraction_field] = new_properties
-        
+
 
     new_properties = nest_schema(properties=properties)
     del schema[Constants.extraction_field]
@@ -187,7 +187,7 @@ def reorganize_schema(schema) -> set:
     definitions dict: Updated definitions field in a json schema
     return drop_keys set: Additional fields to delete from the schema after processing
     """
-    
+
 
     definitions = schema[Constants.nesting_field]
     top_lvl_keys = frozenset(definitions.keys())

tests/download/test_download.py

@@ -4,11 +4,12 @@
 
 import pytest
 import requests
+
 from mikrokondo_tools.download import download
 
 
 @pytest.mark.parametrize("uri",
-                        [ 
+                        [
                         (download.Constants.sketch_url),
                         (download.Constants.dehost_idx),
                         (download.Constants.kraken2_std),
@@ -29,7 +30,7 @@ def test_constants_connection(uri):
                         "PhiPacHum_m2.idx"),
                         ("https://genome-idx.s3.amazonaws.com/kraken/k2_standard_20240112.tar.gz",
                         "k2_standard_20240112.tar.gz"),
-                        ("https://zenodo.org/records/10522951/files/db-light.tar.gz", 
+                        ("https://zenodo.org/records/10522951/files/db-light.tar.gz",
                         "db-light.tar.gz"),
                         ("https://zenodo.org/records/10522951/files/db.tar.gz", "db.tar.gz"),
                         ("db.tar.gz", "db.tar.gz")

tests/format/test_format.py

@@ -62,19 +62,19 @@
                 "seqkit.args.single_end": {"type": "string"},
             },
             {
-            "seqkit": 
+            "seqkit":
                 {
                 "type": "object",
-                "properties": 
+                "properties":
                     {
                     "singularity":
                         {"type": "string"},
                     "docker":
                         {"type": "string"},
-                    "args": 
+                    "args":
                         {
                             "type": "object",
-                            "properties": 
+                            "properties":
                             {
                                 "illumina": {"type": "string"},
                                 "single_end": {"type": "string"},
@@ -93,8 +93,8 @@ def test_nest_schema(test_val, output):
 def test_read_json(real_schema):
     assert real_schema == format.read_json(conftest.SCHEMA_PATH)
 
-@pytest.mark.parametrize("input_,output", 
+@pytest.mark.parametrize("input_,output",
                         [("singularity", "#/definitions/singularity")
                         ])
 def test_create_all_of_ref(input_, output):
-    assert format.create_all_of_ref(input_) == output
+    assert format.create_all_of_ref(input_) == output