Fixing bugs

Author: Roberto Vera Alvarez

.github/workflows/python-app.yml

@@ -24,11 +24,5 @@ jobs:
       run: pip install poetry
     - name: Install dependencies
       run: poetry install
-    - name: Install Flake
-      run: pip install flake8
-    - name: Lint with flake8
-      run: |
-        poetry run flake8 src/ --count --select=E9,F63,F7,F82 --show-source --statistics
-        poetry run flake8 src/ --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
-    - name: Test with pytest
-      run: poetry run pytest
+    - name: Test with tox
+      run: poetry run tox

poetry.lock

@@ -37,6 +37,11 @@ python = ">=3.9,<4.0"
 requests = "*"
 PyYAML = "*"
 
+[tool.poetry.group.dev.dependencies]
+pytest = "*"
+pytest-cov = "*"
+flake8 = "*"
+
 [tool.poetry.scripts]
 bioconda2biocontainer = "bioconda2biocontainer.entry_point:main"
 biocontainers-search = "bioconda2biocontainer.entry_point_search:main"

pyproject.toml

@@ -1,5 +1,5 @@
 import json
-
+from typing import Optional, Union
 import requests
 
 base_url = "http://api.biocontainers.pro/ga4gh/trs/v2/tools"
@@ -14,7 +14,7 @@ def request_url_to_dict(url):
 
 def find_package_by_name(package_name):
     data = request_url_to_dict('{0}/{1}'.format(base_url, package_name))
-    if type(data) == int:
+    if isinstance(data, int):
         return data
     if data and 'versions' in data:
         versions = data['versions']
@@ -44,32 +44,62 @@ def filter_by_container_registry(container_type, registry_host, i):
     return False
 
 
-def find_latest_image(package_name, package_version, all=False, sort_by_size=False,
-                      sort_by_download=False, container_type=None, registry_host=None):
-    data = request_url_to_dict('{0}/{1}/versions/{1}-{2}'.format(
-        base_url, package_name, package_version))
-    if type(data) == dict or type(data) == list:
-        versions = []
-        if data and 'images' in data:
-            for i in data['images']:
-                if filter_by_container_registry(container_type, registry_host, i):
-                    if 'downloads' not in i:
-                        i['downloads'] = 0
-                    if 'updated' not in i:
-                        i['updated'] = ''
-                    if 'size' not in i:
-                        i['size'] = 0
-                    versions.append(i)
-        if sort_by_size:
-            versions = sorted(versions, key=lambda i: i['size'])
-        elif sort_by_download:
-            versions = sorted(versions, key=lambda i: i['downloads'], reverse=True)
-        else:
-            versions = sorted(versions, key=lambda i: i['updated'], reverse=True)
-        if all:
-            return versions
-        if versions:
-            return versions[0]
-        else:
-            return None
-    return data
+def _normalize_image_fields(image: dict) -> dict:
+    """Ensure required fields exist in the image dict."""
+    image.setdefault("downloads", 0)
+    image.setdefault("updated", "")
+    image.setdefault("size", 0)
+    return image
+
+
+def _sort_images(images: list[dict], sort_by_size: bool, sort_by_download: bool) -> list[dict]:
+    """Sort images based on size, downloads, or updated date."""
+    if sort_by_size:
+        return sorted(images, key=lambda i: i["size"])
+    if sort_by_download:
+        return sorted(images, key=lambda i: i["downloads"], reverse=True)
+    return sorted(images, key=lambda i: i["updated"], reverse=True)
+
+
+def find_latest_image(
+        package_name: str,
+        package_version: str,
+        all: bool = False,
+        sort_by_size: bool = False,
+        sort_by_download: bool = False,
+        container_type: Optional[str] = None,
+        registry_host: Optional[str] = None,
+) -> Union[dict, list[dict], int, None]:
+    """
+    Retrieve the latest container image for a given package version.
+
+    Args:
+        package_name: Bioconda package name.
+        package_version: Specific version string.
+        all: If True, return all matching images instead of only the latest.
+        sort_by_size: Sort by image size ascending.
+        sort_by_download: Sort by download count descending.
+        container_type: Optional filter by container type (e.g., 'Docker').
+        registry_host: Optional filter by registry host.
+
+    Returns:
+        The latest image dict, a list of image dicts if all=True,
+        an HTTP status code (int) on error, or None if no images found.
+    """
+    url = f"{base_url}/{package_name}/versions/{package_name}-{package_version}"
+    data = request_url_to_dict(url)
+
+    if not isinstance(data, dict) or "images" not in data:
+        return data
+
+    images = [
+        _normalize_image_fields(i)
+        for i in data["images"]
+        if filter_by_container_registry(container_type, registry_host, i)
+    ]
+
+    sorted_images = _sort_images(images, sort_by_size, sort_by_download)
+
+    if all:
+        return sorted_images
+    return sorted_images[0] if sorted_images else None

requirements/base.txt

@@ -8,7 +8,7 @@
 
 def find_tool(package_name, print_json):
     tool = find_package_by_name(package_name)
-    if type(tool) == dict:
+    if isinstance(tool, dict):
         if print_json:
             print(json.dumps(tool, indent=4))
         else:
@@ -23,7 +23,7 @@ def find_latest_image_main(package_name, package_version, json, all, sort_by_siz
                            sort_by_download, container_type, registry_host):
     images = find_latest_image(package_name, package_version, all, sort_by_size,
                                sort_by_download, container_type, registry_host)
-    if type(images) == list:
+    if isinstance(images, list):
         if json:
             print(json.dumps(images, indent=4))
         elif all:
@@ -38,7 +38,7 @@ def find_latest_image_main(package_name, package_version, json, all, sort_by_siz
                     image_name, updated,
                     size, downloads,
                     image_type))
-    elif type(images) == dict:
+    elif isinstance(images, dict):
         print(images['image_name'])
     else:
         if container_type:

requirements/test.txt

@@ -33,7 +33,7 @@ def main():
     search_term = args.search_term
 
     data = find_package_by_term(search_term)
-    if type(data) == int:
+    if isinstance(data, int):
         print('No packages found with search term: {}'.format(search_term))
     elif args.json:
         print(json.dumps(data, indent=4))

src/bioconda2biocontainer/biocontainer.py

@@ -12,42 +12,43 @@ def test_find_package_by_name(self):
         assert tool['id'] == 'bedtools'
 
     def test_find_latest_image(self):
-        images = find_latest_image('bedtools', '2.27.0', False, False, False, 'Docker', 'quay.io')
+        images = find_latest_image('bedtools', '2.27.1', False, False, False, 'Docker', 'quay.io')
         assert type(images) == dict
 
     def test_find_latest_image_all(self):
-        images = find_latest_image('bedtools', '2.27.0', True, False, False, 'Docker', 'quay.io')
+        images = find_latest_image('bedtools', '2.27.1', True, False, False, 'Docker', 'quay.io')
         flag = False
         for i in images:
-            if i['updated'] == '2019-10-26T00:00:00Z':
+            print(i)
+            if i['updated'] == '2024-12-10T00:00:00Z':
                 flag = True
-                assert i['image_name'] == 'quay.io/biocontainers/bedtools:2.27.0--he513fc3_4'
+                assert i['image_name'] == 'quay.io/biocontainers/bedtools:2.27.1--h077b44d_9'
         assert flag
 
     def test_find_package_by_name_status_code(self):
         tool = find_package_by_name('bedtoolss')
         assert tool == 204
 
     def test_find_latest_image_sort_by_size(self):
-        images = find_latest_image('bedtools', '2.27.0', True, True, False, 'Docker', 'quay.io')
+        images = find_latest_image('bedtools', '2.27.1', True, True, False, 'Docker', 'quay.io')
         flag = False
         for i in images:
-            if i['updated'] == '2019-10-26T00:00:00Z':
+            if i['updated'] == '2024-12-10T00:00:00Z':
                 flag = True
-                assert i['image_name'] == 'quay.io/biocontainers/bedtools:2.27.0--he513fc3_4'
+                assert i['image_name'] == 'quay.io/biocontainers/bedtools:2.27.1--h077b44d_9'
         assert flag
 
     def test_find_latest_image_sort_by_downloads(self):
-        images = find_latest_image('bedtools', '2.27.0', True, False, True, 'Docker', 'quay.io')
+        images = find_latest_image('bedtools', '2.27.1', True, False, True, 'Docker', 'quay.io')
         flag = False
         for i in images:
-            if i['updated'] == '2019-10-26T00:00:00Z':
+            if i['updated'] == '2024-12-10T00:00:00Z':
                 flag = True
-                assert i['image_name'] == 'quay.io/biocontainers/bedtools:2.27.0--he513fc3_4'
+                assert i['image_name'] == 'quay.io/biocontainers/bedtools:2.27.1--h077b44d_9'
         assert flag
 
     def test_find_latest_image_status_code(self):
-        images = find_latest_image('bedtoolss', '2.27.0', True, False, True, 'Docker', 'quay.io')
+        images = find_latest_image('bedtoolss', '2.27.1', True, False, True, 'Docker', 'quay.io')
         assert images == 204
 
     def test_search_tool(self):

src/bioconda2biocontainer/entry_point.py

@@ -4,17 +4,16 @@ envlist = py{39,310,311,312},flake8
 
 [testenv]
 skip_install = true
-commands_pre = poetry install --no-root
-commands = poetry run pytest {posargs:--cov pm4ngs --cov tests --cov-report term-missing --cov-report html --cov-report xml:coverage.xml}
+allowlist_externals = poetry
+commands_pre = poetry install
+commands = poetry run pytest {posargs:--cov bioconda2biocontainer --cov tests --cov-report term-missing --cov-report html --cov-report xml:coverage.xml}
 
 
 [testenv:flake8]
 skip_install = True
+allowlist_externals = poetry
 commands_pre = poetry install --no-root
-deps = flake8
-commands = poetry run flake8 src/ tests/
-
-basepython = python3.9
+commands = poetry run flake8 src/
 
 
 [pytest]