Merge pull request #3548 from JulianFlesch/feature/test-datasets

Feature/test datasets

Author: JulianFlesch

CHANGELOG.md

@@ -62,7 +62,7 @@
 - Changing retrieval of file extension from EDAM ([#3512](https://github.com/nf-core/tools/pull/3512))
 - Refactor adding EDAM ontologies and allowing detect more patterns (e.g., versions.yml) ([#3519](https://github.com/nf-core/tools/pull/3519))
 - Add offline configs test action ([#3524](https://github.com/nf-core/tools/pull/3524))
-- Adds `test-datasets` subcommand for listing/searching files in the nf-core/test-datasets repo from the cli ([#3487](https://github.com/nf-core/tools/issues/3487))
+- Adds `test-datasets` subcommand for listing/searching files in the nf-core/test-datasets repo from the cli ([#3487](https://github.com/nf-core/tools/issues/3487), [#3548](https://github.com/nf-core/tools/pull/3548))
 - Update pre-commit hook astral-sh/ruff-pre-commit to v0.11.2 ([#3521](https://github.com/nf-core/tools/pull/3521))
 - Fix indentation in included_configs API docs ([#3523](https://github.com/nf-core/tools/pull/3523))
 - chore(deps): update python:3.12-slim docker digest to a866731 ([#3527](https://github.com/nf-core/tools/pull/3527))

nf_core/__main__.py

@@ -77,7 +77,7 @@
     "nf-core": [
         {
             "name": "Commands",
-            "commands": ["pipelines", "modules", "subworkflows", "interface", "test-datasets"],
+            "commands": ["pipelines", "modules", "subworkflows", "test-datasets", "interface"],
         },
     ],
     "nf-core pipelines": [
@@ -1762,7 +1762,7 @@ def test_datasets(ctx):
 
 
 # nf-core test-dataset search
-@test_datasets.command("search")
+@test_datasets.command("search", short_help="Search files in the nf-core/test-datasets repository")
 @click.pass_context
 @click.option("-b", "--branch", type=str, help="Branch in the test-datasets repository to reduce search to")
 @click.option(
@@ -1779,11 +1779,13 @@ def test_datasets(ctx):
     default=False,
     help="Auto-generate a github url for downloading the test data file based on the branch and query result",
 )
-def command_test_dataset_search(ctx, branch, generate_nf_path, generate_dl_url):
+@click.argument("query", required=False)
+def command_test_dataset_search(ctx, branch, generate_nf_path, generate_dl_url, query):
     """
-    Interactively search for files on a specified branch in the nf-core/test-datasets repository on github.
+    Search files filtered by QUERY on a specified branch in the nf-core/test-datasets repository.
+    If no QUERY is given or QUERY is ambiguous, an auto-completion form is shown.
     """
-    test_datasets_search(ctx, branch, generate_nf_path, generate_dl_url)
+    test_datasets_search(ctx, branch, generate_nf_path, generate_dl_url, query)
 
 
 # nf-core test-dataset search
@@ -1806,7 +1808,7 @@ def command_test_dataset_search(ctx, branch, generate_nf_path, generate_dl_url):
 )
 def command_test_dataset_list_remote(ctx, branch, generate_nf_path, generate_dl_url):
     """
-    List all data files for a specified branch in the nf-core/test-datasets repository on github.
+    List files on a specified branch in the nf-core/test-datasets repository.
     """
     test_datasets_list_remote(ctx, branch, generate_nf_path, generate_dl_url)
 
@@ -1816,7 +1818,7 @@ def command_test_dataset_list_remote(ctx, branch, generate_nf_path, generate_dl_
 @click.pass_context
 def command_test_datasets_list_branches(ctx):
     """
-    List all remote branches with pipeline or module data in the nf-core/test-dataset repository on github
+    List remote branches with test data in the nf-core/test-dataset repository.
     """
     test_datasets_list_branches(ctx)
 

nf_core/commands_test_datasets.py

@@ -28,11 +28,13 @@ def test_datasets_list_remote(ctx: click.Context, branch: str, generate_nf_path:
     list_datasets(branch, generate_nf_path, generate_dl_url)
 
 
-def test_datasets_search(ctx: click.Context, branch: str, generate_nf_path: bool, generate_dl_url: bool) -> None:
+def test_datasets_search(
+    ctx: click.Context, branch: str, generate_nf_path: bool, generate_dl_url: bool, query: str
+) -> None:
     """
     Search all files on a given branch in the remote nf-core/testdatasets repository on github
     with an interactive autocompleting prompt and print the file matching the query.
     Specifying a branch is required.
     The resulting file can optionally be parsed as a nextflow path or a url for downloading
     """
-    search_datasets(branch, generate_nf_path, generate_dl_url)
+    search_datasets(branch, generate_nf_path=generate_nf_path, generate_dl_url=generate_dl_url, query=query)

nf_core/test_datasets/search.py

@@ -22,6 +22,7 @@ def search_datasets(
     generate_dl_url: bool = False,
     ignored_file_prefixes: List[str] = IGNORED_FILE_PREFIXES,
     plain_text_output: bool = False,
+    query: str = "",
 ) -> None:
     """
     Search all files on a given branch in the remote nf-core/testdatasets repository on github
@@ -39,11 +40,17 @@ def search_datasets(
     files = sum(tree.values(), [])  # flat representation of tree
 
     file_selected = False
+
+    if query:
+        # Check if only one file matches the query and directly return it
+        filtered_files = [f for f in files if query in f]
+        if len(filtered_files) == 1:
+            selection = filtered_files[0]
+            file_selected = True
+
     while not file_selected:
         selection = questionary.autocomplete(
-            "File:",
-            choices=files,
-            style=nfcore_question_style,
+            "File:", choices=files, style=nfcore_question_style, default=query
         ).unsafe_ask()
 
         file_selected = any([selection == file for file in files])

nf_core/test_datasets/test_datasets_utils.py

@@ -1,12 +1,13 @@
 import json
 import logging
 from dataclasses import dataclass
+from pathlib import Path
 from typing import Dict, List, Tuple
 
 import questionary
 import requests
 
-from nf_core.utils import nfcore_question_style
+from nf_core.utils import determine_base_dir, fetch_wf_config, load_tools_config, nfcore_question_style
 
 log = logging.getLogger(__name__)
 
@@ -176,10 +177,26 @@ def get_or_prompt_branch(maybe_branch: str) -> Tuple[str, List[str]]:
     else:
         all_branches = get_remote_branch_names()
 
+        # Find pipeline / modules root directory
+        base_dir: Path = determine_base_dir()
+
+        # Read .nf-core.yml to identify repository_type
+        _, tools_config = load_tools_config(base_dir)
+
+        branch_prefill = ""
+        # either modules or a pipeline branch
+        if tools_config is not None:
+            repo_type = tools_config.get("repository_type", None)
+            if repo_type == MODULES_BRANCH_NAME:
+                branch_prefill = MODULES_BRANCH_NAME
+            elif repo_type == "pipeline":
+                wf_config = fetch_wf_config(base_dir)
+                pipeline_name = wf_config.get("manifest.name", "").split("/")[-1]
+                if pipeline_name in all_branches:
+                    branch_prefill = pipeline_name
+
         branch = questionary.autocomplete(
-            "Branch name:",
-            choices=sorted(all_branches),
-            style=nfcore_question_style,
+            "Branch name:", choices=sorted(all_branches), style=nfcore_question_style, default=branch_prefill
         ).unsafe_ask()
 
         return branch, all_branches