From 64ff23f3c1d26e22742e13e93da29b52a82da8d2 Mon Sep 17 00:00:00 2001
From: "Moritz E. Beber" <midnighter@posteo.net>
Date: Sun, 8 May 2022 15:44:13 +0200
Subject: [PATCH] feat: add module that verifies NCBI settings

---
 modules/sratools/ncbisettings/main.nf         | 20 ++++++++
 modules/sratools/ncbisettings/meta.yml        | 28 +++++++++++
 .../templates/detect_ncbi_settings.sh         | 39 ++++++++++++++++
 tests/config/pytest_modules.yml               |  4 ++
 tests/modules/sratools/ncbisettings/main.nf   | 46 +++++++++++++++++++
 .../sratools/ncbisettings/nextflow.config     | 17 +++++++
 tests/modules/sratools/ncbisettings/test.yml  | 44 ++++++++++++++++++
 7 files changed, 198 insertions(+)
 create mode 100644 modules/sratools/ncbisettings/main.nf
 create mode 100644 modules/sratools/ncbisettings/meta.yml
 create mode 100644 modules/sratools/ncbisettings/templates/detect_ncbi_settings.sh
 create mode 100644 tests/modules/sratools/ncbisettings/main.nf
 create mode 100644 tests/modules/sratools/ncbisettings/nextflow.config
 create mode 100644 tests/modules/sratools/ncbisettings/test.yml

diff --git a/modules/sratools/ncbisettings/main.nf b/modules/sratools/ncbisettings/main.nf
new file mode 100644
index 00000000..9d905f8b
--- /dev/null
+++ b/modules/sratools/ncbisettings/main.nf
@@ -0,0 +1,20 @@
+process SRATOOLS_NCBISETTINGS {
+    tag 'ncbi-settings'
+    label 'process_low'
+
+    conda (params.enable_conda ? 'bioconda::sra-tools=2.11.0' : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/sra-tools:2.11.0--pl5321ha49a11a_3' :
+        'quay.io/biocontainers/sra-tools:2.11.0--pl5321ha49a11a_3' }"
+
+    output:
+    path('user-settings.mkfg'), optional: true, emit: ncbi_settings
+    path 'versions.yml'       , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    shell:
+    config = "/LIBS/GUID = \"${UUID.randomUUID().toString()}\"\\n/libs/cloud/report_instance_identity = \"true\"\\n"
+    template 'detect_ncbi_settings.sh'
+}
diff --git a/modules/sratools/ncbisettings/meta.yml b/modules/sratools/ncbisettings/meta.yml
new file mode 100644
index 00000000..29bcfe34
--- /dev/null
+++ b/modules/sratools/ncbisettings/meta.yml
@@ -0,0 +1,28 @@
+name: "sratools_ncbisettings"
+description: Test for the presence of suitable NCBI settings or create them on the fly.
+keywords:
+  - NCBI
+  - settings
+  - sra-tools
+  - prefetch
+  - fasterq-dump
+tools:
+  - "sratools":
+      description: "SRA Toolkit and SDK from NCBI"
+      homepage: https://github.com/ncbi/sra-tools
+      documentation: https://github.com/ncbi/sra-tools/wiki
+      tool_dev_url: https://github.com/ncbi/sra-tools
+      licence: "['Public Domain']"
+
+output:
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - ncbi_settings:
+      type: file
+      description: An optional, minimal NCBI settings file.
+      pattern: "user-settings.mkfg"
+
+authors:
+  - "@Midnighter"
diff --git a/modules/sratools/ncbisettings/templates/detect_ncbi_settings.sh b/modules/sratools/ncbisettings/templates/detect_ncbi_settings.sh
new file mode 100644
index 00000000..b1b51c1f
--- /dev/null
+++ b/modules/sratools/ncbisettings/templates/detect_ncbi_settings.sh
@@ -0,0 +1,39 @@
+#!/usr/bin/env bash
+
+set -u
+
+
+# Get the expected NCBI settings path and define the environment variable
+# `NCBI_SETTINGS`.
+eval "$(vdb-config -o n NCBI_SETTINGS | sed 's/[" ]//g')"
+
+# If the user settings do not exist yet, create a file suitable for `prefetch`
+# and `fasterq-dump`. If an existing settings file does not contain the required
+# values, error out with a helpful message.
+if [[ ! -f "${NCBI_SETTINGS}" ]]; then
+    printf '!{config}' > 'user-settings.mkfg'
+else
+    prefetch --help &> /dev/null
+    if [[ $? = 78 ]]; then
+        echo "You have an existing vdb-config at '${NCBI_SETTINGS}' but it is"\
+        "missing the required entries for /LIBS/GUID and"\
+        "/libs/cloud/report_instance_identity."\
+        "Feel free to add the following to your settings file:" >&2
+        echo "$(printf '!{config}')" >&2
+        exit 1
+    fi
+    fasterq-dump --help &> /dev/null
+    if [[ $? = 78 ]]; then
+        echo "You have an existing vdb-config at '${NCBI_SETTINGS}' but it is"\
+        "missing the required entries for /LIBS/GUID and"\
+        "/libs/cloud/report_instance_identity."\
+        "Feel free to add the following to your settings file:" >&2
+        echo "$(printf '!{config}')" >&2
+        exit 1
+    fi
+fi
+
+cat <<-END_VERSIONS > versions.yml
+"!{task.process}":
+    sratools: $(vdb-config --version 2>&1 | grep -Eo '[0-9.]+')
+END_VERSIONS
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
index 9afe83fd..4a922a4d 100644
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@@ -1819,6 +1819,10 @@ sratools/fasterqdump:
   - modules/sratools/fasterqdump/**
   - tests/modules/sratools/fasterqdump/**
 
+sratools/ncbisettings:
+  - modules/sratools/ncbisettings/**
+  - tests/modules/sratools/ncbisettings/**
+
 sratools/prefetch:
   - modules/sratools/prefetch/**
   - tests/modules/sratools/prefetch/**
diff --git a/tests/modules/sratools/ncbisettings/main.nf b/tests/modules/sratools/ncbisettings/main.nf
new file mode 100644
index 00000000..08968bb9
--- /dev/null
+++ b/tests/modules/sratools/ncbisettings/main.nf
@@ -0,0 +1,46 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SRATOOLS_NCBISETTINGS } from '../../../../modules/sratools/ncbisettings/main.nf'
+
+workflow test_sratools_ncbisettings_with_good_existing {
+
+    file(params.settings_path).mkdirs()
+    def settings = file(params.settings_file)
+    settings.text = "/LIBS/GUID = \"5b0d4b7d-88c7-4802-98fd-e3afd06feb32\"\n/libs/cloud/report_instance_identity = \"true\"\n"
+
+    SRATOOLS_NCBISETTINGS()
+}
+
+workflow test_sratools_ncbisettings_with_bad_existing {
+
+    file(params.settings_path).mkdirs()
+    def settings = file(params.settings_file)
+    settings.text = '''
+    ## auto-generated configuration file - DO NOT EDIT ##
+
+    config/default = "false"
+    /repository/remote/main/CGI/resolver-cgi = "https://trace.ncbi.nlm.nih.gov/Traces/names/names.fcgi"
+    /repository/remote/protected/CGI/resolver-cgi = "https://trace.ncbi.nlm.nih.gov/Traces/names/names.fcgi"
+    /repository/user/ad/public/apps/file/volumes/flatAd = "."
+    /repository/user/ad/public/apps/refseq/volumes/refseqAd = "."
+    /repository/user/ad/public/apps/sra/volumes/sraAd = "."
+    /repository/user/ad/public/apps/sraPileup/volumes/ad = "."
+    /repository/user/ad/public/apps/sraRealign/volumes/ad = "."
+    /repository/user/ad/public/apps/wgs/volumes/wgsAd = "."
+    /repository/user/ad/public/root = "."
+    /repository/user/default-path = "/root/ncbi"
+    '''.stripIndent()
+
+    SRATOOLS_NCBISETTINGS()
+}
+
+workflow test_sratools_ncbisettings_with_nonexisting {
+
+    file(params.settings_path).mkdirs()
+    def settings = file(params.settings_file)
+    settings.delete()
+
+    SRATOOLS_NCBISETTINGS()
+}
diff --git a/tests/modules/sratools/ncbisettings/nextflow.config b/tests/modules/sratools/ncbisettings/nextflow.config
new file mode 100644
index 00000000..823082c0
--- /dev/null
+++ b/tests/modules/sratools/ncbisettings/nextflow.config
@@ -0,0 +1,17 @@
+params.settings_path = '/tmp/.ncbi'
+params.settings_file = "${params.settings_path}/user-settings.mkfg"
+
+env.NCBI_SETTINGS = params.settings_file
+
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName: SRATOOLS_NCBISETTINGS {
+        containerOptions = {
+            (workflow.containerEngine == 'singularity') ?
+            "-B ${params.settings_path}:${params.settings_path}" :
+            "-v ${params.settings_path}:${params.settings_path}"
+        }
+    }
+}
diff --git a/tests/modules/sratools/ncbisettings/test.yml b/tests/modules/sratools/ncbisettings/test.yml
new file mode 100644
index 00000000..07c45dc7
--- /dev/null
+++ b/tests/modules/sratools/ncbisettings/test.yml
@@ -0,0 +1,44 @@
+- name: "sratools ncbisettings test_sratools_ncbisettings_with_good_existing"
+  command: nextflow run ./tests/modules/sratools/ncbisettings -entry test_sratools_ncbisettings_with_good_existing -c ./tests/config/nextflow.config -c ./tests/modules/sratools/ncbisettings/nextflow.config
+  tags:
+    - "sratools"
+    - "sratools/ncbisettings"
+  files:
+    - path: "output/sratools/user-settings.mkfg"
+      should_exist: false
+    - path: output/sratools/versions.yml
+      contains:
+        - "sratools: 2.11.0"
+
+- name: "sratools ncbisettings test_sratools_ncbisettings_with_bad_existing"
+  command: nextflow run ./tests/modules/sratools/ncbisettings -entry test_sratools_ncbisettings_with_bad_existing -c ./tests/config/nextflow.config -c ./tests/modules/sratools/ncbisettings/nextflow.config
+  tags:
+    - "sratools"
+    - "sratools/ncbisettings"
+  exit_code: 1
+  stdout:
+    contains:
+      - "Command error:"
+      - "missing the required entries"
+      - "/LIBS/GUID"
+      - "/libs/cloud/report_instance_identity"
+      - "Feel free to add the following"
+  files:
+    - path: "output/sratools/user-settings.mkfg"
+      should_exist: false
+    - path: output/sratools/versions.yml
+      should_exist: false
+
+- name: "sratools ncbisettings test_sratools_ncbisettings_with_nonexisting"
+  command: nextflow run ./tests/modules/sratools/ncbisettings -entry test_sratools_ncbisettings_with_nonexisting -c ./tests/config/nextflow.config -c ./tests/modules/sratools/ncbisettings/nextflow.config
+  tags:
+    - "sratools"
+    - "sratools/ncbisettings"
+  files:
+    - path: "output/sratools/user-settings.mkfg"
+      contains:
+        - "/LIBS/GUID"
+        - "/libs/cloud/report_instance_identity"
+    - path: output/sratools/versions.yml
+      contains:
+        - "sratools: 2.11.0"