From a2e8c5d12d55e072f829bc184be03e21893cb524 Mon Sep 17 00:00:00 2001
From: Toni Verbeiren <toni.verbeiren@gmail.com>
Date: Wed, 24 Jul 2024 10:31:36 +0200
Subject: [PATCH] Add concat_text component (#4)

* Add concat_text component
* Minor cleanup
* Update src/concat_text/config.vsh.yaml
* Update in line with most of Dries' comments
* Don't claim to cleanup if it ain't happening...
* Add authorship
* Avoid using deprecated Viash variable

---------

Co-authored-by: Dries Schaumont <5946712+DriesSchaumont@users.noreply.github.com>
---
 src/_authors/dries_schaumont.yaml | 11 +++++
 src/_authors/toni_verbeiren.yaml  |  9 ++++
 src/concat_text/config.vsh.yaml   | 56 +++++++++++++++++++++++++
 src/concat_text/script.sh         | 34 +++++++++++++++
 src/concat_text/test.sh           | 70 +++++++++++++++++++++++++++++++
 5 files changed, 180 insertions(+)
 create mode 100644 src/_authors/dries_schaumont.yaml
 create mode 100644 src/_authors/toni_verbeiren.yaml
 create mode 100644 src/concat_text/config.vsh.yaml
 create mode 100644 src/concat_text/script.sh
 create mode 100644 src/concat_text/test.sh

diff --git a/src/_authors/dries_schaumont.yaml b/src/_authors/dries_schaumont.yaml
new file mode 100644
index 0000000..b267808
--- /dev/null
+++ b/src/_authors/dries_schaumont.yaml
@@ -0,0 +1,11 @@
+name: Dries Schaumont
+info:
+  links:
+    email: dries@data-intuitive.com
+    github: DriesSchaumont
+    orcid: "0000-0002-4389-0440"
+    linkedin: dries-schaumont
+  organizations:
+    - name: Data Intuitive
+      href: https://www.data-intuitive.com
+      role: Data Scientist
diff --git a/src/_authors/toni_verbeiren.yaml b/src/_authors/toni_verbeiren.yaml
new file mode 100644
index 0000000..2f2f851
--- /dev/null
+++ b/src/_authors/toni_verbeiren.yaml
@@ -0,0 +1,9 @@
+name: Toni Verbeiren
+info:
+  links:
+    github: tverbeiren
+    linkedin: verbeiren
+  organizations:
+  - name: Data Intuitive
+    href: https://www.data-intuitive.com
+    role: Data Scientist and CEO
diff --git a/src/concat_text/config.vsh.yaml b/src/concat_text/config.vsh.yaml
new file mode 100644
index 0000000..6a825ac
--- /dev/null
+++ b/src/concat_text/config.vsh.yaml
@@ -0,0 +1,56 @@
+name: concat_text
+description: |
+  Concatenate a number of text files, handle gzipped text files gracefully and
+  optionally gzip the output text file.
+
+  This component is useful for concatening fastq files from different lanes, for instance.
+authors:
+  - __merge__: /src/_authors/toni_verbeiren.yaml
+    roles: [ author, maintainer ]
+  - __merge__: /src/_authors/dries_schaumont.yaml
+    roles: [ reviewer ]
+info:
+  improvements: |
+    This component could be improved in 2 ways:
+      1. Allow for a mix of zipped and plain input files
+      2. Allow to specify a compression algorithm for the output
+argument_groups:
+  - name: Input arguments
+    arguments:
+      - name: --input
+        description: A list of (gzipped) text files.
+        type: file
+        multiple: true
+        required: true
+        example: input?.txt.gz
+  - name: Output arguments
+    arguments:
+      - name: "--gzip_output"
+        type: boolean_true
+        description: Should the output be zipped?
+      - name: --output
+        description: File to write the output to, optionally gzipped.
+        type: file
+        direction: output
+        example: output.txt
+
+resources:
+  - type: bash_script
+    path: script.sh
+test_resources:
+  - type: bash_script
+    path: test.sh
+
+engines:
+  - type: docker
+    image: alpine:latest
+    setup:
+      - type: apk
+        packages:
+          - bash
+          - procps
+          - file
+
+runners:
+  - type: executable
+  - type: nextflow
diff --git a/src/concat_text/script.sh b/src/concat_text/script.sh
new file mode 100644
index 0000000..5efab6f
--- /dev/null
+++ b/src/concat_text/script.sh
@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+
+set -euo pipefail
+
+TMPDIR=$(mktemp -d "$meta_temp_dir/concat_text-XXXXXX")
+function clean_up {
+  [[ -d "$TMPDIR" ]] && rm -r "$TMPDIR"
+}
+trap clean_up EXIT
+
+par_input="$(echo "$par_input" | tr ';' ' ')"
+
+echo -n ">> Check if input is gzipped... "
+set +eo pipefail
+file $par_input | grep -q 'gzip'
+is_zipped="$?"
+set -euo pipefail
+[[ "$is_zipped" == "0" ]] && echo "yes" || echo "no"
+
+if [[ "$is_zipped" == "0" ]]; then
+  echo ">> zcat gzipped files"
+  zcat $par_input > $TMPDIR/contents
+else
+  echo ">> cat plain files"
+  cat $par_input > $TMPDIR/contents
+fi
+
+if [ "$par_gzip_output" == true ]; then
+  echo ">> Zip output file"
+  gzip $TMPDIR/contents
+  mv $TMPDIR/contents.gz $par_output
+else
+  mv $TMPDIR/contents $par_output
+fi
diff --git a/src/concat_text/test.sh b/src/concat_text/test.sh
new file mode 100644
index 0000000..1e75223
--- /dev/null
+++ b/src/concat_text/test.sh
@@ -0,0 +1,70 @@
+#!/usr/bin/env bash
+
+set -euo pipefail
+
+echo ">> Creating test input files file[1-3].txt"
+INPUT_FILE_1="file1.txt"
+INPUT_FILE_2="file2.txt"
+INPUT_FILE_3="file3.txt"
+echo "one" > "$INPUT_FILE_1"
+echo "two" > "$INPUT_FILE_2"
+echo "three" > "$INPUT_FILE_3"
+echo ">> Created input files"
+
+echo ">> Creating zipped versions at file[1-3].txt.gz"
+gzip -k $INPUT_FILE_1
+gzip -k $INPUT_FILE_2
+gzip -k $INPUT_FILE_3
+
+echo ">> Creating expected output file expected_output.txt and zipped version"
+cat > "expected_output.txt" <<EOF
+one
+two
+three
+EOF
+
+gzip -k "expected_output.txt"
+
+echo ">> Run component on 3 plain input files, plain output"
+$meta_executable \
+  --input "$INPUT_FILE_1;$INPUT_FILE_2;$INPUT_FILE_3" \
+  --output "output1.txt"
+
+[[ ! -f "output1.txt" ]] \
+  && echo "Output file output1.txt not found!" && exit 1
+[[ $(cmp "output1.txt" "expected_output.txt") ]] \
+  && echo "Output file output1.txt is not as expected!" && exit 1
+
+echo ">> Run component on 3 zipped input files, plain output"
+$meta_executable \
+  --input "$INPUT_FILE_1.gz;$INPUT_FILE_2.gz;$INPUT_FILE_3.gz" \
+  --output "output2.txt"
+
+[[ ! -f "output2.txt" ]] \
+  && echo "Output file output2.txt not found!" && exit 1
+[[ $(cmp "output2.txt" "expected_output.txt") ]] \
+  && echo "Output file output2.txt is not as expected!" && exit 1
+
+echo ">> Run component on 3 plain input files, zipped output"
+$meta_executable \
+  --input "$INPUT_FILE_1;$INPUT_FILE_2;$INPUT_FILE_3" \
+  --output "output3.txt.gz" \
+  --gzip_output
+
+[[ ! -f "output3.txt.gz" ]] \
+  && echo "Output file output3.txt.gz not found!" && exit 1
+[[ $(cmp "output3.txt.gz" "expected_output.txt.gz") ]] \
+  && echo "Output file output3.txt.gz is not as expected!" && exit 1
+
+echo ">> Run component on 3 zipped input files, zipped output"
+$meta_executable \
+  --input "$INPUT_FILE_1.gz;$INPUT_FILE_2.gz;$INPUT_FILE_3.gz" \
+  --output "output4.txt.gz" \
+  --gzip_output
+
+[[ ! -f "output4.txt.gz" ]] \
+  && echo "Output file output4.txt.gz not found!" && exit 1
+[[ $(cmp "output4.txt.gz" "expected_output.txt.gz") ]] \
+  && echo "Output file output4.txt.gz is not as expected!" && exit 1
+
+echo ">> Tests done"