From 582d955a131c4f0b56fc253ea761a9799e41b952 Mon Sep 17 00:00:00 2001
From: James Hadfield <hadfield.james@gmail.com>
Date: Thu, 20 Apr 2023 16:20:39 +1200
Subject: [PATCH] Add helper script for downloading dataset+sidecars

Adds a small dev-use-only script to download the dataset +
sidecar files from a given nextstrain.org URL and give them
the correct filename so that you can test the same URL pathname
from localhost auspice.

I've wanted this for ages but #1603 finally motivated me to do it.
---
 scripts/get-dataset.sh | 50 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100755 scripts/get-dataset.sh

diff --git a/scripts/get-dataset.sh b/scripts/get-dataset.sh
new file mode 100755
index 000000000..632e5e4e5
--- /dev/null
+++ b/scripts/get-dataset.sh
@@ -0,0 +1,50 @@
+#!/usr/bin/env bash
+# A simple, non-battle-hardened, script to help downloading a dataset
+# and corresponding sidecars for testing locally with auspice.
+#
+# Examples:
+# ./scripts/get-dataset.sh nextstrain.org/zika data
+# ./scripts/get-dataset.sh nextstrain.org/community/joverlee521/nextstrain-testing/flu/seasonal/h1n1pdm/ha/09-17 datasets
+echo "Getting dataset + sidecars associated with $1 and downloading them to $2"
+
+if [ ! -d "${2}" ]; then
+  echo "Directory ${2} doesn't exist!";
+  exit 2;
+fi
+
+dataset_suffix=${1#*nextstrain.org/}
+dataset_underscores=${dataset_suffix//\//_}
+dest="${2%/}/${dataset_underscores}.json"
+
+main="https://nextstrain.org/charon/getDataset?prefix=${dataset_suffix}"
+if [[ $( curl -iI -sw "%{http_code}" "${main}" -o /dev/null ) == 200 ]]; then
+  curl "${main}" --compressed --output "${dest}"
+  echo Downloaded main JSON to "${dest}"
+else
+  echo "Cannot download main JSON. Fatal!"
+  exit 3
+fi
+
+sidecar="tip-frequencies"
+if [[ $( curl -iI -sw "%{http_code}" "${main}&type=${sidecar}" -o /dev/null ) == 200 ]]; then
+  curl "${main}&type=${sidecar}" --compressed --output "${dest%.json}_${sidecar}.json"
+  echo Downloaded ${sidecar} JSON to "${dest%.json}_${sidecar}.json"
+else
+  echo "Cannot download ${sidecar} JSON. Continuing..."
+fi
+
+sidecar="root-sequence"
+if [[ $( curl -iI -sw "%{http_code}" "${main}&type=${sidecar}" -o /dev/null ) == 200 ]]; then
+  curl "${main}&type=${sidecar}" --compressed --output "${dest%.json}_${sidecar}.json"
+  echo Downloaded ${sidecar} JSON to "${dest%.json}_${sidecar}.json"
+else
+  echo "Cannot download ${sidecar} JSON. Continuing..."
+fi
+
+sidecar="measurements"
+if [[ $( curl -iI -sw "%{http_code}" "${main}&type=${sidecar}" -o /dev/null ) == 200 ]]; then
+  curl "${main}&type=${sidecar}" --compressed --output "${dest%.json}_${sidecar}.json"
+  echo Downloaded ${sidecar} JSON to "${dest%.json}_${sidecar}.json"
+else
+  echo "Cannot download ${sidecar} JSON. Continuing..."
+fi
\ No newline at end of file