Ignore cache if Nextclade version is different

Currently just checks Nextclade and dataset versions of the first row of
the nextclade.tsv file and formats them as the intended JSON proposed.
Once the version JSON metadata file is in place, it should be easy to
swap out the check for the separate file rather than the nextclade.tsv.

bin/cache-version

@@ -0,0 +1,19 @@
+#!/bin/bash
+
+s3_url="${1:?An S3 URL is required as the first argument}"
+
+
+trap '' SIGPIPE
+
+(aws s3 cp "${s3_url}" - \
+    | zstd -T0 -dcq \
+    | head -n 2 \
+    | tsv-select -H -f 'nextclade_version,dataset_version' \
+    | tail -n 1 \
+    | jq --raw-input --slurp -c '
+        split("\n")
+        | map(split("\t"))
+        | .[0:-1]
+        | map( { "nextclade_version": .[0], "nextclade_dataset_version": .[1] } )
+        | .[0]') \
+     2> /dev/null

bin/use-nextclade-cache

@@ -1,19 +1,31 @@
 #!/bin/bash
 set -euo pipefail
 
-vendored="$(dirname "$0")"/../vendored
+bin="$(dirname "$0")"
+vendored="$bin"/../vendored
 
 main() {
     s3_dst="${1:?A destination s3:// URL where the renew file is hosted is required as the first argument.}"
     s3_src="${2:?A source s3:// URL where the fallback renew file is hosted is required as the second argument.}"
+    nextclade="${3:?A path to the Nextclade executable is required as the third argument}"
     # Nextclade dataset reference wildcard
-    reference="${3:-}"
+    reference="${4:-}"
+
     if renew-flag-exists; then
         echo "[INFO] Found renew flag" >&2
         echo "false"
         exit 0
     fi
 
+    cache_versions="$(get-cache-version-info)"
+    cache_nextclade_version="$(echo "$cache_versions" | jq -r .nextclade_version)"
+    current_nextclade_version="$("$nextclade" --version)"
+    if [[ "$cache_nextclade_version" != "$current_nextclade_version" ]]; then
+        echo "[INFO] Current Nextclade version (${current_nextclade_version}) is different from cache version (${cache_nextclade_version})" >&2
+        echo "false"
+        exit 0
+    fi
+
     echo "true"
 }
 
@@ -25,4 +37,13 @@ renew-flag-exists() {
     "$vendored"/s3-object-exists "${dst_renew_file}" || "$vendored"/s3-object-exists "${src_renew_file}"
 }
 
+get-cache-version-info() {
+    # TODO: Update check a separate file for version info
+    # Currently just checks the first row of the nextclade.tsv file
+    local version_file="nextclade${reference}.tsv.zst"
+    local dst_version_file="${s3_dst}/${version_file}"
+
+    "$bin"/cache-version "$dst_version_file"
+}
+
 main "$@"

workflow/snakemake_rules/nextclade.smk

@@ -63,6 +63,8 @@ if config.get("s3_dst") and config.get("s3_src"):
     ruleorder: download_previous_alignment_from_s3 > create_empty_nextclade_aligned
 
     rule use_nextclade_cache:
+        input:
+            nextclade="nextclade",
         params:
             dst_source=config["s3_dst"],
             src_source=config["s3_src"],
@@ -73,6 +75,7 @@ if config.get("s3_dst") and config.get("s3_src"):
             ./bin/use-nextclade-cache \
                 {params.dst_source:q} \
                 {params.src_source:q} \
+                {input.nextclade:q} \
                 {wildcards.reference:q} \
                 > {output.use_nextclade_cache}
             """