add function for printing progress. Fixes #72

Dockerfile

@@ -14,7 +14,9 @@ FROM php:8.3-fpm-alpine AS development
 # add other deps for dev here
 RUN apk add --no-cache \
     libzip-dev \
-    bash
+    bash \
+    jq \
+    curl
 
 COPY --from=composer:2.8 /usr/bin/composer /usr/bin/composer
 COPY --from=jsmin /usr/bin/jsmin /usr/bin/jsmin
@@ -30,6 +32,13 @@ RUN composer install --no-interaction
 
 COPY /exe ./exe
 
+RUN chmod +x /opt/findingaid/exe/findingaid-cache-regen/fa-regen \
+        /opt/findingaid/exe/findingaid-cache-regen/fa-full-regen \
+        /opt/findingaid/exe/findingaid-cache-regen/fetch-ead-arks && \
+    ln -s /opt/findingaid/exe/findingaid-cache-regen/fa-regen /usr/local/bin/fa-regen && \
+    ln -s /opt/findingaid/exe/findingaid-cache-regen/fa-full-regen /usr/local/bin/fa-full-regen && \
+    ln -s /opt/findingaid/exe/findingaid-cache-regen/fetch-ead-arks /usr/local/bin/fetch-ead-arks
+
 COPY entrypoint.sh /usr/local/bin/entrypoint.sh
 RUN chmod +x /usr/local/bin/entrypoint.sh
 

exe/findingaid-cache-regen/fa-full-regen

@@ -1,17 +1,90 @@
 #!/bin/bash
 set -euo pipefail
 
-# The environment should provide these variables
-# SOLR_URL: e.g. https://solrhost.example.com/solr/select
+usage() {
+  cat <<'EOF'
+Usage: fa-full-regen [-v|--verbose]
+
+Regenerate the cached page for every finding aid Solr lists as a collection.
+
+Options:
+  -v, --verbose  List every ARK that could not be regenerated. Without it only
+                 the count is reported
+  -h, --help     Show this help.
+
+Environment:
+  SOLR_URL       Required, e.g. https://solrhost.example.com/solr/select
+EOF
+}
+
+VERBOSE=0
+
+while [ $# -gt 0 ]; do
+  case "$1" in
+    -v|--verbose) VERBOSE=1 ;;
+    -h|--help)    usage; exit 0 ;;
+    *)
+      echo "fa-full-regen: unknown option: $1" >&2
+      usage >&2
+      exit 2
+      ;;
+  esac
+  shift
+done
 
 ARKS_FILE="$(mktemp)"
 export ARKS_FILE
 
-trap 'rm -f "$ARKS_FILE"' EXIT
+ERRORS_FILE="$(mktemp)"
+
+trap 'rm -f "$ARKS_FILE" "$ERRORS_FILE"' EXIT
 
-echo "Attempting generation of a list of finding aids to reprocess"
+echo "Attempting generation of a list of finding aids to regenerate"
 fetch-ead-arks
-echo "List generated"
-echo "Reprocessing finding aids"
-xargs -n 1 -P 4 fa-regen < "$ARKS_FILE"
-echo "Finding aids reprocessed"
+FOUND=$(wc -l < "$ARKS_FILE")
+
+PROCESSED=0
+
+if [ "$FOUND" -le 0 ]; then
+  echo "$FOUND findingaid records found. Exiting..."
+  exit 0
+fi
+
+echo "$FOUND findingaid IDs found. Regenerating findingaids..."
+
+function print_progress {
+  PROCESSED=$((PROCESSED + 1))
+  local PROGRESS_PERCENTAGE=$(( 100 * PROCESSED / FOUND ))
+  local PROGRESS_STR="["
+  local HASH_COUNT=$(( PROGRESS_PERCENTAGE / 5 ))
+  local i
+  for i in {1..20}; do
+    if [ "$i" -le $HASH_COUNT ]; then
+      PROGRESS_STR+="#"
+    else
+      PROGRESS_STR+=" "
+    fi
+  done
+  PROGRESS_STR+="] "
+  PROGRESS_STR+="($PROGRESS_PERCENTAGE%)"
+  echo -ne "$PROGRESS_STR\r"
+}
+
+xargs -n 1 -P 4 fa-regen < "$ARKS_FILE" 2> "$ERRORS_FILE" | while read -r _; do
+  print_progress
+done
+printf '\n'
+
+# Missing EADs are a data condition, not a failed run, so report and exit 0.
+if [ -s "$ERRORS_FILE" ]; then
+  MISSING=$(wc -l < "$ERRORS_FILE")
+  if [ "$VERBOSE" -eq 1 ]; then
+    echo "$MISSING of $FOUND finding aids could not be regenerated:" >&2
+    cat "$ERRORS_FILE" >&2
+  else
+    echo "$MISSING of $FOUND finding aids could not be regenerated" \
+         "(re-run with --verbose for the list)" >&2
+  fi
+fi
+
+echo "$((FOUND - MISSING)) finding aids reprocessed"

exe/findingaid-cache-regen/fa-regen

@@ -6,4 +6,15 @@ TARGET_URL=http://web
 ID=$1
 BASE="$TARGET_URL/fa/findingaid/?id=$ID&invalidate_cache=1"
 echo "$BASE"
-curl "$BASE" >/dev/null
+
+# A finding aid whose EAD is missing from the pairtree redirects to /
+STATUS=$(curl -sS -o /dev/null -w '%{http_code}' "$BASE") || STATUS=000
+
+case "$STATUS" in
+    2??) ;;
+    302) echo "$ID: not found (redirected to /); no EAD in the pairtree" >&2 ;;
+    000) echo "$ID: request to $TARGET_URL failed" >&2 ;;
+    *)   echo "$ID: unexpected HTTP $STATUS" >&2 ;;
+esac
+
+exit 0

exe/findingaid-cache-regen/fetch-ead-arks

@@ -1,16 +1,26 @@
 #!/bin/bash
 set -euo pipefail
 
-# The environment should provide these variables
-# SOLR_URL: e.g. https://solrhost.example.com/solr/select
-# ARKS_FILE: where ARKs should be written
+if [[ -z "$SOLR_URL" ]]; then
+  echo "fetch-ead-arks: SOLR_URL must not be empty, e.g. https://solrhost.example.com/solr/select" >&2
+  exit 1
+fi
 
-COUNT_URL="${SOLR_URL}?wt=json&fl=id&fq=format:collections"
-numFound=$(curl -s "$COUNT_URL" | jq -r .response.numFound)
-echo "Found $numFound collections"
+if [[ -z "$ARKS_FILE" ]]; then
+  echo "fetch-ead-arks: ARKS_FILE must not be empty" >&2
+  exit 1
+fi
+
+COUNT_URL="${SOLR_URL}?wt=json&fl=id&fq=format:collections&rows=0"
+NUM_FOUND=$(curl -s "$COUNT_URL" | jq -r '.response.numFound')
+
+if ! [[ "$NUM_FOUND" =~ ^[0-9]+$ ]]; then
+  echo "fetch-ead-arks: unexpected response from Solr: $NUM_FOUND" >&2
+  exit 1
+fi
 
 echo "Fetching ids"
-EADS_URL="${SOLR_URL}?wt=json&fl=id&fq=format:collections&rows=${numFound}"
+EADS_URL="${SOLR_URL}?wt=json&fl=id&fq=format:collections&rows=${NUM_FOUND}"
 curl -s "$EADS_URL" | jq -r '.response.docs[] | .id' > "$ARKS_FILE"
 
-echo "IDs written to $ARKS_FILE"
+echo "List of IDs fetched successfully"