Initial commit

ingest/ingest_pdfs.sh

@@ -0,0 +1,16 @@
+#!/usr/bin/env bash
+set -euo pipefail
+BASE_URL=${MEILI_URL:-http://localhost:7700}
+KEY=${MEILI_KEY:-devkey}
+
+for pdf in "$@"; do
+  title="$(basename "$pdf")"
+  pages=$(pdfinfo "$pdf" | awk '/Pages:/ {print $2}')
+  for p in $(seq 1 "$pages"); do
+    text="$(pdftotext -f $p -l $p -layout "$pdf" - | sed 's/^[[:space:]]*$//' )"
+    [ -z "$text" ] && continue
+    doc=$(jq -nc --arg id "${title}-p${p}" --arg t "$title" --arg src "file://$pdf" --arg txt "$text"           '{id:$id, type:"pdf", title:$t, source:$src, date:"", text:$txt, meta:{page:$id}}')
+    curl -sS -X POST "$BASE_URL/indexes/library/documents"       -H "Authorization: Bearer '"$KEY"'" -H 'Content-Type: application/json'       --data-binary "$doc" >/dev/null
+  done
+  echo "Indexed $title ($pages pages)"
+done