From 7a125382671c638d75d92c89c4b5f2d86ceaabdc Mon Sep 17 00:00:00 2001
From: Michael Genson <71845777+michael-genson@users.noreply.github.com>
Date: Sat, 27 Jun 2026 22:11:55 -0500
Subject: [PATCH] fix: Manually implement nltk.downloader using cURL (#7817)

---
 docker/Dockerfile         |  5 ++--
 docker/setup_nltk_data.sh | 58 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 61 insertions(+), 2 deletions(-)
 create mode 100644 docker/setup_nltk_data.sh

diff --git a/docker/Dockerfile b/docker/Dockerfile
index 24f03cca6..7842c8954 100644
--- a/docker/Dockerfile
+++ b/docker/Dockerfile
@@ -131,6 +131,7 @@ RUN apt-get update \
     iproute2 \
     libldap-common \
     libldap2 \
+    unzip \
     && rm -rf /var/lib/apt/lists/*
 
 # create directory used for Docker Secrets
@@ -141,8 +142,8 @@ COPY --from=venv-builder $VENV_PATH $VENV_PATH
 
 # install nltk data for the ingredient parser
 ENV NLTK_DATA="/nltk_data/"
-RUN mkdir -p $NLTK_DATA
-RUN python -m nltk.downloader -d $NLTK_DATA averaged_perceptron_tagger_eng
+COPY ./docker/setup_nltk_data.sh $MEALIE_HOME/setup_nltk_data.sh
+RUN chmod +x $MEALIE_HOME/setup_nltk_data.sh && $MEALIE_HOME/setup_nltk_data.sh
 
 VOLUME [ "$MEALIE_HOME/data/" ]
 ENV APP_PORT=9000
diff --git a/docker/setup_nltk_data.sh b/docker/setup_nltk_data.sh
new file mode 100644
index 000000000..d523f7d27
--- /dev/null
+++ b/docker/setup_nltk_data.sh
@@ -0,0 +1,58 @@
+#!/bin/bash
+# ------------------------------------------------------------------------------
+# Adapted from: https://github.com/community-scripts/ProxmoxVE/pull/14314
+#
+# Download NLTK data packages directly from GitHub, bypassing Python.
+# Avoids CPU-instruction failures (SIGILL) on older hardware lacking AVX.
+#
+# Intended for use during Docker image build only.
+#
+# Environment:
+#   NLTK_DATA - Target directory for NLTK data (default: /usr/share/nltk_data)
+#
+# Returns: 0 on success, non-zero if any package failed
+# ------------------------------------------------------------------------------
+set -euo pipefail
+
+NLTK_INDEX_URL="https://raw.githubusercontent.com/nltk/nltk_data/gh-pages/index.xml"
+PACKAGES="averaged_perceptron_tagger_eng"
+TARGET_DIR="${NLTK_DATA:-/usr/share/nltk_data}"
+
+echo "Fetching NLTK package index..."
+index_xml=$(curl -fsSL "$NLTK_INDEX_URL")
+
+mkdir -p "$TARGET_DIR"
+
+for pkg in $PACKAGES; do
+  echo "Installing NLTK package: $pkg"
+
+  pkg_line=$(echo "$index_xml" | grep "id=\"${pkg}\"" | head -1)
+  if [[ -z "$pkg_line" ]]; then
+    echo "ERROR: NLTK package not found in index: $pkg" >&2
+    exit 1
+  fi
+
+  subdir=$(echo "$pkg_line" | grep -oP 'subdir="\K[^"]+')
+  pkg_url=$(echo "$pkg_line" | grep -oP 'url="\K[^"]+')
+  do_unzip=$(echo "$pkg_line" | grep -oP 'unzip="\K[^"]+')
+
+  if [[ -z "$subdir" || -z "$pkg_url" ]]; then
+    echo "ERROR: Could not parse NLTK index entry for: $pkg" >&2
+    exit 1
+  fi
+
+  mkdir -p "${TARGET_DIR}/${subdir}"
+  tmp_zip=$(mktemp --suffix=.zip)
+
+  echo "Downloading: $pkg_url"
+  curl -fsSL -o "$tmp_zip" "$pkg_url"
+
+  if [[ "$do_unzip" == "1" ]]; then
+    unzip -q -o "$tmp_zip" -d "${TARGET_DIR}/${subdir}/"
+    rm -f "$tmp_zip"
+  else
+    mv "$tmp_zip" "${TARGET_DIR}/${subdir}/${pkg}.zip"
+  fi
+
+  echo "Done: $pkg"
+done