build(docker): switch to chainguard wolfi-base, upgrade deps to 0.0.93

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 0.0.93
+* Refactored the Dockerfile to use the chainguard/wolfi-base image instead of the unstructured/base-image. This is to align with the recent change in the unstructured repo where the same change was made.
+* upgraded dependancies to address CVEs
+
 ## 0.0.92
 * Upgrade pdfminer-six to 20260107 to fix ~15-18% performance regression from eager f-string evaluation
 

Dockerfile

@@ -1,5 +1,5 @@
 # syntax=docker/dockerfile:experimental
-FROM quay.io/unstructured-io/base-images:wolfi-base-latest as base
+FROM cgr.dev/chainguard/wolfi-base:latest
 
 # NOTE(crag): NB_USER ARG for mybinder.org compat:
 #             https://mybinder.readthedocs.io/en/latest/tutorials/dockerfile.html
@@ -10,25 +10,65 @@ ARG PIPELINE_PACKAGE
 ARG PYTHON_VERSION="3.12"
 
 # Set up environment
-ENV PYTHON python${PYTHON_VERSION}
-ENV PIP ${PYTHON} -m pip
+ENV PYTHON=python${PYTHON_VERSION}
+ENV PIP="${PYTHON} -m pip"
+
+USER root
+
+COPY ./docker/packages/*.apk /tmp/packages/
+
+RUN apk update && \
+    apk add libxml2 python-3.12 python-3.12-base py3.12-pip glib \
+      mesa-gl mesa-libgallium cmake bash libmagic wget git openjpeg \
+      poppler poppler-utils poppler-glib libreoffice tesseract && \
+    apk add --allow-untrusted /tmp/packages/pandoc-3.1.8-r0.apk && \
+    rm -rf /tmp/packages && \
+    git clone --depth 1 https://github.com/tesseract-ocr/tessdata.git /tmp/tessdata && \
+    mkdir -p /usr/local/share/tessdata && \
+    cp /tmp/tessdata/*.traineddata /usr/local/share/tessdata && \
+    rm -rf /tmp/tessdata && \
+    git clone --depth 1 https://github.com/tesseract-ocr/tessconfigs /tmp/tessconfigs && \
+    cp -r /tmp/tessconfigs/configs /usr/local/share/tessdata && \
+    cp -r /tmp/tessconfigs/tessconfigs /usr/local/share/tessdata && \
+    rm -rf /tmp/tessconfigs && \
+    apk cache clean && \
+    ln -s /usr/lib/libreoffice/program/soffice.bin /usr/bin/libreoffice && \
+    ln -s /usr/lib/libreoffice/program/soffice.bin /usr/bin/soffice && \
+    chmod +x /usr/lib/libreoffice/program/soffice.bin && \
+    apk add --no-cache font-ubuntu fontconfig && \
+    apk upgrade --no-cache py3.12-pip && \
+    fc-cache -fv && \
+    ln -sf /usr/bin/$PYTHON /usr/bin/python3 && \
+    addgroup --gid ${NB_UID} ${NB_USER} && \
+    adduser --disabled-password --gecos "" --uid ${NB_UID} -G ${NB_USER} ${NB_USER} && \
+    rm -rf /usr/lib/python3.10 && \
+    rm -rf /usr/lib/python3.11 && \
+    rm -rf /usr/lib/python3.13 && \
+    rm -f /usr/bin/python3.13
+
+ENV USER=${NB_USER}
+ENV HOME=/home/${NB_USER}
+COPY --chown=${NB_USER} scripts/initialize-libreoffice.sh ${HOME}/initialize-libreoffice.sh
 
-WORKDIR ${HOME}
 USER ${NB_USER}
+WORKDIR ${HOME}
+
+# Initialize libreoffice config as non-root user (required for soffice to work properly)
+# See: https://github.com/Unstructured-IO/unstructured/issues/3105
+RUN ./initialize-libreoffice.sh && rm initialize-libreoffice.sh
 
 ENV PYTHONPATH="${PYTHONPATH}:${HOME}"
 ENV PATH="/home/${NB_USER}/.local/bin:${PATH}"
+ENV TESSDATA_PREFIX=/usr/local/share/tessdata
 
-FROM base as python-deps
 COPY --chown=${NB_USER}:${NB_USER} requirements/base.txt requirements-base.txt
-RUN ${PIP} install pip==${PIP_VERSION}
-RUN ${PIP} install --no-cache -r requirements-base.txt
+RUN ${PIP} install pip==${PIP_VERSION} && \
+    ${PIP} install --no-cache -r requirements-base.txt
 
-FROM python-deps as model-deps
 RUN ${PYTHON} -c "from unstructured.nlp.tokenize import download_nltk_packages; download_nltk_packages()" && \
-  ${PYTHON} -c "from unstructured.partition.model_init import initialize; initialize()"
+    ${PYTHON} -c "from unstructured.partition.model_init import initialize; initialize()" && \
+    ${PYTHON} -c "from unstructured_inference.models.tables import UnstructuredTableTransformerModel; model = UnstructuredTableTransformerModel(); model.initialize('microsoft/table-transformer-structure-recognition')"
 
-FROM model-deps as code
 COPY --chown=${NB_USER}:${NB_USER} CHANGELOG.md CHANGELOG.md
 COPY --chown=${NB_USER}:${NB_USER} logger_config.yaml logger_config.yaml
 COPY --chown=${NB_USER}:${NB_USER} prepline_${PIPELINE_PACKAGE}/ prepline_${PIPELINE_PACKAGE}/

prepline_general/api/__version__.py

@@ -1 +1 @@
-__version__ = "0.0.92"  # pragma: no cover
+__version__ = "0.0.93"  # pragma: no cover

preprocessing-pipeline-family.yaml

@@ -1,2 +1,2 @@
 name: general
-version: 0.0.92
+version: 0.0.93