Use ct-python-platform base image

Dockerfile

@@ -1,45 +1,10 @@
-FROM --platform=linux/amd64 python:3.11-slim-bookworm
-# Decord is not available on ARM64 (=OSX), so we are forced to build amd64
-# Future todo: The decord source build instructions here are insufficient and outdated
-# https://github.com/dmlc/decord#installation
-
-
-# Install OS dependencies
-RUN apt-get -qq update && \
- apt-get -qq install --no-install-recommends vim-tiny netcat-openbsd curl git wget ffmpeg build-essential libsm6 libxext6 libxrender-dev automake libtool pkg-config libsdl-pango-dev libicu-dev libcairo2-dev bc libleptonica-dev && \
- apt-get -qq clean autoclean && \
- apt-get -qq autoremove && \
- rm -rf /var/lib/apt/lists/*
-
-# Build stuff for tesseract
-# Based on https://medium.com/quantrium-tech/installing-tesseract-4-on-ubuntu-18-04-b6fcd0cbd78f
-RUN curl -L https://github.com/tesseract-ocr/tesseract/archive/refs/tags/4.1.3.tar.gz | tar xvz
-
-#RUN curl -L https://github.com/tesseract-ocr/tesseract/archive/refs/tags/4.1.1.tar.gz | tar xvz
-# Max threads 2 required for cross platform build on M1 Macs :-( otherwise a build all fails even with 6GB RAM +3GB swap for Docker machine
-ARG MAX_THREADS="2"
-
-WORKDIR /tesseract-4.1.3
-RUN ./autogen.sh && ./configure && make -j ${MAX_THREADS} && make -j ${MAX_THREADS} install && ldconfig
-# Slow! The above line takes 435 seconds on my laptop (1590.8s on a M1 cross compiling to amd64)
-RUN make -j ${MAX_THREADS} training && make -j ${MAX_THREADS} training-install
-# The above line takes 59 seconds on my laptop. 127.3s on my M1 laptop cross compiling
-
-RUN curl -L -o tessdata/eng.traineddata https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata
-RUN curl -L -o tessdata/osd.traineddata https://github.com/tesseract-ocr/tessdata/raw/main/osd.traineddata
-
-ENV TESSDATA_PREFIX=/tesseract-4.1.3/tessdata
-#Disable multi-threading
-ENV OMP_THREAD_LIMIT=1
+FROM classtranscribe/ct-python-platform:staging
+#Tofix: no 'latest' tag; only staging tag exists at the moment
 
 # Install Python dependencies
 WORKDIR /usr/app
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
-# Slow. The above line took 5055.6s on my M1 16GB laptop (cross compiling, 6GB Ram+8cores for docker VM; maybe it was swapping...)
-
-# Additional dependencies for brown corpus/stopwords, wordnet
-RUN python -m nltk.downloader brown stopwords wordnet omw-1.4
 
 # Copy in Python source
 COPY . .

requirements.txt

@@ -8,14 +8,15 @@ jsonschema==3.2.0
 Jinja2
 
 # ai/sci tools
-numpy==1.23.5
-pytesseract==0.3.9
-opencv-python==4.5.5.64
-decord==0.6.0
-scikit-image==0.19.2
-mtcnn-opencv==1.0.2
-nltk==3.7
-prefixspan==0.5.2
+# These are already preinstalled in the base image (see requirements-platform.txt)
+##numpy==1.23.5
+## pytesseract==0.3.9
+## opencv-python==4.5.5.64
+## decord==0.6.0
+## scikit-image==0.19.2
+## mtcnn-opencv==1.0.2
+## nltk==3.7
+## prefixspan==0.5.2
 
 # Flask + Connexion + OpenAPI dependencies
 Flask==2.0.2