Merge pull request #1116 from PyThaiNLP/update-docker-fairseq

Update docker and fairseq

Author: wannaphong

Dockerfile

@@ -2,12 +2,13 @@
 # SPDX-FileType: SOURCE
 # SPDX-License-Identifier: Apache-2.0
 
-FROM python:3.8-slim-buster
+FROM python:3.12
 
 COPY . .
 
-RUN apt-get update && apt-get install -y --no-install-recommends build-essential libicu-dev libicu63 pkg-config && rm -rf /var/lib/apt/lists/*
-
-RUN pip3 install --upgrade pip setuptools
-RUN if [ -f docker_requirements.txt ]; then pip3 install -r docker_requirements.txt; fi
-RUN pip3 install -e .[full] && pip3 cache purge
+RUN apt-get update && apt-get install -y --no-install-recommends build-essential libicu-dev  python3-pip python3-venv pkg-config && rm -rf /var/lib/apt/lists/*
+ENV VIRTUAL_ENV=/opt/venv
+RUN python3 -m venv $VIRTUAL_ENV
+ENV PATH="$VIRTUAL_ENV/bin:$PATH"
+RUN if [ -f docker_requirements.txt ]; then pip install -r docker_requirements.txt; fi
+RUN pip install -e .[full] && pip cache purge

docker_requirements.txt

@@ -1,41 +1,37 @@
-PyYAML==5.4.1
+PyYAML>=5.4.1,<6.0.2
 attacut==1.0.6
-bpemb==0.3.6
+bpemb>=0.3.6,<0.4
 deepcut==0.7.0.0
-emoji==0.6.0
+emoji>=0.6.0,<1
 epitran==1.26.0
-esupar==1.3.9
-fairseq==0.12.2
-fastai==1.0.61
+esupar>=1.3.9,<2
+fairseq>=0.10.0,<0.13;python_version<"3.11"
+fairseq-fixed==0.12.3.1,<0.13;python_version>="3.11"
+fastai>=1.0.61,<2
 fastcoref==2.1.6
-gensim==4.3.3
-h5py==3.13.0
-khanaa==0.0.6
-nlpo3==1.3.1
-nltk==3.6.6
-numpy==1.26.*
-OSKut==1.3
-pandas==2.2.*
+gensim>=4.3.3,<5
+khanaa>=0.1.1,<1
+nlpo3>=1.3.1
+nltk>=3.6.6,<4
+numpy>=1.26.0,<2
+pandas>=2.2.0,<3
 panphon==0.21.2
 phunspell==0.1.6
-protobuf==5.29.3
-pyicu==2.15.2
+pyicu>=2.15.2,<3
 python-crfsuite==0.9.11
-requests==2.32.*
+requests>=2.32.0,<2.33
 sacremoses==0.1.1
-sefr_cut==1.1
-sentence-transformers==2.7.0
+sentence-transformers>=2.7.0,<3
 sentencepiece==0.2.0
 spacy_thai==0.7.8
-spacy==3.5.*
+spacy==3.8.7,<4
 ssg==0.0.8
 symspellpy==6.9.0
-tensorflow==2.18.1
 thai-nner==0.3
-tltk==1.6.8
-torch==1.13.1
+tltk>=1.6.8,<2
+torch>=1.13.1,<3
 transformers==4.52.3
 ufal.chu-liu-edmonds==1.0.3
 wtpsplit==1.3.0
 wunsen==0.0.3
-word2word==1.0.0
+word2word>=1.0.0,<2

setup.py

@@ -80,7 +80,8 @@
     "thai2fit": ["emoji>=0.5.1", "gensim>=4.0.0", "numpy>=1.22"],
     "thai2rom": ["numpy>=1.22", "torch>=1.0.0"],
     "translate": [
-        "fairseq>=0.10.0",
+        'fairseq>=0.10.0,<0.13;python_version<"3.11"',
+        'fairseq-fixed==0.12.3.1,<0.13;python_version>="3.11"',
         "sacremoses>=0.0.41",
         "sentencepiece>=0.1.91",
         "torch>=1.0.0",
@@ -117,7 +118,8 @@
         "bpemb>=0.3.2",
         "emoji>=0.5.1",
         "epitran>=1.1",
-        "fairseq>=0.10.0",
+        'fairseq>=0.10.0,<0.13;python_version<"3.11"',
+        'fairseq-fixed==0.12.3.1,<0.13;python_version>="3.11"',
         "fastai<2.0",
         "fastcoref>=2.1.5",
         "gensim>=4.0.0",