add missing nltk download in CI (#5529)

* add missing nltk download in CI

* fix

* fix

* fix docker run

* upgrade cached-path for bug fix

.github/workflows/ci.yml

@@ -350,6 +350,8 @@ jobs:
         python: ['3.7', '3.8', '3.9']
 
     steps:
+    - uses: actions/checkout@v2
+
     - name: Setup Python
       uses: actions/setup-python@v2
       with:
@@ -369,6 +371,14 @@ jobs:
       run: |
         pip install $(ls dist/*.whl)
 
+    - name: Download NLTK prerequisites
+      run: |
+        make download-extras
+
+    - name: Cleanup workspace
+      run: |
+        rm -rf allennlp/ setup.py tests/ test_fixtures/
+
     - name: Pip freeze
       run: |
         pip freeze
@@ -407,7 +417,7 @@ jobs:
 
     - name: Test image
       run: |
-        make docker-run DOCKER_GPUS='' DOCKER_IMAGE_NAME=$DOCKER_IMAGE_NAME ARGS='test-install'
+        docker run --rm $DOCKER_IMAGE_NAME test-install
 
     - name: Authenticate to Docker Hub
       if: github.event_name == 'release' || github.event_name == 'push'

Dockerfile

@@ -25,6 +25,9 @@ RUN touch allennlp/__init__.py \
 COPY allennlp allennlp
 RUN pip install --no-cache-dir --no-deps -e .
 
+COPY Makefile .
+RUN make download-extras
+
 WORKDIR /app/
 
 # Copy wrapper script to allow beaker to run resumable training workloads.

Makefile

@@ -25,7 +25,7 @@ DOCKER_RUN_CMD = docker run --rm \
 # These nltk packages are used by the 'checklist' module. They are downloaded automatically
 # if not found when `checklist` is imported, but it's good to download the ahead of time
 # to avoid potential race conditions.
-NLTK_DOWNLOAD_CMD = python -c 'import nltk; [nltk.download(p) for p in ("wordnet", "wordnet_ic", "sentiwordnet")]'
+NLTK_DOWNLOAD_CMD = python -c 'import nltk; [nltk.download(p) for p in ("wordnet", "wordnet_ic", "sentiwordnet", "omw-1.4")]'
 
 ifeq ($(shell uname),Darwin)
 ifeq ($(shell which gsed),)

setup.py

@@ -52,7 +52,7 @@
     install_requires=[
         "torch>=1.6.0,<1.11.0",
         "torchvision>=0.8.1,<0.12.0",
-        "cached-path>=1.0.0,<2.0.0",
+        "cached-path>=1.0.2,<2.0.0",
         "fairscale==0.4.3",
         "jsonnet>=0.10.0 ; sys.platform != 'win32'",
         "nltk",