fix(ci): prevent gha caching conflicts (#643)

Author: ryannikolaidis

.github/workflows/ci.yml

@@ -1,14 +1,13 @@
 name: CI
 
 on:
-  # NOTE(robinson) - We are limiting when we run CI avoid exceeding our 2,000 min/month limit.
-  # We can switch to running on push if we make this repo public or are fine with
-  # paying for CI minutes.
   push:
     branches: [ main ]
   pull_request:
     branches: [ main ]
 
+env:
+  GHA_CACHE_KEY_VERSION: "v1"
 
 jobs:
   setup:
@@ -26,7 +25,7 @@ jobs:
         path: |
           .venv
           nltk_data
-        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ hashFiles('requirements/*.txt') }}
+        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ env.GHA_CACHE_KEY_VERSION }}-${{ hashFiles('requirements/*.txt') }}
     - name: Set up Python ${{ matrix.python-version }}
       uses: actions/setup-python@v4
       with:
@@ -36,6 +35,7 @@ jobs:
       run: |
         python${{ matrix.python-version }} -m venv .venv
         source .venv/bin/activate
+        mkdir "$NLTK_DATA"
         make install-ci
 
   check-deps:
@@ -46,13 +46,13 @@ jobs:
     needs: setup
     steps:
     - uses: actions/checkout@v3
-    - uses: actions/cache@v3
+    - uses: actions/cache/restore@v3
       id: virtualenv-cache
       with:
-        path: .venv
-        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ hashFiles('requirements/*.txt') }}
-    # NOTE(robinson) - This is a fallback in case the lint job does not find the cache.
-    # We can take this out when we implement the fix in CORE-99
+        path: |
+          .venv
+          nltk_data
+        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ env.GHA_CACHE_KEY_VERSION }}-${{ hashFiles('requirements/*.txt') }}
     - name: Set up Python ${{ matrix.python-version }}
       uses: actions/setup-python@v4
       with:
@@ -76,13 +76,13 @@ jobs:
     needs: setup
     steps:
     - uses: actions/checkout@v3
-    - uses: actions/cache@v3
+    - uses: actions/cache/restore@v3
       id: virtualenv-cache
       with:
-        path: .venv
-        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ hashFiles('requirements/*.txt') }}
-    # NOTE(robinson) - This is a fallback in case the lint job does not find the cache.
-    # We can take this out when we implement the fix in CORE-99
+        path: |
+          .venv
+          nltk_data
+        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ env.GHA_CACHE_KEY_VERSION }}-${{ hashFiles('requirements/*.txt') }}
     - name: Set up Python ${{ matrix.python-version }}
       uses: actions/setup-python@v4
       with:
@@ -115,20 +115,19 @@ jobs:
     needs: [setup, lint]
     steps:
     - uses: actions/checkout@v3
-    - uses: actions/cache@v3
+    - uses: actions/cache/restore@v3
       id: virtualenv-cache
       with:
         path: |
           .venv
           nltk_data
-        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ hashFiles('requirements/*.txt') }}
-    # NOTE(robinson) - This is a fallback in case the job does not find the cache.
-    # We can take this out when we implement the fix in CORE-99
+        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ env.GHA_CACHE_KEY_VERSION }}-${{ hashFiles('requirements/*.txt') }}
     - name: Setup virtual environment (no cache hit)
       if: steps.virtualenv-cache.outputs.cache-hit != 'true'
       run: |
         python${{ matrix.python-version}} -m venv .venv
         source .venv/bin/activate
+        mkdir "$NLTK_DATA"
         make install-ci
     - name: Test
       run: |
@@ -152,20 +151,19 @@ jobs:
     needs: [setup, lint]
     steps:
     - uses: actions/checkout@v3
-    - uses: actions/cache@v3
+    - uses: actions/cache/restore@v3
       id: virtualenv-cache
       with:
         path: |
           .venv
           nltk_data
-        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ hashFiles('requirements/*.txt') }}
-    # NOTE(robinson) - This is a fallback in case the lint job does not find the cache.
-    # We can take this out when we implement the fix in CORE-99
+        key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ env.GHA_CACHE_KEY_VERSION }}-${{ hashFiles('requirements/*.txt') }}
     - name: Setup virtual environment (no cache hit)
       if: steps.virtualenv-cache.outputs.cache-hit != 'true'
       run: |
         python${{ matrix.python-version}} -m venv .venv
         source .venv/bin/activate
+        mkdir "$NLTK_DATA"
         make install-ci
     - name: Test
       env:
@@ -210,13 +208,13 @@ jobs:
     needs: [ setup, lint ]
     steps:
       - uses: actions/checkout@v3
-      - uses: actions/cache@v3
+      - uses: actions/cache/restore@v3
         id: virtualenv-cache
         with:
           path: |
             .venv
             nltk_data
-          key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ hashFiles('requirements/*.txt') }}
+          key: unstructured-${{ runner.os }}-${{ matrix.python-version }}-${{ env.GHA_CACHE_KEY_VERSION }}-${{ hashFiles('requirements/*.txt') }}
       - name: Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v4
         with: