From 8369c59dcc43d7901eb71c5e91563523654b27b3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cem=20G=C3=B6kmen?= <cgokmen@stanford.edu>
Date: Sat, 27 Jan 2024 18:34:21 -0800
Subject: [PATCH] Cherry-pick: Updates to workflow

---
 .github/workflows/pull-sheets.yml | 54 +++++++++++++++++++++++++++----
 1 file changed, 48 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/pull-sheets.yml b/.github/workflows/pull-sheets.yml
index 14d1c40e..5ea1b1a5 100644
--- a/.github/workflows/pull-sheets.yml
+++ b/.github/workflows/pull-sheets.yml
@@ -14,10 +14,18 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - name: Checkout code
+    - name: Clone BDDL repo
       uses: actions/checkout@v3
       with:
+        path: bddl
         ref: develop
+
+    - name: Clone ig pipeline repo
+      uses: actions/checkout@v3
+      with:
+        repository: StanfordVL/ig_pipeline
+        path: ig_pipeline
+        token: ${{ secrets.REPO_TOKEN }}
     
     - name: Setup python
       uses: actions/setup-python@v2
@@ -25,23 +33,57 @@ jobs:
         python-version: "3.8"
         architecture: x64
 
+    - name: Authenticate on Google Cloud
+      uses: 'google-github-actions/auth@v1'
+      with:
+        credentials_json: '${{ secrets.GCP_CREDENTIALS }}'
+
+    # See if we need to re-pull any ig_pipeline data from DVC.
+    - name: Check cache for ig_pipeline data
+      id: cache-pipeline
+      uses: actions/cache@v3
+      with:
+        key: ig_pipeline-${{ hashFiles('ig_pipeline/dvc.lock') }}
+        path: |
+          ig_pipeline/artifacts/pipeline/combined_room_object_list.json
+          ig_pipeline/artifacts/pipeline/combined_room_object_list_future.json
+          ig_pipeline/artifacts/pipeline/object_inventory.json
+          ig_pipeline/artifacts/pipeline/object_inventory_future.json
+
+    - if: ${{ steps.cache-pipeline.outputs.cache-hit != 'true' }}
+      name: Install dvc
+      run: pip install dvc[gs]
+
+    - if: ${{ steps.cache-pipeline.outputs.cache-hit != 'true' }}
+      name: Pull dvc data
+      working-directory: ig_pipeline
+      run: dvc pull combined_room_object_list combined_room_object_list_future object_inventory object_inventory_future
+
+    - if: ${{ steps.cache-pipeline.outputs.cache-hit != 'true' }}
+      name: Unprotect data
+      working-directory: ig_pipeline
+      run: dvc unprotect artifacts/pipeline/combined_room_object_list.json artifacts/pipeline/combined_room_object_list_future.json artifacts/pipeline/object_inventory.json artifacts/pipeline/object_inventory_future.json
+
+    - name: Copy over ig_pipeline files
+      run: cp ig_pipeline/artifacts/pipeline/{combined_room_object_list,combined_room_object_list_future,object_inventory,object_inventory_future}.json bddl/bddl/generated_data
+
     - name: Install BDDL
+      working-directory: bddl
       run: pip install -e .
 
     - name: Install dev requirements
+      working-directory: bddl
       run: pip install -r requirements-dev.txt
 
-    - name: Authenticate on Google Cloud
-      uses: 'google-github-actions/auth@v1'
-      with:
-        credentials_json: '${{ secrets.GCP_CREDENTIALS }}'
-
     - name: Refresh sheets data
+      working-directory: bddl
       run: python -m bddl.data_generation.pull_sheets
 
     - name: Refresh derivative data
+      working-directory: bddl
       run: python -m bddl.data_generation.generate_datafiles
 
     - uses: stefanzweifel/git-auto-commit-action@v4
       with:
         commit_message: "Sync Google Sheets data"
+        repository: bddl