pbtxt -> pb in submission workflow (open-reaction-database#53)

* pbtxt -> pb in submission workflow * update ord-schema version * remove pbtxt * Update submission.yml * Update validation.yml
skearnes · Feb 8, 2021 · 4728834 · 4728834
1 parent c96a1e6
commit 4728834
Show file tree

Hide file tree

Showing 8 changed files with 11 additions and 5,135,284 deletions.
diff --git a/.github/workflows/submission.yml b/.github/workflows/submission.yml
@@ -10,7 +10,7 @@ name: Submission
 on: [pull_request, push]
 
 env:
-  ORD_SCHEMA_TAG: v0.2.4
+  ORD_SCHEMA_TAG: v0.2.7
 
 jobs:
   count_reactions:
@@ -129,13 +129,12 @@ jobs:
         echo "Found $(wc -l < changed_files.txt | tr -d ' ') changed files"
         cat changed_files.txt
         # Use `|| (( $? == 1 ))` in case no lines match (exit code is nonzero).
-        grep -E "\.pbtxt$" changed_files.txt > changed_pbtxt_files.txt || (( $? == 1 ))
         grep -E "\.pb$" changed_files.txt > changed_pb_files.txt || (( $? == 1 ))
         # Use LOCAL_NUM_CHANGED since ::set-env values are not available immediately.
-        LOCAL_NUM_CHANGED="$(wc -l < changed_pbtxt_files.txt | tr -d ' ')"
+        LOCAL_NUM_CHANGED="$(wc -l < changed_pb_files.txt | tr -d ' ')"
         echo "NUM_CHANGED_FILES=${LOCAL_NUM_CHANGED}" >> $GITHUB_ENV
-        echo "Found ${LOCAL_NUM_CHANGED} changed pbtxt files"
-        cat changed_pbtxt_files.txt
+        echo "Found ${LOCAL_NUM_CHANGED} changed pb files"
+        cat changed_pb_files.txt
     - name: Install miniconda
       uses: s-weigand/setup-conda@v1
       with:
@@ -151,25 +150,15 @@ jobs:
     - name: Validate submission
       run: |
         cd "${GITHUB_WORKSPACE}"
-        python ./ord-schema/ord_schema/process_dataset.py \
-          --input_file=changed_pbtxt_files.txt \
+        python ./ord-schema/ord_schema/scripts/process_dataset.py \
+          --input_file=changed_pb_files.txt \
           --base=upstream/main
-        if test "$( wc -l < changed_pb_files.txt )" -gt 0; then
-          python ./ord-schema/ord_schema/process_dataset.py \
-            --input_file=changed_pb_files.txt
-          for filename in $(cat changed_pb_files.txt | awk '{print $2}'); do
-            echo "${filename}"
-            python ./ord-schema/ord_schema/scripts/check_pb.py \
-              --pb="${filename}" \
-              --pbtxt="${filename}txt"
-          done
-        fi
       if: env.NUM_CHANGED_FILES != '0'
     - name: Update submission
       run: |
         cd "${GITHUB_WORKSPACE}"
-        python ./ord-schema/ord_schema/process_dataset.py \
-          --input_file=changed_pbtxt_files.txt \
+        python ./ord-schema/ord_schema/scripts/process_dataset.py \
+          --input_file=changed_pb_files.txt \
           --update \
           --cleanup \
           --base=upstream/main \

diff --git a/.github/workflows/validation.yml b/.github/workflows/validation.yml
@@ -15,7 +15,7 @@ on:
       - '.github/workflows/validation.yml'
 
 env:
-  ORD_SCHEMA_TAG: v0.2.4
+  ORD_SCHEMA_TAG: v0.2.7
 
 jobs:
   validate_database:
@@ -39,22 +39,8 @@ jobs:
         pip install -r requirements.txt
         conda install -c rdkit rdkit
         python setup.py install
-    - name: Validate *.pbtxt datasets
+    - name: Validate datasets
       run: |
         cd "${GITHUB_WORKSPACE}"
-        python ./ord-schema/ord_schema/process_dataset.py \
-          --input_pattern="data/*/*.pbtxt"
-    - name: Validate *.pb datasets
-      run: |
-        cd "${GITHUB_WORKSPACE}"
-        python ./ord-schema/ord_schema/process_dataset.py \
+        python ./ord-schema/ord_schema/scripts/process_dataset.py \
           --input_pattern="data/*/*.pb"
-    - name: Check pbtxt <-> pb encoding
-      run: |
-        cd "${GITHUB_WORKSPACE}"
-        for filename in data/*/*.pb; do
-          echo "${filename}"
-          python ./ord-schema/ord_schema/scripts/check_pb.py \
-            --pb="${filename}" \
-            --pbtxt="${filename}txt"
-        done