Add USPTO-480K dataset from https://doi.org/10.1039/C8SC04228D (#61) (#…

…64)

* Add USPTO-480K dataset from https://doi.org/10.1039/C8SC04228D (#61)

* Add USPTO-480K dataset from https://doi.org/10.1039/C8SC04228D

* trigger

* bump ord-schema version

* update to just use reaction smiles

* bump ord-schema version

* trigger

* remove LFS placeholder

* REACTION_SMILES

* fix typo

* Update submission

* Update badges

* Update validation.yml

* Update submission

* bump ord-schema version

* Update submission

Co-authored-by: github-actions <[email protected]>

.gitattributes

@@ -1,2 +1,3 @@
 ord_data-* filter=lfs diff=lfs merge=lfs -text
 *.pb filter=lfs diff=lfs merge=lfs -text
+*.pb.gz filter=lfs diff=lfs merge=lfs -text

.github/workflows/submission.yml

@@ -10,7 +10,7 @@ name: Submission
 on: [pull_request, push]
 
 env:
-  ORD_SCHEMA_TAG: v0.2.7
+  ORD_SCHEMA_TAG: v0.2.12
 
 jobs:
   count_reactions:
@@ -68,7 +68,7 @@ jobs:
       run: |
         cd "${GITHUB_WORKSPACE}"
         git diff --name-only upstream/main > changed_files.txt
-        grep -vE '\.pb(txt)?$' changed_files.txt && \
+        grep -vE '\.pb(txt)?(.gz)?$' changed_files.txt && \
           echo "Error: submissions must only contain *.pbtxt or *.pb files" && \
           exit 1 || true
       if: >-
@@ -78,7 +78,7 @@ jobs:
       run: |
         cd "${GITHUB_WORKSPACE}"
         git diff --name-only upstream/main > changed_files.txt
-        grep -vE '\.pbtxt$|\.pb$|\.svg$' changed_files.txt && \
+        grep -vE '\.pbtxt(.gz)?$|\.pb(.gz)?$' changed_files.txt && \
           echo "Error: submission contains unusual file types" && \
           exit 1 || true
       if: >-
@@ -133,7 +133,7 @@ jobs:
         echo "Found $(wc -l < changed_files.txt | tr -d ' ') changed files"
         cat changed_files.txt
         # Use `|| (( $? == 1 ))` in case no lines match (exit code is nonzero).
-        grep -E "\.pb$" changed_files.txt > changed_pb_files.txt || (( $? == 1 ))
+        grep -E "\.pb(.gz)?$" 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_pb_files.txt | tr -d ' ')"
         echo "NUM_CHANGED_FILES=${LOCAL_NUM_CHANGED}" >> $GITHUB_ENV

.github/workflows/validation.yml

@@ -6,16 +6,13 @@ on:
   push:
     branches:
       - main
-  schedule:
-    # Runs every night at midnight.
-    - cron: '0 0 * * *'
   pull_request:
     paths:
       # Runs when this file is modified in a PR.
       - '.github/workflows/validation.yml'
 
 env:
-  ORD_SCHEMA_TAG: v0.2.7
+  ORD_SCHEMA_TAG: v0.2.12
 
 jobs:
   validate_database:
@@ -45,4 +42,4 @@ jobs:
       run: |
         cd "${GITHUB_WORKSPACE}"
         python ./ord-schema/ord_schema/scripts/process_dataset.py \
-          --input_pattern="data/*/*.pb"
+          --input_pattern="data/*/*.pb*"