listgen, support: attempt to improve data quality

Author: woodruffw

.github/workflows/listgen.yml

@@ -2,6 +2,11 @@ name: Generate stdlib lists
 
 on:
   workflow_dispatch:
+    inputs:
+      dry-run:
+        description: "Don't create any PRs for this run"
+        type: boolean
+        default: false
   schedule:
     - cron: "0 0 * * 2"
 
@@ -95,13 +100,20 @@ jobs:
       - name: walk modules
         env:
           LISTGEN_PYTHON_VERSION: ${{ matrix.python }}
+          LISTGEN_DRY_RUN: ${{ inputs.dry-run }}
         run: |
           python -V
           python ./support/walk-modules.py "${LISTGEN_PYTHON_VERSION}.txt"
           sort -o "${LISTGEN_PYTHON_VERSION}.txt" "${LISTGEN_PYTHON_VERSION}.txt"
-          mv "${LISTGEN_PYTHON_VERSION}.txt" ./stdlib_list/lists/
+
+          if [[ "${LISTGEN_DRY_RUN}" == "true" ]]; then
+            diff ./stdlib_list/lists/"${LISTGEN_PYTHON_VERSION}.txt" "${LISTGEN_PYTHON_VERSION}.txt"
+          else
+            mv "${LISTGEN_PYTHON_VERSION}.txt" ./stdlib_list/lists/
+          fi
 
       - name: create PR
+        if: ${{ !inputs.dry-run }}
         uses: peter-evans/create-pull-request@v5
         with:
           commit-message: "[BOT] update list for ${{ matrix.python }}"

support/walk-modules.py

@@ -83,4 +83,9 @@ def walk(mod_name, io):
                 walk(mod_name, io)
         else:
             for mod_name in sys.stdin:
+                # Our precomputed list might not start at the root, since it
+                # might be a package rather than a module.
+                if "." in mod_name:
+                    top_mod = mod_name.split(".")[0]
+                    walk(top_mod, io)
                 walk(mod_name.rstrip("\n"), io)