Add SJP dictionary to Polish spellcheck

Author: m-aciek

.github/workflows/spellcheck.yml

@@ -49,7 +49,7 @@ jobs:
       - name: Install dependencies
         run: |
           sudo apt-get update
-          sudo apt-get install -y hunspell hunspell-pl
+          sudo apt-get install -y curl hunspell hunspell-pl unzip
           make venv
           # Pospell needs newer Docutils than the Sphinx used by older branches.
           uv venv "$RUNNER_TEMP/spellcheck-venv"
@@ -64,6 +64,26 @@ jobs:
           sed -i 's/^SET .*/SET UTF-8/' "$RUNNER_TEMP/hunspell-utf8/pl_PL.aff"
           iconv -f "$dictionary_encoding" -t UTF-8 /usr/share/hunspell/pl_PL.dic \
             > "$RUNNER_TEMP/hunspell-utf8/pl_PL.dic"
+      - name: Download SJP Polish dictionary
+        run: |
+          curl --fail --location --retry 3 \
+            https://sjp.pl/sl/ort/sjp-hunspell-pl-20260901.zip \
+            --output "$RUNNER_TEMP/sjp-hunspell.zip"
+          echo "47a5ab02cf3def374994b1b03efd9fed2de868349b8ddcebac281b484353d157  $RUNNER_TEMP/sjp-hunspell.zip" \
+            | sha256sum --check --strict
+          # SJP wraps the dictionary files in a second ZIP archive.
+          unzip -p "$RUNNER_TEMP/sjp-hunspell.zip" pl_PL.zip \
+            > "$RUNNER_TEMP/sjp-pl_PL.zip"
+          unzip -p "$RUNNER_TEMP/sjp-pl_PL.zip" pl_PL.aff \
+            > "$RUNNER_TEMP/sjp-pl_PL.aff"
+          unzip -p "$RUNNER_TEMP/sjp-pl_PL.zip" pl_PL.dic \
+            > "$RUNNER_TEMP/sjp-pl_PL.dic"
+          dictionary_encoding=$(sed -n 's/^SET //p' "$RUNNER_TEMP/sjp-pl_PL.aff")
+          iconv -f "$dictionary_encoding" -t UTF-8 "$RUNNER_TEMP/sjp-pl_PL.aff" \
+            > "$RUNNER_TEMP/hunspell-utf8/pl_SJP.aff"
+          sed -i 's/^SET .*/SET UTF-8/' "$RUNNER_TEMP/hunspell-utf8/pl_SJP.aff"
+          iconv -f "$dictionary_encoding" -t UTF-8 "$RUNNER_TEMP/sjp-pl_PL.dic" \
+            > "$RUNNER_TEMP/hunspell-utf8/pl_SJP.dic"
       - name: Download Polish terminology glossary
         working-directory: cpython/Doc/locales/pl/LC_MESSAGES
         env:
@@ -90,7 +110,7 @@ jobs:
         env:
           DICPATH: ${{ runner.temp }}/hunspell-utf8
         run: >
-          "$RUNNER_TEMP/spellcheck-venv/bin/pospell" --language pl_PL
+          "$RUNNER_TEMP/spellcheck-venv/bin/pospell" --language pl_PL,pl_SJP
           --personal-dict "$RUNNER_TEMP/hunspell-glossary.dic"
           --glob 'locales/pl/LC_MESSAGES/**/*.po'
       # PO updates exist only in this disposable checkout; nothing is committed.