Add roundtrip enwik8 hash to CI tests.

Author: BrianPugh

.github/workflows/tests.yaml

@@ -104,6 +104,15 @@ jobs:
           path: ~/.cache/pre-commit/
           key: pre-commit-${{ runner.os }}-${{ env.pythonLocation }}-${{ hashFiles('.pre-commit-config.yaml') }}
 
+      - name: Cache enwik8 dataset
+        uses: actions/cache@v4
+        with:
+          path: build/enwik8.zip
+          key: enwik8-dataset-${{ hashFiles('Makefile') }}
+
+      - name: Download enwik8 dataset
+        run: make download-enwik8
+
       - name: Pre-commit run
         run: |
           source ${{ env.ACTIVATE_PYTHON_VENV }}
@@ -116,6 +125,23 @@ jobs:
           belay run micropython -X heapsize=200M -m unittest tests/*.py
           coverage report
 
+      - name: Verify CLI compression/decompression integrity
+        run: |
+          source ${{ env.ACTIVATE_PYTHON_VENV }}
+          make download-enwik8
+          original_hash=$(sha256sum build/enwik8 | cut -d' ' -f1)
+          roundtrip_hash=$(tamp compress build/enwik8 | tamp decompress | sha256sum | cut -d' ' -f1)
+          if [ "$original_hash" = "$roundtrip_hash" ]; then
+            echo "✓ Compression/decompression integrity verified"
+            echo "Original hash: $original_hash"
+            echo "Roundtrip hash: $roundtrip_hash"
+          else
+            echo "✗ Compression/decompression integrity check failed"
+            echo "Original hash: $original_hash"
+            echo "Roundtrip hash: $roundtrip_hash"
+            exit 1
+          fi
+
       - name: Upload coverage to Codecov
         if: steps.check_test_files.outputs.files_exists == 'true'
         uses: codecov/codecov-action@v4

Makefile

@@ -53,13 +53,20 @@ clean: clean-cython
 	@rm -rf build
 	@rm -rf dist
 
-build/enwik8:
-	if [ ! -f build/enwik8 ]; then \
+build/enwik8.zip:
+	if [ ! -f build/enwik8.zip ]; then \
 		mkdir -p build; \
 		cd build; \
 		curl -O https://mattmahoney.net/dc/enwik8.zip; \
+		cd ..; \
+	fi
+
+download-enwik8-zip: build/enwik8.zip
+
+build/enwik8: build/enwik8.zip
+	if [ ! -f build/enwik8 ]; then \
+		cd build; \
 		unzip -q enwik8.zip; \
-		rm enwik8.zip; \
 		cd ..; \
 	fi