First attempt at adding Narwhals to test suite. (#918)

DrTodd13 · Todd A. Anderson · web-flow · commit ad7f1a5f34bd · 2025-11-07T14:16:28.000-08:00
Co-authored-by: Todd A. Anderson &lt;toddanderson@bodo.ai&gt;
diff --git a/.github/workflows/_test_python_source.yml b/.github/workflows/_test_python_source.yml
@@ -23,11 +23,15 @@ on:
         required: false
         default: 'ubuntu-latest'
         type: string
-      # Options: DF_LIB, DF_LIB_NO_JIT, SPAWN, NORMAL
+      # Options: DF_LIB, DF_LIB_NO_JIT, SPAWN, NORMAL, NARWHALS
       test-type:
         type: string
         description: The kind of tests to run e.g. spawn tests
         required: true
+      bodo-version:
+        type: string
+        description: The version string of Bodo.
+        required: true
 
 jobs:
   run:
@@ -43,6 +47,8 @@ jobs:
           uses: ./.github/actions/build-source
           with:
             build-all: true
+          env:
+            SETUPTOOLS_SCM_PRETEND_VERSION: ${{ inputs.bodo-version }}
         - name: Load Hadoop from Cache
           if: runner.os != 'Windows'
           uses: actions/cache/restore@v4
@@ -66,6 +72,10 @@ jobs:
           run: |
             pip install transformers
             pip install --no-deps 'git+https://github.com/apache/polaris.git@release/1.0.x#subdirectory=client/python'
+        - name: Install Narwhals Test dependencies
+          if: ${{ inputs.test-type == 'NARWHALS' }}
+          run: |
+            pip install hypothesis polars sqlframe ibis-framework pyarrow-hotfix
 
         # Run Tests
         - name: Run Tests
@@ -81,40 +91,50 @@ jobs:
             fi
 
             set +eo pipefail
-            # Bodo Tests: Coverage is collected
-            pytest -s -v -Wignore \
-              --cov-report= --cov=bodo \
-              --splits=${{ inputs.total-batches }} --group=${{ inputs.batch }} \
-              --store-durations --clean-durations \
-              --durations-path=buildscripts/github/test_dur_bodo.json \
-              -m "$PYTEST_MARKER" $PYTEST_IGNORE bodo/tests/
 
-            # Save Exit Code for Later
-            python_test_exit_code=$?
+            if [[ "${{ inputs.test-type }}" == "NARWHALS" ]]; then
+              # Checkout narwhals fork
+              git clone https://github.com/bodo-ai/narwhals.git
+              cd narwhals
+              python -c "import bodo; print('Detected Bodo version:', bodo.__version__)"
+              pytest -s -v -Wignore -k bodo
+              exit $?
+            else
+              # Bodo Tests: Coverage is collected
+              pytest -s -v -Wignore \
+                --cov-report= --cov=bodo \
+                --splits=${{ inputs.total-batches }} --group=${{ inputs.batch }} \
+                --store-durations --clean-durations \
+                --durations-path=buildscripts/github/test_dur_bodo.json \
+                -m "$PYTEST_MARKER" $PYTEST_IGNORE bodo/tests/
 
-            # BodoSQL Tests
-            cd BodoSQL
-            pytest -s -v -Wignore \
-              --splits=${{ inputs.total-batches }} --group=${{ inputs.batch }} \
-              --store-durations --clean-durations \
-              --durations-path=../buildscripts/github/test_dur_bodosql.json \
-              -m "$PYTEST_MARKER" bodosql/tests/
-            sql_test_exit_code=$?
+              # Save Exit Code for Later
+              python_test_exit_code=$?
 
-            # Pytest exits code 5 if no tests are run. Some markers are only in python or sql
-            # so we suppress this.
-            if [ "$python_test_exit_code" -eq 5 ]; then
-              python_test_exit_code=0
-            fi
-            if [ "$sql_test_exit_code" -eq 5 ]; then
-              sql_test_exit_code=0
-            fi
+              # BodoSQL Tests
+              cd BodoSQL
+              pytest -s -v -Wignore \
+                --splits=${{ inputs.total-batches }} --group=${{ inputs.batch }} \
+                --store-durations --clean-durations \
+                --durations-path=../buildscripts/github/test_dur_bodosql.json \
+                -m "$PYTEST_MARKER" bodosql/tests/
+              sql_test_exit_code=$?
 
-            # Merge Exit Codes.
-            # If neither Bodo nor BodoSQL fails, the exit code will be 0
-            # If one fails, the exit code will be 1
-            # If both fail, the exit code will be 2
-            exit $((python_test_exit_code + sql_test_exit_code))
+              # Pytest exits code 5 if no tests are run. Some markers are only in python or sql
+              # so we suppress this.
+              if [ "$python_test_exit_code" -eq 5 ]; then
+                python_test_exit_code=0
+              fi
+              if [ "$sql_test_exit_code" -eq 5 ]; then
+                sql_test_exit_code=0
+              fi
+
+              # Merge Exit Codes.
+              # If neither Bodo nor BodoSQL fails, the exit code will be 0
+              # If one fails, the exit code will be 1
+              # If both fail, the exit code will be 2
+              exit $((python_test_exit_code + sql_test_exit_code))
+            fi
           shell: bash
           env:
             BODO_TESTING_ONLY_RUN_1D_VAR: true
@@ -128,6 +148,7 @@ jobs:
             BODO_TEST_SPAWN_MODE: ${{ inputs.test-type == 'SPAWN' && '1' || '0' }}
             # Disabling the DataFrame library for spawn tests since some of the tests
             # create Pandas manager states for testing that are not fully functional.
+            BODO_DATAFRAME_LIBRARY_WARN: ${{ inputs.test-type == 'NARWHALS' && '0' || '1' }}
             BODO_ENABLE_DATAFRAME_LIBRARY: ${{ inputs.test-type != 'SPAWN' && '1' || '0' }}
             BODO_ENABLE_TEST_DATAFRAME_LIBRARY: ${{ startsWith(inputs.test-type, 'DF_LIB') && '1' || '0' }}
             BODOSQL_PY4J_GATEWAY_PORT: "auto"
diff --git a/.github/workflows/pr_ci.yml b/.github/workflows/pr_ci.yml
@@ -16,6 +16,25 @@ concurrency:
   cancel-in-progress: true
 
 jobs:
+  get_version:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout Code
+        uses: actions/checkout@v5
+        with:
+          # For setuptools_scm to be able to determine version,
+          # we need to fetch the entire history
+          fetch-depth: 0
+      - name: Get Version
+        id: get_version
+        run: |
+          set -exo pipefail
+          pip install setuptools_scm
+          # Print to the logs
+          echo "bodo_version=$(python -m setuptools_scm)"
+          echo "bodo_version=$(python -m setuptools_scm)" >> "$GITHUB_OUTPUT"
+    outputs:
+      bodo_version: ${{ steps.get_version.outputs.bodo_version }}
 
   # 1) Validate changes and commit message
   validate:
@@ -107,7 +126,7 @@ jobs:
 
   # 4) Actually run tests
   pr-ci:
-    needs: [compile-bodo]
+    needs: [compile-bodo, get_version]
     name: Test Compiler
     strategy:
       matrix:
@@ -122,10 +141,11 @@ jobs:
       collect-coverage: true
       os: ${{ inputs.runner_os || 'ubuntu-latest' }}
       test-type: NORMAL
+      bodo-version: ${{ needs.get_version.outputs.bodo_version }}
     secrets: inherit
 
   spawn-ci:
-    needs: [compile-bodo]
+    needs: [compile-bodo, get_version]
     name: Test Spawn
     uses: ./.github/workflows/_test_python_source.yml
     with:
@@ -135,10 +155,11 @@ jobs:
       collect-coverage: true
       os: ${{ inputs.runner_os || 'ubuntu-latest' }}
       test-type: SPAWN
+      bodo-version: ${{ needs.get_version.outputs.bodo_version }}
     secrets: inherit
 
   df-lib-non-jit-ci:
-    needs: [compile-bodo]
+    needs: [compile-bodo, get_version]
     name: Test DF Library Non-JIT
     uses: ./.github/workflows/_test_python_source.yml
     with:
@@ -148,10 +169,11 @@ jobs:
       collect-coverage: true
       os: ${{ inputs.runner_os || 'ubuntu-latest' }}
       test-type: "DF_LIB_NO_JIT"
+      bodo-version: ${{ needs.get_version.outputs.bodo_version }}
     secrets: inherit
 
   df-lib-ci:
-    needs: [compile-bodo]
+    needs: [compile-bodo, get_version]
     name: Test DF Library
     strategy:
       matrix:
@@ -166,10 +188,25 @@ jobs:
       collect-coverage: true
       os: ${{ inputs.runner_os || 'ubuntu-latest' }}
       test-type: "DF_LIB"
+      bodo-version: ${{ needs.get_version.outputs.bodo_version }}
+    secrets: inherit
+
+  narwhals-ci:
+    needs: [compile-bodo, get_version]
+    name: Test Narwhals Bodo Integration
+    uses: ./.github/workflows/_test_python_source.yml
+    with:
+      batch: 1
+      total-batches: 1
+      pytest-marker: ""
+      collect-coverage: false
+      os: ${{ inputs.runner_os || 'ubuntu-latest' }}
+      test-type: NARWHALS
+      bodo-version: ${{ needs.get_version.outputs.bodo_version }}
     secrets: inherit
 
   bodosql-cpp-ci:
-    needs: [compile-bodo]
+    needs: [compile-bodo, get_version]
     name: Test BodoSQL C++ Backend
     uses: ./.github/workflows/_test_python_source.yml
     with:
@@ -179,6 +216,7 @@ jobs:
       collect-coverage: true
       os: ${{ inputs.runner_os || 'ubuntu-latest' }}
       test-type: "BODOSQL_CPP"
+      bodo-version: ${{ needs.get_version.outputs.bodo_version }}
     secrets: inherit
 
   java-ci:
@@ -268,6 +306,7 @@ jobs:
       - collect-results
       - df-lib-ci
       - df-lib-non-jit-ci
+      - narwhals-ci
       - bodosql-cpp-ci
       - validate
     runs-on: ubuntu-latest
