Add test case for reading non-pandas parquet file (#870)

* Add test case for reading non-pandas parquet file

* daskver

* update pandas build flag

* also dep

* move dep to conda

* another dep

* localserver too

---------

Co-authored-by: Martin Durant <martin.durant@alumni.utoronto.ca>

.github/workflows/main.yaml

@@ -85,7 +85,7 @@ jobs:
       - name: Setup conda
         uses: mamba-org/provision-with-micromamba@main
         with:
-          environment-file: ci/environment-py38.yml
+          environment-file: ci/environment-py39.yml
 
       - name: pip-install
         shell: bash -l {0}
@@ -122,12 +122,12 @@ jobs:
         run: |
           pip install Cython
           pip install hypothesis
-          pip install pytest-xdist pytest-asyncio # Needed for a pandas test, also runs tests in parallel
+          pip install pytest-httpserver pytest-localserver pytest-xdist pytest-asyncio
           pip install -e . --no-deps # Install fastparquet
           git clone https://github.com/pandas-dev/pandas
           cd pandas
           python setup.py build_ext -j 4
-          pip install -e . --no-build-isolation --no-use-pep517
+          pip install -e . --no-build-isolation
 
       - name: Run Tests
         shell: bash -l {0}

ci/environment-py310.yml

@@ -19,3 +19,5 @@ dependencies:
   - ujson
   - python-rapidjson
   - versioneer
+  - meson-python
+  - pyarrow

ci/environment-py38.yml

@@ -17,3 +17,4 @@ dependencies:
   - orjson
   - ujson
   - python-rapidjson
+  - pyarrow

ci/environment-py38win.yml

@@ -17,3 +17,4 @@ dependencies:
   - orjson
   - ujson
   - python-rapidjson
+  - pyarrow

ci/environment-py39.yml

@@ -18,3 +18,4 @@ dependencies:
   - orjson
   - ujson
   - python-rapidjson
+  - pyarrow

fastparquet/test/test_api.py

@@ -1525,3 +1525,17 @@ def test_select_or_iter():
 
     assert df1["id"].tolist() == dfs[0]["id"].tolist() == list(range(32))
 
+
+def test_read_a_non_pandas_parquet_file(tempdir):
+    pa = pytest.importorskip("pyarrow")
+    pq = pytest.importorskip("pyarrow.parquet")
+
+    fn = os.path.join(tempdir, "test.parquet")
+
+    test_table = pa.table({"foo": [0, 1], "bar": ["a", "b"]})
+    pq.write_table(test_table, fn)
+
+    parquet_file = ParquetFile(fn)
+
+    assert parquet_file.count() == 2
+    assert parquet_file.head(1).equals(pd.DataFrame({"foo": [0], "bar": ["a"]}))