ARROW-7936: [Python] Fix and exercise tests on python 3.5

- added bindings for the Codec class
- support to check compiled/available compression types
- test on ubuntu 16.04 with python 3.5.2 (needs to pin dependencies because of python typing incompatibilities)
- skip the failing Filesystem.from_uri tests on python 3.5 because pathlib.Path.resolve checks that the path exists (which is not)
- run a github actions build with python 3.5.2 test for each commit

Closes #6503 from kszucs/py35 and squashes the following commits:

a5460d2 <Krisztián Szűcs> Docstrings and tests
629e03b <Krisztián Szűcs> trigger ci
94af5a0 <Krisztián Szűcs> OrderedDict
2c85212 <Krisztián Szűcs> fix workflow syntax
649c0b5 <Krisztián Szűcs> ARROW-7936:  Fix and exercise tests on python 3.5

Authored-by: Krisztián Szűcs <szucs.krisztian@gmail.com>
Signed-off-by: Krisztián Szűcs <szucs.krisztian@gmail.com>

.github/workflows/python.yml

@@ -34,11 +34,15 @@ on:
 jobs:
 
   ubuntu:
-    name: AMD64 Ubuntu 18.04 Python 3
+    name: AMD64 Ubuntu ${{ matrix.ubuntu }} Python 3
     runs-on: ubuntu-latest
     if: github.event_name == 'push'
+    strategy:
+      fail-fast: false
+      matrix:
+        ubuntu: [16.04, 18.04]
     env:
-      UBUNTU: 18.04
+      UBUNTU: ${{ matrix.ubuntu }}
     steps:
       - name: Checkout Arrow
         uses: actions/checkout@v1

ci/docker/ubuntu-16.04-cpp.dockerfile

@@ -44,7 +44,6 @@ RUN apt-get update -y -q && \
         libgoogle-glog-dev \
         liblz4-dev \
         libre2-dev \
-        libsnappy-dev \
         libssl-dev \
         llvm-7-dev \
         make \
@@ -65,8 +64,8 @@ RUN apt-get update -y -q && \
 #   unit tests, so doing vendored build by default
 ENV ARROW_BUILD_BENCHMARKS=OFF \
     ARROW_BUILD_TESTS=ON \
-    ARROW_DEPENDENCY_SOURCE=SYSTEM \
     ARROW_DATASET=ON \
+    ARROW_DEPENDENCY_SOURCE=SYSTEM \
     ARROW_GANDIVA_JAVA=OFF \
     ARROW_GANDIVA=ON \
     ARROW_HOME=/usr/local \
@@ -85,9 +84,10 @@ ENV ARROW_BUILD_BENCHMARKS=OFF \
     gRPC_SOURCE=BUNDLED \
     GTest_SOURCE=BUNDLED \
     ORC_SOURCE=BUNDLED \
-    PARQUET_BUILD_EXECUTABLES=ON \
     PARQUET_BUILD_EXAMPLES=ON \
+    PARQUET_BUILD_EXECUTABLES=ON \
     PATH=/usr/lib/ccache/:$PATH \
     Protobuf_SOURCE=BUNDLED \
     RapidJSON_SOURCE=BUNDLED \
+    Snappy_SOURCE=BUNDLED \
     Thrift_SOURCE=BUNDLED

ci/scripts/cpp_build.sh

@@ -120,6 +120,7 @@ cmake -G "${CMAKE_GENERATOR:-Ninja}" \
       -DProtobuf_SOURCE=${Protobuf_SOURCE:-AUTO} \
       -DRapidJSON_SOURCE=${RapidJSON_SOURCE:-AUTO} \
       -DRE2_SOURCE=${RE2_SOURCE:-AUTO} \
+      -DSnappy_SOURCE=${Snappy_SOURCE:-AUTO} \
       -DThrift_SOURCE=${Thrift_SOURCE:-AUTO} \
       -DZSTD_SOURCE=${ZSTD_SOURCE:-AUTO} \
       ${CMAKE_ARGS} \

dev/tasks/python-wheels/osx-build.sh

@@ -58,7 +58,6 @@ function build_wheel {
     pushd cpp
     mkdir build
     pushd build
-
     cmake -DARROW_BUILD_SHARED=ON \
           -DARROW_BUILD_TESTS=OFF \
           -DARROW_DATASET=ON \

python/pyarrow/__init__.py

@@ -105,7 +105,7 @@ def parse_git(root, **kwargs):
 
 # Buffers, allocation
 from pyarrow.lib import (Buffer, ResizableBuffer, foreign_buffer, py_buffer,
-                         compress, decompress, allocate_buffer)
+                         Codec, compress, decompress, allocate_buffer)
 
 from pyarrow.lib import (MemoryPool, LoggingMemoryPool, ProxyMemoryPool,
                          total_allocated_bytes, set_memory_pool,

python/pyarrow/includes/libarrow.pxd

@@ -1694,27 +1694,29 @@ cdef extern from 'arrow/python/benchmark.h' namespace 'arrow::py::benchmark':
 
 
 cdef extern from 'arrow/util/compression.h' namespace 'arrow' nogil:
-    enum CompressionType" arrow::Compression::type":
-        CompressionType_UNCOMPRESSED" arrow::Compression::UNCOMPRESSED"
-        CompressionType_SNAPPY" arrow::Compression::SNAPPY"
-        CompressionType_GZIP" arrow::Compression::GZIP"
-        CompressionType_BROTLI" arrow::Compression::BROTLI"
-        CompressionType_ZSTD" arrow::Compression::ZSTD"
-        CompressionType_LZ4" arrow::Compression::LZ4"
-        CompressionType_BZ2" arrow::Compression::BZ2"
+    enum CCompressionType" arrow::Compression::type":
+        CCompressionType_UNCOMPRESSED" arrow::Compression::UNCOMPRESSED"
+        CCompressionType_SNAPPY" arrow::Compression::SNAPPY"
+        CCompressionType_GZIP" arrow::Compression::GZIP"
+        CCompressionType_BROTLI" arrow::Compression::BROTLI"
+        CCompressionType_ZSTD" arrow::Compression::ZSTD"
+        CCompressionType_LZ4" arrow::Compression::LZ4"
+        CCompressionType_BZ2" arrow::Compression::BZ2"
 
     cdef cppclass CCodec" arrow::util::Codec":
         @staticmethod
-        CResult[unique_ptr[CCodec]] Create(CompressionType codec)
+        CResult[unique_ptr[CCodec]] Create(CCompressionType codec)
+
+        @staticmethod
+        c_bool IsAvailable(CCompressionType codec)
 
         CResult[int64_t] Decompress(int64_t input_len, const uint8_t* input,
                                     int64_t output_len,
                                     uint8_t* output_buffer)
-
         CResult[int64_t] Compress(int64_t input_len, const uint8_t* input,
                                   int64_t output_buffer_len,
                                   uint8_t* output_buffer)
-
+        const char* name() const
         int64_t MaxCompressedLen(int64_t input_len, const uint8_t* input)