PARQUET-932: Add option to build parquet library with minimal dependency

If users want to build only the parquet library, thirdparty headers are sufficient and GTEST and GBENCHMARK are not required.

Author: Deepak Majeti <deepak.majeti@hpe.com>

Closes #279 from majetideepak/ParquetMinimal and squashes the following commits:

cdb4135 [Deepak Majeti] correct arrow.h with api.h
100a47a [Deepak Majeti] add to README
dacc908 [Deepak Majeti] add message
016d055 [Deepak Majeti] Support minimal dependency build

Author: Deepak Majeti

CMakeLists.txt

@@ -113,10 +113,14 @@ if ("${CMAKE_SOURCE_DIR}" STREQUAL "${CMAKE_CURRENT_SOURCE_DIR}")
   option(PARQUET_RPATH_ORIGIN
     "Build Parquet libraries with RATH set to \$ORIGIN"
     OFF)
+  option(PARQUET_MINIMAL_DEPENDENCY
+    "Depend only on Thirdparty headers to build libparquet. Always OFF if building binaries"
+    OFF)
 endif()
 
 if (PARQUET_BUILD_TESTS OR PARQUET_BUILD_EXECUTABLES OR PARQUET_BUILD_BENCHMARKS)
   set(PARQUET_BUILD_STATIC ON)
+  set(PARQUET_MINIMAL_DEPENDENCY OFF)
 endif()
 
 # If build in-source, create the latest symlink. If build out-of-source, which is
@@ -315,6 +319,12 @@ enable_testing()
 # Dependencies
 ############################################################
 
+if (PARQUET_MINIMAL_DEPENDENCY)
+    set(IGNORE_OPTIONAL_PACKAGES ON)
+    message(STATUS "Build using minimal dependencies")
+else()
+    set(IGNORE_OPTIONAL_PACKAGES OFF)
+endif()
 include(ThirdpartyToolchain)
 
 # Thrift requires these definitions for some types that we use
@@ -574,16 +584,6 @@ set(BUNDLED_STATIC_LIBS
   zlibstatic
 )
 
-# These are libraries that we will link privately with parquet_shared (as they
-# do not need to be linked transitively by other linkers), but publicly with
-# parquet_static (because internal users need to transitively link all
-# dependencies)
-set(LIBPARQUET_INTERFACE_LINK_LIBS
-  ${ARROW_LINK_LIBS}
-  ${BOOST_LINK_LIBS}
-  ${BUNDLED_STATIC_LIBS}
-)
-
 add_library(parquet_objlib OBJECT
   ${LIBPARQUET_SRCS}
 )
@@ -594,10 +594,21 @@ add_custom_target(thrift-deps ALL
   DEPENDS ${THRIFT_OUTPUT_FILES})
 add_dependencies(parquet_objlib thrift-deps)
 
+if (NOT PARQUET_MINIMAL_DEPENDENCY)
+# These are libraries that we will link privately with parquet_shared (as they
+# do not need to be linked transitively by other linkers), but publicly with
+# parquet_static (because internal users need to transitively link all
+# dependencies)
+  set(LIBPARQUET_INTERFACE_LINK_LIBS
+    ${ARROW_LINK_LIBS}
+    ${BOOST_LINK_LIBS}
+    ${BUNDLED_STATIC_LIBS}
+  )
 # Although we don't link parquet_objlib against anything, we need it to depend
 # on these libs as we may generate their headers via ExternalProject_Add
-add_dependencies(parquet_objlib
-  ${LIBPARQUET_INTERFACE_LINK_LIBS})
+  add_dependencies(parquet_objlib
+    ${LIBPARQUET_INTERFACE_LINK_LIBS})
+endif()
 
 set_property(TARGET parquet_objlib PROPERTY POSITION_INDEPENDENT_CODE 1)
 

README.md

@@ -102,6 +102,10 @@ To disable the testing (which requires `googletest`), pass
 For release-level builds (enable optimizations and disable debugging), pass
 `-DCMAKE_BUILD_TYPE=Release` to `cmake`.
 
+To build only the library with minimal dependencies, pass
+`-DPARQUET_MINIMAL_DEPENDENCY=ON` to `cmake`.
+Note that the executables, tests, and benchmarks should be disabled as well.
+
 Incremental builds can be done afterwords with just `make`.
 
 ## Using with Apache Arrow

cmake_modules/FindArrow.cmake

@@ -51,8 +51,10 @@ find_library(ARROW_IO_LIB_PATH NAMES arrow_io
   ${ARROW_SEARCH_LIB_PATH}
   NO_DEFAULT_PATH)
 
-if (ARROW_INCLUDE_DIR AND ARROW_LIB_PATH)
+if (ARROW_INCLUDE_DIR AND (PARQUET_MINIMAL_DEPENDENCY OR ARROW_LIB_PATH))
   set(ARROW_FOUND TRUE)
+  set(ARROW_HEADER_NAME arrow/api.h)
+  set(ARROW_HEADER ${ARROW_INCLUDE_DIR}/${ARROW_HEADER_NAME})
   set(ARROW_LIB_NAME libarrow)
   set(ARROW_IO_LIB_NAME libarrow_io)
 
@@ -63,8 +65,12 @@ if (ARROW_INCLUDE_DIR AND ARROW_LIB_PATH)
   set(ARROW_IO_STATIC_LIB ${ARROW_LIBS}/${ARROW_IO_LIB_NAME}.a)
   set(ARROW_IO_SHARED_LIB ${ARROW_LIBS}/${ARROW_IO_LIB_NAME}${CMAKE_SHARED_LIBRARY_SUFFIX})
   if (NOT Arrow_FIND_QUIETLY)
-    message(STATUS "Found the Arrow core library: ${ARROW_LIB_PATH}")
-    message(STATUS "Found the Arrow IO library: ${ARROW_IO_LIB_PATH}")
+    if (PARQUET_MINIMAL_DEPENDENCY)
+      message(STATUS "Found the Arrow core and IO header: ${ARROW_HEADER}")
+    else ()
+      message(STATUS "Found the Arrow core library: ${ARROW_LIB_PATH}")
+      message(STATUS "Found the Arrow IO library: ${ARROW_IO_LIB_PATH}")
+    endif ()
   endif ()
 else ()
   if (NOT Arrow_FIND_QUIETLY)

cmake_modules/FindBrotli.cmake

@@ -60,9 +60,11 @@ endif ()
 set(BROTLI_LIBRARIES ${BROTLI_LIBRARY_ENC} ${BROTLI_LIBRARY_DEC}
     ${BROTLI_LIBRARY_COMMON})
 
-if (BROTLI_INCLUDE_DIR AND BROTLI_LIBRARIES)
+if (BROTLI_INCLUDE_DIR AND (PARQUET_MINIMAL_DEPENDENCY OR BROTLI_LIBRARIES))
   set(BROTLI_FOUND TRUE)
   get_filename_component( BROTLI_LIBS ${BROTLI_LIBRARY_ENC} PATH )
+  set(BROTLI_HEADER_NAME brotli.h)
+  set(BROTLI_HEADER ${BROTLI_INCLUDE_DIR}/${BROTLI_HEADER_NAME})
   set(BROTLI_LIB_NAME libbrotli)
   set(BROTLI_STATIC_LIB
       ${BROTLI_LIBS}/${BROTLI_LIB_NAME}enc.a
@@ -78,7 +80,11 @@ endif ()
 
 if (BROTLI_FOUND)
   if (NOT Brotli_FIND_QUIETLY)
-    message(STATUS "Found the Brotli library: ${BROTLI_LIBRARIES}")
+    if (PARQUET_MINIMAL_DEPENDENCY)
+      message(STATUS "Found the Brotli header: ${BROTLI_HEADER}")
+    else ()
+      message(STATUS "Found the Brotli library: ${BROTLI_LIBRARIES}")
+    endif ()
   endif ()
 else ()
   if (NOT Brotli_FIND_QUIETLY)

cmake_modules/FindSnappy.cmake

@@ -49,9 +49,11 @@ find_library( SNAPPY_LIBRARIES NAMES snappy PATHS
   NO_DEFAULT_PATH
   PATH_SUFFIXES "lib")
 
-if (SNAPPY_INCLUDE_DIR AND SNAPPY_LIBRARIES)
+if (SNAPPY_INCLUDE_DIR AND (PARQUET_MINIMAL_DEPENDENCY OR SNAPPY_LIBRARIES))
   set(SNAPPY_FOUND TRUE)
   get_filename_component( SNAPPY_LIBS ${SNAPPY_LIBRARIES} PATH )
+  set(SNAPPY_HEADER_NAME snappy.h)
+  set(SNAPPY_HEADER ${SNAPPY_INCLUDE_DIR}/${SNAPPY_HEADER_NAME})
   set(SNAPPY_LIB_NAME libsnappy)
   set(SNAPPY_STATIC_LIB ${SNAPPY_LIBS}/${SNAPPY_LIB_NAME}.a)
   set(SNAPPY_SHARED_LIB ${SNAPPY_LIBS}/${SNAPPY_LIB_NAME}${CMAKE_SHARED_LIBRARY_SUFFIX})
@@ -61,7 +63,11 @@ endif ()
 
 if (SNAPPY_FOUND)
   if (NOT Snappy_FIND_QUIETLY)
-    message(STATUS "Found the Snappy library: ${SNAPPY_LIBRARIES}")
+    if (PARQUET_MINIMAL_DEPENDENCY)
+      message(STATUS "Found the Snappy header: ${SNAPPY_HEADER}")
+    else ()
+      message(STATUS "Found the Snappy library: ${SNAPPY_LIBRARIES}")
+    endif ()
   endif ()
 else ()
   if (NOT Snappy_FIND_QUIETLY)

cmake_modules/FindZLIB.cmake

@@ -22,7 +22,7 @@
 #
 #  ZLIB_HOME - When set, this path is inspected instead of standard library
 #             locations as the root of the ZLIB installation.
-#             The environment variable ZLIB_HOME overrides this veriable.
+#             The environment variable ZLIB_HOME overrides this variable.
 #
 # - Find ZLIB (zlib.h, libz.a, libz.so, and libz.so.1)
 # This module defines
@@ -54,9 +54,11 @@ else ()
 endif ()
 
 
-if (ZLIB_INCLUDE_DIR AND ZLIB_LIBRARIES)
+if (ZLIB_INCLUDE_DIR AND (PARQUET_MINIMAL_DEPENDENCY OR ZLIB_LIBRARIES))
   set(ZLIB_FOUND TRUE)
   get_filename_component( ZLIB_LIBS ${ZLIB_LIBRARIES} PATH )
+  set(ZLIB_HEADER_NAME zlib.h)
+  set(ZLIB_HEADER ${ZLIB_INCLUDE_DIR}/${ZLIB_HEADER_NAME})
   set(ZLIB_LIB_NAME libz)
   set(ZLIB_STATIC_LIB ${ZLIB_LIBS}/${ZLIB_LIB_NAME}.a)
   set(ZLIB_SHARED_LIB ${ZLIB_LIBS}/${ZLIB_LIB_NAME}${CMAKE_SHARED_LIBRARY_SUFFIX})
@@ -66,7 +68,11 @@ endif ()
 
 if (ZLIB_FOUND)
   if (NOT ZLIB_FIND_QUIETLY)
-    message(STATUS "Found the ZLIB library: ${ZLIB_LIBRARIES}")
+    if (PARQUET_MINIMAL_DEPENDENCY)
+      message(STATUS "Found the ZLIB header: ${ZLIB_HEADER}")
+    else()
+      message(STATUS "Found the ZLIB library: ${ZLIB_LIBRARIES}")
+    endif ()
   endif ()
 else ()
   if (NOT ZLIB_FIND_QUIETLY)

cmake_modules/ThirdpartyToolchain.cmake

@@ -251,7 +251,7 @@ if (ZLIB_VENDORED)
 endif()
 
 ## GTest
-if(PARQUET_BUILD_TESTS)
+if(PARQUET_BUILD_TESTS AND NOT IGNORE_OPTIONAL_PACKAGES)
   add_custom_target(unittest ctest -L unittest)
 
   if("$ENV{GTEST_HOME}" STREQUAL "")
@@ -308,7 +308,7 @@ if ("$ENV{GBENCHMARK_HOME}" STREQUAL "")
   set(GBENCHMARK_HOME ${THIRDPARTY_DIR}/installed)
 endif()
 
-if(PARQUET_BUILD_BENCHMARKS)
+if(PARQUET_BUILD_BENCHMARKS AND NOT IGNORE_OPTIONAL_PACKAGES)
   add_custom_target(runbenchmark ctest -L benchmark)