feat: Support loading parquet files from local disk

libs/CMakeLists.txt

@@ -394,3 +394,41 @@ set_property(TARGET usearch PROPERTY INTERFACE_INCLUDE_DIRECTORIES
   ${CMAKE_CURRENT_SOURCE_DIR}/usearch/include
   ${CMAKE_CURRENT_SOURCE_DIR}/usearch/fp16/include
   ${CMAKE_CURRENT_SOURCE_DIR}/usearch/stringzilla/include)
+
+macro(build_arrow)
+  set(ARROW_ROOT ${CMAKE_CURRENT_SOURCE_DIR}/arrow)
+  set(ARROW_LIBDIR "${ARROW_ROOT}/lib")
+
+  add_external_project(arrow
+    SOURCE_DIR ${ARROW_ROOT}/cpp
+    CMAKE_ARGS
+      -DCMAKE_INSTALL_PREFIX=${ARROW_ROOT}
+      -DCMAKE_INSTALL_LIBDIR=lib
+      -DARROW_PARQUET=ON
+      -DARROW_WITH_SNAPPY=ON
+      -DARROW_FILESYSTEM=ON
+      -DARROW_BUILD_STATIC=ON
+      -DARROW_BUILD_SHARED=OFF
+      -DARROW_MIMALLOC=OFF
+      -DARROW_WITH_RE2=OFF
+      -DARROW_WITH_UTF8PROC=OFF
+      -DARROW_DEPENDENCY_SOURCE=BUNDLED
+    BUILD_BYPRODUCTS
+      "${ARROW_LIBDIR}/libarrow.a" "${ARROW_LIBDIR}/libarrowd.a"
+      "${ARROW_LIBDIR}/libparquet.a" "${ARROW_LIBDIR}/libparquetd.a"
+      "${ARROW_LIBDIR}/libarrow_bundled_dependencies.a"
+  )
+
+  # handle debug build with `d` suffix
+  set(_d "$<$<CONFIG:Debug>:d>")
+  set(_arrow_libs "${ARROW_LIBDIR}/libparquet${_d}.a;${ARROW_LIBDIR}/libarrow${_d}.a;${ARROW_LIBDIR}/libarrow_bundled_dependencies.a")
+
+  add_library(arrow INTERFACE IMPORTED GLOBAL)
+  set_target_properties(arrow PROPERTIES
+    INTERFACE_INCLUDE_DIRECTORIES "${ARROW_ROOT}/include"
+    INTERFACE_LINK_LIBRARIES "${_arrow_libs}"
+  )
+  add_dependencies(arrow arrow-proj)
+  file(MAKE_DIRECTORY ${ARROW_ROOT}/include)
+endmacro()
+build_arrow()

libs/setup.sh

@@ -156,6 +156,7 @@ declare -A primary_urls=(
   ["nuraft"]="http://$local_cache_host/git/NuRaft.git"
   ["mgcxx"]="http://$local_cache_host/git/mgcxx.git"
   ["usearch"]="http://$local_cache_host/git/usearch.git"
+  ["arrow"]="http://$local_cache_host/git/arrow.git"
 )
 
 # The goal of secondary urls is to have links to the "source of truth" of
@@ -175,6 +176,7 @@ declare -A secondary_urls=(
   ["nuraft"]="https://github.com/eBay/NuRaft.git"
   ["mgcxx"]="https://github.com/memgraph/mgcxx.git"
   ["usearch"]="https://github.com/unum-cloud/usearch.git"
+  ["arrow"]="https://github.com/apache/arrow"
 )
 
 # Skip download if we are under the latest toolchains (>= 6).
@@ -268,3 +270,7 @@ repo_clone_try_double "${primary_urls[usearch]}" "${secondary_urls[usearch]}" "u
 pushd usearch
 git submodule update --init --recursive
 popd
+
+# arrow (load parquet)
+arrow_ref="apache-arrow-21.0.0"
+repo_clone_try_double "${primary_urls[arrow]}" "${secondary_urls[arrow]}" "arrow" "$arrow_ref" true

src/audit/log.cpp

@@ -70,7 +70,7 @@ inline nlohmann::json BoltValueToJson(const communication::bolt::Value &value) {
     }
     case Date: {
       std::stringstream ss;
-      ss << utils::Date(value.ValueDate().MicrosecondsSinceEpoch());
+      ss << utils::Date(std::chrono::microseconds{value.ValueDate().MicrosecondsSinceEpoch()});
       ret = ss.str();
       break;
     }

src/glue/communication.cpp

@@ -397,7 +397,7 @@ Value ToBoltValue(const storage::PropertyValue &value, const storage::Storage &s
       const auto &type = value.ValueTemporalData();
       switch (type.type) {
         case storage::TemporalType::Date:
-          return Value(utils::Date(type.microseconds));
+          return {utils::Date{std::chrono::microseconds{type.microseconds}}};
         case storage::TemporalType::LocalTime:
           return Value(utils::LocalTime(type.microseconds));
         case storage::TemporalType::LocalDateTime:

src/query/CMakeLists.txt

@@ -31,6 +31,7 @@ add_library(mg-query STATIC)
 target_sources(mg-query
     PRIVATE
     auth_query_handler.cpp
+    arrow_parquet/reader.cpp
     common.cpp
     context.cpp
     cypher_query_interpreter.cpp
@@ -86,6 +87,7 @@ target_sources(mg-query
     PUBLIC
     FILE_SET HEADERS
     FILES
+    arrow_parquet/reader.hpp
     dependant_symbol_visitor.hpp
     edge_accessor.hpp
     frontend/ast/ast.hpp
@@ -111,13 +113,13 @@ target_sources(mg-query
     vertex_accessor.hpp
 )
 
-
 target_link_libraries(mg-query
     PUBLIC
     dl
     cppitertools::cppitertools
     range-v3::range-v3
     Python3::Python
+    arrow
     mg-integrations-pulsar
     mg-integrations-kafka
     mg::storage