[arrow] Fix dataset feature (#28969)

2024-11-24 12:09:00 +08:00 · 2023-01-16 21:53:41 +01:00 · 2023-01-16 21:53:41 +01:00 · 9a469723ae
commit 9a469723ae
parent 021ba042f2
6 changed files with 138 additions and 2 deletions
--- a/ports/arrow/fix-dataset.patch
+++ b/ports/arrow/fix-dataset.patch
@ -0,0 +1,29 @@
+diff --git a/cpp/src/arrow/dataset/CMakeLists.txt b/cpp/src/arrow/dataset/CMakeLists.txt
+index b80370e63985..04f80ab6d1b3 100644
+--- a/cpp/src/arrow/dataset/CMakeLists.txt
+++ b/cpp/src/arrow/dataset/CMakeLists.txt
+@@ -31,6 +31,12 @@ set(ARROW_DATASET_SRCS
+     scanner.cc
+     scan_node.cc)
+ 
+if(ARROW_PARQUET)
+  set(ARROW_DATASET_PKG_CONFIG_REQUIRES parquet)
+else()
+  set(ARROW_DATASET_PKG_CONFIG_REQUIRES arrow)
+endif()
+
+ set(ARROW_DATASET_STATIC_LINK_LIBS)
+ set(ARROW_DATASET_SHARED_LINK_LIBS)
+ set(ARROW_DATASET_STATIC_INSTALL_INTERFACE_LIBS)
+diff --git a/cpp/src/arrow/dataset/arrow-dataset.pc.in b/cpp/src/arrow/dataset/arrow-dataset.pc.in
+index 3e474e3825c5..d587fd110c08 100644
+--- a/cpp/src/arrow/dataset/arrow-dataset.pc.in
+++ b/cpp/src/arrow/dataset/arrow-dataset.pc.in
+@@ -22,6 +22,6 @@ libdir=@ARROW_PKG_CONFIG_LIBDIR@
+ Name: Apache Arrow Dataset
+ Description: Apache Arrow Dataset provides an API to read and write semantic datasets stored in different locations and formats.
+ Version: @ARROW_VERSION@
+-Requires: arrow parquet
+Requires: @ARROW_DATASET_PKG_CONFIG_REQUIRES@
+ Libs: -L${libdir} -larrow_dataset
+ Cflags.private: -DARROW_DS_STATIC
--- a/ports/arrow/fix-pkgconfig-windows.patch
+++ b/ports/arrow/fix-pkgconfig-windows.patch
@ -0,0 +1,100 @@
+From 109a40a12019e28ecd60bbf9210f9c949c827d31 Mon Sep 17 00:00:00 2001
+From: Luke Elliott <luke.b.elliott@gmail.com>
+Date: Wed, 7 Dec 2022 11:57:57 +0000
+Subject: [PATCH] GH-14869: [C++] Add Cflags.private defining <lib>_STATIC to
+ <lib>.pc.in.
+
+Required for Windows usage of static builds via vcpkg & pkgconfig.
+---
+ cpp/src/arrow/arrow-testing.pc.in               | 1 +
+ cpp/src/arrow/arrow.pc.in                       | 1 +
+ cpp/src/arrow/dataset/arrow-dataset.pc.in       | 1 +
+ cpp/src/arrow/engine/arrow-substrait.pc.in      | 1 +
+ cpp/src/arrow/flight/arrow-flight.pc.in         | 1 +
+ cpp/src/arrow/flight/sql/arrow-flight-sql.pc.in | 1 +
+ cpp/src/gandiva/gandiva.pc.in                   | 1 +
+ cpp/src/parquet/parquet.pc.in                   | 1 +
+ python/pyarrow/src/arrow-python.pc.in           | 1 +
+ 9 files changed, 9 insertions(+)
+
+diff --git a/cpp/src/arrow/arrow-testing.pc.in b/cpp/src/arrow/arrow-testing.pc.in
+index 5a991e796d83..9a452a5d113f 100644
+--- a/cpp/src/arrow/arrow-testing.pc.in
+++ b/cpp/src/arrow/arrow-testing.pc.in
+@@ -27,3 +27,4 @@ Version: @ARROW_VERSION@
+ Requires: arrow
+ Libs: -L${libdir} -larrow_testing
+ Cflags: -I${gtest_includedir}
+Cflags.private: -DARROW_TESTING_STATIC
+diff --git a/cpp/src/arrow/arrow.pc.in b/cpp/src/arrow/arrow.pc.in
+index 3a5710ab6b88..90c98bf32b2b 100644
+--- a/cpp/src/arrow/arrow.pc.in
+++ b/cpp/src/arrow/arrow.pc.in
+@@ -30,3 +30,4 @@ Requires.private:@ARROW_PC_REQUIRES_PRIVATE@
+ Libs: -L${libdir} -larrow
+ Libs.private:@ARROW_PC_LIBS_PRIVATE@
+ Cflags: -I${includedir}
+Cflags.private: -DARROW_STATIC
+diff --git a/cpp/src/arrow/dataset/arrow-dataset.pc.in b/cpp/src/arrow/dataset/arrow-dataset.pc.in
+index 7c433321c135..3e474e3825c5 100644
+--- a/cpp/src/arrow/dataset/arrow-dataset.pc.in
+++ b/cpp/src/arrow/dataset/arrow-dataset.pc.in
+@@ -24,3 +24,4 @@ Description: Apache Arrow Dataset provides an API to read and write semantic dat
+ Version: @ARROW_VERSION@
+ Requires: arrow parquet
+ Libs: -L${libdir} -larrow_dataset
+Cflags.private: -DARROW_DS_STATIC
+diff --git a/cpp/src/arrow/engine/arrow-substrait.pc.in b/cpp/src/arrow/engine/arrow-substrait.pc.in
+index 77ac510933f5..173e1c91efd8 100644
+--- a/cpp/src/arrow/engine/arrow-substrait.pc.in
+++ b/cpp/src/arrow/engine/arrow-substrait.pc.in
+@@ -24,3 +24,4 @@ Description: Apache Arrow's Substrait Consumer.
+ Version: @ARROW_VERSION@
+ Requires: arrow
+ Libs: -L${libdir} -larrow_substrait
+Cflags.private: -DARROW_ENGINE_STATIC
+diff --git a/cpp/src/arrow/flight/arrow-flight.pc.in b/cpp/src/arrow/flight/arrow-flight.pc.in
+index 955fc19b0df4..5a57a8922377 100644
+--- a/cpp/src/arrow/flight/arrow-flight.pc.in
+++ b/cpp/src/arrow/flight/arrow-flight.pc.in
+@@ -24,3 +24,4 @@ Description: Apache Arrow's RPC system built on gRPC
+ Version: @ARROW_VERSION@
+ Requires: arrow
+ Libs: -L${libdir} -larrow_flight
+Cflags.private: -DARROW_FLIGHT_STATIC
+diff --git a/cpp/src/arrow/flight/sql/arrow-flight-sql.pc.in b/cpp/src/arrow/flight/sql/arrow-flight-sql.pc.in
+index 4f8398bdd259..b76964b97c89 100644
+--- a/cpp/src/arrow/flight/sql/arrow-flight-sql.pc.in
+++ b/cpp/src/arrow/flight/sql/arrow-flight-sql.pc.in
+@@ -24,3 +24,4 @@ Description: Apache Arrow Flight SQL extension
+ Version: @ARROW_VERSION@
+ Requires: arrow-flight
+ Libs: -L${libdir} -larrow_flight_sql
+Cflags.private: -DARROW_FLIGHT_SQL_STATIC
+diff --git a/cpp/src/gandiva/gandiva.pc.in b/cpp/src/gandiva/gandiva.pc.in
+index 60034c285790..49260092d297 100644
+--- a/cpp/src/gandiva/gandiva.pc.in
+++ b/cpp/src/gandiva/gandiva.pc.in
+@@ -25,3 +25,4 @@ Version: @GANDIVA_VERSION@
+ Requires: arrow
+ Libs: -L${libdir} -lgandiva
+ Cflags: -I${includedir}
+Cflags.private: -DGANDIVA_STATIC
+diff --git a/cpp/src/parquet/parquet.pc.in b/cpp/src/parquet/parquet.pc.in
+index e510bb1d6bfc..b992eba81998 100644
+--- a/cpp/src/parquet/parquet.pc.in
+++ b/cpp/src/parquet/parquet.pc.in
+@@ -29,3 +29,4 @@ Version: @ARROW_VERSION@
+ Requires: arrow
+ Libs: -L${libdir} -lparquet
+ Cflags: -I${includedir}
+Cflags.private: -DPARQUET_STATIC
+diff --git a/python/pyarrow/src/arrow-python.pc.in b/python/pyarrow/src/arrow-python.pc.in
+index c077c7dc84c3..0efd0ad98309 100644
+--- a/python/pyarrow/src/arrow-python.pc.in
+++ b/python/pyarrow/src/arrow-python.pc.in
+@@ -25,3 +25,4 @@ Version: @ARROW_VERSION@
+ Requires: arrow
+ Libs: -L${libdir} -larrow_python
+ Cflags: -I${includedir} -I@PYTHON_INCLUDE_DIRS@
+Cflags.private: -DARROW_PYTHON_STATIC
--- a/ports/arrow/portfile.cmake
+++ b/ports/arrow/portfile.cmake
@ -12,6 +12,8 @@ vcpkg_extract_source_archive(
        msvc-static-name.patch
        thrift.patch
        utf8proc.patch
+        fix-pkgconfig-windows.patch # needed for fix-dataset.patch (https://github.com/apache/arrow/pull/14900)
+        fix-dataset.patch # https://github.com/apache/arrow/pull/33665
 )

 vcpkg_check_features(OUT_FEATURE_OPTIONS FEATURE_OPTIONS
--- a/ports/arrow/vcpkg.json
+++ b/ports/arrow/vcpkg.json
@ -1,7 +1,7 @@
 {
  "name": "arrow",
  "version": "10.0.1",
-  "port-version": 2,
+  "port-version": 3,
  "description": "Cross-language development platform for in-memory analytics",
  "homepage": "https://arrow.apache.org",
  "license": "Apache-2.0",
--- a/versions/a-/arrow.json
+++ b/versions/a-/arrow.json
@ -1,5 +1,10 @@
 {
  "versions": [
+    {
+      "git-tree": "3829a0697ffe3ba8cb1274e2edd7cfbcbc48482b",
+      "version": "10.0.1",
+      "port-version": 3
+    },
    {
      "git-tree": "ea9873418e3de2180518389baa3562084a566702",
      "version": "10.0.1",
--- a/versions/baseline.json
+++ b/versions/baseline.json
@ -198,7 +198,7 @@
    },
    "arrow": {
      "baseline": "10.0.1",
-      "port-version": 2
+      "port-version": 3
    },
    "ashes": {
      "baseline": "2022-06-08",