From 01e1d05256b924873492e5903402961e9f3362c7 Mon Sep 17 00:00:00 2001
From: Juan Miguel Carceller <22276694+jmcarcell@users.noreply.github.com>
Date: Thu, 2 Nov 2023 09:15:38 +0100
Subject: [PATCH] Use the cmake ExternalData module to manage test data (#508)

* Use the cmake ExternalData to manage test data

* Remove the macro and script to download files
---
 cmake/podioTest.cmake                         | 24 ---------
 tests/CMakeLists.txt                          |  8 ++-
 tests/input_files/v00-13-example.root.md5     |  1 +
 tests/input_files/v00-16-02-example.root.md5  |  1 +
 tests/input_files/v00-16-02-example.sio.md5   |  1 +
 .../v00-16-02-example_frame.root.md5          |  1 +
 .../v00-16-02-example_frame.sio.md5           |  1 +
 tests/input_files/v00-16-04-example.root.md5  |  1 +
 tests/input_files/v00-16-04-example.sio.md5   |  1 +
 .../v00-16-04-example_frame.root.md5          |  1 +
 .../v00-16-04-example_frame.sio.md5           |  1 +
 tests/input_files/v00-16-05-example.root.md5  |  1 +
 tests/input_files/v00-16-05-example.sio.md5   |  1 +
 .../v00-16-05-example_frame.root.md5          |  1 +
 .../v00-16-05-example_frame.sio.md5           |  1 +
 tests/input_files/v00-16-06-example.root.md5  |  1 +
 tests/input_files/v00-16-06-example.sio.md5   |  1 +
 .../v00-16-06-example_frame.root.md5          |  1 +
 .../v00-16-06-example_frame.sio.md5           |  1 +
 tests/input_files/v00-16-example.root.md5     |  1 +
 tests/input_files/v00-16-example.sio.md5      |  1 +
 .../input_files/v00-16-example_frame.root.md5 |  1 +
 .../input_files/v00-16-example_frame.sio.md5  |  1 +
 tests/root_io/CMakeLists.txt                  | 49 +++++++++----------
 tests/scripts/get_test_inputs.sh              | 23 ---------
 25 files changed, 51 insertions(+), 74 deletions(-)
 create mode 100644 tests/input_files/v00-13-example.root.md5
 create mode 100644 tests/input_files/v00-16-02-example.root.md5
 create mode 100644 tests/input_files/v00-16-02-example.sio.md5
 create mode 100644 tests/input_files/v00-16-02-example_frame.root.md5
 create mode 100644 tests/input_files/v00-16-02-example_frame.sio.md5
 create mode 100644 tests/input_files/v00-16-04-example.root.md5
 create mode 100644 tests/input_files/v00-16-04-example.sio.md5
 create mode 100644 tests/input_files/v00-16-04-example_frame.root.md5
 create mode 100644 tests/input_files/v00-16-04-example_frame.sio.md5
 create mode 100644 tests/input_files/v00-16-05-example.root.md5
 create mode 100644 tests/input_files/v00-16-05-example.sio.md5
 create mode 100644 tests/input_files/v00-16-05-example_frame.root.md5
 create mode 100644 tests/input_files/v00-16-05-example_frame.sio.md5
 create mode 100644 tests/input_files/v00-16-06-example.root.md5
 create mode 100644 tests/input_files/v00-16-06-example.sio.md5
 create mode 100644 tests/input_files/v00-16-06-example_frame.root.md5
 create mode 100644 tests/input_files/v00-16-06-example_frame.sio.md5
 create mode 100644 tests/input_files/v00-16-example.root.md5
 create mode 100644 tests/input_files/v00-16-example.sio.md5
 create mode 100644 tests/input_files/v00-16-example_frame.root.md5
 create mode 100644 tests/input_files/v00-16-example_frame.sio.md5
 delete mode 100644 tests/scripts/get_test_inputs.sh

diff --git a/cmake/podioTest.cmake b/cmake/podioTest.cmake
index 5230af7a3..fb92ddce7 100644
--- a/cmake/podioTest.cmake
+++ b/cmake/podioTest.cmake
@@ -26,27 +26,3 @@ function(CREATE_PODIO_TEST sourcefile additional_libs)
   target_link_libraries(${name} PRIVATE TestDataModel ExtensionDataModel ${additional_libs})
   PODIO_SET_TEST_ENV(${name})
 endfunction()
-
-#--- utility macro to facilitate the downloading of legacy input data
-macro(PODIO_DOWNLOAD_LEGACY_INPUTS legacy_versions)
-  # Avoid fetching these everytime cmake is run by caching the directory the first
-  # time the inputs are fetched or if the expected file does not exist in the
-  # expected directory
-  if (NOT DEFINED CACHE{PODIO_TEST_INPUT_DATA_DIR} OR NOT EXISTS ${PODIO_TEST_INPUT_DATA_DIR}/v00-16-05/example_frame.root)
-    message(STATUS "Getting test input files")
-    execute_process(
-      COMMAND bash ${PROJECT_SOURCE_DIR}/tests/scripts/get_test_inputs.sh ${legacy_versions}
-      OUTPUT_VARIABLE podio_test_input_data_dir
-      RESULT_VARIABLE test_inputs_available
-    )
-    if (NOT "${test_inputs_available}" STREQUAL "0")
-      message(WARNING "Could not get test input files. Will skip some tests that depend on these")
-      # Catch cases where the variable is cached but the file no longer exists
-      unset(PODIO_TEST_INPUT_DATA_DIR CACHE)
-    else()
-      message(STATUS "Test inputs stored in: " ${podio_test_input_data_dir})
-      set(PODIO_TEST_INPUT_DATA_DIR ${podio_test_input_data_dir} CACHE INTERNAL "input dir for test inputs fetched from remote sources")
-      mark_as_advanced(PODIO_TEST_INPUT_DATA_DIR)
-    endif()
-  endif()
-endmacro()
diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
index b09c6efa6..9a1583f47 100644
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -43,8 +43,10 @@ set(legacy_test_versions
   v00-16-06
 )
 
-### Define the actual tests
-PODIO_DOWNLOAD_LEGACY_INPUTS("${legacy_test_versions}")
+include(ExternalData)
+list(APPEND ExternalData_URL_TEMPLATES
+ "https://key4hep.web.cern.ch:443/testFiles/podio/%(hash)"
+ )
 
 add_executable(check_benchmark_outputs check_benchmark_outputs.cpp)
 target_link_libraries(check_benchmark_outputs PRIVATE ROOT::Tree)
@@ -63,3 +65,5 @@ CREATE_PODIO_TEST(write_ascii.cpp "")
 
 # Customize CTest to potentially disable some of the tests with known problems
 configure_file(CTestCustom.cmake ${PROJECT_BINARY_DIR}/CTestCustom.cmake @ONLY)
+
+ExternalData_Add_Target(legacy_test_cases)
diff --git a/tests/input_files/v00-13-example.root.md5 b/tests/input_files/v00-13-example.root.md5
new file mode 100644
index 000000000..23d937ff7
--- /dev/null
+++ b/tests/input_files/v00-13-example.root.md5
@@ -0,0 +1 @@
+86a33a765483c6e9f0babe798ef5a8c5
diff --git a/tests/input_files/v00-16-02-example.root.md5 b/tests/input_files/v00-16-02-example.root.md5
new file mode 100644
index 000000000..9c7491c84
--- /dev/null
+++ b/tests/input_files/v00-16-02-example.root.md5
@@ -0,0 +1 @@
+153ca8cc2f6110076c16319f425e214d
diff --git a/tests/input_files/v00-16-02-example.sio.md5 b/tests/input_files/v00-16-02-example.sio.md5
new file mode 100644
index 000000000..5bde6daa1
--- /dev/null
+++ b/tests/input_files/v00-16-02-example.sio.md5
@@ -0,0 +1 @@
+2b38723420c408517ab34449efdab8f6
diff --git a/tests/input_files/v00-16-02-example_frame.root.md5 b/tests/input_files/v00-16-02-example_frame.root.md5
new file mode 100644
index 000000000..9ff725c82
--- /dev/null
+++ b/tests/input_files/v00-16-02-example_frame.root.md5
@@ -0,0 +1 @@
+44fded47f2b01b168d3aece05a52fd43
diff --git a/tests/input_files/v00-16-02-example_frame.sio.md5 b/tests/input_files/v00-16-02-example_frame.sio.md5
new file mode 100644
index 000000000..8a29ef9d5
--- /dev/null
+++ b/tests/input_files/v00-16-02-example_frame.sio.md5
@@ -0,0 +1 @@
+616ca9ed47a64e34840ad8e77f75b8e1
diff --git a/tests/input_files/v00-16-04-example.root.md5 b/tests/input_files/v00-16-04-example.root.md5
new file mode 100644
index 000000000..162b877af
--- /dev/null
+++ b/tests/input_files/v00-16-04-example.root.md5
@@ -0,0 +1 @@
+3c31525e669e0d278826409f08eba48a
diff --git a/tests/input_files/v00-16-04-example.sio.md5 b/tests/input_files/v00-16-04-example.sio.md5
new file mode 100644
index 000000000..784ea3d16
--- /dev/null
+++ b/tests/input_files/v00-16-04-example.sio.md5
@@ -0,0 +1 @@
+6a9096303a61e3a09ef99dbf60af4568
diff --git a/tests/input_files/v00-16-04-example_frame.root.md5 b/tests/input_files/v00-16-04-example_frame.root.md5
new file mode 100644
index 000000000..0fb43e947
--- /dev/null
+++ b/tests/input_files/v00-16-04-example_frame.root.md5
@@ -0,0 +1 @@
+8bd4fa838c6165a6539be86edc5c31ca
diff --git a/tests/input_files/v00-16-04-example_frame.sio.md5 b/tests/input_files/v00-16-04-example_frame.sio.md5
new file mode 100644
index 000000000..eea5e2339
--- /dev/null
+++ b/tests/input_files/v00-16-04-example_frame.sio.md5
@@ -0,0 +1 @@
+7f1895b8ee5c6e8030038cf525018274
diff --git a/tests/input_files/v00-16-05-example.root.md5 b/tests/input_files/v00-16-05-example.root.md5
new file mode 100644
index 000000000..0d7473a29
--- /dev/null
+++ b/tests/input_files/v00-16-05-example.root.md5
@@ -0,0 +1 @@
+dc6181be4825cd49225e17d655134dba
diff --git a/tests/input_files/v00-16-05-example.sio.md5 b/tests/input_files/v00-16-05-example.sio.md5
new file mode 100644
index 000000000..b96e9569a
--- /dev/null
+++ b/tests/input_files/v00-16-05-example.sio.md5
@@ -0,0 +1 @@
+c308cef481593063212b4f0d763129ec
diff --git a/tests/input_files/v00-16-05-example_frame.root.md5 b/tests/input_files/v00-16-05-example_frame.root.md5
new file mode 100644
index 000000000..27f3d1077
--- /dev/null
+++ b/tests/input_files/v00-16-05-example_frame.root.md5
@@ -0,0 +1 @@
+4851c2cd392d2dff52d0c0804640a581
diff --git a/tests/input_files/v00-16-05-example_frame.sio.md5 b/tests/input_files/v00-16-05-example_frame.sio.md5
new file mode 100644
index 000000000..9d0aedb51
--- /dev/null
+++ b/tests/input_files/v00-16-05-example_frame.sio.md5
@@ -0,0 +1 @@
+d9b999a6184e0e8649a732b049c9109a
diff --git a/tests/input_files/v00-16-06-example.root.md5 b/tests/input_files/v00-16-06-example.root.md5
new file mode 100644
index 000000000..df926199f
--- /dev/null
+++ b/tests/input_files/v00-16-06-example.root.md5
@@ -0,0 +1 @@
+db390543bab1a89f52968516a2648af7
diff --git a/tests/input_files/v00-16-06-example.sio.md5 b/tests/input_files/v00-16-06-example.sio.md5
new file mode 100644
index 000000000..0dc78675b
--- /dev/null
+++ b/tests/input_files/v00-16-06-example.sio.md5
@@ -0,0 +1 @@
+c74b9634b1fa1305d3813ed23c30cc27
diff --git a/tests/input_files/v00-16-06-example_frame.root.md5 b/tests/input_files/v00-16-06-example_frame.root.md5
new file mode 100644
index 000000000..310c3f7a1
--- /dev/null
+++ b/tests/input_files/v00-16-06-example_frame.root.md5
@@ -0,0 +1 @@
+e64e8720a2746ae30c363aa0d4f5935e
diff --git a/tests/input_files/v00-16-06-example_frame.sio.md5 b/tests/input_files/v00-16-06-example_frame.sio.md5
new file mode 100644
index 000000000..aea3f2634
--- /dev/null
+++ b/tests/input_files/v00-16-06-example_frame.sio.md5
@@ -0,0 +1 @@
+22504d7cd722cde81202c1bdfbe44308
diff --git a/tests/input_files/v00-16-example.root.md5 b/tests/input_files/v00-16-example.root.md5
new file mode 100644
index 000000000..9155b356e
--- /dev/null
+++ b/tests/input_files/v00-16-example.root.md5
@@ -0,0 +1 @@
+fc47e06f8c42b249dc0fbcbff09ef562
diff --git a/tests/input_files/v00-16-example.sio.md5 b/tests/input_files/v00-16-example.sio.md5
new file mode 100644
index 000000000..a422c9463
--- /dev/null
+++ b/tests/input_files/v00-16-example.sio.md5
@@ -0,0 +1 @@
+8146876f0121a8d07fdfd80c2408ed48
diff --git a/tests/input_files/v00-16-example_frame.root.md5 b/tests/input_files/v00-16-example_frame.root.md5
new file mode 100644
index 000000000..daa114f0c
--- /dev/null
+++ b/tests/input_files/v00-16-example_frame.root.md5
@@ -0,0 +1 @@
+a1065876927fa04f70687ed507dba5a2
diff --git a/tests/input_files/v00-16-example_frame.sio.md5 b/tests/input_files/v00-16-example_frame.sio.md5
new file mode 100644
index 000000000..6fcad659f
--- /dev/null
+++ b/tests/input_files/v00-16-example_frame.sio.md5
@@ -0,0 +1 @@
+c3a6ef4d13499c4ac5f1249a40d03e9c
diff --git a/tests/root_io/CMakeLists.txt b/tests/root_io/CMakeLists.txt
index 509749643..598046e3a 100644
--- a/tests/root_io/CMakeLists.txt
+++ b/tests/root_io/CMakeLists.txt
@@ -27,7 +27,6 @@ foreach( sourcefile ${root_dependent_tests} )
 endforeach()
 
 
-
 #--- set some dependencies between the different tests to ensure input generating ones are run first
 set_property(TEST read PROPERTY DEPENDS write)
 set_property(TEST read-multiple PROPERTY DEPENDS write)
@@ -51,34 +50,34 @@ endif()
 add_test(NAME check_benchmark_outputs COMMAND check_benchmark_outputs write_benchmark_root.root read_benchmark_root.root)
 set_property(TEST check_benchmark_outputs PROPERTY DEPENDS read_timed write_timed)
 
+message(STATUS "Test inputs will be stored in: ${ExternalData_OBJECT_STORES} if they are not already present")
+add_executable(read-legacy-files-root read-legacy-files-root.cpp)
+target_link_libraries(read-legacy-files-root PRIVATE TestDataModel TestDataModelDict podio::podioRootIO)
 
-# If the variable is cached and defined now, we have inputs and can add the
-# legacy file read test
-if (DEFINED CACHE{PODIO_TEST_INPUT_DATA_DIR})
-  message(STATUS "Using test inputs stored in: "  ${PODIO_TEST_INPUT_DATA_DIR})
-  add_executable(read-legacy-files-root read-legacy-files-root.cpp)
-  target_link_libraries(read-legacy-files-root PRIVATE TestDataModel TestDataModelDict podio::podioRootIO)
+# Add a legacy test case based on a base executable and a version for which an
+# input file exists
+macro(ADD_PODIO_LEGACY_TEST version base_test input_file)
+  ExternalData_Add_Test(legacy_test_cases
+    NAME ${base_test}_${version}
+    COMMAND ${base_test} ${CMAKE_CURRENT_BINARY_DIR}/../input_files/${input_file}
+    DATA{${PROJECT_SOURCE_DIR}/tests/input_files/${input_file}}
+  )
+  set_property(TEST ${base_test}_${version} PROPERTY ENVIRONMENT
+    LD_LIBRARY_PATH=${PROJECT_BINARY_DIR}/tests:${PROJECT_BINARY_DIR}/src:$ENV{LD_LIBRARY_PATH}
+    # Clear the ROOT_INCLUDE_PATH for the tests, to avoid potential conflicts
+    # with existing headers from other installations
+    ROOT_INCLUDE_PATH=
+  )
 
-  # Add a legacy test case based on a base executable and a version for which an
-  # input file exists
-  macro(ADD_PODIO_LEGACY_TEST version base_test input_file)
-    add_test(NAME ${base_test}_${version} COMMAND ${base_test} ${PODIO_TEST_INPUT_DATA_DIR}/${version}/${input_file})
-    set_property(TEST ${base_test}_${version} PROPERTY ENVIRONMENT
-      LD_LIBRARY_PATH=${PROJECT_BINARY_DIR}/tests:${PROJECT_BINARY_DIR}/src:$ENV{LD_LIBRARY_PATH}
-      # Clear the ROOT_INCLUDE_PATH for the tests, to avoid potential conflicts
-      # with existing headers from other installations
-      ROOT_INCLUDE_PATH=
-    )
-  endmacro()
+endmacro()
 
-  ADD_PODIO_LEGACY_TEST(v00-13 read-legacy-files-root example.root legacy_test_cases)
+ADD_PODIO_LEGACY_TEST(v00-13 read-legacy-files-root v00-13-example.root legacy_test_cases)
 
-  foreach(version IN LISTS legacy_test_versions)
-    ADD_PODIO_LEGACY_TEST(${version} read-legacy-files-root example.root legacy_test_cases)
-    ADD_PODIO_LEGACY_TEST(${version} read_frame_root example_frame.root legacy_test_cases)
-    ADD_PODIO_LEGACY_TEST(${version} read_frame_legacy_root example.root legacy_test_cases)
-  endforeach()
-endif()
+foreach(version IN LISTS legacy_test_versions)
+  ADD_PODIO_LEGACY_TEST(${version} read-legacy-files-root ${version}-example.root legacy_test_cases)
+  ADD_PODIO_LEGACY_TEST(${version} read_frame_root ${version}-example_frame.root legacy_test_cases)
+  ADD_PODIO_LEGACY_TEST(${version} read_frame_legacy_root ${version}-example.root legacy_test_cases)
+endforeach()
 
 #--- Write via python and the ROOT backend and see if we can read it back in in
 #--- c++
diff --git a/tests/scripts/get_test_inputs.sh b/tests/scripts/get_test_inputs.sh
deleted file mode 100644
index 523c3d0cc..000000000
--- a/tests/scripts/get_test_inputs.sh
+++ /dev/null
@@ -1,23 +0,0 @@
-#!/usr/bin/env bash
-
-set -euo pipefail
-
-# Create a temporay folder to get the data to
-PODIO_TEST_INPUT_DATA_DIR=$(mktemp -d -p $(pwd) test_input_data_XXXXXXXX)
-export PODIO_TEST_INPUT_DATA_DIR
-
-# Get a legacy file with the oldest version that we still support
-cd ${PODIO_TEST_INPUT_DATA_DIR}
-mkdir v00-13 && cd v00-13
-wget https://key4hep.web.cern.ch:443/testFiles/podio/v00-13/example.root > /dev/null 2>&1
-
-for version in $@; do
-    cd ${PODIO_TEST_INPUT_DATA_DIR}
-    mkdir ${version} && cd ${version}
-    for fileName in example.root example_frame.root; do
-        wget https://key4hep.web.cern.ch:443/testFiles/podio/${version}/${fileName} > /dev/null 2>&1
-    done
-done
-
-# Announce where we store variables to the outside
-echo -n ${PODIO_TEST_INPUT_DATA_DIR}