[bug#58261] [PATCH v2 13/13] gnu: Add rdkit.

[Liliana Marie Prikler <liliana.prikler@gmail.com>]

Am Freitag, dem 07.10.2022 um 15:22 +0000 schrieb David Elsing:
> * gnu/packages/chemistry.scm (rdkit): New variable.
> ---
>  gnu/packages/chemistry.scm                    | 178
> ++++++++++++++++++
>  .../rdkit-cmake-external-dependencies.patch   | 102 ++++++++++
>  2 files changed, 280 insertions(+)
>  create mode 100644 gnu/packages/patches/rdkit-cmake-external-
> dependencies.patch
> 
> diff --git a/gnu/packages/chemistry.scm b/gnu/packages/chemistry.scm
> index 43953238cd..4b34a38797 100644
> --- a/gnu/packages/chemistry.scm
> +++ b/gnu/packages/chemistry.scm
> @@ -34,6 +34,7 @@ (define-module (gnu packages chemistry)
>    #:use-module (gnu packages algebra)
>    #:use-module (gnu packages autotools)
>    #:use-module (gnu packages backup)
> +  #:use-module (gnu packages base)
>    #:use-module (gnu packages bison)
>    #:use-module (gnu packages boost)
>    #:use-module (gnu packages check)
> @@ -41,10 +42,12 @@ (define-module (gnu packages chemistry)
>    #:use-module (gnu packages cpp)
>    #:use-module (gnu packages documentation)
>    #:use-module (gnu packages flex)
> +  #:use-module (gnu packages fonts)
>    #:use-module (gnu packages fontutils)
>    #:use-module (gnu packages gcc)
>    #:use-module (gnu packages gl)
>    #:use-module (gnu packages graphviz)
> +  #:use-module (gnu packages gtk)
>    #:use-module (gnu packages gv)
>    #:use-module (gnu packages image)
>    #:use-module (gnu packages maths)
> @@ -56,6 +59,7 @@ (define-module (gnu packages chemistry)
>    #:use-module (gnu packages qt)
>    #:use-module (gnu packages serialization)
>    #:use-module (gnu packages sphinx)
> +  #:use-module (gnu packages sqlite)
>    #:use-module (gnu packages stb)
>    #:use-module (gnu packages tex)
>    #:use-module (gnu packages web)
> @@ -1101,3 +1105,177 @@ (define-public ringdecomposerlib
>  unique ring families, relevant cycles, the smallest set of smallest
> rings and
>  other ring topology descriptions.")
>      (license license:bsd-3)))
> +
> +(define-public rdkit
> +  (package
> +    (name "rdkit")
> +    (version "2022.03.5")
> +    (source (origin
> +              (method git-fetch)
> +              (uri (git-reference
> +                    (url "https://github.com/rdkit/rdkit")
> +                    (commit
> +                     (string-append
> +                      "Release_" (string-replace-substring version
> "." "_")))))
> +              (file-name (git-file-name name version))
> +              (sha256
> +               (base32
> +               
> "19idgilabh04cbr1qj6zgrgsfjm248mmfz6fsr0smrd68d0xnml9"))
> +              (patches
> +               (search-patches "rdkit-cmake-external-
> dependencies.patch"))
> +              (modules '((guix build utils)))
> +              (snippet
> +               #~(begin
> +                   ;; Remove pickle files (only used in tests),
> +                   ;; as they are compiled programs
> +                   (for-each
> +                    (lambda (name)
> +                      (display (string-append name "\n"))
> +                      (delete-file name))
> +                    (find-files "." "\\.pkl(\\.gz)?$"))
> +                   ;; Remove SQLite data files (can be generated)
> +                   (delete-file "Data/RDData.sqlt")
> +                   (delete-file "Data/RDTests.sqlt")))))
> +    (build-system cmake-build-system)
> +    (arguments
> +     (list
> +      #:imported-modules (append %cmake-build-system-modules
> +                                 '((guix build python-build-
> system)))
> +      #:modules '((guix build cmake-build-system)
> +                  (guix build utils)
> +                  ((guix build python-build-system)
> +                   #:select (add-installed-pythonpath)))
> +      #:configure-flags
> +      #~(list
> +         "-DRDK_BUILD_AVALON_SUPPORT=ON"
> +         "-DRDK_BUILD_CAIRO_SUPPORT=ON"
> +         "-DRDK_BUILD_CPP_TESTS=ON"
> +         "-DRDK_BUILD_FREESASA_SUPPORT=ON"
> +         "-DRDK_BUILD_INCHI_SUPPORT=ON"
> +         "-DRDK_BUILD_YAEHMOP_SUPPORT=ON"
Are any of these implied or do they have to be explicitly enabled?
> +         (string-append "-DCATCH_DIR="
> +                        #$(this-package-native-input "catch2")
> +                        "/include/catch2")
> +         "-DRDK_INSTALL_DEV_COMPONENT=ON"
> +         "-DRDK_INSTALL_INTREE=OFF"
> +         "-DRDK_INSTALL_STATIC_LIBS=OFF"
> +         (string-append
> +          "-DRDK_OPTIMIZE_POPCNT="
> +          #$(let ((system (or (%current-target-system)
> +                              (%current-system))))
> +              (cond
> +               ((string-prefix? "x86_64" system) "ON")
> +               ((string-prefix? "i686" system) "ON")
> +               (else "OFF"))))
> +         "-DRDK_USE_BOOST_IOSTREAMS=ON"
> +         "-DRDK_USE_FLEXBISON=ON")
> +      #:phases
> +      #~(modify-phases %standard-phases
> +          (add-after 'unpack 'copy-external-dependencies
> +            (lambda _
> +              (symlink
> +               (string-append
> +                #$(this-package-input "font-comic-neue")
> +                "/share/fonts/truetype/ComicNeue-Regular.ttf")
> +               "Data/Fonts/ComicNeue-Regular.ttf")))
Try (search-input-file) instead.
> +          (add-after 'copy-external-dependencies 'patch-source
> +            (lambda _
> +              (substitute* "External/AvalonTools/CMakeLists.txt"
> +                (("^if\\(NOT RDK_BUILD" orig)
> +                 (string-append "if(0)\n" orig))
> +                (("include_directories.*AVALON_SRC_PATH.*")
> +                 (string-append "endif()\ninclude_directories("
> +                                #$avalontoolkit
> "/include/avalontoolkit)\n"))
> +                (("LINK_LIBRARIES avalon_clib")
> +                 (string-append
> +                  "LINK_LIBRARIES -L" #$avalontoolkit "/lib -
> lavalontoolkit")))
> +              (let ((port
> +                     (open-file "External/RingFamilies/dummy.cpp"
> "w")))
> +                (display "" port)
> +                (close port))
> +              (let ((port
> +                     (open-file
> "External/RingFamilies/CMakeLists.txt" "w")))
> +                (display (string-append
> +                          "rdkit_library(RingDecomposerLib dummy.cpp
> SHARED"
> +                          " LINK_LIBRARIES -lRingDecomposerLib)\n"
> +                          "set(RDK_URF_LIBS RingDecomposerLib CACHE
> STRING \"\" FORCE)\n")
> +                         port)
> +                (close port))
> +              (substitute*
> "Code/GraphMol/MolInterchange/CMakeLists.txt"
> +                (("if\\(NOT" orig)
> +                 (string-append "if(0)\n" orig))
> +                (("^include_directories.*") "endif()\n"))
> +              (substitute* "External/YAeHMOP/EHTTools.cpp"
> +                (("<yaehmop/tightbind/bind.h>")
> "<yaehmop/bind.h>"))))
I think you might want to unbundle these with a patch.
> +          (add-after 'unpack 'fix-inchi-include
> +            (lambda _
> +              (substitute* "Code/cmake/Modules/FindInchi.cmake"
> +                (("inchi_api.h.*\\)") "inchi/inchi_api.h)")
> +                (("INCHI_LIBRARY NAMES.*\\)")
> +                 "INCHI_LIBRARY NAMES inchi PATH_SUFFIXES inchi)")
> +                (("find_library" prev)
> +                 (string-append
> +                  "list(APPEND CMAKE_FIND_LIBRARY_SUFFIXES .so.1)\n"
> +                  prev)))
> +              (substitute* "External/INCHI-API/inchi.cpp"
> +                (("<inchi_api.h>") "<inchi/inchi_api.h>"))))
> +          (add-before 'build 'enable-bytecode-determinism
> +              (lambda _
> +                (setenv "PYTHONHASHSEED" "0")
> +                (setenv "PYTHONDONTWRITEBYTECODE" "1")))
> +          (add-after 'install 'pre-check
> +            (lambda* (#:key inputs outputs #:allow-other-keys)
> +              (with-directory-excursion "../source"
> +                (invoke "sqlite3" "Data/RDData.sqlt"
> +                        ".read rdkit/Dbase/test_data/RDData.sqlite")
> +                (invoke "sqlite3" "Data/RDTests.sqlt"
> +                        ".read
> rdkit/Dbase/test_data/RDTests.sqlite")
> +                (setenv "RDBASE" (canonicalize-path ".")))
> +              (add-installed-pythonpath inputs outputs)))
> +          (delete 'check)
> +          (add-after 'pre-check 'check
> +            (lambda* (#:key tests? parallel-tests? #:allow-other-
> keys)
> +              (when tests?
> +                (let ((job-count (number->string
> +                                  (if parallel-tests? (parallel-job-
> count) 1))))
> +                  (invoke
> +                   "ctest" "-j" job-count
> +                   "-E" (string-append
> +                         "(" (string-join
> +                              '(;; need pickled data
> +                                "(pyDiscreteValueVect"
> "pySparseIntVect"
> +                                "graphmoltestPickler"
> "pyPartialCharges"
> +                                "substructLibraryTest" "pyFeatures"
> +                                "pythonTestDirML"
> "pythonTestDirChem"
> +                                ;; Catching Python exception fails
> +                                "pyRanker)") "|") ")")))))))))
I think you have two layers of brackets now – delete the inner ones.
> +    (native-inputs
> +     (list bison
> +           boost
> +           catch2
> +           eigen
> +           flex
> +           freesasa
> +           pkg-config
> +           rapidjson
> +           tar))
> +    (inputs
> +     (list avalontoolkit
> +           cairo
> +           coordgenlibs
> +           font-comic-neue
> +           freetype
> +           inchi
> +           maeparser
> +           python
> +           ringdecomposerlib
> +           sqlite
> +           yaehmop))
> +    (propagated-inputs
> +     (list python-numpy python-cairocffi python-pillow))
> +    (home-page "https://rdkit.org/")
> +    (synopsis "Collection of cheminformatics software")
> +    (description "RDKit is a C++ and Python library for
> cheminformatics, which
> +includes (among other things) the analysis and modification of
> molecules in 2D
> +and 3D and descriptor generation for machine learning.")
> +    (license license:bsd-3)))
> diff --git a/gnu/packages/patches/rdkit-cmake-external-
> dependencies.patch b/gnu/packages/patches/rdkit-cmake-external-
> dependencies.patch
> new file mode 100644
> index 0000000000..5b73864f51
> --- /dev/null
> +++ b/gnu/packages/patches/rdkit-cmake-external-dependencies.patch
> @@ -0,0 +1,102 @@
> +diff -ur a/External/FreeSASA/CMakeLists.txt
> b/External/FreeSASA/CMakeLists.txt
> +--- a/External/FreeSASA/CMakeLists.txt
> ++++ b/External/FreeSASA/CMakeLists.txt
> +@@ -10,7 +10,7 @@
> +   endif()
> + endif()
> + 
> +-if(needDownload)
> ++if(FALSE)
> +   # don't actually use the md5 here
> +   set(FREESASA_VERSION "2.0.3")
> +   set(FREESASA_SRC_DIR "${CMAKE_CURRENT_SOURCE_DIR}/freesasa-
> ${FREESASA_VERSION}")
> +@@ -66,24 +66,24 @@
> + endif()
> + 
> + set (freesasa_h ${FREESASA_SRC_DIR}/src/freesasa.h)
> +-file(READ ${freesasa_h} freesasa_h_data)
> ++#file(READ ${freesasa_h} freesasa_h_data)
> + string(REGEX REPLACE "(#include <stdio.h>)" "\\1\n#include
> <RDGeneral/export.h>" freesasa_h_data "${freesasa_h_data}")
> + string(REGEX REPLACE
> "([^R][^D][^K][^I][^T][^_][^F][^R][^E][^E][^S][^A][^S][^A][^_][^C][^L
> ][^I][^B][^_][^E][^X][^P][^O][^R][^T][^ ])(extern const)"
> "\\1RDKIT_FREESASA_CLIB_EXPORT \\2" freesasa_h_data
> "${freesasa_h_data}")
> +-file(WRITE ${freesasa_h} "${freesasa_h_data}")
> ++#file(WRITE ${freesasa_h} "${freesasa_h_data}")
> + 
> + add_definitions(-DUSE_THREADS=0)
> + add_definitions(-DUSE_JSON=0)
> + add_definitions(-DUSE_XML=0)
> +-rdkit_library(freesasa_clib ${freesasa_clib_srcs})
> +-target_compile_definitions(freesasa_clib PRIVATE
> RDKIT_FREESASALIB_BUILD)
> ++#rdkit_library(freesasa_clib ${freesasa_clib_srcs})
> ++#target_compile_definitions(freesasa_clib PRIVATE
> RDKIT_FREESASALIB_BUILD)
> + 
> + if((MSVC AND RDK_INSTALL_DLLS_MSVC) OR ((NOT MSVC) AND WIN32))
> +   set_target_properties(freesasa_clib PROPERTIES
> WINDOWS_EXPORT_ALL_SYMBOLS TRUE)
> + endif()
> +-include_directories("${FREESASA_SRC_DIR}/src")
> ++#include_directories("${FREESASA_SRC_DIR}/src")
> + 
> + rdkit_library(FreeSASALib RDFreeSASA.cpp SHARED 
> +-    LINK_LIBRARIES freesasa_clib GraphMol )
> ++    LINK_LIBRARIES -lfreesasa GraphMol )
> + target_compile_definitions(FreeSASALib PRIVATE
> RDKIT_FREESASALIB_BUILD)
> + 
> + rdkit_headers(RDFreeSASA.h DEST GraphMol)
> +diff -ur a/External/YAeHMOP/CMakeLists.txt
> b/External/YAeHMOP/CMakeLists.txt
> +--- a/External/YAeHMOP/CMakeLists.txt
> ++++ b/External/YAeHMOP/CMakeLists.txt
> +@@ -18,32 +18,32 @@
> + 
> + include_directories( ${RDKit_ExternalDir}/YAeHMOP )
> + 
> +-ExternalProject_Add(yaehmop_project
> +-  GIT_REPOSITORY https://github.com/greglandrum/yaehmop.git
> +-  GIT_TAG master
> +-  UPDATE_COMMAND ""
> +-  PATCH_COMMAND ""
> +-  PREFIX ${CMAKE_CURRENT_SOURCE_DIR}
> +-  SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/yaehmop"
> +-  SOURCE_SUBDIR "tightbind"
> +-  CMAKE_ARGS -DUSE_BLAS_LAPACK=OFF -
> DCMAKE_INSTALL_PREFIX=${PROJECT_BINARY_DIR} -
> DCMAKE_C_FLAGS=${CMAKE_C_FLAGS} -
> DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
> +-  TEST_COMMAND "")
> +-
> +-include_directories(${PROJECT_BINARY_DIR}/include)
> +-link_directories(${PROJECT_BINARY_DIR}/lib)
> +-link_directories(${CMAKE_CURRENT_SOURCE_DIR}/src/yaehmop_project-
> build)
> +-
> +-set(EHT_PARAM_FILE
> ${CMAKE_CURRENT_SOURCE_DIR}/yaehmop/tightbind/eht_parms.dat )
> +-install(FILES ${EHT_PARAM_FILE}
> +-        DESTINATION ${RDKit_ShareDir}/Data
> +-        COMPONENT data)
> ++#ExternalProject_Add(yaehmop_project
> ++#  GIT_REPOSITORY https://github.com/greglandrum/yaehmop.git
> ++#  GIT_TAG master
> ++#  UPDATE_COMMAND ""
> ++#  PATCH_COMMAND ""
> ++#  PREFIX ${CMAKE_CURRENT_SOURCE_DIR}
> ++#  SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/yaehmop"
> ++#  SOURCE_SUBDIR "tightbind"
> ++#  CMAKE_ARGS -DUSE_BLAS_LAPACK=OFF -
> DCMAKE_INSTALL_PREFIX=${PROJECT_BINARY_DIR} -
> DCMAKE_C_FLAGS=${CMAKE_C_FLAGS} -
> DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
> ++#  TEST_COMMAND "")
> ++
> ++#include_directories(${PROJECT_BINARY_DIR}/include)
> ++#link_directories(${PROJECT_BINARY_DIR}/lib)
> ++#link_directories(${CMAKE_CURRENT_SOURCE_DIR}/src/yaehmop_project-
> build)
> ++
> ++#set(EHT_PARAM_FILE
> ${CMAKE_CURRENT_SOURCE_DIR}/yaehmop/tightbind/eht_parms.dat )
> ++#install(FILES ${EHT_PARAM_FILE}
> ++#        DESTINATION ${RDKit_ShareDir}/Data
> ++#        COMPONENT data)
> + 
> +-message("YAeHMOP include_dirs: ${PROJECT_BINARY_DIR}/include")
> +-message("YAeHMOP link_dirs: ${PROJECT_BINARY_DIR}/lib
> ${CMAKE_CURRENT_SOURCE_DIR}/src/yaehmop_project-build")
> ++#message("YAeHMOP include_dirs: ${PROJECT_BINARY_DIR}/include")
> ++#message("YAeHMOP link_dirs: ${PROJECT_BINARY_DIR}/lib
> ${CMAKE_CURRENT_SOURCE_DIR}/src/yaehmop_project-build")
> + 
> +-rdkit_library(EHTLib EHTTools.cpp SHARED LINK_LIBRARIES yaehmop_eht
> GraphMol )
> ++rdkit_library(EHTLib EHTTools.cpp SHARED LINK_LIBRARIES -
> lyaehmop_eht GraphMol )
> + target_compile_definitions(EHTLib PRIVATE RDKIT_EHTLIB_BUILD)
> +-add_dependencies(EHTLib yaehmop_project)
> ++#add_dependencies(EHTLib yaehmop_project)
> + rdkit_headers(EHTTools.h DEST GraphMol)
> + rdkit_catch_test(testEHTLib1 test1.cpp
> +            LINK_LIBRARIES EHTLib FileParsers SmilesParse )
Cheers