Use GLOBAL_SRCS to pass sbom information

8b17a7d46ec1a950b59da80e66713dc21714248b

6 files changed, 59 insertions, 51 deletions

diff --git a/build/conf/go.conf b/build/conf/go.conf
index a12d7b8e6a..298b1caa2d 100644
--- a/build/conf/go.conf
+++ b/build/conf/go.conf
@@ -635,7 +635,7 @@ otherwise {
 ###
 ### The base module of all golang modules. Defines common properties, dependencies and rules for go build.
 module _GO_BASE_UNIT: _BASE_UNIT {
-    .EXTS=.go .gosrc .o .obj .a .symabis .mf
+    .EXTS=.go .gosrc .o .obj .sbom .a .symabis .mf
     .CMD=TOUCH_UNIT
     .NODE_TYPE=Bundle
     .GLOBAL=GO_LDFLAGS OBJADDE_LIB
@@ -959,6 +959,7 @@ module GO_TEST: GO_PROGRAM {
     SET(_GO_TOOL_MODE test)
     SET(MODULE_TYPE PROGRAM)
     ENABLE(GO_TEST_MODULE)
+    DISABLE(_NEED_SBOM_INFO)
     SET(GO_COVER_INFO_VALUE)
 
     # We have to add uncoditionaly PEERDIR to ${GOSTD}/os here thouth it is really
diff --git a/build/conf/license.conf b/build/conf/license.conf
index d83e81528d..76133b0c1b 100644
--- a/build/conf/license.conf
+++ b/build/conf/license.conf
@@ -27,16 +27,7 @@ macro LICENSE(Flags...) {
     SET(LICENSE_EXPRESSION $Flags)
     SET(LICENSE_NAMES $Flags)
     # TODO(YMAKE-1136) avoid abusing LICENSE
-    # NOTICE: final value of MODVER might not be set yet if VERSION macro is called after LICENSE. Var expansion is
-    #         escaped here to prevent eager value substitution by SET but allow delayed value substitution when
-    #         adding linking SBOM data command to graph.
-    when ($_NEED_SBOM_INFO == "yes") {
-        _SBOM_INFO_LOCAL="path=${MODDIR};ver=${join=.:MODVER};lang=${MODULE_LANG}"
-    }
-    otherwise {
-        _SBOM_INFO_LOCAL=
-    }
-    SET_APPEND(_SBOM_INFO_GLOBAL $_SBOM_INFO_LOCAL)
+    _CONTRIB_MODULE_HOOKS()
 }
 
 ### @usage LICENSE_RESTRICTION(ALLOW_ONLY|DENY LicenseProperty...)
diff --git a/build/conf/proto.conf b/build/conf/proto.conf
index 7f87cd8e6f..96ae17212a 100644
--- a/build/conf/proto.conf
+++ b/build/conf/proto.conf
@@ -698,6 +698,7 @@ multimodule PROTO_LIBRARY {
         .SEM=JAVA_PROTO_LIBRARY_SEM
         SET(PEERDIR_TAGS JAVA_PROTO)
         ENABLE(JAVA_PROTO)
+        DISABLE(_NEED_SBOM_INFO)
         PEERDIR+=$JAVA_PROTOBUF_PEERS
 
         when ($KOTLIN_PROTO == "yes") {
@@ -724,6 +725,7 @@ multimodule PROTO_LIBRARY {
         .SEM=IGNORED
         SET(PEERDIR_TAGS PY2 PY_PROTO)
         ENABLE(PY_PROTO)
+        DISABLE(_NEED_SBOM_INFO)
         OPTIMIZE_PY_PROTOS()
         OBJ_SUF=.py2
         # Can not use NO_LINT(), because is not allowed outside of contrib directory
@@ -748,6 +750,7 @@ multimodule PROTO_LIBRARY {
         .SEM=IGNORED
         SET(PEERDIR_TAGS PY3 PY3_PROTO)
         ENABLE(PY3_PROTO)
+        DISABLE(_NEED_SBOM_INFO)
         OPTIMIZE_PY_PROTOS()
         OBJ_SUF=.py3
         # Can not use NO_LINT(), because is not allowed outside of contrib directory
@@ -789,11 +792,13 @@ multimodule PROTO_LIBRARY {
         .INCLUDE_TAG=no
         .EPILOGUE=_TS_CONFIG_EPILOGUE
         .PEERDIRSELF=TS_PREPARE_DEPS
+        DISABLE(_NEED_SBOM_INFO)
     }
 
     module TS_PREPARE_DEPS: _PREPARE_DEPS_BASE {
         .INCLUDE_TAG=no
         .IGNORED=PEERDIR
+        DISABLE(_NEED_SBOM_INFO)
     }
 
     module DESC_PROTO: _BARE_UNIT {
@@ -805,6 +810,7 @@ multimodule PROTO_LIBRARY {
 
         SET(PEERDIR_TAGS DESC_PROTO)
         ENABLE(DESC_PROTO)
+        DISABLE(_NEED_SBOM_INFO)
         MODULE_SUFFIX=.self.protodesc
         SET(MODULE_TYPE LIBRARY)
 
diff --git a/build/internal/scripts/gen_sbom.py b/build/internal/scripts/gen_sbom.py
new file mode 100644
index 0000000000..2aec699931
--- /dev/null
+++ b/build/internal/scripts/gen_sbom.py
@@ -0,0 +1,38 @@
+import argparse
+import json
+import os
+
+
+def deduce_name(path):
+    name = os.path.basename(path)
+    for prefix in ['contrib/libs/', 'contrib/python/py2/', 'contrib/python/py3/', 'contrib/python/']:
+        if path.startswith(prefix):
+            name = path[len(prefix):].replace('/', '-')
+            break
+    return name
+
+
+def main():
+    parser = argparse.ArgumentParser(description='Generate single SBOM component JSON object for current third-party library')
+    parser.add_argument('-o', '--output', type=argparse.FileType('w', encoding='UTF-8'), help='resulting SBOM component file', required=True)
+    parser.add_argument('--path', type=str, help='Path to module in arcadia', required=True)
+    parser.add_argument('--ver', type=str, help='Version of the contrib module', required=True)
+    parser.add_argument('--lang', type=str, help='Language of the library', required=True)
+
+    args = parser.parse_args()
+
+    res = {}
+    res['type'] = 'library'
+    res['name'] = deduce_name(args.path)
+    res['version'] = args.ver
+    res["properties"] = [
+        {'name': 'arcadia_module_subdir', 'value': args.path},
+        {'name': 'language', 'value': args.lang}
+    ]
+
+    json.dump(res, args.output)
+    args.output.close()
+
+
+if __name__ == '__main__':
+    main()
diff --git a/build/internal/scripts/link_sbom.py b/build/internal/scripts/link_sbom.py
index fa366007f4..3f1dd71246 100644
--- a/build/internal/scripts/link_sbom.py
+++ b/build/internal/scripts/link_sbom.py
@@ -3,45 +3,12 @@ import json
 import os
 
 
-def parse_kv_arr(val):
-    res = {}
-    for kv in val.split(';'):
-        k, v = kv.split('=')
-        res[k] = v
-    return res
-
-
-def deduce_name(path):
-    name = os.path.basename(path)
-    for prefix in ['contrib/libs/', 'contrib/python/py2/', 'contrib/python/py3/', 'contrib/python/']:
-        if path.startswith(prefix):
-            name = path[len(prefix):].replace('/', '-')
-            break
-    return name
-
-
-def parse_componenet(component):
-    props = parse_kv_arr(component)
-    path = props['path']
-    ver = props['ver']
-
-    res = {}
-    res['type'] = 'library'
-    res['name'] = deduce_name(path)
-    res['version'] = ver
-    res["properties"] = [
-        {'name': 'arcadia_module_subdir', 'value': path},
-        {'name': 'language', 'value': props['lang']}
-    ]
-    return res
-
-
 def main():
     parser = argparse.ArgumentParser(description='Generate SBOM data from used contribs info')
     parser.add_argument('-o', '--output', type=argparse.FileType('w', encoding='UTF-8'), help='resulting SBOM file', required=True)
     parser.add_argument('--vcs-info', type=argparse.FileType('r', encoding='UTF-8'), help='VCS information file', required=True)
     parser.add_argument('--mod-path', type=str, help='Path to module in arcadia', required=True)
-    parser.add_argument('libinfo', metavar='N', type=str, nargs='*', help='libraries info for components section')
+    parser.add_argument('components', metavar='N', type=argparse.FileType('r', encoding='UTF-8'), nargs='*', help='dependencies info in SBOM component JSON format')
 
     args = parser.parse_args()
 
@@ -52,7 +19,7 @@ def main():
     res["bomFormat"] = "CycloneDX"
     res["specVersion"] = "1.5"
     res["version"] = 1
-    res["components"] = [parse_componenet(lib) for lib in args.libinfo]
+    res["components"] = [json.load(dep) for dep in args.components]
     res["properties"] = [
         {'name': 'commit_hash', 'value': vcs['ARCADIA_SOURCE_HG_HASH']},
         {'name': 'arcadia_module_subdir', 'value': args.mod_path}
diff --git a/build/ymake.core.conf b/build/ymake.core.conf
index 9444bfaddb..9c0ae75a4a 100644
--- a/build/ymake.core.conf
+++ b/build/ymake.core.conf
@@ -595,7 +595,7 @@ module GEN_LIBRARY: _BARE_UNIT {
 ### The base of all LIBRARY/PROGRAM modules describing common logic for all modules.
 ### To avoid surprises, all buildable modules are better to be inherited from it or its descendants.
 module _BASE_UNIT: _BARE_UNIT {
-    .GLOBAL=_FBS_NAMESPACE_MAP _SBOM_INFO
+    .GLOBAL=_FBS_NAMESPACE_MAP
 
     PEERDIR_TAGS=CPP_PROTO CPP_FBS CPP_ROS H_IDL PY2 PY2_NATIVE YQL_UDF_STATIC __EMPTY__ RESOURCE_LIB DLL_LIB
 
@@ -1071,7 +1071,7 @@ macro LINK_EXEC_DYN_LIB_IMPL(WHOLE_ARCHIVE_PEERS...) {
 ###
 ### The base of all linkable modules: programs, DLLs etc. Describes common linking logic.
 module _LINK_UNIT: _BASE_UNIT {
-    .EXTS=.o .obj .supp .tidyjson .ld .a .lib
+    .EXTS=.o .obj .supp .sbom .tidyjson .ld .a .lib
     .CMD=LINK_EXE
     .ALLOWED=EXTRALIBS OBJADDE_GLOBAL RESOURCE_FILES
     .NODE_TYPE=Program
@@ -1804,7 +1804,7 @@ module _LIBRARY: _BASE_UNIT {
     .CMD=LINK_LIB
     .NODE_TYPE=Library
     .PEERDIR_POLICY=as_include
-    .EXTS=.o .obj .a .mf .supp .tidyjson .ld .lib
+    .EXTS=.o .obj .a .mf .supp .sbom .tidyjson .ld .lib
     .ALLOWED=GRPC USE_SKIFF EXTRALIBS OBJADDE_GLOBAL RESOURCE_FILES
     .GLOBAL=USER_CFLAGS USER_CXXFLAGS USER_CONLYFLAGS LDFLAGS SRCS _WHOLE_ARCHIVE_LIBS_VALUE RPATH OBJADDE_LIB
     .RESTRICTED=ALLOCATOR SIZE TAG DATA TEST_DATA DEPENDS FORK_TESTS FORK_SUBTESTS SPLIT_FACTOR TEST_CWD RUN TIMEOUT SPLIT_DWARF
@@ -1988,8 +1988,8 @@ module RESOURCES_LIBRARY: _BARE_UNIT {
     .ALIASES=EXTRALIBS=PY_EXTRALIBS
     .NODE_TYPE=Library
     .PEERDIR_POLICY=as_include
-    .EXTS=.o .obj .a .mf .supp .tidyjson .ld .lib
-    .GLOBAL=USER_CFLAGS USER_CXXFLAGS USER_CONLYFLAGS LDFLAGS _WHOLE_ARCHIVE_LIBS_VALUE RPATH OBJADDE_LIB
+    .EXTS=.o .obj .a .mf .supp .sbom .tidyjson .ld .lib
+    .GLOBAL=USER_CFLAGS USER_CXXFLAGS USER_CONLYFLAGS LDFLAGS _WHOLE_ARCHIVE_LIBS_VALUE RPATH OBJADDE_LIB SRCS
 
 
     SET(MODULE_TAG RESOURCE_LIB)
@@ -5915,3 +5915,8 @@ macro CLANG_EMIT_AST_CXX_RUN_TOOL(SOURCES[], OPTS[], Tool, IN{input}[], IN_NOPAR
     .CMD=${cwd:BINDIR} $YMAKE_PYTHON ${input:"build/scripts/clang_wrapper.py"} $WINDOWS ${CLANG_RESOURCE_GLOBAL}/bin/clang++ ${pre=-I:_C__INCLUDE} $CXXFLAGS $C_FLAGS_PLATFORM $LLVM_OPTS -emit-ast -c ${input:SOURCES} ${hide;tmp;noext;nopath:SOURCES.ast} $OPTS ${hide;kv:"p ST"} ${hide;kv:"pc light-green"} && ${cwd:CWD} ${env:ENV} ${tool:Tool} $Args ${hide;input:IN} ${input;context=TEXT;hide:IN_NOPARSE} ${hide;input:IN_DEPS} ${output_include;hide:OUTPUT_INCLUDES} $INDUCED_DEPS ${hide;tool:TOOL} ${hide;output:OUT} ${hide;noauto;output:OUT_NOAUTO} ${output;stdout:STDOUT} ${output;stdout;noauto:STDOUT_NOAUTO} ${hide;kv:"p PR"} ${hide;kv:"pc yellow"} ${hide;kv:"show_out"}
     PEERDIR(build/platform/clang)
 }
+
+### Hooks for contrib modules which can be transparenctly extended in repo internal python plugins
+macro _CONTRIB_MODULE_HOOKS() {
+    ENABLE(UNUSED_MACRO)
+}