Merge pull request #17 from LLNL/CYT-334-Java-support

Add Java support

pyproject.toml

@@ -47,7 +47,8 @@ dependencies = [
     "defusedxml",
     "spdx-tools>=0.7.1,==0.7.*",
     "pluggy",
-    "click"
+    "click",
+    "javatools>=1.6.0"
 ]
 dynamic = ["version"]
 

surfactant/infoextractors/java_file.py

@@ -0,0 +1,80 @@
+from typing import Any, Dict
+
+import javatools.jarinfo
+
+import surfactant.plugin
+from surfactant.sbomtypes import SBOM, Software
+
+# TODO: Add documentation about how to install javatools
+# swig and libssl-dev needs to be installed on Ubuntu
+# https://gitlab.com/m2crypto/m2crypto/-/blob/master/INSTALL.rst
+
+
+def supports_file(filetype: str) -> bool:
+    return filetype in ("JAVACLASS", "JAR", "WAR", "EAR")
+
+
+@surfactant.plugin.hookimpl
+def extract_file_info(sbom: SBOM, software: Software, filename: str, filetype: str) -> object:
+    if not supports_file(filetype):
+        return None
+    return extract_java_info(filename, filetype)
+
+
+# Map from internal major number to Java SE version
+# https://docs.oracle.com/javase/specs/jvms/se20/html/jvms-4.html#jvms-4.1-200-B.2
+_JAVA_VERSION_MAPPING = {
+    45: "1.1",
+    46: "1.2",
+    47: "1.3",
+    48: "1.4",
+    49: "5.0",
+    50: "6",
+    51: "7",
+    52: "8",
+    53: "9",
+    54: "10",
+    55: "11",
+    56: "12",
+    57: "13",
+    58: "14",
+    59: "15",
+    60: "16",
+    61: "17",
+    62: "18",
+    63: "19",
+    64: "20",
+}
+
+
+def handle_java_class(info: Dict[str, Any], class_info: javatools.JavaClassInfo):
+    # This shouldn't happen but just in-case it does don't overwrite information
+    if class_info.get_this() in info["javaClasses"]:
+        return
+    info["javaClasses"][class_info.get_this()] = {}
+    add_to = info["javaClasses"][class_info.get_this()]
+    (major_version, _) = class_info.get_version()
+    if major_version in _JAVA_VERSION_MAPPING:
+        add_to["javaMinSEVersion"] = _JAVA_VERSION_MAPPING[major_version]
+    add_to["javaExports"] = [*class_info.get_provides()]
+    # I've seen this fail for some reason; catch errors on it and just ignore
+    # them if it fails
+    try:
+        add_to["javaImports"] = [*class_info.get_requires()]
+    except IndexError:
+        # Should this be set to "Unknown" or similar?
+        add_to["javaImports"] = []
+
+
+def extract_java_info(filename: str, filetype: str) -> object:
+    info = {"javaClasses": {}}
+    if filetype in ("JAR", "EAR", "WAR"):
+        with javatools.jarinfo.JarInfo(filename) as jarinfo:
+            for class_ in jarinfo.get_classes():
+                handle_java_class(info, jarinfo.get_classinfo(class_))
+    elif filetype == "JAVACLASS":
+        with open(filename, "rb") as f:
+            class_info = javatools.JavaClassInfo()
+            class_info.unpack(javatools.unpack(f))
+        handle_java_class(info, class_info)
+    return info

surfactant/plugin/manager.py

@@ -15,13 +15,15 @@ def _register_plugins(pm: pluggy.PluginManager) -> None:
         a_out_file,
         coff_file,
         elf_file,
+        java_file,
         ole_file,
         pe_file,
     )
     from surfactant.output import csv_writer, cytrics_writer, spdx_writer
     from surfactant.relationships import (
         dotnet_relationship,
         elf_relationship,
+        java_relationship,
         pe_relationship,
     )
 
@@ -31,10 +33,12 @@ def _register_plugins(pm: pluggy.PluginManager) -> None:
         a_out_file,
         coff_file,
         elf_file,
+        java_file,
         pe_file,
         ole_file,
         dotnet_relationship,
         elf_relationship,
+        java_relationship,
         pe_relationship,
         csv_writer,
         cytrics_writer,

surfactant/relationships/java_relationship.py

@@ -0,0 +1,50 @@
+from typing import List, Optional
+
+import surfactant.plugin
+from surfactant.sbomtypes import SBOM, Relationship, Software
+
+
+def has_required_fields(metadata) -> bool:
+    return "javaClasses" in metadata
+
+
+class _ExportDict:
+    created = False
+    supplied_by = {}
+
+    @classmethod
+    def create_export_dict(cls, sbom: SBOM):
+        if cls.created:
+            return
+        for software_entry in sbom.software:
+            for metadata in software_entry.metadata:
+                if "javaClasses" in metadata:
+                    for class_info in metadata["javaClasses"].values():
+                        for export in class_info["javaExports"]:
+                            cls.supplied_by[export] = software_entry.UUID
+        cls.created = True
+
+    @classmethod
+    def get_supplier(cls, import_name: str) -> Optional[str]:
+        if import_name in cls.supplied_by:
+            return cls.supplied_by[import_name]
+        return None
+
+
+@surfactant.plugin.hookimpl
+def establish_relationships(
+    sbom: SBOM, software: Software, metadata
+) -> Optional[List[Relationship]]:
+    if not has_required_fields(metadata):
+        return None
+    _ExportDict.create_export_dict(sbom)
+    relationships = []
+    dependant_uuid = software.UUID
+    for class_info in metadata["javaClasses"].values():
+        for import_ in class_info["javaImports"]:
+            if supplier_uuid := _ExportDict.get_supplier(import_):
+                if supplier_uuid != dependant_uuid:
+                    rel = Relationship(dependant_uuid, supplier_uuid, "Uses")
+                    if rel not in relationships:
+                        relationships.append(rel)
+    return relationships

tests/relationships/test_java.py

@@ -0,0 +1,43 @@
+# Copyright 2023 Lawrence Livermore National Security, LLC
+# See the top-level LICENSE file for details.
+#
+# SPDX-License-Identifier: MIT
+
+from surfactant.plugin.manager import get_plugin_manager
+from surfactant.sbomtypes import SBOM, Relationship, Software
+
+sbom = SBOM(
+    software=[
+        Software(
+            UUID="supplier",
+            fileName=["supplier"],
+            installPath=["supplier"],
+            metadata=[{"javaClasses": {"dummy": {"javaExports": ["someFunc():void"]}}}],
+        ),
+        Software(
+            UUID="consumer",
+            fileName=["consumer"],
+            installPath=["consumer"],
+            metadata=[
+                {
+                    "javaClasses": {
+                        "dummy": {
+                            "javaExports": [],
+                            "javaImports": ["someFunc():void"],
+                        },
+                    },
+                },
+            ],
+        ),
+    ],
+    relationships=[],
+)
+
+
+def test_java_relationship():
+    javaPlugin = get_plugin_manager().get_plugin("surfactant.relationships.java_relationship")
+    sw = sbom.software[1]
+    md = sw.metadata[0]
+    assert javaPlugin.establish_relationships(sbom, sw, md) == [
+        Relationship("consumer", "supplier", "Uses")
+    ]