instantiate get_spdx_licensing() in a singleton module

this getter takes quite some time and should be called as few times as possible

Signed-off-by: Armin Tänzer <armin.taenzer@tngtech.com>

Author: armintaenzertng

examples/spdx2_document_from_scratch.py

@@ -5,8 +5,7 @@
 from datetime import datetime
 from typing import List
 
-from license_expression import get_spdx_licensing
-
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.model import (
     Actor,
     ActorType,
@@ -65,9 +64,9 @@
         Checksum(ChecksumAlgorithm.SHA1, "d6a770ba38583ed4bb4525bd96e50461655d2758"),
         Checksum(ChecksumAlgorithm.MD5, "624c1abb3664f4b35547e7c73864ad24"),
     ],
-    license_concluded=get_spdx_licensing().parse("GPL-2.0-only OR MIT"),
-    license_info_from_files=[get_spdx_licensing().parse("GPL-2.0-only"), get_spdx_licensing().parse("MIT")],
-    license_declared=get_spdx_licensing().parse("GPL-2.0-only AND MIT"),
+    license_concluded=spdx_licensing.parse("GPL-2.0-only OR MIT"),
+    license_info_from_files=[spdx_licensing.parse("GPL-2.0-only"), spdx_licensing.parse("MIT")],
+    license_declared=spdx_licensing.parse("GPL-2.0-only AND MIT"),
     license_comment="license comment",
     copyright_text="Copyright 2022 Jane Doe",
     description="package description",
@@ -100,8 +99,8 @@
         Checksum(ChecksumAlgorithm.SHA1, "d6a770ba38583ed4bb4525bd96e50461655d2758"),
         Checksum(ChecksumAlgorithm.MD5, "624c1abb3664f4b35547e7c73864ad24"),
     ],
-    license_concluded=get_spdx_licensing().parse("MIT"),
-    license_info_in_file=[get_spdx_licensing().parse("MIT")],
+    license_concluded=spdx_licensing.parse("MIT"),
+    license_info_in_file=[spdx_licensing.parse("MIT")],
     copyright_text="Copyright 2022 Jane Doe",
 )
 file2 = File(
@@ -110,7 +109,7 @@
     checksums=[
         Checksum(ChecksumAlgorithm.SHA1, "d6a770ba38583ed4bb4525bd96e50461655d2759"),
     ],
-    license_concluded=get_spdx_licensing().parse("GPL-2.0-only"),
+    license_concluded=spdx_licensing.parse("GPL-2.0-only"),
 )
 
 # Assuming the package contains those two files, we create two CONTAINS relationships.

src/spdx_tools/common/spdx_licensing.py

@@ -0,0 +1,7 @@
+# SPDX-FileCopyrightText: 2023 spdx contributors
+#
+# SPDX-License-Identifier: Apache-2.0
+from license_expression import get_spdx_licensing
+
+# this getter takes quite long so we only call it once in this singleton module
+spdx_licensing = get_spdx_licensing()

src/spdx_tools/spdx/parser/rdf/license_expression_parser.py

@@ -2,10 +2,11 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 from beartype.typing import Optional, Union
-from license_expression import LicenseExpression, get_spdx_licensing
+from license_expression import LicenseExpression
 from rdflib import RDF, Graph
 from rdflib.term import BNode, Identifier, Node, URIRef
 
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.parser.logger import Logger
 from spdx_tools.spdx.parser.rdf.graph_parsing_functions import get_value_from_graph, remove_prefix
 from spdx_tools.spdx.rdfschema.namespace import LICENSE_NAMESPACE, SPDX_NAMESPACE
@@ -19,7 +20,7 @@ def parse_license_expression(
 ) -> LicenseExpression:
     if not logger:
         logger = Logger()
-    spdx_licensing = get_spdx_licensing()
+
     expression = ""
     if license_expression_node.startswith(LICENSE_NAMESPACE):
         expression = remove_prefix(license_expression_node, LICENSE_NAMESPACE)

src/spdx_tools/spdx/validation/license_expression_validator.py

@@ -3,8 +3,9 @@
 # SPDX-License-Identifier: Apache-2.0
 
 from beartype.typing import List, Optional, Union
-from license_expression import ExpressionError, ExpressionParseError, LicenseExpression, get_spdx_licensing
+from license_expression import ExpressionError, ExpressionParseError, LicenseExpression
 
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.model import Document, SpdxNoAssertion, SpdxNone
 from spdx_tools.spdx.validation.validation_message import SpdxElementType, ValidationContext, ValidationMessage
 
@@ -40,7 +41,7 @@ def validate_license_expression(
     validation_messages = []
     license_ref_ids: List[str] = [license_ref.license_id for license_ref in document.extracted_licensing_info]
 
-    for non_spdx_token in get_spdx_licensing().validate(license_expression).invalid_symbols:
+    for non_spdx_token in spdx_licensing.validate(license_expression).invalid_symbols:
         if non_spdx_token not in license_ref_ids:
             validation_messages.append(
                 ValidationMessage(
@@ -51,14 +52,14 @@ def validate_license_expression(
             )
 
     try:
-        get_spdx_licensing().parse(str(license_expression), validate=True, strict=True)
+        spdx_licensing.parse(str(license_expression), validate=True, strict=True)
     except ExpressionParseError as err:
         # This error is raised when an exception symbol is used as a license symbol and vice versa.
         # So far, it only catches the first such error in the provided string.
         validation_messages.append(ValidationMessage(f"{err}. for license_expression: {license_expression}", context))
     except ExpressionError:
         # This error is raised for invalid symbols within the license_expression, but it provides only a string of
-        # these. On the other hand, get_spdx_licensing().validate() gives an actual list of invalid symbols, so this is
+        # these. On the other hand, spdx_licensing.validate() gives an actual list of invalid symbols, so this is
         # handled above.
         pass
 

src/spdx_tools/spdx/writer/rdf/license_expression_writer.py

@@ -3,18 +3,11 @@
 # SPDX-License-Identifier: Apache-2.0
 from beartype.typing import List, Union
 from boolean import Expression
-from license_expression import (
-    AND,
-    OR,
-    ExpressionInfo,
-    LicenseExpression,
-    LicenseSymbol,
-    LicenseWithExceptionSymbol,
-    get_spdx_licensing,
-)
+from license_expression import AND, OR, ExpressionInfo, LicenseExpression, LicenseSymbol, LicenseWithExceptionSymbol
 from rdflib import RDF, BNode, Graph, URIRef
 from rdflib.term import Literal, Node
 
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.model import SpdxNoAssertion, SpdxNone
 from spdx_tools.spdx.rdfschema.namespace import LICENSE_NAMESPACE, SPDX_NAMESPACE
 
@@ -75,7 +68,7 @@ def add_license_expression_to_graph(
 
 
 def license_or_exception_is_on_spdx_licensing_list(license_symbol: LicenseSymbol) -> bool:
-    symbol_info: ExpressionInfo = get_spdx_licensing().validate(license_symbol)
+    symbol_info: ExpressionInfo = spdx_licensing.validate(license_symbol)
     return not symbol_info.errors
 
 

src/spdx_tools/spdx3/bump_from_spdx2/license_expression.py

@@ -2,15 +2,9 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 from beartype.typing import List, Union
-from license_expression import (
-    AND,
-    OR,
-    LicenseExpression,
-    LicenseSymbol,
-    LicenseWithExceptionSymbol,
-    get_spdx_licensing,
-)
+from license_expression import AND, OR, LicenseExpression, LicenseSymbol, LicenseWithExceptionSymbol
 
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx3.model.licensing import (
     AnyLicenseInfo,
     ConjunctiveLicenseSet,
@@ -61,7 +55,7 @@ def bump_license_expression(
             subject_addition=bump_license_exception(license_expression.exception_symbol, extracted_licensing_info),
         )
     if isinstance(license_expression, LicenseSymbol):
-        if not get_spdx_licensing().validate(license_expression).invalid_symbols:
+        if not spdx_licensing.validate(license_expression).invalid_symbols:
             return ListedLicense(license_expression.key, license_expression.obj, "blank")
         else:
             for licensing_info in extracted_licensing_info:
@@ -80,7 +74,7 @@ def bump_license_expression(
 def bump_license_exception(
     license_exception: LicenseSymbol, extracted_licensing_info: List[ExtractedLicensingInfo]
 ) -> LicenseAddition:
-    if not get_spdx_licensing().validate(license_exception).invalid_symbols:
+    if not spdx_licensing.validate(license_exception).invalid_symbols:
         return ListedLicenseException(license_exception.key, "", "")
     else:
         for licensing_info in extracted_licensing_info:

tests/spdx/examples/test_spdx2_document_from_scratch.py

@@ -5,8 +5,7 @@
 from datetime import datetime
 from typing import List
 
-from license_expression import get_spdx_licensing
-
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.model import (
     Actor,
     ActorType,
@@ -67,9 +66,9 @@ def test_spdx2_document_from_scratch():
             Checksum(ChecksumAlgorithm.SHA1, "d6a770ba38583ed4bb4525bd96e50461655d2758"),
             Checksum(ChecksumAlgorithm.MD5, "624c1abb3664f4b35547e7c73864ad24"),
         ],
-        license_concluded=get_spdx_licensing().parse("GPL-2.0-only OR MIT"),
-        license_info_from_files=[get_spdx_licensing().parse("GPL-2.0-only"), get_spdx_licensing().parse("MIT")],
-        license_declared=get_spdx_licensing().parse("GPL-2.0-only AND MIT"),
+        license_concluded=spdx_licensing.parse("GPL-2.0-only OR MIT"),
+        license_info_from_files=[spdx_licensing.parse("GPL-2.0-only"), spdx_licensing.parse("MIT")],
+        license_declared=spdx_licensing.parse("GPL-2.0-only AND MIT"),
         license_comment="license comment",
         copyright_text="Copyright 2022 Jane Doe",
         description="package description",
@@ -102,8 +101,8 @@ def test_spdx2_document_from_scratch():
             Checksum(ChecksumAlgorithm.SHA1, "d6a770ba38583ed4bb4525bd96e50461655d2758"),
             Checksum(ChecksumAlgorithm.MD5, "624c1abb3664f4b35547e7c73864ad24"),
         ],
-        license_concluded=get_spdx_licensing().parse("MIT"),
-        license_info_in_file=[get_spdx_licensing().parse("MIT")],
+        license_concluded=spdx_licensing.parse("MIT"),
+        license_info_in_file=[spdx_licensing.parse("MIT")],
         copyright_text="Copyright 2022 Jane Doe",
     )
     file2 = File(
@@ -112,7 +111,7 @@ def test_spdx2_document_from_scratch():
         checksums=[
             Checksum(ChecksumAlgorithm.SHA1, "d6a770ba38583ed4bb4525bd96e50461655d2759"),
         ],
-        license_concluded=get_spdx_licensing().parse("GPL-2.0-only"),
+        license_concluded=spdx_licensing.parse("GPL-2.0-only"),
     )
 
     # Assuming the package contains those two files, we create two CONTAINS relationships.

tests/spdx/fixtures.py

@@ -3,8 +3,7 @@
 # SPDX-License-Identifier: Apache-2.0
 from datetime import datetime
 
-from license_expression import get_spdx_licensing
-
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.constants import DOCUMENT_SPDX_ID
 from spdx_tools.spdx.model import (
     Actor,
@@ -88,7 +87,7 @@ def file_fixture(
     spdx_id="SPDXRef-File",
     checksums=None,
     file_types=None,
-    license_concluded=get_spdx_licensing().parse("MIT and GPL-2.0"),
+    license_concluded=spdx_licensing.parse("MIT and GPL-2.0"),
     license_info_in_file=None,
     license_comment="licenseComment",
     copyright_text="copyrightText",
@@ -100,7 +99,7 @@ def file_fixture(
     checksums = [checksum_fixture()] if checksums is None else checksums
     file_types = [FileType.TEXT] if file_types is None else file_types
     license_info_in_file = (
-        [get_spdx_licensing().parse("MIT"), get_spdx_licensing().parse("GPL-2.0"), SpdxNoAssertion()]
+        [spdx_licensing.parse("MIT"), spdx_licensing.parse("GPL-2.0"), SpdxNoAssertion()]
         if license_info_in_file is None
         else license_info_in_file
     )
@@ -135,9 +134,9 @@ def package_fixture(
     checksums=None,
     homepage="https://homepage.com",
     source_info="sourceInfo",
-    license_concluded=get_spdx_licensing().parse("MIT and GPL-2.0"),
+    license_concluded=spdx_licensing.parse("MIT and GPL-2.0"),
     license_info_from_files=None,
-    license_declared=get_spdx_licensing().parse("MIT and GPL-2.0"),
+    license_declared=spdx_licensing.parse("MIT and GPL-2.0"),
     license_comment="packageLicenseComment",
     copyright_text="packageCopyrightText",
     summary="packageSummary",
@@ -152,7 +151,7 @@ def package_fixture(
 ) -> Package:
     checksums = [checksum_fixture()] if checksums is None else checksums
     license_info_from_files = (
-        [get_spdx_licensing().parse("MIT"), get_spdx_licensing().parse("GPL-2.0"), SpdxNoAssertion()]
+        [spdx_licensing.parse("MIT"), spdx_licensing.parse("GPL-2.0"), SpdxNoAssertion()]
         if license_info_from_files is None
         else license_info_from_files
     )
@@ -208,7 +207,7 @@ def snippet_fixture(
     file_spdx_id="SPDXRef-File",
     byte_range=(1, 2),
     line_range=(3, 4),
-    license_concluded=get_spdx_licensing().parse("MIT and GPL-2.0"),
+    license_concluded=spdx_licensing.parse("MIT and GPL-2.0"),
     license_info_in_snippet=None,
     license_comment="snippetLicenseComment",
     copyright_text="licenseCopyrightText",
@@ -217,7 +216,7 @@ def snippet_fixture(
     attribution_texts=None,
 ) -> Snippet:
     license_info_in_snippet = (
-        [get_spdx_licensing().parse("MIT"), get_spdx_licensing().parse("GPL-2.0"), SpdxNone()]
+        [spdx_licensing.parse("MIT"), spdx_licensing.parse("GPL-2.0"), SpdxNone()]
         if license_info_in_snippet is None
         else license_info_in_snippet
     )

tests/spdx/parser/jsonlikedict/test_license_expression_parser.py

@@ -4,8 +4,8 @@
 from unittest import TestCase
 
 import pytest
-from license_expression import get_spdx_licensing
 
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.model import SpdxNoAssertion, SpdxNone
 from spdx_tools.spdx.parser.error import SPDXParsingError
 from spdx_tools.spdx.parser.jsonlikedict.license_expression_parser import LicenseExpressionParser
@@ -14,8 +14,8 @@
 @pytest.mark.parametrize(
     "license_expression_str, expected_license",
     [
-        ("First License", get_spdx_licensing().parse("First License")),
-        ("Second License", get_spdx_licensing().parse("Second License")),
+        ("First License", spdx_licensing.parse("First License")),
+        ("Second License", spdx_licensing.parse("Second License")),
         ("NOASSERTION", SpdxNoAssertion()),
         ("NONE", SpdxNone()),
     ],

tests/spdx/parser/rdf/test_file_parser.py

@@ -5,9 +5,9 @@
 from unittest import TestCase
 
 import pytest
-from license_expression import get_spdx_licensing
 from rdflib import RDF, BNode, Graph, URIRef
 
+from spdx_tools.common.spdx_licensing import spdx_licensing
 from spdx_tools.spdx.model import Checksum, ChecksumAlgorithm, FileType, SpdxNoAssertion
 from spdx_tools.spdx.parser.error import SPDXParsingError
 from spdx_tools.spdx.parser.rdf.file_parser import parse_file
@@ -29,10 +29,10 @@ def test_parse_file():
     assert file.comment == "fileComment"
     assert file.copyright_text == "copyrightText"
     assert file.contributors == ["fileContributor"]
-    assert file.license_concluded == get_spdx_licensing().parse("MIT AND GPL-2.0")
+    assert file.license_concluded == spdx_licensing.parse("MIT AND GPL-2.0")
     TestCase().assertCountEqual(
         file.license_info_in_file,
-        [get_spdx_licensing().parse("MIT"), get_spdx_licensing().parse("GPL-2.0"), SpdxNoAssertion()],
+        [spdx_licensing.parse("MIT"), spdx_licensing.parse("GPL-2.0"), SpdxNoAssertion()],
     )
     assert file.license_comment == "licenseComment"
     assert file.notice == "fileNotice"