[issue-185] squash review commits

Signed-off-by: Meret Behrens <meret.behrens@tngtech.com>

examples/write_tv.py

@@ -15,7 +15,7 @@
     from spdx.review import Review
     from spdx.package import Package
     from spdx.file import File, FileType
-    from spdx.checksum import Algorithm
+    from spdx.checksum import Checksum
     from spdx.utils import SPDXNone, NoAssert, UnKnown
 
     doc = Document()
@@ -36,7 +36,7 @@
     testfile1.type = FileType.BINARY
     testfile1.spdx_id = "TestFilet#SPDXRef-FILE"
     testfile1.comment = "This is a test file."
-    testfile1.chksum = Algorithm("SHA1", "c537c5d99eca5333f23491d47ededd083fefb7ad")
+    testfile1.chksum = Checksum("SHA1", "c537c5d99eca5333f23491d47ededd083fefb7ad")
     testfile1.conc_lics = License.from_identifier("BSD-2-Clause")
     testfile1.add_lics(License.from_identifier("BSD-2-Clause"))
     testfile1.copyright = SPDXNone()
@@ -48,7 +48,7 @@
     testfile2.type = FileType.SOURCE
     testfile2.spdx_id = "TestFile2#SPDXRef-FILE"
     testfile2.comment = "This is a test file."
-    testfile2.chksum = Algorithm("SHA1", "bb154f28d1cf0646ae21bb0bec6c669a2b90e113")
+    testfile2.chksum = Checksum("SHA1", "bb154f28d1cf0646ae21bb0bec6c669a2b90e113")
     testfile2.conc_lics = License.from_identifier("Apache-2.0")
     testfile2.add_lics(License.from_identifier("Apache-2.0"))
     testfile2.copyright = NoAssert()
@@ -62,7 +62,7 @@
     package.file_name = "twt.jar"
     package.spdx_id = 'TestPackage#SPDXRef-PACKAGE'
     package.download_location = "http://www.tagwritetest.test/download"
-    package.checksum = Algorithm("SHA1", "c537c5d99eca5333f23491d47ededd083fefb7ad")
+    package.checksum = Checksum("SHA1", "c537c5d99eca5333f23491d47ededd083fefb7ad")
     package.homepage = SPDXNone()
     package.verif_code = "4e3211c67a2d28fced849ee1bb76e7391b93feba"
     license_set = LicenseConjunction(

spdx/checksum.py

@@ -10,10 +10,10 @@
 # limitations under the License.
 import re
 from enum import Enum, auto
+from typing import Optional
 
 
-
-class ChecksumAlgorithmIdentifier(Enum):
+class ChecksumAlgorithm(Enum):
     SHA1 = auto()
     SHA224 = auto()
     SHA256 = auto()
@@ -32,27 +32,47 @@ class ChecksumAlgorithmIdentifier(Enum):
     MD6 = auto()
     ADLER32 = auto()
 
-    def checksum_to_rdf(self):
-        return "checksumAlgorithm_" + self.name.lower()
+    def algorithm_to_rdf_representation(self) -> str:
+        if self.name.startswith("BLAKE2B"):
+            return "checksumAlgorithm_" + self.name.replace("_", "").lower()
+        else:
+            return "checksumAlgorithm_" + self.name.lower()
 
     @classmethod
-    def checksum_from_rdf(cls, identifier: str) -> str:
+    def checksum_from_rdf(cls, identifier: str) -> 'ChecksumAlgorithm':
         identifier = identifier.split('_', 1)[-1].upper()
-        blake_checksum = re.compile(r"(BLAKE2B)\s*(256|384|512)", re.UNICODE)
+        blake_checksum = re.compile(r"^(BLAKE2B)(256|384|512)$", re.UNICODE)
         match = blake_checksum.match(identifier)
         if match:
-            return match[1] + '_' + match[2]
-        return identifier
+            identifier = match[1] + '_' + match[2]
+        if identifier not in ChecksumAlgorithm.__members__:
+            raise ValueError(f"Invalid algorithm for checksum: {identifier}")
+        return ChecksumAlgorithm[identifier]
+
+    @classmethod
+    def checksum_algorithm_from_string(cls, identifier: str) -> 'ChecksumAlgorithm':
+        identifier.replace("-", "_").upper()
+        if identifier not in ChecksumAlgorithm.__members__:
+            raise ValueError(f"Invalid algorithm for checksum: {identifier}")
+        return ChecksumAlgorithm[identifier]
 
 
-class Algorithm(object):
+class Checksum(object):
     """Generic checksum algorithm."""
 
-    def __init__(self, identifier: str, value):
-        if identifier.upper().replace('-', '_') not in ChecksumAlgorithmIdentifier.__members__:
-            raise ValueError('Invalid algorithm for Checksum: {}'.format(identifier))
+    def __init__(self, identifier: ChecksumAlgorithm, value: str):
         self.identifier = identifier
         self.value = value
 
+    @classmethod
+    def checksum_from_string(cls, value: str) -> 'Checksum':
+        CHECKSUM_RE = re.compile("(ADLER32|BLAKE2b-256|BLAKE2b-384|BLAKE2b-512|BLAKE3|MD2|MD4|MD5|MD6|" \
+                                 "SHA1|SHA224|SHA256|SHA384|SHA512|SHA3-256|SHA3-384|SHA3-512):\\s*([a-fA-F0-9]*)")
+        match = CHECKSUM_RE.match(value)
+        if match is None or match.group(1) is None or match.group(2) is None:
+            raise ValueError(f"Invalid checksum: {value}")
+        identifier = ChecksumAlgorithm.checksum_algorithm_from_string(match.group(1))
+        return Checksum(identifier=identifier, value=match.group(2))
+
     def to_tv(self):
-        return "{0}: {1}".format(self.identifier, self.value)
+        return "{0}: {1}".format(self.identifier.name, self.value)

spdx/cli_tools/parser.py

@@ -52,7 +52,7 @@ def main(file, force):
         )
         print("Package Homepage: {0}".format(package.homepage))
         for checksum in doc.package.checksums.values():
-            print("Package Checksum: {0} {1}".format(checksum.identifier, checksum.value))
+            print("Package Checksum: {0} {1}".format(checksum.identifier.name, checksum.value))
         print("Package Attribution Text: {0}".format(package.attribution_text))
         print("Package verification code: {0}".format(package.verif_code))
         print(
@@ -82,8 +82,8 @@ def main(file, force):
         print("\tFile name: {0}".format(f.name))
         for file_type in f.file_types:
             print("\tFile type: {0}".format(file_type.name))
-        for file_checksum in f.checksums:
-            print("\tFile Checksum: {0} {1}".format(file_checksum.identifier, file_checksum.value))
+        for file_checksum in f.checksums.values():
+            print("\tFile Checksum: {0} {1}".format(file_checksum.identifier.name, file_checksum.value))
         print("\tFile license concluded: {0}".format(f.conc_lics))
         print(
             "\tFile license info in file: {0}".format(

spdx/document.py

@@ -34,25 +34,25 @@ class ExternalDocumentRef(object):
     """
 
     def __init__(
-        self, external_document_id=None, spdx_document_uri=None, check_sum=None
+        self, external_document_id=None, spdx_document_uri=None, checksum=None
     ):
         self.external_document_id = external_document_id
         self.spdx_document_uri = spdx_document_uri
-        self.check_sum = check_sum
+        self.checksum = checksum
 
     def __eq__(self, other):
         return (
             isinstance(other, ExternalDocumentRef)
             and self.external_document_id == other.external_document_id
             and self.spdx_document_uri == other.spdx_document_uri
-            and self.check_sum == other.check_sum
+            and self.checksum == other.checksum
         )
 
     def __lt__(self, other):
-        return (self.external_document_id, self.spdx_document_uri, self.check_sum) < (
+        return (self.external_document_id, self.spdx_document_uri, self.checksum) < (
             other.external_document_id,
             other.spdx_document_uri,
-            other.check_sum,
+            other.checksum,
         )
 
     def validate(self, messages: ErrorMessages) -> ErrorMessages:
@@ -74,7 +74,7 @@ def validate_spdx_doc_uri(self, messages):
             messages.append("ExternalDocumentRef has no SPDX Document URI.")
 
     def validate_checksum(self, messages):
-        if not self.check_sum:
+        if not self.checksum:
             messages.append("ExternalDocumentRef has no Checksum.")
 
 

spdx/file.py

@@ -14,10 +14,11 @@
 from functools import total_ordering
 import hashlib
 
-from spdx.checksum import Algorithm, ChecksumAlgorithmIdentifier
-from spdx import utils
 from spdx.license import License
+from spdx import utils
+from spdx.checksum import Checksum, ChecksumAlgorithm
 from spdx.parsers.builderexceptions import SPDXValueError
+from spdx.parsers.loggers import ErrorMessages
 
 
 class FileType(Enum):
@@ -59,7 +60,7 @@ class File(object):
     referenced by other elements. Mandatory, one. Type: str.
     - comment: File comment str, Optional zero or one.
     - file_types: list of file types. Cardinality 0..*
-    - checksums: Dict with identifier of spdx.checksum.Algorithm as key and spdx.checksum.Algorithm as value,
+    - checksums: Dict with checksum.ChecksumAlgorithm as key and checksum.Checksum as value,
     there must be a SHA1 hash, at least.
     - conc_lics: Mandatory one. license.License or utils.NoAssert or utils.SPDXNone.
     - licenses_in_file: list of licenses found in file, mandatory one or more.
@@ -101,22 +102,22 @@ def __lt__(self, other):
         return self.name < other.name
 
     @property
-    def chk_sum(self):
+    def checksum(self):
         """
         Backwards compatibility, return SHA1 checksum.
         """
         warnings.warn("This property is deprecated. Use get_checksum instead.")
-        return self.get_checksum(ChecksumAlgorithmIdentifier.SHA1)
+        return self.get_checksum(ChecksumAlgorithm.SHA1)
 
-    @chk_sum.setter
-    def chk_sum(self, value):
+    @checksum.setter
+    def checksum(self, value):
         """
         Backwards compatibility, set checksum.
         """
         warnings.warn("This property is deprecated. Use set_checksum instead.")
         if isinstance(value, str):
-            self.set_checksum(Algorithm("SHA1", value))
-        elif isinstance(value, Algorithm):
+            self.set_checksum(Checksum("SHA1", value))
+        elif isinstance(value, Checksum):
             self.set_checksum(value)
 
     def add_lics(self, lics):
@@ -144,7 +145,7 @@ def validate(self, messages):
         messages.push_context(self.name)
         self.validate_concluded_license(messages)
         self.validate_file_types(messages)
-        self.validate_checksum(messages)
+        self.validate_checksums(messages)
         self.validate_licenses_in_file(messages)
         self.validate_copyright(messages)
         self.validate_artifacts(messages)
@@ -211,13 +212,16 @@ def validate_file_types(self, messages):
                 messages.append(f"{file_type} is not of type FileType.")
         return messages
 
-    def validate_checksum(self, messages):
-        if self.get_checksum(ChecksumAlgorithmIdentifier.SHA1) is None:
+    def validate_checksums(self, messages: ErrorMessages):
+        for checksum in self.checksums.values():
+            if not isinstance(checksum, Checksum):
+                messages.append("File checksum must be instance of spdx.checksum.Checksum.")
+
+        if self.get_checksum(ChecksumAlgorithm.SHA1) is None:
             messages.append("At least one file checksum algorithm must be SHA1")
-            return messages
 
     def calculate_checksum(self, hash_algorithm='SHA1'):
-        if hash_algorithm not in ChecksumAlgorithmIdentifier.__members__:
+        if hash_algorithm not in ChecksumAlgorithm.__members__:
             raise ValueError
         BUFFER_SIZE = 65536
 
@@ -231,14 +235,13 @@ def calculate_checksum(self, hash_algorithm='SHA1'):
 
         return file_hash.hexdigest()
 
-    def get_checksum(self, hash_algorithm: ChecksumAlgorithmIdentifier = ChecksumAlgorithmIdentifier.SHA1) -> Algorithm:
-        return self.checksums[hash_algorithm.name]
+    def get_checksum(self, hash_algorithm: ChecksumAlgorithm = ChecksumAlgorithm.SHA1) -> Checksum:
+        return self.checksums[hash_algorithm]
 
-    def set_checksum(self, new_checksum: Algorithm):
-        if not isinstance(new_checksum, Algorithm):
+    def set_checksum(self, new_checksum: Checksum):
+        if not isinstance(new_checksum, Checksum):
             raise SPDXValueError
-        if new_checksum.identifier in self.checksums:
-            return
+
         self.checksums[new_checksum.identifier] = new_checksum
 
     def has_optional_field(self, field):

spdx/package.py

@@ -17,7 +17,7 @@
 from spdx import creationinfo
 from spdx import license
 from spdx import utils
-from spdx.checksum import Algorithm, ChecksumAlgorithmIdentifier
+from spdx.checksum import Checksum, ChecksumAlgorithm
 from spdx.parsers.builderexceptions import SPDXValueError
 from spdx.parsers.loggers import ErrorMessages
 
@@ -60,7 +60,7 @@ class Package(object):
      whenever files_analyzed is True or None (omitted) and Must be None (omitted)
      if files_analyzed is False. However, as a convenience within this library,
      we allow this to be Optional even when files_analyzed is True/None.
-     - checksums: Optional, Dict with identifier of spdx.checksum.Algorithm as key and spdx.checksum.Algorithm as value.
+     - checksums: Optional, Dict with checksum.ChecksumAlgorithm as key and checksum.Checksum as value.
      - source_info: Optional string.
      - conc_lics: Mandatory license.License or utils.SPDXNone or
      utils.NoAssert.
@@ -120,22 +120,22 @@ def __init__(
         self.valid_until_date: Optional[datetime] = None
 
     @property
-    def check_sum(self):
+    def checksum(self):
         """
         Backwards compatibility, return SHA1 checksum.
         """
         warnings.warn("This property is deprecated. Use get_checksum instead.")
-        return self.get_checksum(ChecksumAlgorithmIdentifier.SHA1)
+        return self.get_checksum(ChecksumAlgorithm.SHA1)
 
-    @check_sum.setter
-    def check_sum(self, value):
+    @checksum.setter
+    def checksum(self, value):
         """
         Backwards compatibility, set SHA1 checksum.
         """
         warnings.warn("This property is deprecated. Use set_checksum instead.")
         if isinstance(value, str):
-            self.set_checksum(Algorithm("SHA1", value))
-        elif isinstance(value, Algorithm):
+            self.set_checksum(Checksum("SHA1", value))
+        elif isinstance(value, Checksum):
             self.set_checksum(value)
 
     @property
@@ -160,7 +160,7 @@ def validate(self, messages):
         """
         messages.push_context(self.name)
         self.validate_files_analyzed(messages)
-        self.validate_checksum(messages)
+        self.validate_checksums(messages)
         self.validate_optional_str_fields(messages)
         self.validate_mandatory_str_fields(messages)
         self.validate_pkg_ext_refs(messages)
@@ -291,23 +291,20 @@ def validate_str_fields(self, fields, optional, messages):
 
         return messages
 
-    def validate_checksum(self, messages):
+    def validate_checksums(self, messages: ErrorMessages):
         if not self.checksums:
-            return messages
-        try:
-            self.get_checksum(ChecksumAlgorithmIdentifier.SHA1)
-        except KeyError:
-            messages.append("At least one package checksum algorithm must be SHA1")
-            return messages
-
-    def get_checksum(self, hash_algorithm: ChecksumAlgorithmIdentifier = ChecksumAlgorithmIdentifier.SHA1) -> Algorithm:
-        return self.checksums[hash_algorithm.name]
-
-    def set_checksum(self, new_checksum: Algorithm):
-        if not isinstance(new_checksum, Algorithm):
-            raise SPDXValueError("Package::Checksum")
-        if new_checksum.identifier in self.checksums:
             return
+        for checksum in self.checksums.values():
+            if not isinstance(checksum, Checksum):
+                messages.append("Package checksum must be instance of spdx.checksum.Checksum")
+
+    def get_checksum(self, hash_algorithm: ChecksumAlgorithm = ChecksumAlgorithm.SHA1) -> Checksum:
+        return self.checksums[hash_algorithm]
+
+    def set_checksum(self, new_checksum: Checksum):
+        if not isinstance(new_checksum, Checksum):
+            raise SPDXValueError("Package::Checksum")
+
         self.checksums[new_checksum.identifier] = new_checksum
 
     def has_optional_field(self, field):

spdx/parsers/jsonyamlxml.py

@@ -10,7 +10,7 @@
 # limitations under the License.
 from datetime import datetime
 from enum import Enum, auto
-from typing import List, Dict, Tuple, Callable
+from typing import List, Dict, Tuple, Callable, Optional
 
 from spdx import document
 from spdx import utils
@@ -1093,10 +1093,10 @@ def _handle_file_dependency(self, file_dependency):
             return None
         return None
 
-    def parse_file_checksums(self, file_checksums):
+    def parse_file_checksums(self, file_checksums: List[Dict]) -> Optional[bool]:
         """
         Parse File checksums
-        - file_checksum: Python List
+        - file_checksums: Python List
         """
         if isinstance(file_checksums, list):
             for checksum in file_checksums:
@@ -1593,10 +1593,10 @@ def parse_pkg_files(self, pkg_has_files: List[str], method_to_parse_relationship
         elif pkg_has_files is not None:
             self.value_error("PKG_HAS_FILES", pkg_has_files)
 
-    def parse_pkg_checksums(self, pkg_checksums):
+    def parse_pkg_checksums(self, pkg_checksums: List[Dict]) -> Optional[bool]:
         """
-        Parse Package checksum
-        - pkg_chksum: Python str/unicode
+        Parse Package checksums
+        - pkg_chksums: Python List
         """
         if isinstance(pkg_checksums, list):
             for checksum in pkg_checksums: