Merge pull request #74 from E-ARK-Software/feat/validating-files

Validation of file entries

eark_validator/infopacks/checksummer.py

@@ -0,0 +1,50 @@
+# -*- coding: utf-8 -*-
+from pathlib import Path
+
+from eark_validator.const import NO_PATH, NOT_FILE
+from eark_validator.model import Checksum, ChecksumAlg
+
+class Checksummer:
+    def __init__(self, algorithm: ChecksumAlg | str):
+        if isinstance(algorithm, ChecksumAlg):
+            self._algorithm: ChecksumAlg = algorithm
+        else:
+            self._algorithm: ChecksumAlg = ChecksumAlg.from_string(algorithm)
+
+    @property
+    def algorithm(self) -> ChecksumAlg:
+        """Return the checksum algorithm used by this checksummer."""
+        return self._algorithm
+
+    def hash_file(self, path: Path) -> 'Checksum':
+        """Calculate the checksum of a file.
+
+        Args:
+            path (Path): A path to a file to checksum.
+
+        Raises:
+            FileNotFoundError: If the path parameter is found.
+            ValueError: If the path parameter resolves to a directory.
+
+        Returns:
+            Checksum: A Checksum object containing the Hexadecimal digest of the file.
+        """
+        if not path.exists():
+            raise FileNotFoundError(NO_PATH.format(path))
+        if not path.is_file():
+            raise ValueError(NOT_FILE.format(path))
+        implemenation: ChecksumAlg = ChecksumAlg.get_implementation(self._algorithm)
+        with open(path, 'rb') as file:
+            for chunk in iter(lambda: file.read(4096), b''):
+                implemenation.update(chunk)
+        return Checksum.model_validate({
+                'algorithm': self._algorithm,
+                'value': implemenation.hexdigest()
+                }, strict=True
+            )
+
+    @classmethod
+    def from_file(cls, path: Path, algorithm: 'ChecksumAlg') -> 'Checksum':
+        """Create a Checksum from an etree element."""
+        # Get the child flocat element and grab the href attribute.
+        return Checksummer(algorithm).hash_file(path)

eark_validator/infopacks/information_package.py

@@ -82,14 +82,16 @@ def from_path(package_path: Path) -> InformationPackage:
         if not package_path.exists():
             raise FileNotFoundError(NO_PATH.format(package_path))
         handler: PackageHandler = PackageHandler()
-        to_parse:Path = handler.prepare_package(package_path)
+        to_parse: Path = handler.prepare_package(package_path)
         mets_path: Path = to_parse.joinpath(METS_FILE)
         if not mets_path.is_file():
             raise ValueError('No METS file found in package')
-        mets: MetsFile = MetsFiles.from_file(to_parse.joinpath(METS_FILE))
+
+        mets: MetsFile = MetsFiles.from_file(mets_path)
+        details: PackageDetails = InformationPackages.details_from_mets_file(mets_path)
         return InformationPackage.model_validate({
             METS: mets,
-            'details': InformationPackages.details_from_mets_file(to_parse.joinpath(METS_FILE))
+            'details': details
         })
 
     @staticmethod

eark_validator/infopacks/manifest.py

@@ -34,52 +34,7 @@
 from eark_validator.model.manifest import SourceType
 from eark_validator.model.metadata import FileEntry
 from eark_validator.utils import get_path
-
-
-class Checksummer:
-    def __init__(self, algorithm: ChecksumAlg | str):
-        if isinstance(algorithm, ChecksumAlg):
-            self._algorithm: ChecksumAlg = algorithm
-        else:
-            self._algorithm: ChecksumAlg = ChecksumAlg.from_string(algorithm)
-
-    @property
-    def algorithm(self) -> ChecksumAlg:
-        """Return the checksum algorithm used by this checksummer."""
-        return self._algorithm
-
-    def hash_file(self, path: Path) -> 'Checksum':
-        """Calculate the checksum of a file.
-
-        Args:
-            path (Path): A path to a file to checksum.
-
-        Raises:
-            FileNotFoundError: If the path parameter is found.
-            ValueError: If the path parameter resolves to a directory.
-
-        Returns:
-            Checksum: A Checksum object containing the Hexadecimal digest of the file.
-        """
-        if not path.exists():
-            raise FileNotFoundError(NO_PATH.format(path))
-        if not path.is_file():
-            raise ValueError(NOT_FILE.format(path))
-        implemenation: ChecksumAlg = ChecksumAlg.get_implementation(self._algorithm)
-        with open(path, 'rb') as file:
-            for chunk in iter(lambda: file.read(4096), b''):
-                implemenation.update(chunk)
-        return Checksum.model_validate({
-                'algorithm': self._algorithm,
-                'value': implemenation.hexdigest()
-                }, strict=True
-            )
-
-    @classmethod
-    def from_file(cls, path: Path, algorithm: 'ChecksumAlg') -> 'Checksum':
-        """Create a Checksum from an etree element."""
-        # Get the child flocat element and grab the href attribute.
-        return Checksummer(algorithm).hash_file(path)
+from eark_validator.infopacks.checksummer import Checksummer
 
 class ManifestEntries:
     @staticmethod

eark_validator/ipxml/resources/schematron/V2.0.4/CSIP/mets_amdSec_rules.xml

@@ -19,7 +19,7 @@
       <assert id="CSIP37" role="ERROR" test="@xlink:type = 'simple'">Attribute MUST be used with the value “simple”. Value list is maintained by the xlink standard.</assert>
       <assert id="CSIP38" role="ERROR" test="@xlink:href">MUST record the actual location of the resource. This specification recommends recording a URL type filepath within this attribute.</assert>
       <assert id="CSIP39" role="ERROR" test="@MDTYPE">MUST record the type of metadata at the referenced location.</assert>
-      <assert id="CSIP40" role="ERROR" test="@MIMETYPE_IANA_test">MUST record the MIME type of the referenced file.</assert>
+      <assert id="CSIP40" role="ERROR" test="@MIMETYPE">MUST record the MIME type of the referenced file.</assert>
       <assert id="CSIP41" role="ERROR" test="@SIZE">MUST record the size in bytes of the referenced file.</assert>
       <assert id="CSIP42" role="ERROR" test="@CREATED">MUST record the date the referenced file was created.</assert>
       <assert id="CSIP43" role="ERROR" test="@CHECKSUM">MUST record the checksum of the referenced file.</assert>

eark_validator/ipxml/resources/schematron/V2.0.4/CSIP/mets_dmdSec_rules.xml

@@ -16,7 +16,7 @@
       <assert id="CSIP23" role="ERROR" test="@xlink:type = 'simple'">Attribute used with the value “simple”. Value list is maintained by the xlink standard.</assert>
       <assert id="CSIP24" role="ERROR" test="@xlink:href">The actual location of the resource. This specification recommends recording a URL type filepath in this attribute.</assert>
       <assert id="CSIP25" role="ERROR" test="@MDTYPE">Specifies the type of metadata in the referenced file. Values are taken from the list provided by the METS.</assert>
-      <assert id="CSIP26" role="ERROR" test="@MIMETYPE_IANA_test">MUST hold the IANA mime type of the referenced file.</assert>
+      <assert id="CSIP26" role="ERROR" test="@MIMETYPE">MUST hold the IANA mime type of the referenced file.</assert>
       <assert id="CSIP27" role="ERROR" test="@SIZE">MUST hold the size of the referenced file in bytes.</assert>
       <assert id="CSIP28" role="ERROR" test="@CREATED">MUST hold the creation date of the referenced file.</assert>
       <assert id="CSIP29" role="ERROR" test="@CHECKSUM">MUST hold the checksum of the referenced file.</assert>

eark_validator/ipxml/resources/schematron/V2.1.0/CSIP/mets_amdSec_rules.xml

@@ -19,7 +19,7 @@
       <assert id="CSIP37" role="ERROR" test="@xlink:type = 'simple'">Attribute MUST be used with the value “simple”. Value list is maintained by the xlink standard.</assert>
       <assert id="CSIP38" role="ERROR" test="@xlink:href">MUST record the actual location of the resource. This specification recommends recording a URL type filepath within this attribute.</assert>
       <assert id="CSIP39" role="ERROR" test="@MDTYPE">MUST record the type of metadata at the referenced location.</assert>
-      <assert id="CSIP40" role="ERROR" test="@MIMETYPE_IANA_test">MUST record the MIME type of the referenced file.</assert>
+      <assert id="CSIP40" role="ERROR" test="@MIMETYPE">MUST record the MIME type of the referenced file.</assert>
       <assert id="CSIP41" role="ERROR" test="@SIZE">MUST record the size in bytes of the referenced file.</assert>
       <assert id="CSIP42" role="ERROR" test="@CREATED">MUST record the date and time the referenced file was created.</assert>
       <assert id="CSIP43" role="ERROR" test="@CHECKSUM">MUST record the checksum of the referenced file.</assert>

eark_validator/ipxml/resources/schematron/V2.1.0/CSIP/mets_dmdSec_rules.xml

@@ -16,7 +16,7 @@
       <assert id="CSIP23" role="ERROR" test="@xlink:type = 'simple'">Attribute used with the value “simple”. Value list is maintained by the xlink standard.</assert>
       <assert id="CSIP24" role="ERROR" test="@xlink:href">The actual location of the resource. This specification recommends recording a URL type filepath in this attribute.</assert>
       <assert id="CSIP25" role="ERROR" test="@MDTYPE">Specifies the type of metadata in the referenced file. Values are taken from the list provided by the METS.</assert>
-      <assert id="CSIP26" role="ERROR" test="@MIMETYPE_IANA_test">MUST hold the IANA mime type of the referenced file.</assert>
+      <assert id="CSIP26" role="ERROR" test="@MIMETYPE">MUST hold the IANA mime type of the referenced file.</assert>
       <assert id="CSIP27" role="ERROR" test="@SIZE">MUST hold the size of the referenced file in bytes.</assert>
       <assert id="CSIP28" role="ERROR" test="@CREATED">MUST hold the creation date and time of the referenced file.</assert>
       <assert id="CSIP29" role="ERROR" test="@CHECKSUM">MUST hold the checksum of the referenced file.</assert>