dev

Jenkinsfile

@@ -965,6 +965,21 @@ pipeline {
                                                 }
                                             }
                                         }
+                                        stage('pyDocStyle'){
+                                            steps{
+                                                catchError(buildResult: 'SUCCESS', message: 'Did not pass all pyDocStyle tests', stageResult: 'UNSTABLE') {
+                                                    sh(
+                                                        label: 'Run pydocstyle',
+                                                        script: 'uv run pydocstyle src/uiucprescon/imagevalidate > reports/pydocstyle-report.txt'
+                                                    )
+                                                }
+                                            }
+                                            post {
+                                                always{
+                                                    recordIssues(tools: [pyDocStyle(pattern: 'reports/pydocstyle-report.txt')])
+                                                }
+                                            }
+                                        }
                                         stage('Run Doctest Tests'){
                                            steps {
                                                catchError(buildResult: 'SUCCESS', message: 'Doctest found issues', stageResult: 'UNSTABLE') {
@@ -1019,6 +1034,24 @@ pipeline {
                                                 }
                                             }
                                         }
+                                        stage('Run Pylint Static Analysis') {
+                                            steps{
+                                                catchError(buildResult: 'SUCCESS', message: 'Pylint found issues', stageResult: 'UNSTABLE') {
+                                                    sh(
+                                                        script: '''mkdir -p logs
+                                                                   mkdir -p reports
+                                                                   uv run pylint src/uiucprescon/imagevalidate -r n --msg-template="{path}:{line}: [{msg_id}({symbol}), {obj}] {msg}" > reports/pylint.txt
+                                                                ''',
+                                                        label: 'Running pylint'
+                                                    )
+                                                }
+                                            }
+                                            post{
+                                                always {
+                                                    recordIssues(tools: [pyLint(pattern: 'reports/pylint.txt')])
+                                                }
+                                            }
+                                        }
                                     }
                                     post{
                                         always{
@@ -1071,12 +1104,12 @@ pipeline {
                                                     if (env.CHANGE_ID){
                                                         sh(
                                                             label: 'Running Sonar Scanner',
-                                                            script: "uv run pysonar -t \$token -Dsonar.projectVersion=\$VERSION -Dsonar.buildString=\"${env.BUILD_TAG}\" -Dsonar.pullrequest.key=${env.CHANGE_ID} -Dsonar.pullrequest.base=${env.CHANGE_TARGET} -Dsonar.cfamily.cache.enabled=false -Dsonar.cfamily.threads=\$(grep -c ^processor /proc/cpuinfo) -Dsonar.cfamily.build-wrapper-output=build/build_wrapper_output_directory"
+                                                            script: "uv run pysonar -t \$token -Dsonar.projectVersion=\$VERSION -Dsonar.buildString=\"${env.BUILD_TAG}\" -Dsonar.pullrequest.key=${env.CHANGE_ID} -Dsonar.pullrequest.base=${env.CHANGE_TARGET} -Dsonar.cfamily.cache.enabled=false -Dsonar.cfamily.threads=\$(grep -c ^processor /proc/cpuinfo) -Dsonar.cfamily.build-wrapper-output=build/build_wrapper_output_directory -Dsonar.python.pylint.reportPaths=reports/pylint.txt"
                                                             )
                                                     } else {
                                                         sh(
                                                             label: 'Running Sonar Scanner',
-                                                            script: "uv run pysonar -t \$token -Dsonar.projectVersion=\$VERSION -Dsonar.buildString=\"${env.BUILD_TAG}\" -Dsonar.branch.name=${env.BRANCH_NAME} -Dsonar.cfamily.cache.enabled=false -Dsonar.cfamily.threads=\$(grep -c ^processor /proc/cpuinfo) -Dsonar.cfamily.build-wrapper-output=build/build_wrapper_output_directory"
+                                                            script: "uv run pysonar -t \$token -Dsonar.projectVersion=\$VERSION -Dsonar.buildString=\"${env.BUILD_TAG}\" -Dsonar.branch.name=${env.BRANCH_NAME} -Dsonar.cfamily.cache.enabled=false -Dsonar.cfamily.threads=\$(grep -c ^processor /proc/cpuinfo) -Dsonar.cfamily.build-wrapper-output=build/build_wrapper_output_directory -Dsonar.python.pylint.reportPaths=reports/pylint.txt"
                                                             )
                                                     }
                                                 }

docs/source/conf.py

@@ -38,7 +38,7 @@ def get_project_metadata():
 author = metadata['authors'][0]['name']
 
 # The short X.Y version
-version_extractor = re.compile("\d+[.]\d+[.]\d+")
+version_extractor = re.compile(r"\d+[.]\d+[.]\d+")
 version = version_extractor.search(metadata["version"]).group(0)
 # The full version, including alpha/beta/rc tags.
 release = metadata["version"]

pyproject.toml

@@ -102,6 +102,11 @@ norecursedirs = "build"
 markers = "integration"
 junit_family="xunit2"
 
+[tool.pylint.MAIN]
+init-hook = "import sys; sys.path.append('src')"  # Replace 'src' with your module directory
+extension-pkg-allow-list = ["py3exiv2bind.core", "uiucprescon.imagevalidate.openjp2wrap"]
+
+
 [tool.cibuildwheel]
 test-groups = ["test"]
 test-command = "pytest {project}/tests"

src/uiucprescon/imagevalidate/__init__.py

@@ -2,7 +2,9 @@
 
 from .issues import IssueCategory
 from .report import Report
-from .profile import Profile, available_profiles, get_profile, get_profile_classes
+from .profile import (
+    Profile, available_profiles, get_profile, get_profile_classes
+)
 from . import profiles
 
 __all__ = [

src/uiucprescon/imagevalidate/common.py

@@ -12,6 +12,7 @@ class InvalidStrategy(Exception):
 
 
 class AbsColorSpaceExtractor(metaclass=abc.ABCMeta):
+    # pylint: disable=too-few-public-methods
     """Base class for extracting the color space from an image file."""
 
     @abc.abstractmethod
@@ -29,6 +30,7 @@ def check(self, image: str) -> str:
 
 
 class ExtractColorSpace:
+    # pylint: disable=too-few-public-methods
     """Strategy context for extract color space from a file."""
 
     def __init__(self, strategy: AbsColorSpaceExtractor) -> None:
@@ -55,6 +57,7 @@ def check(self, image: str) -> str:
 
 
 class ColorSpaceIccDeviceModelCheck(AbsColorSpaceExtractor):
+    # pylint: disable=too-few-public-methods
     """Extract color space by reading the device_model tag in the ICC profile.
 
     Useful for identifying sRGB.
@@ -74,8 +77,8 @@ def check(self, image: str) -> str:
         exiv_image = py3exiv2bind.Image(image)
         try:
             icc = exiv_image.icc()
-        except py3exiv2bind.core.NoICCError:
-            raise InvalidStrategy("Unable to get ICC profile.")
+        except py3exiv2bind.core.NoICCError as error:
+            raise InvalidStrategy("Unable to get ICC profile.") from error
 
         device_model = icc.get('device_model')
         if not device_model or \
@@ -85,6 +88,7 @@ def check(self, image: str) -> str:
 
 
 class ColorSpaceIccPrefCcmCheck(AbsColorSpaceExtractor):
+    # pylint: disable=too-few-public-methods
     """Extract color space from reading pref_ccm in the ICC profile header."""
 
     def check(self, image: str) -> str:
@@ -102,8 +106,9 @@ def check(self, image: str) -> str:
         try:
             icc = exiv2_image.icc()
         except py3exiv2bind.core.NoICCError as error:
-            raise InvalidStrategy("Unable to get ICC profile."
-                                  "Reason: {}".format(error))
+            raise InvalidStrategy(
+                f"Unable to get ICC profile.Reason: {error}"
+            ) from error
 
         pref_ccm = icc.get("pref_ccm")
         if not pref_ccm or pref_ccm.value.decode("ascii").rstrip(' \0') == '':
@@ -112,6 +117,7 @@ def check(self, image: str) -> str:
 
 
 class ColorSpaceOJPCheck(AbsColorSpaceExtractor):
+    # pylint: disable=too-few-public-methods
     """Color space extractor using openjpeg library."""
 
     def check(self, image: str) -> str:

src/uiucprescon/imagevalidate/issues.py

@@ -1,7 +1,11 @@
+"""Module for defining issue categories related to image validation."""
+
 from enum import Enum
 
 
 class IssueCategory(Enum):
+    """Enum class defining issue categories."""
+
     VALID = 0
     EMPTY_DATA = 1
     MISSING_FIELD = 2

src/uiucprescon/imagevalidate/messages.py

@@ -6,6 +6,7 @@
 
 
 class AbsMessage(metaclass=abc.ABCMeta):
+    # pylint: disable=too-few-public-methods
     """Base class for messages."""
 
     @abc.abstractmethod
@@ -14,6 +15,7 @@ def generate_message(self, field: str, data: report.Result) -> str:
 
 
 class InvalidData(AbsMessage):
+    # pylint: disable=too-few-public-methods
     """Invalid data."""
 
     def generate_message(self, field: str, data: report.Result) -> str:
@@ -24,6 +26,7 @@ def generate_message(self, field: str, data: report.Result) -> str:
 
 
 class EmptyData(AbsMessage):
+    # pylint: disable=too-few-public-methods
     """Empty data."""
 
     def generate_message(self, field: str, data: report.Result) -> str:
@@ -32,6 +35,7 @@ def generate_message(self, field: str, data: report.Result) -> str:
 
 
 class MissingField(AbsMessage):
+    # pylint: disable=too-few-public-methods
     """Missing fields."""
 
     def generate_message(self, field: str, data: report.Result) -> str:
@@ -40,6 +44,7 @@ def generate_message(self, field: str, data: report.Result) -> str:
 
 
 class MessageGenerator:
+    # pylint: disable=too-few-public-methods
     """Message Generator."""
 
     def __init__(self, strategy: AbsMessage) -> None:

src/uiucprescon/imagevalidate/profile.py

@@ -1,15 +1,19 @@
 """Profile for validating images."""
+from __future__ import annotations
 
 import os
 import inspect
-from typing import Type, Set, Dict
-from uiucprescon import imagevalidate
+from typing import Type, Set, Dict, TYPE_CHECKING
 from . import profiles as profile_pkg
 
+if TYPE_CHECKING:
+    from uiucprescon import imagevalidate
+
 known_profiles: Dict[str, Type[profile_pkg.AbsProfile]] = {}
 
 
 class Profile:
+    # pylint: disable=too-few-public-methods
     """Profile loader for validating embedded metadata in image files."""
 
     def __init__(self, validation_profile: profile_pkg.AbsProfile) -> None:
@@ -33,6 +37,7 @@ def validate(self, file: str) -> imagevalidate.Report:
         if not os.path.exists(file):
             raise FileNotFoundError(f"Unable to locate {file}")
         return self._profile.validate(file)
+# pylint: enable=too-few-public-methods
 
 
 def available_profiles() -> Set[str]:
@@ -55,6 +60,7 @@ def get_profile(name: str) -> profile_pkg.AbsProfile:
 
 
 def get_profile_classes():
+    """Get all available profiles."""
     known_package_profiles: Dict[str, Type[profile_pkg.AbsProfile]] = {}
     profiles = \
         inspect.getmembers(
@@ -66,4 +72,5 @@ def get_profile_classes():
         known_package_profiles[profile[1].profile_name()] = profile[1]
     return known_package_profiles
 
+
 known_profiles = get_profile_classes()

src/uiucprescon/imagevalidate/profiles/absProfile.py

@@ -1,21 +1,27 @@
 """Abstract class for creating a profile."""
+from __future__ import annotations
 
 import abc
 
-from typing import Dict, List, Optional, Set
+from typing import Dict, List, Optional, Set, TYPE_CHECKING
+
 import py3exiv2bind
-from uiucprescon.imagevalidate import Report, IssueCategory, messages
+from uiucprescon.imagevalidate import messages
+from uiucprescon.imagevalidate.issues import IssueCategory
 from uiucprescon.imagevalidate.report import Result, ResultCategory
 
+if TYPE_CHECKING:
+    from uiucprescon.imagevalidate.report import Report
+
 
 class AbsProfile(metaclass=abc.ABCMeta):
     """Base class for metadata validation.
 
     Implement the validate method when creating new profile
     """
 
-    expected_metadata_constants: Dict[str, str] = dict()
-    expected_metadata_any_value: List[str] = list()
+    expected_metadata_constants: Dict[str, str] = {}
+    expected_metadata_any_value: List[str] = []
     valid_extensions: Set[str] = set()
 
     @staticmethod
@@ -38,7 +44,7 @@
     def _get_metadata_static_values(cls, image: py3exiv2bind.Image) \
             -> Dict[str, Result]:
 
-        data = dict()
+        data = {}
         for key, value in cls.expected_metadata_constants.items():
             data[key] = Result(
                 expected=value,
@@ -50,7 +56,7 @@
     def _get_metadata_has_values(cls, image: py3exiv2bind.Image) -> \
             Dict[str, Result]:
 
-        data = dict()
+        data = {}
         for key in cls.expected_metadata_any_value:
             data[key] = Result(
                 expected=ResultCategory.ANY,
@@ -75,7 +81,7 @@
 
             return message_generator.generate_message(field, report_data)
 
-        return "Unknown error with {}".format(field)
+        return f"Unknown error with {field}"
 
     @staticmethod
     def analyze_data_for_issues(result: Result) -> Optional[IssueCategory]:
@@ -98,7 +104,7 @@
             -> Dict[str, Result]:
         """Access data from image."""
         image = py3exiv2bind.Image(filename)
-        data: Dict[str, Result] = dict()
+        data: Dict[str, Result] = {}
         data.update(cls._get_metadata_has_values(image))
         data.update(cls._get_metadata_static_values(image))
         return data

src/uiucprescon/imagevalidate/profiles/hathi_common.py

@@ -0,0 +1,77 @@
+"""Shared values for HathiTrust profiles."""
+from __future__ import annotations
+
+import collections
+import typing
+from abc import ABC
+
+from uiucprescon.imagevalidate import Report
+from . import AbsProfile
+
+if typing.TYPE_CHECKING:
+    from uiucprescon.imagevalidate import IssueCategory
+
+__all__ = [
+    "SHARED_EXPECTED_METADATA_ANY_VALUE",
+    "SHARED_EXPECT_RESOLUTION_CONSTANTS"
+]
+
+SHARED_EXPECTED_METADATA_ANY_VALUE = [
+        'Xmp.dc.creator',
+
+        # Address
+        'Xmp.iptc.CreatorContactInfo/Iptc4xmpCore:CiAdrExtadr',
+
+        # City
+        'Xmp.iptc.CreatorContactInfo/Iptc4xmpCore:CiAdrCity',
+
+        # State
+        'Xmp.iptc.CreatorContactInfo/Iptc4xmpCore:CiAdrRegion',
+
+        # Zip code
+        'Xmp.iptc.CreatorContactInfo/Iptc4xmpCore:CiAdrPcode',
+
+        # Country
+        'Xmp.iptc.CreatorContactInfo/Iptc4xmpCore:CiAdrCtry',
+
+        # phone number
+        'Xmp.iptc.CreatorContactInfo/Iptc4xmpCore:CiTelWork',
+    ]
+
+SHARED_EXPECT_RESOLUTION_CONSTANTS = {
+    "Exif.Image.XResolution": "400/1",
+    "Exif.Image.YResolution": "400/1",
+}
+
+
+class AbsValidateHathiTrustProfile(AbsProfile, ABC):
+    """Profile for validating files for HathiTrust."""
+
+    def validate(self, file: str) -> Report:
+        """Validate the image file as a HathiTrust image.
+
+        Args:
+            file:
+                File path to an image file
+
+        Returns:
+            Returns a report of the results.
+
+        """
+        report = Report()
+        report.filename = file
+        report_data = self.get_data_from_image(file)
+        report._properties = report_data
+
+        analysis: typing.Dict[IssueCategory, list] = \
+            collections.defaultdict(list)
+
+        for key, result in report_data.items():
+            issue_category = self.analyze_data_for_issues(result)
+            if issue_category:
+                message = self.generate_error_msg(issue_category, key, result)
+                analysis[issue_category].append(message)
+
+        report._data.update(analysis)
+
+        return report