Merge pull request #40 from UIUCLibrary/json-decode-error

feat: --verbose option added to tripwire metadata validate

Author: henryborchers

Jenkinsfile

@@ -26,20 +26,27 @@ documentation = "https://uiuclibrary.github.io/tripwire/"
 tripwire = "uiucprescon.tripwire.main:main"
 
 [dependency-groups]
-test = ["pytest>7", "coverage"]
+test = ["pytest>7"]
 docs = ["sphinx"]
 audit-dependencies = ['uv-secure']
+type-checking = ["mypy", "types-tqdm"]
+lint = ["ruff"]
+tox = ["tox"]
+tox-uv = [
+    {include-group = "tox"},
+    "tox-uv-bare"
+]
 dev = [
     "lxml",
-    "mypy",
     "pre-commit",
     "pytest",
-    "ruff",
-    "tox",
-    "types-tqdm",
     {include-group = "audit-dependencies"},
     {include-group = "docs"},
+    {include-group = "lint"},
     {include-group = "test"},
+    {include-group = "tox"},
+    {include-group = "type-checking"},
+    "coverage"
     ]
 deploy = ["twine"]
 ci = ["pysonar", {include-group = "dev"}]
@@ -68,6 +75,7 @@ pre_bump_hooks = [
 [tool.ruff]
 line-length = 79
 extend-exclude = ["tests"]
+include = ["src/**/*.py"]
 
 [tool.ruff.lint]
 extend-select = [
@@ -90,6 +98,44 @@ description = "Run test under {base_python}"
 commands = [["{env_bin_dir}{/}pytest"]]
 dependency_groups = ['test']
 
+[tool.tox.env.type-checking]
+description = "Run mypy type checking"
+dependency_groups = ['type-checking']
+commands = [["{env_bin_dir}{/}mypy", "-p", "uiucprescon.tripwire"]]
+
+[tool.tox.env.lint]
+description = "Run ruff linting"
+dependency_groups = ['lint']
+skip_install = true
+commands = [["{env_bin_dir}{/}ruff", "check"]]
+
+[tool.tox.env.docs]
+description = "Build documentation"
+dependency_groups = ['docs']
+commands = [
+    [
+        "{env_bin_dir}{/}sphinx-build", "docs", "{temp_dir}{/}build{/}docs",
+        "--fail-on-warning",
+        "--keep-going",
+        "--write-all",
+        "--fresh-env"
+    ]
+]
+
+[tool.tox.env.docs-linkcheck]
+description = "Build documentation"
+dependency_groups = ['docs']
+commands = [
+    [
+        "{env_bin_dir}{/}sphinx-build", "docs", "{temp_dir}{/}build{/}docs-link-check",
+        "--fail-on-warning",
+        "--keep-going",
+        "--write-all",
+        "--fresh-env",
+        "--builder=linkcheck"
+    ]
+]
+
 [tool.mypy]
 mypy_path = "src"
 

pyproject.toml

@@ -0,0 +1,35 @@
+"""Exceptions used in the tripwire package.
+
+.. versionadded:: 0.3.7
+    Added Exceptions module and relocated InvalidFileFormat to here
+"""
+
+
+class TripwireException(Exception):
+    """Base class for all exceptions raised by this module."""
+
+
+class InvalidFileFormat(TripwireException):
+    """Invalid file format exception.
+
+    .. versionchanged:: 0.3.7
+        relocate from uiucprescon.tripwire.files to here
+    """
+
+    def __init__(self, file: str = "", details: str = "") -> None:
+        """Initialize exception.
+
+        Args:
+            file: path of the file that caused the exception. Optional.
+            details: details of the exception. Optional.
+        """
+        message = (
+            f"Invalid file format. File: {file}"
+            if file
+            else "Invalid file format"
+        )
+        if details:
+            message = f"{message}. Details: {details}"
+        super().__init__(message)
+        self.file_name = file
+        self.details = details

src/uiucprescon/tripwire/exceptions.py

@@ -1,4 +1,9 @@
-"""File handling for files used by tripwire."""
+"""File handling for files used by tripwire.
+
+.. versionchanged:: 0.3.7
+   InvalidFileFormat now located in uiucprescon.tripwire.exceptions
+
+"""
 
 import abc
 import collections
@@ -35,22 +40,6 @@
 logger.setLevel(logging.INFO)
 
 
-class InvalidFileFormat(Exception):
-    """Invalid file format exception."""
-
-    def __init__(self, file: str = "", details: str = "") -> None:
-        message = (
-            f"Invalid file format. File: {file}"
-            if file
-            else "Invalid file format"
-        )
-        if details:
-            message = f"{message}. Details: {details}"
-        super().__init__(message)
-        self.file_name = file
-        self.details = details
-
-
 @dataclasses.dataclass(frozen=True)
 class TableRow:
     """Table row."""

src/uiucprescon/tripwire/files.py

@@ -0,0 +1,34 @@
+"""Introspection utilities."""
+
+from importlib import metadata
+
+__all__ = ["get_application_info"]
+
+
+def get_application_info() -> str:
+    """Get application info."""
+    description = get_application_description()
+    packages_report = (
+        f"\nInstalled Python packages:\n\n{get_install_packages_info()}"
+    )
+    return "\n".join(list(filter(None, [description, packages_report])))
+
+
+def get_application_description() -> str:
+    return (
+        "Tripwire is a tool for validating and processing media manifests. "
+        "It is developed by the University of Illinois Urbana-Champaign "
+        "Library's Preservation Services Department."
+    )
+
+
+def get_install_packages_info() -> str:
+    return "\n".join(
+        [
+            f"  {x.metadata['Name']}, version: {x.metadata['Version']}"
+            for x in sorted(
+                metadata.distributions(),
+                key=lambda x: x.metadata["Name"].upper(),
+            )
+        ]
+    )

src/uiucprescon/tripwire/introspection.py

@@ -2,14 +2,21 @@
 # PYTHON_ARGCOMPLETE_OK
 
 import argparse
+import contextlib
 import functools
 import logging
 import pathlib
 import sys
 from typing import Callable, Any, Dict, Tuple, Optional
 
-from uiucprescon.tripwire import validation, utils, manifest_check, metadata
-from uiucprescon.tripwire.files import InvalidFileFormat
+from uiucprescon.tripwire import (
+    validation,
+    utils,
+    manifest_check,
+    metadata,
+    introspection,
+)
+from uiucprescon.tripwire.exceptions import InvalidFileFormat
 import argcomplete
 
 logger = logging.getLogger(__name__)
@@ -121,10 +128,22 @@ def get_arg_parser() -> Tuple[
 
     metadata_show_show.add_argument("glob", type=str)
 
-    metadata_validate = metadata_parser.add_parser("validate")
+    metadata_validate = metadata_parser.add_parser(
+        "validate", help="validate files from mediaconch policy"
+    )
     metadata_validate.add_argument("policy_file", type=pathlib.Path)
     metadata_validate.add_argument("glob", type=str)
-
+    metadata_validate.add_argument(
+        "-v",
+        "--verbose",
+        action="count",
+        default=1,
+        help="increase output verbosity",
+        dest="verbosity",
+    )
+    sub_commands.add_parser(
+        "info", help="get information about current version of tripwire"
+    )
     return (
         parser,
         {
@@ -142,14 +161,56 @@ def metadata_show_command(args: argparse.Namespace) -> None:
     metadata.show_metadata(args.glob, search_path=pathlib.Path("."))
 
 
+@contextlib.contextmanager
+def module_logging_verbosity(logger, verbosity=logging.INFO):
+    """Set logging level to verbosity for a module.
+
+    This is useful for setting the logging level for a module to a specific
+    level for the duration of a block of code, and then resetting it back after
+    leaving the decorated function's scope.
+    """
+    starting_verbosity = logger.getEffectiveLevel()
+    handler_levels = {}
+    try:
+        logger.setLevel(verbosity)
+        for handler in logger.handlers:
+            handler_levels[handler.name] = handler.level
+            handler.setLevel(verbosity)
+        yield
+    finally:
+        logger.setLevel(starting_verbosity)
+        for handler in logger.handlers:
+            if handler.name in handler_levels:
+                handler.setLevel(handler_levels[handler.name])
+
+
 @capture_log(logger=metadata.logger)
-def metadata_validate_command(args: argparse.Namespace) -> None:
+def metadata_validate_command(
+    args: argparse.Namespace,
+    validate_metadata_strategy=metadata.validate_metadata,
+) -> None:
     """Run metadata validate command."""
-    if not metadata.validate_metadata(
-        args.glob, policy_xml_file=args.policy_file
+
+    def get_log_level(verbosity: int) -> int:
+        if verbosity > 2:
+            print("verbosity level to max, defaulting to DEBUG")
+            return logging.DEBUG
+        else:
+            match args.verbosity:
+                case 1:
+                    return logging.INFO
+                case 2:
+                    return logging.DEBUG
+            return logging.INFO
+
+    with module_logging_verbosity(
+        metadata.logger, verbosity=get_log_level(args.verbosity)
     ):
-        print("failed metadata validation")
-        sys.exit(1)
+        if not validate_metadata_strategy(
+            args.glob, policy_xml_file=args.policy_file
+        ):
+            print("failed metadata validation")
+            sys.exit(1)
 
     print("passed metadata validation")
 
@@ -165,6 +226,11 @@ def metadata_command(args: argparse.Namespace, subcommand: str) -> None:
             raise ValueError(f"Unknown metadata subcommand: {subcommand}")
 
 
+def show_info_command() -> None:
+    """Show info about application."""
+    print(introspection.get_application_info())
+
+
 def main() -> None:
     """Main entry point for the Tripwire command line interface."""
     parser, print_help_commands = get_arg_parser()
@@ -182,6 +248,8 @@ def main() -> None:
             )
         case "metadata":
             metadata_command(args, args.metadata_command)
+        case "info":
+            show_info_command()
 
 
 if __name__ == "__main__":

src/uiucprescon/tripwire/main.py

@@ -16,6 +16,8 @@
 )
 import logging
 from uiucprescon.tripwire import files as tripwire_files
+from uiucprescon.tripwire.exceptions import InvalidFileFormat
+
 from tqdm import tqdm
 
 __all__ = ["locate_manifest_files"]
@@ -349,9 +351,7 @@ def locate_manifest_files_fp(
     )
     manifest = tripwire_files.TSVManifest(manifest_tsv_fp)
     if not manifest.is_valid_file():
-        raise tripwire_files.InvalidFileFormat(
-            details="Not a valid TSV manifest file."
-        )
+        raise InvalidFileFormat(details="Not a valid TSV manifest file.")
 
     scanner = PackageScanner(search_path)
     for row in (
@@ -389,8 +389,8 @@ def locate_manifest_files(
             unexpected_files = locate_manifest_files_fp(
                 fp, search_path, manifest_type=manifest_type
             )
-    except tripwire_files.InvalidFileFormat as e:
-        raise tripwire_files.InvalidFileFormat(
+    except InvalidFileFormat as e:
+        raise InvalidFileFormat(
             file=manifest_tsv.name, details=e.details
         ) from e
     if unexpected_files: