✨ New DatcmpJob

xtl.config.settings:DependenciesSettings
- Added new dependency for ATSAS

xtl.saxs.jobs.atsas_utils:DatcmpOptions
- Configuration for ATSAS datcmp

xtl.saxs.jobs.atsas:DatcmpJob
- Job for ATSAS datcmp

xtl.automate.batchfile:BatchFile
- Fixed a bug in .add_commands() where if a single command was passed as
  a string it would not have been properly propagated
- Fixed a bug in .dependencies where the missing dependencies where not
  properly reported

xtl.jobs.config:BatchConfig
- Added .dependencies attribute
- Updated the .get_batch() method to propagate the dependencies to the
  BatchFile constructor
- Updated the .shell setter to also accept WslShell instances

Author: dtriand

src/xtl/automate/batchfile.py

@@ -108,7 +108,10 @@ def dependencies(self) -> list['DependencySettings']:
         for dname, dep in settings.dependencies:
             if dname in self._dependencies:
                 deps.append(dep)
-            else:
+
+        # Report missing dependencies
+        for dname in self._dependencies:
+            if dname not in settings.dependencies.to_dict():
                 logger.warning('Dependency %(name)s not found in xtl.settings, '
                                'skipping', {'name': dname})
         return deps
@@ -148,6 +151,8 @@ def add_commands(self, *commands: str | Iterable[str]) -> None:
         """
         if len(commands) == 1 and isinstance(commands[0], Iterable):  # unpack a list or tuple of commands
             commands = commands[0]
+        if isinstance(commands, str):
+            commands = [commands]
         for command in commands:
             if not isinstance(command, str):
                 raise TypeError(f'\'command\' must be a str, not {type(command)}')

src/xtl/config/settings.py

@@ -126,6 +126,13 @@ class DependenciesSettings(Settings):
     """
 
     # Model attributes
+    atsas: DependencySettings = \
+        Option(
+            default=DependencySettings(
+                provides={'datcmp'}
+            )
+        )
+
     autoproc: DependencySettings = \
         Option(
             default=DependencySettings(

src/xtl/jobs/config.py

@@ -6,8 +6,8 @@
 from pydantic import computed_field, PrivateAttr, model_validator
 
 from xtl import Logger
-from xtl.automate.shells import Shell, DefaultShell
-from xtl.automate.sites import ComputeSiteType, LocalSite, BiotixHPC
+from xtl.automate.shells import Shell, DefaultShell, WslShell
+from xtl.automate.sites import ComputeSiteType, LocalSite
 from xtl.common.options import Option, Options
 from xtl.common.os import FilePermissions
 from xtl.common.serializers import PermissionOctal
@@ -23,24 +23,43 @@ class BatchConfig(Options):
     """
     Configuration for execution of batch files.
     """
-    filename: str = Option(default='batch_job', desc='Batch file name (without '
-                                                     'extension)')
-    directory: Path | str = Option(default_factory=lambda: Path(tempfile.gettempdir()),
-                                   desc='Directory for dumping batch file and logs',
-                                   cast_as=Path)
-    permissions: FilePermissions | str | int = Option(default=FilePermissions(0o700),
-                                                      desc='Permissions for the batch '
-                                                           'file in octal format '
-                                                           '(e.g., 700)',
-                                                      cast_as=FilePermissions,
-                                                      formatter=PermissionOctal)
-    compute_site: ComputeSiteType = Option(default=LocalSite(), desc='Compute site')
-    default_shell: Optional[Shell] = Option(default=None,
-                                            desc='Default shell to use for batch '
-                                                 'execution.')
-    compatible_shells: set[Shell] = Option(default_factory=set,
-                                           desc='List of compatible shell types for '
-                                                'this batch job')
+    filename: str = \
+        Option(
+            default='batch_job',
+            desc='Batch file name (without extension)'
+        )
+    directory: Path = \
+        Option(
+            default_factory=lambda: Path(tempfile.mkdtemp()),
+            desc='Directory for dumping batch file and logs'
+        )
+    permissions: FilePermissions | str | int = \
+        Option(
+            default=FilePermissions(0o700),
+            desc='Permissions for the batch file in octal format (e.g., 700)',
+            cast_as=FilePermissions,
+            formatter=PermissionOctal
+        )
+    compute_site: ComputeSiteType = \
+        Option(
+            default=LocalSite(),
+            desc='Compute site'
+        )
+    default_shell: Optional[Shell] = \
+        Option(
+            default=None,
+            desc='Default shell to use for batch execution.'
+        )
+    compatible_shells: set[Shell] = \
+        Option(
+            default_factory=set,
+            desc='List of compatible shell types for this batch job'
+        )
+    dependencies: set[str] = \
+        Option(
+            default_factory=set,
+            desc='List of dependencies required for this batch job'
+        )
 
     _shell: Shell | None = PrivateAttr(None)
 
@@ -104,6 +123,7 @@ def stderr(self) -> Path:
         return self.directory / f
 
     @computed_field
+    @property
     def shell(self) -> Shell:
         """
         Returns the shell that will be used to execute the batch file.
@@ -115,8 +135,8 @@ def shell(self, value: Shell):
         """
         Sets the shell to be used for executing the batch file.
         """
-        if not isinstance(value, Shell):
-            raise ValueError(f'Shell must be an instance of {Shell.__name__}')
+        if not isinstance(value, (Shell, WslShell)):
+            raise ValueError(f'shell must be an instance of {Shell.__name__}')
         self._shell = value
 
     def get_batch(self) -> 'BatchFile':
@@ -126,7 +146,8 @@ def get_batch(self) -> 'BatchFile':
         from xtl.automate.batchfile import BatchFile
         batch = BatchFile(filename=self.directory/self.filename,
                           compute_site=self.compute_site,
-                          shell=self.shell)
+                          shell=self.shell,
+                          dependencies=self.dependencies)
         batch.permissions = self.permissions
         return batch
 

src/xtl/jobs/jobs.py

@@ -340,9 +340,10 @@ def get_logger(cls, job_id: str, config: LoggerConfig = None) -> logging.Logger:
 
         return logger
 
-    async def _execute_batch(self, commands: list[str], args: list[str] = None,
-                             filename: str = None, stdout_log: Path = None,
-                             stderr_log: Path = None) -> JobResults:
+    async def _execute_batch(self, commands: str | Iterable[str],
+                             args: Iterable[str] = None, filename: str = None,
+                             stdout_log: Path = None, stderr_log: Path = None) -> \
+            JobResults:
         """
         Execute a batch file with the specified commands.
 

src/xtl/saxs/__init__.py

@@ -0,0 +1,61 @@
+import abc
+from pathlib import Path
+
+from xtl.common.options import Option
+from xtl.jobs.jobs import Job
+from xtl.jobs.config import JobConfig
+from xtl.saxs.jobs.atsas_utils import ATSASOptions, DatcmpOptions
+
+
+class ATSASJobConfig(JobConfig, abc.ABC):
+    """
+    Base configuration for ATSAS jobs.
+    """
+    options: ATSASOptions
+
+    @abc.abstractmethod
+    def get_command(self, as_list: bool = False) -> list[str] | str:
+        ...
+
+
+class DatcmpJobConfig(ATSASJobConfig):
+    files: list[Path] = \
+        Option(
+            ...,
+            desc='List of data files to compare',
+            min_length=2,
+            path_exists=True)
+    options: DatcmpOptions = \
+        Option(
+            default_factory=DatcmpOptions,
+            desc='Options for `datcmp`'
+        )
+
+    def get_command(self, as_list: bool = False) -> list[str] | str:
+        parts = list(map(str, [self.options.executable,
+                               *self.files,
+                               *self.options.get_args()]))
+        if as_list:
+            return parts
+        return ' '.join(parts)
+
+
+class DatcmpJob(Job[DatcmpJobConfig]):
+    """
+    Job to compare SAXS datasets using ATSAS datcmp.
+    """
+
+    async def _execute(self):
+        if self.config is None:
+            self.logger.error('Job is not configured.')
+            raise ValueError('Job is not configured.')
+
+        # Execute the command
+        results = await self._execute_batch(self.config.get_command())
+
+        if not results.success or 'stdout' not in results.data:
+            self.logger.error('An error occurred during execution of the batch file %s',
+                              results.error)
+            return results
+
+        return results.data['stdout']

src/xtl/saxs/jobs/__init__.py

@@ -0,0 +1,61 @@
+from pydantic import PrivateAttr
+
+from xtl.common.options import Option, Options
+
+
+class ATSASOptions(Options):
+    """
+    Base class for ATSAS job options.
+    """
+    _executable: str = PrivateAttr()
+
+    @property
+    def executable(self) -> str:
+        return self._executable
+
+    def get_args(self) -> list[str]:
+        """
+        Returns the command-line arguments for the ATSAS executable.
+        """
+        args = []
+        for key, value in self.to_dict(by_alias=True).items():
+            if value is not None:
+                args.append(f'--{key}={value}')
+        return args
+
+
+class DatcmpOptions(ATSASOptions):
+    """
+    Configuration for an ATSAS datcmp job.
+    """
+    _executable: str = PrivateAttr(default='datcmp')
+
+    mode: str | None = \
+        Option(
+            default='PAIRWISE',
+            desc='Comparison mode',
+            choices={'PAIRWISE', 'INDEPENDENT', None}
+        )
+    test: str | None = \
+        Option(
+            default='CORMAP',
+            desc='Test name',
+            choices={'CORMAP', 'CHI-SQUARE', 'ANDERSON-DARLING', None}
+        )
+    adjust: str | None = \
+        Option(
+            default='FWER',
+            desc='Adjustment for multiple testing',
+            choices={'FWER', 'FDR', None}
+        )
+    alpha: float | None = \
+        Option(
+            default=0.01,
+            desc='Significance level for clique search'
+        )
+    format: str | None = \
+        Option(
+            default='FULL',
+            desc='Output format',
+            choices={'FULL', 'CSV', None}
+        )