Use datalad containers-run for single-container jobs

babs/bootstrap.py

@@ -227,10 +227,43 @@ def babs_bootstrap(
         )
         # into `analysis/containers` folder
 
-        # Create initial container for sanity check
-        container = Container(container_ds, container_name, container_config)
+        # Discover container image path from the container dataset:
+        containers_path = op.join(self.analysis_path, 'containers')
+        result = dlapi.containers_list(dataset=containers_path, result_renderer='disabled')
+        container_info = [
+            r for r in result if r['action'] == 'containers' and r['name'] == container_name
+        ]
+        if not container_info:
+            available = [r['name'] for r in result if r['action'] == 'containers']
+            raise ValueError(
+                f"Container '{container_name}' not found in container dataset. "
+                f'Available: {available}'
+            )
+        image_path_in_ds = op.relpath(container_info[0]['path'], containers_path)
+        container_image_path = op.join('containers', image_path_in_ds)
+
+        # Build call_fmt from user's singularity_args:
+        with open(container_config) as f:
+            user_config = yaml.safe_load(f)
+        singularity_args = user_config.get('singularity_args', [])
+        singularity_args_str = ' '.join(singularity_args) if singularity_args else ''
+        call_fmt = f'singularity run -B $PWD --pwd $PWD {singularity_args_str} {{img}} {{cmd}}'
+
+        # Register container at analysis level so datalad containers-run works:
+        print(f'\nRegistering container at analysis level: {container_image_path}')
+        dlapi.containers_add(
+            dataset=self.analysis_path,
+            name=container_name,
+            image=container_image_path,
+            call_fmt=call_fmt,
+        )
 
-        # sanity check of container ds:
+        container = Container(
+            container_ds,
+            container_name,
+            container_config,
+            container_image_path=container_image_path,
+        )
         container.sanity_check(self.analysis_path)
 
         # ==============================================================
@@ -254,9 +287,18 @@ def babs_bootstrap(
             container = containers[0]
         else:
             self._bootstrap_single_app_scripts(
-                container_ds, container_name, container_config, system
+                container_ds,
+                container_name,
+                container_config,
+                system,
+                container_image_path=container_image_path,
+            )
+            container = Container(
+                container_ds,
+                container_name,
+                container_config,
+                container_image_path=container_image_path,
             )
-            container = Container(container_ds, container_name, container_config)
 
         # Copy in any other files needed:
         self._init_import_files(container.config.get('imported_files', []))
@@ -394,21 +436,30 @@ def babs_bootstrap(
         print('`babs init` was successful!')
 
     def _bootstrap_single_app_scripts(
-        self, container_ds, container_name, container_config, system
+        self,
+        container_ds,
+        container_name,
+        container_config,
+        system,
+        container_image_path=None,
     ):
         """Bootstrap scripts for single BIDS app configuration."""
-        container = Container(container_ds, container_name, container_config)
+        container = Container(
+            container_ds,
+            container_name,
+            container_config,
+            container_image_path=container_image_path,
+        )
 
         # Generate `<containerName>_zip.sh`: ----------------------------------
-        # which is a bash script of singularity run + zip
-        # in folder: `analysis/code`
-        print('\nGenerating a bash script for running container and zipping the outputs...')
-        print('This bash script will be named as `' + container_name + '_zip.sh`')
+        # Zip-only script (container execution is now handled by containers-run
+        # in participant_job.sh)
+        print('\nGenerating zip script: ' + container_name + '_zip.sh')
         bash_path = op.join(self.analysis_path, 'code', container_name + '_zip.sh')
-        container.generate_bash_run_bidsapp(bash_path, self.input_datasets, self.processing_level)
+        container.generate_bash_zip_outputs(bash_path, self.processing_level)
         self.datalad_save(
             path='code/' + container_name + '_zip.sh',
-            message='Generate script of running container',
+            message='Generate zip script',
         )
 
         # make another folder within `code` for test jobs:

babs/container.py

@@ -6,15 +6,19 @@
 import yaml
 from jinja2 import Environment, PackageLoader, StrictUndefined
 
-from babs.generate_bidsapp_runscript import generate_bidsapp_runscript
+from babs.generate_bidsapp_runscript import (
+    bids_app_args_from_config,
+    generate_bidsapp_runscript,
+    get_output_zipping_cmds,
+)
 from babs.generate_submit_script import generate_submit_script, generate_test_submit_script
 from babs.utils import app_output_settings_from_config
 
 
 class Container:
     """This class is for the BIDS App Container"""
 
-    def __init__(self, container_ds, container_name, config_yaml_file):
+    def __init__(self, container_ds, container_name, config_yaml_file, container_image_path=None):
         """
         This is to initialize Container class.
 
@@ -67,9 +71,7 @@ def __init__(self, container_ds, container_name, config_yaml_file):
         with open(self.config_yaml_file) as f:
             self.config = yaml.safe_load(f)
 
-        self.container_path_relToAnalysis = op.join(
-            'containers', '.datalad', 'environments', self.container_name, 'image'
-        )
+        self.container_path_relToAnalysis = container_image_path
 
     def sanity_check(self, analysis_path):
         """
@@ -101,6 +103,28 @@ def sanity_check(self, analysis_path):
             + "'."
         )
 
+    def generate_bash_zip_outputs(self, bash_path, processing_level):
+        """Generate a bash script that only zips BIDS App outputs."""
+        dict_zip_foldernames, _ = app_output_settings_from_config(self.config)
+        cmd_zip = get_output_zipping_cmds(dict_zip_foldernames, processing_level)
+
+        env = Environment(
+            loader=PackageLoader('babs', 'templates'),
+            trim_blocks=True,
+            lstrip_blocks=True,
+            autoescape=False,
+            undefined=StrictUndefined,
+        )
+        template = env.get_template('zip_outputs.sh.jinja2')
+        script_content = template.render(
+            processing_level=processing_level,
+            cmd_zip=cmd_zip,
+        )
+
+        with open(bash_path, 'w') as f:
+            f.write(script_content)
+        os.chmod(bash_path, 0o700)
+
     def generate_bash_run_bidsapp(self, bash_path, input_ds, processing_level):
         """
         This is to generate a bash script that runs the BIDS App singularity image.
@@ -165,16 +189,34 @@ def generate_bash_participant_job(
             Shown in the script error message when PROJECT_ROOT is unset.
         """
 
+        input_datasets = input_ds.as_records()
+        _, bids_app_output_dir = app_output_settings_from_config(self.config)
+
+        raw_bids_app_args = self.config.get('bids_app_args', None)
+        if raw_bids_app_args:
+            bids_app_args, subject_selection_flag, _, _, bids_app_input_dir = (
+                bids_app_args_from_config(raw_bids_app_args, input_datasets)
+            )
+        else:
+            bids_app_args = []
+            subject_selection_flag = '--participant-label'
+            bids_app_input_dir = input_datasets[0]['unzipped_path_containing_subject_dirs']
+
         script_content = generate_submit_script(
             queue_system=system.type,
             cluster_resources_config=self.config['cluster_resources'],
             script_preamble=self.config['script_preamble'],
             job_scratch_directory=self.config['job_compute_space'],
-            input_datasets=input_ds.as_records(),
+            input_datasets=input_datasets,
             processing_level=processing_level,
             container_name=self.container_name,
             zip_foldernames=self.config['zip_foldernames'],
             project_root=project_root,
+            container_image_path=self.container_path_relToAnalysis,
+            bids_app_args=bids_app_args,
+            bids_app_input_dir=bids_app_input_dir,
+            bids_app_output_dir=bids_app_output_dir,
+            subject_selection_flag=subject_selection_flag,
         )
 
         with open(bash_path, 'w') as f:

babs/generate_submit_script.py

@@ -30,6 +30,11 @@ def generate_submit_script(
     container_images=None,
     datalad_run_message=None,
     project_root=None,
+    container_image_path=None,
+    bids_app_args=None,
+    bids_app_input_dir=None,
+    bids_app_output_dir=None,
+    subject_selection_flag=None,
 ):
     """
     Generate a bash script that runs the BIDS App singularity image.
@@ -122,6 +127,11 @@ def generate_submit_script(
         container_images=container_images,
         datalad_run_message=datalad_run_message,
         project_root=project_root,
+        container_image_path=container_image_path,
+        bids_app_args=bids_app_args or [],
+        bids_app_input_dir=bids_app_input_dir or '',
+        bids_app_output_dir=bids_app_output_dir or '',
+        subject_selection_flag=subject_selection_flag or '',
     )
 
 

babs/templates/participant_job.sh.jinja2

@@ -70,6 +70,7 @@ if ! git sparse-checkout init --cone; then
 fi
 
 git sparse-checkout set \
+  .datalad \
   code \
   containers \
 {% for input_dataset in input_datasets %}
@@ -107,9 +108,8 @@ fi
 {{ zip_locator_text }}
 
 # Link to shared container image so each job does not re-clone the same image.
-# If shared path is not available (e.g. Slurm Docker workers), retrieve image in this clone.
-CONTAINER_SHARED="${PROJECT_ROOT}/analysis/containers/.datalad/environments/{{ container_name }}/image"
-CONTAINER_JOB="containers/.datalad/environments/{{ container_name }}/image"
+CONTAINER_SHARED="${PROJECT_ROOT}/analysis/{{ container_image_path }}"
+CONTAINER_JOB="{{ container_image_path }}"
 
 if [ ! -L "${CONTAINER_SHARED}" ]; then
   echo "ERROR: shared container image not found at ${CONTAINER_SHARED}" >&2
@@ -125,35 +125,52 @@ if [ ! -L "${CONTAINER_JOB}" ]; then
   exit 1
 fi
 
-# datalad run:
-datalad run \
-	-i "{{ run_script_relpath if run_script_relpath else 'code/' + container_name + '_zip.sh' }}" \
+# Step 1: Run BIDS app via containers-run
+datalad containers-run \
+	-n {{ container_name }} \
+	--explicit \
 {% for input_dataset in input_datasets %}
 {% if not input_dataset['is_zipped'] %}
-	-i "{{ input_dataset['unzipped_path_containing_subject_dirs'] }}/{% raw %}${subid}{% endraw %}{% if processing_level == 'session' %}/{% raw %}${sesid}{% endraw %}{% endif %}" \
-	-i "{{ input_dataset['path_in_babs'] }}/dataset_description.json" \
+	--input "{{ input_dataset['unzipped_path_containing_subject_dirs'] }}/{% raw %}${subid}{% endraw %}{% if processing_level == 'session' %}/{% raw %}${sesid}{% endraw %}{% endif %}" \
+	--input "{{ input_dataset['path_in_babs'] }}/dataset_description.json" \
 {% else %}
-	-i "${%raw%}{{%endraw%}{{ input_dataset['name'].upper() }}_ZIP{%raw%}}{%endraw%}" \
+	--input "${%raw%}{{%endraw%}{{ input_dataset['name'].upper() }}_ZIP{%raw%}}{%endraw%}" \
 {% endif %}
 {% endfor %}
-{% if container_images %}
-{% for image_path in container_images %}
-	-i "{{ image_path }}" \
-{% endfor %}
-{% else %}
-	-i "containers/.datalad/environments/{{container_name}}/image" \
-{% endif %}
 {% if datalad_expand_inputs %}
 	--expand inputs \
 {% endif %}
+	--output "{{ bids_app_output_dir }}" \
+	-m "{{ container_name }} {% raw %}${subid}{% endraw %}{% if processing_level == 'session' %} {% raw %}${sesid}{% endraw %}{% endif %}" \
+	-- \
+	{{ bids_app_input_dir }} \
+	{{ bids_app_output_dir }} \
+	participant \
+{% for bids_app_arg in bids_app_args %}
+{% if bids_app_arg %}
+	{{ bids_app_arg }} \
+{% endif %}
+{% endfor %}
+	{{ subject_selection_flag }} "{% raw %}${subid}{% endraw %}"
+
+# Unlock outputs so zip gets real content, not annex symlinks
+datalad unlock {{ bids_app_output_dir }}
+
+# Step 2: Zip outputs
+datalad run \
 	--explicit \
 {% if zip_foldernames is not none %}
 {% for key, value in zip_foldernames.items() %}
 	-o "{% raw %}${subid}{% endraw %}{% if processing_level == 'session' %}_{% raw %}${sesid}{% endraw %}{% endif %}_{{ key }}-{{ value }}.zip" \
 {% endfor %}
 {% endif %}
-	-m "{{ (datalad_run_message if datalad_run_message is defined and datalad_run_message else container_name) }} {% raw %}${subid}{% endraw %}{% if processing_level == 'session' %} {% raw %}${sesid}{% endraw %}{% endif %}" \
-    "bash ./{{ run_script_relpath if run_script_relpath else 'code/' + container_name + '_zip.sh' }} {% raw %}${subid}{% endraw %} {% if processing_level == 'session' %} {% raw %}${sesid}{% endraw %}{% endif %}{% for input_dataset in input_datasets %}{% if input_dataset['is_zipped'] %} ${%raw%}{{%endraw%}{{ input_dataset['name'].upper() }}_ZIP{%raw%}}{%endraw%}{%endif%}{%endfor%}"
+	-m "Zip {% raw %}${subid}{% endraw %}{% if processing_level == 'session' %} {% raw %}${sesid}{% endraw %}{% endif %}" \
+	-- \
+	bash ./code/{{ container_name }}_zip.sh "{% raw %}${subid}{% endraw %}"{% if processing_level == 'session' %} "{% raw %}${sesid}{% endraw %}"{% endif %}
+
+# Step 3: Remove raw outputs (datalad run --explicit doesn't track deletions)
+git rm -rf --sparse {{ bids_app_output_dir }}
+git commit -m "Remove raw outputs for {% raw %}${subid}{% endraw %}{% if processing_level == 'session' %} {% raw %}${sesid}{% endraw %}{% endif %} (zipped)"
 
 # Finish up:
 # push result file content to output RIA storage:

babs/templates/zip_outputs.sh.jinja2

@@ -0,0 +1,9 @@
+#!/bin/bash
+set -e -u -x
+
+subid="$1"
+{% if processing_level == 'session' %}
+sesid="$2"
+{% endif %}
+
+{{ cmd_zip }}