Warning

This document is for an in-development version of Galaxy. You can alternatively view this page in the latest release if it exists or view the top of the latest release's documentation.

Source code for galaxy.tool_shed.metadata.metadata_generator

import json
import logging
import os
import tempfile

from sqlalchemy import and_

from galaxy import util
from galaxy.tool_shed.repository_type import (
    REPOSITORY_DEPENDENCY_DEFINITION_FILENAME,
    TOOL_DEPENDENCY_DEFINITION_FILENAME,
)
from galaxy.tool_shed.tools.tool_validator import ToolValidator
from galaxy.tool_shed.util import (
    shed_util_common as suc,
    tool_dependency_util,
    tool_util,
)
from galaxy.tool_shed.util.basic_util import remove_dir, strip_path
from galaxy.tool_shed.util.hg_util import get_config_from_disk
from galaxy.tool_shed.util.metadata_util import get_updated_changeset_revisions_from_tool_shed
from galaxy.tool_shed.util.repository_util import get_repository_for_dependency_relationship
from galaxy.tool_util.loader_directory import looks_like_a_tool
from galaxy.tool_util.parser.interface import TestCollectionDef
from galaxy.tools.data_manager.manager import DataManager
from galaxy.tools.repositories import ValidationContext
from galaxy.util.tool_shed.common_util import (
    generate_clone_url_for_installed_repository,
    generate_clone_url_for_repository_in_tool_shed,
    remove_protocol_and_user_from_clone_url,
    remove_protocol_from_tool_shed_url,
)
from galaxy.util.tool_shed.xml_util import parse_xml
from galaxy.web import url_for

log = logging.getLogger(__name__)


[docs]class MetadataGenerator(object):
[docs] def __init__(self, app, repository=None, changeset_revision=None, repository_clone_url=None, shed_config_dict=None, relative_install_dir=None, repository_files_dir=None, resetting_all_metadata_on_repository=False, updating_installed_repository=False, persist=False, metadata_dict=None, user=None): self.app = app self.user = user self.repository = repository if self.app.name == 'galaxy': if changeset_revision is None and self.repository is not None: self.changeset_revision = self.repository.changeset_revision else: self.changeset_revision = changeset_revision if repository_clone_url is None and self.repository is not None: self.repository_clone_url = generate_clone_url_for_installed_repository(self.app, self.repository) else: self.repository_clone_url = repository_clone_url if shed_config_dict is None: if self.repository is not None: self.shed_config_dict = self.repository.get_shed_config_dict(self.app) else: self.shed_config_dict = {} else: self.shed_config_dict = shed_config_dict if relative_install_dir is None and self.repository is not None: tool_path, relative_install_dir = self.repository.get_tool_relative_path(self.app) if repository_files_dir is None and self.repository is not None: repository_files_dir = self.repository.repo_files_directory(self.app) if metadata_dict is None: # Shed related tool panel configs are only relevant to Galaxy. self.metadata_dict = {'shed_config_filename': self.shed_config_dict.get('config_filename', None)} else: self.metadata_dict = metadata_dict else: # We're in the Tool Shed. if changeset_revision is None and self.repository is not None: self.changeset_revision = self.repository.tip() else: self.changeset_revision = changeset_revision if repository_clone_url is None and self.repository is not None: self.repository_clone_url = \ generate_clone_url_for_repository_in_tool_shed(self.user, self.repository) else: self.repository_clone_url = repository_clone_url if shed_config_dict is None: self.shed_config_dict = {} else: self.shed_config_dict = shed_config_dict if relative_install_dir is None and self.repository is not None: relative_install_dir = self.repository.repo_path(self.app) if repository_files_dir is None and self.repository is not None: repository_files_dir = self.repository.repo_path(self.app) if metadata_dict is None: self.metadata_dict = {} else: self.metadata_dict = metadata_dict self.relative_install_dir = relative_install_dir self.repository_files_dir = repository_files_dir self.resetting_all_metadata_on_repository = resetting_all_metadata_on_repository self.updating_installed_repository = updating_installed_repository self.persist = persist self.invalid_file_tups = [] self.sa_session = app.model.context.current self.NOT_TOOL_CONFIGS = [suc.DATATYPES_CONFIG_FILENAME, REPOSITORY_DEPENDENCY_DEFINITION_FILENAME, TOOL_DEPENDENCY_DEFINITION_FILENAME, suc.REPOSITORY_DATA_MANAGER_CONFIG_FILENAME]
[docs] def generate_data_manager_metadata(self, repo_dir, data_manager_config_filename, metadata_dict, shed_config_dict=None): """ Update the received metadata_dict with information from the parsed data_manager_config_filename. """ if data_manager_config_filename is None: return metadata_dict repo_path = self.repository.repo_path(self.app) try: # Galaxy Side. repo_files_directory = self.repository.repo_files_directory(self.app) repo_dir = repo_files_directory except AttributeError: # Tool Shed side. repo_files_directory = repo_path relative_data_manager_dir = util.relpath(os.path.split(data_manager_config_filename)[0], repo_dir) rel_data_manager_config_filename = os.path.join(relative_data_manager_dir, os.path.split(data_manager_config_filename)[1]) data_managers = {} invalid_data_managers = [] data_manager_metadata = {'config_filename': rel_data_manager_config_filename, 'data_managers': data_managers, 'invalid_data_managers': invalid_data_managers, 'error_messages': []} metadata_dict['data_manager'] = data_manager_metadata tree, error_message = parse_xml(data_manager_config_filename) if tree is None: # We are not able to load any data managers. data_manager_metadata['error_messages'].append(error_message) return metadata_dict tool_path = None if shed_config_dict: tool_path = shed_config_dict.get('tool_path', None) tools = {} for tool in metadata_dict.get('tools', []): tool_conf_name = tool['tool_config'] if tool_path: tool_conf_name = os.path.join(tool_path, tool_conf_name) tools[tool_conf_name] = tool root = tree.getroot() data_manager_tool_path = root.get('tool_path', None) if data_manager_tool_path: relative_data_manager_dir = os.path.join(relative_data_manager_dir, data_manager_tool_path) for i, data_manager_elem in enumerate(root.findall('data_manager')): tool_file = data_manager_elem.get('tool_file', None) data_manager_id = data_manager_elem.get('id', None) if data_manager_id is None: log.error('Data Manager entry is missing id attribute in "%s".' % (data_manager_config_filename)) invalid_data_managers.append({'index': i, 'error_message': 'Data Manager entry is missing id attribute'}) continue # FIXME: default behavior is to fall back to tool.name. data_manager_name = data_manager_elem.get('name', data_manager_id) version = data_manager_elem.get('version', DataManager.DEFAULT_VERSION) guid = self.generate_guid_for_object(DataManager.GUID_TYPE, data_manager_id, version) data_tables = [] if tool_file is None: log.error('Data Manager entry is missing tool_file attribute in "%s".' % (data_manager_config_filename)) invalid_data_managers.append({'index': i, 'error_message': 'Data Manager entry is missing tool_file attribute'}) continue else: bad_data_table = False for data_table_elem in data_manager_elem.findall('data_table'): data_table_name = data_table_elem.get('name', None) if data_table_name is None: log.error('Data Manager data_table entry is missing name attribute in "%s".' % (data_manager_config_filename)) invalid_data_managers.append({'index': i, 'error_message': 'Data Manager entry is missing name attribute'}) bad_data_table = True break else: data_tables.append(data_table_name) if bad_data_table: continue data_manager_metadata_tool_file = os.path.normpath(os.path.join(relative_data_manager_dir, tool_file)) tool_metadata_tool_file = os.path.join(repo_files_directory, data_manager_metadata_tool_file) tool = tools.get(tool_metadata_tool_file, None) if tool is None: log.error("Unable to determine tools metadata for '%s'." % (data_manager_metadata_tool_file)) invalid_data_managers.append({'index': i, 'error_message': 'Unable to determine tools metadata'}) continue data_managers[data_manager_id] = {'id': data_manager_id, 'name': data_manager_name, 'guid': guid, 'version': version, 'tool_config_file': data_manager_metadata_tool_file, 'data_tables': data_tables, 'tool_guid': tool['guid']} log.debug('Loaded Data Manager tool_files: %s' % (tool_file)) return metadata_dict
[docs] def generate_datatypes_metadata(self, tv, repository_files_dir, datatypes_config, metadata_dict): """Update the received metadata_dict with information from the parsed datatypes_config.""" tree, error_message = parse_xml(datatypes_config) if tree is None: return metadata_dict root = tree.getroot() repository_datatype_code_files = [] datatype_files = root.find('datatype_files') if datatype_files is not None: for elem in datatype_files.findall('datatype_file'): name = elem.get('name', None) repository_datatype_code_files.append(name) metadata_dict['datatype_files'] = repository_datatype_code_files datatypes = [] registration = root.find('registration') if registration is not None: for elem in registration.findall('datatype'): converters = [] display_app_containers = [] datatypes_dict = {} # Handle defined datatype attributes. display_in_upload = elem.get('display_in_upload', None) if display_in_upload: datatypes_dict['display_in_upload'] = display_in_upload dtype = elem.get('type', None) if dtype: datatypes_dict['dtype'] = dtype extension = elem.get('extension', None) if extension: datatypes_dict['extension'] = extension max_optional_metadata_filesize = elem.get('max_optional_metadata_filesize', None) if max_optional_metadata_filesize: datatypes_dict['max_optional_metadata_filesize'] = max_optional_metadata_filesize mimetype = elem.get('mimetype', None) if mimetype: datatypes_dict['mimetype'] = mimetype subclass = elem.get('subclass', None) if subclass: datatypes_dict['subclass'] = subclass # Handle defined datatype converters and display applications. for sub_elem in elem: if sub_elem.tag == 'converter': # <converter file="bed_to_gff_converter.xml" target_datatype="gff"/> tool_config = sub_elem.attrib['file'] target_datatype = sub_elem.attrib['target_datatype'] # Parse the tool_config to get the guid. tool_config_path = get_config_from_disk(tool_config, repository_files_dir) full_path = os.path.abspath(tool_config_path) tool, valid, error_message = \ tv.load_tool_from_config(self.app.security.encode_id(self.repository.id), full_path) if tool is None: guid = None else: guid = suc.generate_tool_guid(self.repository_clone_url, tool) converter_dict = dict(tool_config=tool_config, guid=guid, target_datatype=target_datatype) converters.append(converter_dict) elif sub_elem.tag == 'display': # <display file="ucsc/bigwig.xml" /> # Should we store more than this? display_file = sub_elem.attrib['file'] display_app_dict = dict(display_file=display_file) display_app_containers.append(display_app_dict) if converters: datatypes_dict['converters'] = converters if display_app_containers: datatypes_dict['display_app_containers'] = display_app_containers if datatypes_dict: datatypes.append(datatypes_dict) if datatypes: metadata_dict['datatypes'] = datatypes return metadata_dict
[docs] def generate_environment_dependency_metadata(self, elem, valid_tool_dependencies_dict): """ The value of env_var_name must match the value of the "set_environment" type in the tool config's <requirements> tag set, or the tool dependency will be considered an orphan. """ # The value of the received elem looks something like this: # <set_environment version="1.0"> # <environment_variable name="JAVA_JAR_PATH" action="set_to">$INSTALL_DIR</environment_variable> # </set_environment> for env_elem in elem: # <environment_variable name="JAVA_JAR_PATH" action="set_to">$INSTALL_DIR</environment_variable> env_name = env_elem.get('name', None) if env_name: requirements_dict = dict(name=env_name, type='set_environment') if 'set_environment' in valid_tool_dependencies_dict: valid_tool_dependencies_dict['set_environment'].append(requirements_dict) else: valid_tool_dependencies_dict['set_environment'] = [requirements_dict] return valid_tool_dependencies_dict
[docs] def generate_guid_for_object(self, guid_type, obj_id, version): tmp_url = remove_protocol_and_user_from_clone_url(self.repository_clone_url) return '%s/%s/%s/%s' % (tmp_url, guid_type, obj_id, version)
[docs] def generate_metadata_for_changeset_revision(self): """ Generate metadata for a repository using its files on disk. To generate metadata for changeset revisions older than the repository tip, the repository will have been cloned to a temporary location and updated to a specified changeset revision to access that changeset revision's disk files, so the value of self.repository_files_dir will not always be self.repository.repo_path( self.app ) (it could be an absolute path to a temporary directory containing a clone). If it is an absolute path, the value of self.relative_install_dir must contain repository.repo_path( self.app ). The value of self.persist will be True when the installed repository contains a valid tool_data_table_conf.xml.sample file, in which case the entries should ultimately be persisted to the file referred to by self.app.config.shed_tool_data_table_config. """ if self.shed_config_dict is None: self.shed_config_dict = {} if self.updating_installed_repository: # Keep the original tool shed repository metadata if setting metadata on a repository # installed into a local Galaxy instance for which we have pulled updates. original_repository_metadata = self.repository.metadata else: original_repository_metadata = None readme_file_names = _get_readme_file_names(str(self.repository.name)) if self.app.name == 'galaxy': # Shed related tool panel configs are only relevant to Galaxy. metadata_dict = {'shed_config_filename': self.shed_config_dict.get('config_filename')} else: metadata_dict = {} readme_files = [] invalid_tool_configs = [] if self.resetting_all_metadata_on_repository: if not self.relative_install_dir: raise Exception("The value of self.repository.repo_path must be set when resetting all metadata on a repository.") # Keep track of the location where the repository is temporarily cloned so that we can # strip the path when setting metadata. The value of self.repository_files_dir is the # full path to the temporary directory to which self.repository was cloned. work_dir = self.repository_files_dir files_dir = self.repository_files_dir # Since we're working from a temporary directory, we can safely copy sample files included # in the repository to the repository root. else: # Use a temporary working directory to copy all sample files. work_dir = tempfile.mkdtemp(prefix="tmp-toolshed-gmfcr") # All other files are on disk in the repository's repo_path, which is the value of # self.relative_install_dir. files_dir = self.relative_install_dir if self.shed_config_dict.get('tool_path'): files_dir = os.path.join(self.shed_config_dict['tool_path'], files_dir) # Create ValidationContext to load and validate tools, data tables and datatypes with ValidationContext.from_app(app=self.app, work_dir=work_dir) as validation_context: tv = ToolValidator(validation_context) # Handle proprietary datatypes, if any. datatypes_config = get_config_from_disk(suc.DATATYPES_CONFIG_FILENAME, files_dir) if datatypes_config: metadata_dict = self.generate_datatypes_metadata(tv, files_dir, datatypes_config, metadata_dict) # Get the relative path to all sample files included in the repository for storage in # the repository's metadata. sample_file_metadata_paths, sample_file_copy_paths = \ self.get_sample_files_from_disk(repository_files_dir=files_dir, tool_path=self.shed_config_dict.get('tool_path'), relative_install_dir=self.relative_install_dir) if sample_file_metadata_paths: metadata_dict['sample_files'] = sample_file_metadata_paths # Copy all sample files included in the repository to a single directory location so we # can load tools that depend on them. data_table_conf_xml_sample_files = [] for sample_file in sample_file_copy_paths: tool_util.copy_sample_file(self.app, sample_file, dest_path=work_dir) # If the list of sample files includes a tool_data_table_conf.xml.sample file, load # its table elements into memory. relative_path, filename = os.path.split(sample_file) if filename == 'tool_data_table_conf.xml.sample': data_table_conf_xml_sample_files.append(sample_file) for data_table_conf_xml_sample_file in data_table_conf_xml_sample_files: # We create a new ToolDataTableManager to avoid adding entries to the app-wide # tool data tables. This is only used for checking that the data table is valid. new_table_elems, error_message = \ validation_context.tool_data_tables.add_new_entries_from_config_file(config_filename=data_table_conf_xml_sample_file, tool_data_path=work_dir, shed_tool_data_table_config=work_dir, persist=False) if error_message: self.invalid_file_tups.append((filename, error_message)) for root, dirs, files in os.walk(files_dir): if root.find('.hg') < 0 and root.find('hgrc') < 0: if '.hg' in dirs: dirs.remove('.hg') for name in files: # See if we have a repository dependencies defined. if name == REPOSITORY_DEPENDENCY_DEFINITION_FILENAME: path_to_repository_dependencies_config = os.path.join(root, name) metadata_dict, error_message = \ self.generate_repository_dependency_metadata(path_to_repository_dependencies_config, metadata_dict) if error_message: self.invalid_file_tups.append((name, error_message)) # See if we have one or more READ_ME files. elif name.lower() in readme_file_names: relative_path_to_readme = self.get_relative_path_to_repository_file(root, name, self.relative_install_dir, work_dir, self.shed_config_dict) readme_files.append(relative_path_to_readme) # See if we have a tool config. elif looks_like_a_tool(os.path.join(root, name), invalid_names=self.NOT_TOOL_CONFIGS): full_path = str(os.path.abspath(os.path.join(root, name))) # why the str, seems very odd element_tree, error_message = parse_xml(full_path) if element_tree is None: is_tool = False else: element_tree_root = element_tree.getroot() is_tool = element_tree_root.tag == 'tool' if is_tool: tool, valid, error_message = \ tv.load_tool_from_config(self.app.security.encode_id(self.repository.id), full_path) if tool is None: if not valid: invalid_tool_configs.append(name) self.invalid_file_tups.append((name, error_message)) else: invalid_files_and_errors_tups = \ tv.check_tool_input_params(files_dir, name, tool, sample_file_copy_paths) can_set_metadata = True for tup in invalid_files_and_errors_tups: if name in tup: can_set_metadata = False invalid_tool_configs.append(name) break if can_set_metadata: relative_path_to_tool_config = \ self.get_relative_path_to_repository_file(root, name, self.relative_install_dir, work_dir, self.shed_config_dict) metadata_dict = self.generate_tool_metadata(relative_path_to_tool_config, tool, metadata_dict) else: for tup in invalid_files_and_errors_tups: self.invalid_file_tups.append(tup) # Find all exported workflows. elif name.endswith('.ga'): relative_path = os.path.join(root, name) if os.path.getsize(os.path.abspath(relative_path)) > 0: fp = open(relative_path, 'rb') workflow_text = fp.read() fp.close() if workflow_text: valid_exported_galaxy_workflow = True try: exported_workflow_dict = json.loads(workflow_text) except Exception: log.exception("Skipping file %s since it does not seem to be a valid exported Galaxy workflow", str(relative_path)) valid_exported_galaxy_workflow = False if valid_exported_galaxy_workflow and \ 'a_galaxy_workflow' in exported_workflow_dict and \ exported_workflow_dict['a_galaxy_workflow'] == 'true': metadata_dict = self.generate_workflow_metadata(relative_path, exported_workflow_dict, metadata_dict) # Handle any data manager entries data_manager_config = get_config_from_disk(suc.REPOSITORY_DATA_MANAGER_CONFIG_FILENAME, files_dir) metadata_dict = self.generate_data_manager_metadata(files_dir, data_manager_config, metadata_dict, shed_config_dict=self.shed_config_dict) if readme_files: metadata_dict['readme_files'] = readme_files # This step must be done after metadata for tools has been defined. tool_dependencies_config = get_config_from_disk(TOOL_DEPENDENCY_DEFINITION_FILENAME, files_dir) if tool_dependencies_config: metadata_dict, error_message = \ self.generate_tool_dependency_metadata(tool_dependencies_config, metadata_dict, original_repository_metadata=original_repository_metadata) if error_message: self.invalid_file_tups.append((TOOL_DEPENDENCY_DEFINITION_FILENAME, error_message)) if invalid_tool_configs: metadata_dict['invalid_tools'] = invalid_tool_configs self.metadata_dict = metadata_dict remove_dir(work_dir)
[docs] def generate_package_dependency_metadata(self, elem, valid_tool_dependencies_dict, invalid_tool_dependencies_dict): """ Generate the metadata for a tool dependencies package defined for a repository. The value of package_name must match the value of the "package" type in the tool config's <requirements> tag set. """ # TODO: make this function a class. repository_dependency_is_valid = True repository_dependency_tup = [] requirements_dict = {} error_message = '' package_name = elem.get('name', None) package_version = elem.get('version', None) if package_name and package_version: requirements_dict['name'] = package_name requirements_dict['version'] = package_version requirements_dict['type'] = 'package' for sub_elem in elem: if sub_elem.tag == 'readme': requirements_dict['readme'] = sub_elem.text elif sub_elem.tag == 'repository': # We have a complex repository dependency. If the returned value of repository_dependency_is_valid # is True, the tool dependency definition will be set as invalid. This is currently the only case # where a tool dependency definition is considered invalid. repository_dependency_tup, repository_dependency_is_valid, error_message = \ self.handle_repository_elem(repository_elem=sub_elem, only_if_compiling_contained_td=False) elif sub_elem.tag == 'install': package_install_version = sub_elem.get('version', '1.0') if package_install_version == '1.0': # Complex repository dependencies can be defined within the last <actions> tag set contained in an # <actions_group> tag set. Comments, <repository> tag sets and <readme> tag sets will be skipped # in tool_dependency_util.parse_package_elem(). actions_elem_tuples = tool_dependency_util.parse_package_elem(sub_elem, platform_info_dict=None, include_after_install_actions=False) if actions_elem_tuples: # We now have a list of a single tuple that looks something like: # [(True, <Element 'actions' at 0x104017850>)] actions_elem_tuple = actions_elem_tuples[0] in_actions_group, actions_elem = actions_elem_tuple if in_actions_group: # Since we're inside an <actions_group> tag set, inspect the actions_elem to see if a complex # repository dependency is defined. By definition, complex repository dependency definitions # contained within the last <actions> tag set within an <actions_group> tag set will have the # value of "only_if_compiling_contained_td" set to True in for action_elem in actions_elem: if action_elem.tag == 'package': # <package name="libgtextutils" version="0.6"> # <repository name="package_libgtextutils_0_6" owner="test" prior_installation_required="True" /> # </package> ae_package_name = action_elem.get('name', None) ae_package_version = action_elem.get('version', None) if ae_package_name and ae_package_version: for sub_action_elem in action_elem: if sub_action_elem.tag == 'repository': # We have a complex repository dependency. repository_dependency_tup, repository_dependency_is_valid, error_message = \ self.handle_repository_elem(repository_elem=sub_action_elem, only_if_compiling_contained_td=True) elif action_elem.tag == 'action': # <action type="set_environment_for_install"> # <repository changeset_revision="b107b91b3574" name="package_readline_6_2" owner="devteam" prior_installation_required="True" toolshed="http://localhost:9009"> # <package name="readline" version="6.2" /> # </repository> # </action> for sub_action_elem in action_elem: if sub_action_elem.tag == 'repository': # We have a complex repository dependency. repository_dependency_tup, repository_dependency_is_valid, error_message = \ self.handle_repository_elem(repository_elem=sub_action_elem, only_if_compiling_contained_td=True) if requirements_dict: dependency_key = '%s/%s' % (package_name, package_version) if repository_dependency_is_valid: valid_tool_dependencies_dict[dependency_key] = requirements_dict else: # Append the error message to the requirements_dict. requirements_dict['error'] = error_message invalid_tool_dependencies_dict[dependency_key] = requirements_dict return valid_tool_dependencies_dict, \ invalid_tool_dependencies_dict, \ repository_dependency_tup, \ repository_dependency_is_valid, \ error_message
[docs] def generate_repository_dependency_metadata(self, repository_dependencies_config, metadata_dict): """ Generate a repository dependencies dictionary based on valid information defined in the received repository_dependencies_config. This method is called from the tool shed as well as from Galaxy. """ # Make sure we're looking at a valid repository_dependencies.xml file. tree, error_message = parse_xml(repository_dependencies_config) if tree is None: xml_is_valid = False else: root = tree.getroot() xml_is_valid = root.tag == 'repositories' if xml_is_valid: invalid_repository_dependencies_dict = dict(description=root.get('description')) invalid_repository_dependency_tups = [] valid_repository_dependencies_dict = dict(description=root.get('description')) valid_repository_dependency_tups = [] for repository_elem in root.findall('repository'): repository_dependency_tup, repository_dependency_is_valid, err_msg = \ self.handle_repository_elem(repository_elem, only_if_compiling_contained_td=False) if repository_dependency_is_valid: valid_repository_dependency_tups.append(repository_dependency_tup) else: # Append the error_message to the repository dependencies tuple. toolshed, name, owner, changeset_revision, prior_installation_required, only_if_compiling_contained_td = \ repository_dependency_tup repository_dependency_tup = (toolshed, name, owner, changeset_revision, prior_installation_required, only_if_compiling_contained_td, err_msg) invalid_repository_dependency_tups.append(repository_dependency_tup) error_message += err_msg if invalid_repository_dependency_tups: invalid_repository_dependencies_dict['repository_dependencies'] = invalid_repository_dependency_tups metadata_dict['invalid_repository_dependencies'] = invalid_repository_dependencies_dict if valid_repository_dependency_tups: valid_repository_dependencies_dict['repository_dependencies'] = valid_repository_dependency_tups metadata_dict['repository_dependencies'] = valid_repository_dependencies_dict return metadata_dict, error_message
[docs] def generate_tool_metadata(self, tool_config, tool, metadata_dict): """Update the received metadata_dict with changes that have been applied to the received tool.""" # Generate the guid. guid = suc.generate_tool_guid(self.repository_clone_url, tool) # Handle tool.requirements. tool_requirements = [] for tool_requirement in tool.requirements: name = str(tool_requirement.name) tool_type = str(tool_requirement.type) version = str(tool_requirement.version) if tool_requirement.version else None requirement_dict = dict(name=name, type=tool_type, version=version) tool_requirements.append(requirement_dict) # Handle tool.tests. tool_tests = [] if tool.tests: for ttb in tool.tests: required_files = [] for required_file in ttb.required_files: value, extra = required_file required_files.append((value)) inputs = [] for param_name, values in ttb.inputs.items(): # Handle improperly defined or strange test parameters and values. if param_name is not None: if values in [None, False]: # An example is the third test in http://testtoolshed.g2.bx.psu.edu/view/devteam/samtools_rmdup # which is defined as: # <test> # <param name="input1" value="1.bam" ftype="bam" /> # <param name="bam_paired_end_type_selector" value="PE" /> # <param name="force_se" /> # <output name="output1" file="1.bam" ftype="bam" sort="True" /> # </test> inputs.append((param_name, values)) else: if isinstance(values, TestCollectionDef): # Nested required files are being populated correctly, # not sure we need the value here to be anything else? collection_type = values.collection_type metadata_display_value = "%s collection" % collection_type inputs.append((param_name, metadata_display_value)) else: try: if len(values) == 1: inputs.append((param_name, values[0])) continue except TypeError: log.exception('Expected a list of values for tool "%s" parameter "%s", got %s: %s', tool.id, param_name, type(values), values) inputs.append((param_name, values)) outputs = [] for output in ttb.outputs: name, file_name, extra = output outputs.append((name, strip_path(file_name) if file_name else None)) if file_name not in required_files and file_name is not None: required_files.append(file_name) test_dict = dict(name=str(ttb.name), required_files=required_files, inputs=inputs, outputs=outputs) tool_tests.append(test_dict) # Determine if the tool should be loaded into the tool panel. Examples of valid tools that # should not be displayed in the tool panel are datatypes converters and DataManager tools # (which are of type 'manage_data'). datatypes = metadata_dict.get('datatypes', None) add_to_tool_panel_attribute = self.set_add_to_tool_panel_attribute_for_tool(tool=tool, guid=guid, datatypes=datatypes) tool_dict = dict(id=tool.id, guid=guid, name=tool.name, version=tool.version, profile=tool.profile, description=tool.description, version_string_cmd=tool.version_string_cmd, tool_config=tool_config, tool_type=tool.tool_type, requirements=tool_requirements, tests=tool_tests, add_to_tool_panel=add_to_tool_panel_attribute) if 'tools' in metadata_dict: metadata_dict['tools'].append(tool_dict) else: metadata_dict['tools'] = [tool_dict] return metadata_dict
[docs] def generate_tool_dependency_metadata(self, tool_dependencies_config, metadata_dict, original_repository_metadata=None): """ If the combination of name, version and type of each element is defined in the <requirement> tag for at least one tool in self.repository, then update the received metadata_dict with information from the parsed tool_dependencies_config. """ error_message = '' if original_repository_metadata: # Keep a copy of the original tool dependencies dictionary and the list of tool # dictionaries in the metadata. original_valid_tool_dependencies_dict = original_repository_metadata.get('tool_dependencies', None) else: original_valid_tool_dependencies_dict = None tree, error_message = parse_xml(tool_dependencies_config) if tree is None: return metadata_dict, error_message root = tree.getroot() class RecurserValueStore(object): pass rvs = RecurserValueStore() rvs.valid_tool_dependencies_dict = {} rvs.invalid_tool_dependencies_dict = {} valid_repository_dependency_tups = [] invalid_repository_dependency_tups = [] description = root.get('description') def _check_elem_for_dep(elems): error_messages = [] for elem in elems: if elem.tag == 'package': rvs.valid_tool_dependencies_dict, rvs.invalid_tool_dependencies_dict, \ repository_dependency_tup, repository_dependency_is_valid, \ message = self.generate_package_dependency_metadata(elem, rvs.valid_tool_dependencies_dict, rvs.invalid_tool_dependencies_dict) if repository_dependency_is_valid: if repository_dependency_tup and repository_dependency_tup not in valid_repository_dependency_tups: # We have a valid complex repository dependency. valid_repository_dependency_tups.append(repository_dependency_tup) else: if repository_dependency_tup and repository_dependency_tup not in invalid_repository_dependency_tups: # We have an invalid complex repository dependency, so mark the tool dependency as invalid. # Append the error message to the invalid repository dependency tuple. toolshed, name, owner, changeset_revision, \ prior_installation_required, \ only_if_compiling_contained_td = \ repository_dependency_tup repository_dependency_tup = \ (toolshed, name, owner, changeset_revision, prior_installation_required, only_if_compiling_contained_td, message) invalid_repository_dependency_tups.append(repository_dependency_tup) error_messages.append('%s %s' % (error_message, message)) elif elem.tag == 'set_environment': rvs.valid_tool_dependencies_dict = \ self.generate_environment_dependency_metadata(elem, rvs.valid_tool_dependencies_dict) error_messages += _check_elem_for_dep(elem) return error_messages error_message = "\n".join([error_message] + _check_elem_for_dep(root)) if rvs.valid_tool_dependencies_dict: if original_valid_tool_dependencies_dict: # We're generating metadata on an update pulled to a tool shed repository installed # into a Galaxy instance, so handle changes to tool dependencies appropriately. irm = self.app.installed_repository_manager updated_tool_dependency_names, deleted_tool_dependency_names = \ irm.handle_existing_tool_dependencies_that_changed_in_update(self.repository, original_valid_tool_dependencies_dict, rvs.valid_tool_dependencies_dict) metadata_dict['tool_dependencies'] = rvs.valid_tool_dependencies_dict if rvs.invalid_tool_dependencies_dict: metadata_dict['invalid_tool_dependencies'] = rvs.invalid_tool_dependencies_dict if valid_repository_dependency_tups: metadata_dict = \ self.update_repository_dependencies_metadata(metadata=metadata_dict, repository_dependency_tups=valid_repository_dependency_tups, is_valid=True, description=description) if invalid_repository_dependency_tups: metadata_dict = \ self.update_repository_dependencies_metadata(metadata=metadata_dict, repository_dependency_tups=invalid_repository_dependency_tups, is_valid=False, description=description) return metadata_dict, error_message
[docs] def generate_workflow_metadata(self, relative_path, exported_workflow_dict, metadata_dict): """ Update the received metadata_dict with changes that have been applied to the received exported_workflow_dict. """ if 'workflows' in metadata_dict: metadata_dict['workflows'].append((relative_path, exported_workflow_dict)) else: metadata_dict['workflows'] = [(relative_path, exported_workflow_dict)] return metadata_dict
[docs] def get_invalid_file_tups(self): return self.invalid_file_tups
[docs] def get_metadata_dict(self): return self.metadata_dict
[docs] def get_relative_path_to_repository_file(self, root, name, relative_install_dir, work_dir, shed_config_dict): if self.resetting_all_metadata_on_repository: full_path_to_file = os.path.join(root, name) stripped_path_to_file = full_path_to_file.replace(work_dir, '') if stripped_path_to_file.startswith('/'): stripped_path_to_file = stripped_path_to_file[1:] relative_path_to_file = os.path.join(relative_install_dir, stripped_path_to_file) else: relative_path_to_file = os.path.join(root, name) if relative_install_dir and \ shed_config_dict.get('tool_path') and \ relative_path_to_file.startswith(os.path.join(shed_config_dict.get('tool_path'), relative_install_dir)): relative_path_to_file = relative_path_to_file[len(shed_config_dict.get('tool_path')) + 1:] return relative_path_to_file
[docs] def get_sample_files_from_disk(self, repository_files_dir, tool_path=None, relative_install_dir=None): work_dir = '' if self.resetting_all_metadata_on_repository: # Keep track of the location where the repository is temporarily cloned so that we can strip # it when setting metadata. work_dir = repository_files_dir sample_file_metadata_paths = [] sample_file_copy_paths = [] for root, dirs, files in os.walk(repository_files_dir): if root.find('.hg') < 0: for name in files: if name.endswith('.sample'): if self.resetting_all_metadata_on_repository: full_path_to_sample_file = os.path.join(root, name) stripped_path_to_sample_file = full_path_to_sample_file.replace(work_dir, '') if stripped_path_to_sample_file.startswith('/'): stripped_path_to_sample_file = stripped_path_to_sample_file[1:] relative_path_to_sample_file = os.path.join(relative_install_dir, stripped_path_to_sample_file) if os.path.exists(relative_path_to_sample_file): sample_file_copy_paths.append(relative_path_to_sample_file) else: sample_file_copy_paths.append(full_path_to_sample_file) else: relative_path_to_sample_file = os.path.join(root, name) sample_file_copy_paths.append(relative_path_to_sample_file) if tool_path and relative_install_dir: if relative_path_to_sample_file.startswith(os.path.join(tool_path, relative_install_dir)): relative_path_to_sample_file = relative_path_to_sample_file[len(tool_path) + 1:] sample_file_metadata_paths.append(relative_path_to_sample_file) return sample_file_metadata_paths, sample_file_copy_paths
[docs] def handle_repository_elem(self, repository_elem, only_if_compiling_contained_td=False): """ Process the received repository_elem which is a <repository> tag either from a repository_dependencies.xml file or a tool_dependencies.xml file. If the former, we're generating repository dependencies metadata for a repository in the Tool Shed. If the latter, we're generating package dependency metadata within Galaxy or the Tool Shed. """ is_valid = True error_message = '' toolshed = repository_elem.get('toolshed', None) name = repository_elem.get('name', None) owner = repository_elem.get('owner', None) changeset_revision = repository_elem.get('changeset_revision', None) prior_installation_required = str(repository_elem.get('prior_installation_required', False)) repository_dependency_tup = [toolshed, name, owner, changeset_revision, prior_installation_required, str(only_if_compiling_contained_td)] if self.app.name == 'galaxy': if self.updating_installed_repository: pass else: # We're installing a repository into Galaxy, so make sure its contained repository # dependency definition is valid. if toolshed is None or name is None or owner is None or changeset_revision is None: # Several packages exist in the Tool Shed that contain invalid repository # definitions, but will still install. We will report these errors to the # installing user. Previously, we would: # Raise an exception here instead of returning an error_message to keep the # installation from proceeding. Reaching here implies a bug in the Tool Shed # framework. error_message = 'Installation encountered an invalid repository dependency definition:\n' error_message += util.xml_to_string(repository_elem, pretty=True) log.error(error_message) return repository_dependency_tup, False, error_message if not toolshed: # Default to the current tool shed. toolshed = str(url_for('/', qualified=True)).rstrip('/') repository_dependency_tup[0] = toolshed toolshed = remove_protocol_from_tool_shed_url(toolshed) if self.app.name == 'galaxy': # We're in Galaxy. We reach here when we're generating the metadata for a tool # dependencies package defined for a repository or when we're generating metadata # for an installed repository. See if we can locate the installed repository via # the changeset_revision defined in the repository_elem (it may be outdated). If # we're successful in locating an installed repository with the attributes defined # in the repository_elem, we know it is valid. repository = get_repository_for_dependency_relationship(self.app, toolshed, name, owner, changeset_revision) if repository: return repository_dependency_tup, is_valid, error_message else: # Send a request to the tool shed to retrieve appropriate additional changeset # revisions with which the repository # may have been installed. text = get_updated_changeset_revisions_from_tool_shed(self.app, toolshed, name, owner, changeset_revision) if text: updated_changeset_revisions = util.listify(text) for updated_changeset_revision in updated_changeset_revisions: repository = get_repository_for_dependency_relationship(self.app, toolshed, name, owner, updated_changeset_revision) if repository: return repository_dependency_tup, is_valid, error_message if self.updating_installed_repository: # The repository dependency was included in an update to the installed # repository, so it will not yet be installed. Return the tuple for later # installation. return repository_dependency_tup, is_valid, error_message if self.updating_installed_repository: # The repository dependency was included in an update to the installed repository, # so it will not yet be installed. Return the tuple for later installation. return repository_dependency_tup, is_valid, error_message # Don't generate an error message for missing repository dependencies that are required # only if compiling the dependent repository's tool dependency. if not only_if_compiling_contained_td: # We'll currently default to setting the repository dependency definition as invalid # if an installed repository cannot be found. This may not be ideal because the tool # shed may have simply been inaccessible when metadata was being generated for the # installed tool shed repository. error_message = "Ignoring invalid repository dependency definition for tool shed %s, name %s, owner %s, " % \ (toolshed, name, owner) error_message += "changeset revision %s." % changeset_revision log.debug(error_message) is_valid = False return repository_dependency_tup, is_valid, error_message else: # We're in the tool shed. if suc.tool_shed_is_this_tool_shed(toolshed): try: user = self.sa_session.query(self.app.model.User) \ .filter(self.app.model.User.table.c.username == owner) \ .one() except Exception: error_message = "Ignoring repository dependency definition for tool shed %s, name %s, owner %s, " % \ (toolshed, name, owner) error_message += "changeset revision %s because the owner is invalid." % changeset_revision log.debug(error_message) is_valid = False return repository_dependency_tup, is_valid, error_message try: repository = self.sa_session.query(self.app.model.Repository) \ .filter(and_(self.app.model.Repository.table.c.name == name, self.app.model.Repository.table.c.user_id == user.id)) \ .one() except Exception: error_message = "Ignoring repository dependency definition for tool shed %s, name %s, owner %s, " % \ (toolshed, name, owner) error_message += "changeset revision %s because the name is invalid. " % changeset_revision log.debug(error_message) is_valid = False return repository_dependency_tup, is_valid, error_message repo = repository.hg_repo # The received changeset_revision may be None since defining it in the dependency definition is optional. # If this is the case, the default will be to set its value to the repository dependency tip revision. # This probably occurs only when handling circular dependency definitions. tip_ctx = repo[repo.changelog.tip()] # Make sure the repo.changlog includes at least 1 revision. if changeset_revision is None and tip_ctx.rev() >= 0: changeset_revision = str(tip_ctx) repository_dependency_tup = [toolshed, name, owner, changeset_revision, prior_installation_required, str(only_if_compiling_contained_td)] return repository_dependency_tup, is_valid, error_message else: # Find the specified changeset revision in the repository's changelog to see if it's valid. found = False for changeset in repo.changelog: changeset_hash = str(repo[changeset]) if changeset_hash == changeset_revision: found = True break if not found: error_message = "Ignoring repository dependency definition for tool shed %s, name %s, owner %s, " % \ (toolshed, name, owner) error_message += "changeset revision %s because the changeset revision is invalid. " % changeset_revision log.debug(error_message) is_valid = False return repository_dependency_tup, is_valid, error_message else: # Repository dependencies are currently supported within a single tool shed. error_message = "Repository dependencies are currently supported only within the same tool shed. Ignoring " error_message += "repository dependency definition for tool shed %s, name %s, owner %s, changeset revision %s. " % \ (toolshed, name, owner, changeset_revision) log.debug(error_message) is_valid = False return repository_dependency_tup, is_valid, error_message return repository_dependency_tup, is_valid, error_message
[docs] def set_add_to_tool_panel_attribute_for_tool(self, tool, guid, datatypes): """ Determine if a tool should be loaded into the Galaxy tool panel. Examples of valid tools that should not be displayed in the tool panel are datatypes converters and DataManager tools. """ if hasattr(tool, 'tool_type'): if tool.tool_type in ['manage_data']: # We have a DataManager tool. return False if datatypes: for datatype_dict in datatypes: converters = datatype_dict.get('converters', None) # [{"converters": # [{"target_datatype": "gff", # "tool_config": "bed_to_gff_converter.xml", # "guid": "localhost:9009/repos/test/bed_to_gff_converter/CONVERTER_bed_to_gff_0/2.0.0"}], # "display_in_upload": "true", # "dtype": "galaxy.datatypes.interval:Bed", # "extension": "bed"}] if converters: for converter_dict in converters: converter_guid = converter_dict.get('guid', None) if converter_guid: if converter_guid == guid: # We have a datatypes converter. return False return True
[docs] def set_changeset_revision(self, changeset_revision): self.changeset_revision = changeset_revision
[docs] def set_relative_install_dir(self, relative_install_dir): self.relative_install_dir = relative_install_dir
[docs] def set_repository(self, repository, relative_install_dir=None, changeset_revision=None): self.repository = repository # Shed related tool panel configs are only relevant to Galaxy. if self.app.name == 'galaxy': if relative_install_dir is None and self.repository is not None: tool_path, relative_install_dir = self.repository.get_tool_relative_path(self.app) if changeset_revision is None and self.repository is not None: self.set_changeset_revision(self.repository.changeset_revision) else: self.set_changeset_revision(changeset_revision) self.shed_config_dict = repository.get_shed_config_dict(self.app) self.metadata_dict = {'shed_config_filename': self.shed_config_dict.get('config_filename', None)} else: if relative_install_dir is None and self.repository is not None: relative_install_dir = repository.repo_path(self.app) if changeset_revision is None and self.repository is not None: self.set_changeset_revision(self.repository.tip()) else: self.set_changeset_revision(changeset_revision) self.shed_config_dict = {} self.metadata_dict = {} self.set_relative_install_dir(relative_install_dir) self.set_repository_files_dir() self.resetting_all_metadata_on_repository = False self.updating_installed_repository = False self.persist = False self.invalid_file_tups = []
[docs] def set_repository_clone_url(self, repository_clone_url): self.repository_clone_url = repository_clone_url
[docs] def set_repository_files_dir(self, repository_files_dir=None): self.repository_files_dir = repository_files_dir
[docs] def update_repository_dependencies_metadata(self, metadata, repository_dependency_tups, is_valid, description): if is_valid: repository_dependencies_dict = metadata.get('repository_dependencies', None) else: repository_dependencies_dict = metadata.get('invalid_repository_dependencies', None) for repository_dependency_tup in repository_dependency_tups: if repository_dependencies_dict: repository_dependencies = repository_dependencies_dict.get('repository_dependencies', []) for repository_dependency_tup in repository_dependency_tups: if repository_dependency_tup not in repository_dependencies: repository_dependencies.append(repository_dependency_tup) repository_dependencies_dict['repository_dependencies'] = repository_dependencies else: repository_dependencies_dict = dict(description=description, repository_dependencies=repository_dependency_tups) if repository_dependencies_dict: if is_valid: metadata['repository_dependencies'] = repository_dependencies_dict else: metadata['invalid_repository_dependencies'] = repository_dependencies_dict return metadata
def _get_readme_file_names(repository_name): """Return a list of file names that will be categorized as README files for the received repository_name.""" readme_files = ['readme', 'read_me', 'install'] valid_filenames = ['%s.txt' % f for f in readme_files] valid_filenames.extend(['%s.rst' % f for f in readme_files]) valid_filenames.extend(readme_files) valid_filenames.append('%s.txt' % repository_name) valid_filenames.append('%s.rst' % repository_name) return valid_filenames