Merge pull request #14617 from jmchilton/remove_propertiary_datatypes

Remove legacy tool shed datatype handling
This commit is contained in:
Björn Grüning
2022-09-15 18:21:43 +02:00
committed by GitHub
7 changed files with 238 additions and 706 deletions
+1 -14
View File
@@ -327,19 +327,9 @@ class ConfiguresGalaxyMixin:
if exc.errno != errno.ENOENT or self.config.is_set("shed_tool_data_table_config"):
raise
def _configure_datatypes_registry(
self, installed_repository_manager=None, use_display_applications=True, use_converters=True
):
def _configure_datatypes_registry(self, use_display_applications=True, use_converters=True):
# Create an empty datatypes registry.
self.datatypes_registry = Registry(self.config)
if installed_repository_manager and self.config.load_tool_shed_datatypes:
# Load proprietary datatypes defined in datatypes_conf.xml files in all installed tool shed repositories. We
# load proprietary datatypes before datatypes in the distribution because Galaxy's default sniffers include some
# generic sniffers (eg text,xml) which catch anything, so it's impossible for proprietary sniffers to be used.
# However, if there is a conflict (2 datatypes with the same extension) between a proprietary datatype and a datatype
# in the Galaxy distribution, the datatype in the Galaxy distribution will take precedence. If there is a conflict
# between 2 proprietary datatypes, the datatype from the repository that was installed earliest will take precedence.
installed_repository_manager.load_proprietary_datatypes()
# Load the data types in the Galaxy distribution, which are defined in self.config.datatypes_config.
datatypes_configs = self.config.datatypes_config
for datatypes_config in listify(datatypes_configs):
@@ -583,7 +573,6 @@ class GalaxyManagerApplication(MinimalManagerApp, MinimalGalaxyApplication):
)
self.dynamic_tool_manager = self._register_singleton(DynamicToolManager)
self._configure_datatypes_registry(
self.installed_repository_manager,
use_converters=use_converters,
use_display_applications=use_display_applications,
)
@@ -659,8 +648,6 @@ class UniverseApplication(StructuredApp, GalaxyManagerApplication):
self.update_repository_manager = self._register_singleton(
UpdateRepositoryManager, UpdateRepositoryManager(self)
)
# Load proprietary datatype converters and display applications.
self.installed_repository_manager.load_proprietary_converters_and_display_applications()
# Load datatype display applications defined in local datatypes_conf.xml
self.datatypes_registry.load_display_applications(self)
# Load datatype converters defined in local datatypes_conf.xml
@@ -466,15 +466,6 @@ galaxy:
# <config_dir>.
#tool_sheds_config_file: tool_sheds_conf.xml
# This option controls whether legacy datatypes are loaded from
# installed tool shed repositories. We're are in the process of
# disabling Tool Shed datatypes. This option with a default of true
# will be added in 22.01, we will disable the datatypes on the big
# public servers during that release. This option will be switched to
# False by default in 22.05 and this broken functionality will be
# removed all together during some future release.
#load_tool_shed_datatypes: true
# Monitor the tools and tool directories listed in any tool config
# file specified in tool_config_file option. If changes are found,
# tools are automatically reloaded. Watchdog (
@@ -443,17 +443,6 @@ mapping:
File containing the Galaxy Tool Sheds that should be made available to
install from in the admin interface (.sample used if default does not exist).
load_tool_shed_datatypes:
type: bool
default: true
required: false
desc: |
This option controls whether legacy datatypes are loaded from installed tool shed repositories.
We're are in the process of disabling Tool Shed datatypes. This option with a default of true
will be added in 22.01, we will disable the datatypes on the big public servers during that
release. This option will be switched to False by default in 22.05 and this broken functionality
will be removed all together during some future release.
watch_tools:
type: str
default: 'false'
+237 -457
View File
@@ -68,8 +68,6 @@ class Registry:
# Converters defined in local datatypes_conf.xml
self.converters = []
self.converter_tools = set()
# Converters defined in datatypes_conf.xml included in installed tool shed repositories.
self.proprietary_converters = []
self.converter_deps = {}
self.available_tracks = []
self.set_external_metadata_tool = None
@@ -77,9 +75,6 @@ class Registry:
self.upload_file_formats = []
# Datatype elements defined in local datatypes_conf.xml that contain display applications.
self.display_app_containers = []
# Datatype elements in datatypes_conf.xml included in installed
# tool shed repositories that contain display applications.
self.proprietary_display_app_containers = []
# Map a display application id to a display application
self.display_applications: Dict[str, DisplayApplication] = {}
# The following 2 attributes are used in the to_xml_file()
@@ -91,8 +86,6 @@ class Registry:
# The 'default' display_path defined in local datatypes_conf.xml
self.display_applications_path = None
self.inherit_display_application_by_class = []
# Keep a list of imported proprietary datatype class modules.
self.imported_modules = []
self.datatype_elems = []
self.datatype_info_dicts = []
self.sniffer_elems = []
@@ -109,7 +102,6 @@ class Registry:
self,
root_dir=None,
config=None,
deactivate=False,
override=True,
use_converters=True,
use_display_applications=True,
@@ -117,10 +109,7 @@ class Registry:
):
"""
Parse a datatypes XML file located at root_dir/config (if processing the Galaxy distributed config) or contained within
an installed Tool Shed repository. If deactivate is True, an installed Tool Shed repository that includes custom datatypes
is being deactivated or uninstalled, so appropriate loaded datatypes will be removed from the registry. The value of
override will be False when a Tool Shed repository is being installed. Since installation is occurring after the datatypes
registry has been initialized at server startup, its contents cannot be overridden by newly introduced conflicting data types.
an installed Tool Shed repository.
"""
def __import_module(full_path: str, datatype_module: str):
@@ -134,24 +123,13 @@ class Registry:
return module
if root_dir and config:
# If handling_proprietary_datatypes is determined as True below, we'll have an elem that looks something like this:
# <datatype display_in_upload="true"
# extension="blastxml"
# mimetype="application/xml"
# proprietary_datatype_module="blast"
# proprietary_path="[cloned repository path]"
# type="galaxy.datatypes.blast:BlastXml" />
compressed_sniffers = {}
handling_proprietary_datatypes = False
if isinstance(config, (str, Path)):
# Parse datatypes_conf.xml
tree = galaxy.util.parse_xml(config)
root = tree.getroot()
# Load datatypes and converters from config
if deactivate:
self.log.debug(f"Deactivating datatypes from {config}")
else:
self.log.debug(f"Loading datatypes from {config}")
self.log.debug(f"Loading datatypes from {config}")
else:
root = config
registration = root.find("registration")
@@ -182,15 +160,7 @@ class Registry:
os.path.join(os.path.dirname(__file__), "display_applications", "configs")
)
self.display_applications_path = self.display_path_attr
# Proprietary datatype's <registration> tag may have special attributes, proprietary_converter_path and proprietary_display_path.
proprietary_converter_path = registration.get("proprietary_converter_path", None)
proprietary_display_path = registration.get("proprietary_display_path", None)
if (
proprietary_converter_path is not None
or proprietary_display_path is not None
and not handling_proprietary_datatypes
):
handling_proprietary_datatypes = True
for elem in registration.findall("datatype"):
# Keep a status of the process steps to enable stopping the process of handling the datatype if necessary.
ok = True
@@ -217,263 +187,192 @@ class Registry:
if edam_data and not make_subclass:
self.log.warning("Cannot specify edam_data without setting subclass to True, skipping datatype.")
continue
# Proprietary datatypes included in installed tool shed repositories will include two special attributes
# (proprietary_path and proprietary_datatype_module) if they depend on proprietary datatypes classes.
# The value of proprietary_path is the path to the cloned location of the tool shed repository's contained
# datatypes_conf.xml file.
proprietary_path = elem.get("proprietary_path", None)
proprietary_datatype_module = elem.get("proprietary_datatype_module", None)
if (
proprietary_path is not None
or proprietary_datatype_module is not None
and not handling_proprietary_datatypes
):
handling_proprietary_datatypes = True
if deactivate:
# We are deactivating or uninstalling an installed tool shed repository, so eliminate the datatype
# elem from the in-memory list of datatype elems.
for in_memory_elem in self.datatype_elems:
in_memory_extension = in_memory_elem.get("extension", None)
if in_memory_extension == extension:
in_memory_dtype = elem.get("type", None)
in_memory_type_extension = elem.get("type_extension", None)
in_memory_mimetype = elem.get("mimetype", None)
in_memory_display_in_upload = galaxy.util.string_as_bool(
elem.get("display_in_upload", False)
)
in_memory_make_subclass = galaxy.util.string_as_bool(elem.get("subclass", False))
if (
in_memory_dtype == dtype
and in_memory_type_extension == type_extension
and in_memory_mimetype == mimetype
and in_memory_display_in_upload == display_in_upload
and in_memory_make_subclass == make_subclass
):
self.datatype_elems.remove(in_memory_elem)
if extension is not None and extension in self.datatypes_by_extension:
# We are deactivating or uninstalling an installed tool shed repository, so eliminate the datatype
# from the registry. TODO: Handle deactivating datatype converters, etc before removing from
# self.datatypes_by_extension.
del self.datatypes_by_extension[extension]
if extension in self.upload_file_formats:
self.upload_file_formats.remove(extension)
self.log.debug(f"Removed datatype with extension '{extension}' from the registry.")
else:
# We are loading new datatype, so we'll make sure it is correctly defined before proceeding.
can_process_datatype = False
if extension is not None:
if dtype is not None or type_extension is not None:
if override or extension not in self.datatypes_by_extension:
can_process_datatype = True
if can_process_datatype:
if dtype is not None:
try:
fields = dtype.split(":")
datatype_module = fields[0]
datatype_class_name = fields[1]
except Exception:
self.log.exception("Error parsing datatype definition for dtype %s", str(dtype))
ok = False
if ok:
datatype_class = None
if proprietary_path and proprietary_datatype_module and datatype_class_name:
# TODO: previously comments suggested this needs to be locked because it modifies
# the sys.path, probably true but the previous lock wasn't doing that.
try:
imported_module = __import_module(proprietary_path, proprietary_datatype_module)
if imported_module not in self.imported_modules:
self.imported_modules.append(imported_module)
if hasattr(imported_module, datatype_class_name):
datatype_class = getattr(imported_module, datatype_class_name)
except Exception as e:
full_path = os.path.join(proprietary_path, proprietary_datatype_module)
self.log.debug(
"Exception importing proprietary code file %s: %s",
full_path,
e,
)
# Either the above exception was thrown because the proprietary_datatype_module is not derived from a class
# in the repository, or we are loading Galaxy's datatypes. In either case we'll look in the registry.
if datatype_class is None:
try:
# The datatype class name must be contained in one of the datatype modules in the Galaxy distribution.
fields = datatype_module.split(".")[1:]
module = __import__(datatype_module)
for mod in fields:
module = getattr(module, mod)
datatype_class = getattr(module, datatype_class_name)
self.log.debug(
f"Retrieved datatype module {str(datatype_module)}:{datatype_class_name} from the datatype registry for extension {extension}."
)
except Exception:
self.log.exception("Error importing datatype module %s", str(datatype_module))
ok = False
elif type_extension is not None:
try:
datatype_class = self.datatypes_by_extension[type_extension].__class__
self.log.debug(
f"Retrieved datatype module {str(datatype_class.__name__)} from type_extension {type_extension} for extension {extension}."
)
except Exception:
self.log.exception(
"Error determining datatype_class for type_extension %s", str(type_extension)
)
ok = False
# We are loading new datatype, so we'll make sure it is correctly defined before proceeding.
can_process_datatype = False
if extension is not None:
if dtype is not None or type_extension is not None:
if override or extension not in self.datatypes_by_extension:
can_process_datatype = True
if can_process_datatype:
if dtype is not None:
try:
fields = dtype.split(":")
datatype_module = fields[0]
datatype_class_name = fields[1]
except Exception:
self.log.exception("Error parsing datatype definition for dtype %s", str(dtype))
ok = False
if ok:
if not deactivate:
# A new tool shed repository that contains custom datatypes is being installed, and since installation is
# occurring after the datatypes registry has been initialized at server startup, its contents cannot be
# overridden by new introduced conflicting data types unless the value of override is True.
if extension in self.datatypes_by_extension:
# Because of the way that the value of can_process_datatype was set above, we know that the value of
# override is True.
datatype_class = None
if datatype_class is None:
try:
# The datatype class name must be contained in one of the datatype modules in the Galaxy distribution.
fields = datatype_module.split(".")[1:]
module = __import__(datatype_module)
for mod in fields:
module = getattr(module, mod)
datatype_class = getattr(module, datatype_class_name)
self.log.debug(
"Overriding conflicting datatype with extension '%s', using datatype from %s."
% (str(extension), str(config))
f"Retrieved datatype module {str(datatype_module)}:{datatype_class_name} from the datatype registry for extension {extension}."
)
if make_subclass:
datatype_class = type(datatype_class_name, (datatype_class,), {})
if edam_format:
datatype_class.edam_format = edam_format
if edam_data:
datatype_class.edam_data = edam_data
datatype_class.is_subclass = make_subclass
description = elem.get("description", None)
description_url = elem.get("description_url", None)
datatype_instance = datatype_class()
self.datatypes_by_extension[extension] = datatype_instance
if mimetype is None:
# Use default mimetype per datatype specification.
mimetype = self.datatypes_by_extension[extension].get_mime()
self.mimetypes_by_extension[extension] = mimetype
if datatype_class.track_type:
self.available_tracks.append(extension)
if display_in_upload and extension not in self.upload_file_formats:
self.upload_file_formats.append(extension)
# Max file size cut off for setting optional metadata.
self.datatypes_by_extension[extension].max_optional_metadata_filesize = elem.get(
"max_optional_metadata_filesize", None
)
infer_from_suffixes = []
# read from element instead of attribute so we can customize references to
# compressed files in the future (e.g. maybe some day faz will be a compressed fasta
# or something along those lines)
for infer_from in elem.findall("infer_from"):
suffix = infer_from.get("suffix", None)
if suffix is None:
raise Exception("Failed to parse infer_from datatype element")
infer_from_suffixes.append(suffix)
self.datatypes_by_suffix_inferences[suffix] = datatype_instance
for converter in elem.findall("converter"):
# Build the list of datatype converters which will later be loaded into the calling app's toolbox.
converter_config = converter.get("file", None)
target_datatype = converter.get("target_datatype", None)
depends_on = converter.get("depends_on", None)
if depends_on is not None and target_datatype is not None:
if extension not in self.converter_deps:
self.converter_deps[extension] = {}
self.converter_deps[extension][target_datatype] = depends_on.split(",")
if converter_config and target_datatype:
if proprietary_converter_path:
self.proprietary_converters.append(
(converter_config, extension, target_datatype)
)
else:
self.converters.append((converter_config, extension, target_datatype))
# Add composite files.
for composite_file in elem.findall("composite_file"):
name = composite_file.get("name", None)
if name is None:
self.log.warning(
f"You must provide a name for your composite_file ({composite_file})."
)
optional = composite_file.get("optional", False)
mimetype = composite_file.get("mimetype", None)
self.datatypes_by_extension[extension].add_composite_file(
name, optional=optional, mimetype=mimetype
)
for _display_app in elem.findall("display"):
if proprietary_display_path:
if elem not in self.proprietary_display_app_containers:
self.proprietary_display_app_containers.append(elem)
else:
if elem not in self.display_app_containers:
self.display_app_containers.append(elem)
datatype_info_dict = {
except Exception:
self.log.exception("Error importing datatype module %s", str(datatype_module))
ok = False
elif type_extension is not None:
try:
datatype_class = self.datatypes_by_extension[type_extension].__class__
self.log.debug(
f"Retrieved datatype module {str(datatype_class.__name__)} from type_extension {type_extension} for extension {extension}."
)
except Exception:
self.log.exception(
"Error determining datatype_class for type_extension %s", str(type_extension)
)
ok = False
if ok:
# A new tool shed repository that contains custom datatypes is being installed, and since installation is
# occurring after the datatypes registry has been initialized at server startup, its contents cannot be
# overridden by new introduced conflicting data types unless the value of override is True.
if extension in self.datatypes_by_extension:
# Because of the way that the value of can_process_datatype was set above, we know that the value of
# override is True.
self.log.debug(
"Overriding conflicting datatype with extension '%s', using datatype from %s."
% (str(extension), str(config))
)
if make_subclass:
datatype_class = type(datatype_class_name, (datatype_class,), {})
if edam_format:
datatype_class.edam_format = edam_format
if edam_data:
datatype_class.edam_data = edam_data
datatype_class.is_subclass = make_subclass
description = elem.get("description", None)
description_url = elem.get("description_url", None)
datatype_instance = datatype_class()
self.datatypes_by_extension[extension] = datatype_instance
if mimetype is None:
# Use default mimetype per datatype specification.
mimetype = self.datatypes_by_extension[extension].get_mime()
self.mimetypes_by_extension[extension] = mimetype
if datatype_class.track_type:
self.available_tracks.append(extension)
if display_in_upload and extension not in self.upload_file_formats:
self.upload_file_formats.append(extension)
# Max file size cut off for setting optional metadata.
self.datatypes_by_extension[extension].max_optional_metadata_filesize = elem.get(
"max_optional_metadata_filesize", None
)
infer_from_suffixes = []
# read from element instead of attribute so we can customize references to
# compressed files in the future (e.g. maybe some day faz will be a compressed fasta
# or something along those lines)
for infer_from in elem.findall("infer_from"):
suffix = infer_from.get("suffix", None)
if suffix is None:
raise Exception("Failed to parse infer_from datatype element")
infer_from_suffixes.append(suffix)
self.datatypes_by_suffix_inferences[suffix] = datatype_instance
for converter in elem.findall("converter"):
# Build the list of datatype converters which will later be loaded into the calling app's toolbox.
converter_config = converter.get("file", None)
target_datatype = converter.get("target_datatype", None)
depends_on = converter.get("depends_on", None)
if depends_on is not None and target_datatype is not None:
if extension not in self.converter_deps:
self.converter_deps[extension] = {}
self.converter_deps[extension][target_datatype] = depends_on.split(",")
if converter_config and target_datatype:
self.converters.append((converter_config, extension, target_datatype))
# Add composite files.
for composite_file in elem.findall("composite_file"):
name = composite_file.get("name", None)
if name is None:
self.log.warning(f"You must provide a name for your composite_file ({composite_file}).")
optional = composite_file.get("optional", False)
mimetype = composite_file.get("mimetype", None)
self.datatypes_by_extension[extension].add_composite_file(
name, optional=optional, mimetype=mimetype
)
for _display_app in elem.findall("display"):
if elem not in self.display_app_containers:
self.display_app_containers.append(elem)
datatype_info_dict = {
"display_in_upload": display_in_upload,
"extension": extension,
"description": description,
"description_url": description_url,
}
composite_files = datatype_instance.get_composite_files()
if composite_files:
_composite_files = []
for name, composite_file in composite_files.items():
_composite_file = composite_file.dict()
_composite_file["name"] = name
_composite_files.append(_composite_file)
datatype_info_dict["composite_files"] = _composite_files
self.datatype_info_dicts.append(datatype_info_dict)
for auto_compressed_type in auto_compressed_types:
compressed_extension = f"{extension}.{auto_compressed_type}"
upper_compressed_type = auto_compressed_type[0].upper() + auto_compressed_type[1:]
auto_compressed_type_name = datatype_class_name + upper_compressed_type
attributes = {}
if auto_compressed_type == "gz":
dynamic_parent = binary.GzDynamicCompressedArchive
elif auto_compressed_type == "bz2":
dynamic_parent = binary.Bz2DynamicCompressedArchive
else:
raise Exception(f"Unknown auto compression type [{auto_compressed_type}]")
attributes["file_ext"] = compressed_extension
attributes["uncompressed_datatype_instance"] = datatype_instance
compressed_datatype_class = type(
auto_compressed_type_name,
(
datatype_class,
dynamic_parent,
),
attributes,
)
if edam_format:
compressed_datatype_class.edam_format = edam_format
if edam_data:
compressed_datatype_class.edam_data = edam_data
compressed_datatype_instance = compressed_datatype_class()
self.datatypes_by_extension[compressed_extension] = compressed_datatype_instance
for suffix in infer_from_suffixes:
self.datatypes_by_suffix_inferences[
f"{suffix}.{auto_compressed_type}"
] = compressed_datatype_instance
if display_in_upload and compressed_extension not in self.upload_file_formats:
self.upload_file_formats.append(compressed_extension)
self.datatype_info_dicts.append(
{
"display_in_upload": display_in_upload,
"extension": extension,
"extension": compressed_extension,
"description": description,
"description_url": description_url,
}
composite_files = datatype_instance.get_composite_files()
if composite_files:
_composite_files = []
for name, composite_file in composite_files.items():
_composite_file = composite_file.dict()
_composite_file["name"] = name
_composite_files.append(_composite_file)
datatype_info_dict["composite_files"] = _composite_files
self.datatype_info_dicts.append(datatype_info_dict)
for auto_compressed_type in auto_compressed_types:
compressed_extension = f"{extension}.{auto_compressed_type}"
upper_compressed_type = auto_compressed_type[0].upper() + auto_compressed_type[1:]
auto_compressed_type_name = datatype_class_name + upper_compressed_type
attributes = {}
if auto_compressed_type == "gz":
dynamic_parent = binary.GzDynamicCompressedArchive
elif auto_compressed_type == "bz2":
dynamic_parent = binary.Bz2DynamicCompressedArchive
else:
raise Exception(f"Unknown auto compression type [{auto_compressed_type}]")
attributes["file_ext"] = compressed_extension
attributes["uncompressed_datatype_instance"] = datatype_instance
compressed_datatype_class = type(
auto_compressed_type_name,
(
datatype_class,
dynamic_parent,
),
attributes,
)
if auto_compressed_type == "gz":
self.converters.append(
(
f"uncompressed_to_{auto_compressed_type}.xml",
extension,
compressed_extension,
)
if edam_format:
compressed_datatype_class.edam_format = edam_format
if edam_data:
compressed_datatype_class.edam_data = edam_data
compressed_datatype_instance = compressed_datatype_class()
self.datatypes_by_extension[compressed_extension] = compressed_datatype_instance
for suffix in infer_from_suffixes:
self.datatypes_by_suffix_inferences[
f"{suffix}.{auto_compressed_type}"
] = compressed_datatype_instance
if display_in_upload and compressed_extension not in self.upload_file_formats:
self.upload_file_formats.append(compressed_extension)
self.datatype_info_dicts.append(
{
"display_in_upload": display_in_upload,
"extension": compressed_extension,
"description": description,
"description_url": description_url,
}
)
if auto_compressed_type == "gz":
self.converters.append(
(
f"uncompressed_to_{auto_compressed_type}.xml",
extension,
compressed_extension,
)
)
self.converters.append(
(f"{auto_compressed_type}_to_uncompressed.xml", compressed_extension, extension)
)
if datatype_class not in compressed_sniffers:
compressed_sniffers[datatype_class] = []
if sniff_compressed_types:
compressed_sniffers[datatype_class].append(compressed_datatype_instance)
# Processing the new datatype elem is now complete, so make sure the element defining it is retained by appending
# the new datatype to the in-memory list of datatype elems to enable persistence.
self.datatype_elems.append(elem)
)
self.converters.append(
(f"{auto_compressed_type}_to_uncompressed.xml", compressed_extension, extension)
)
if datatype_class not in compressed_sniffers:
compressed_sniffers[datatype_class] = []
if sniff_compressed_types:
compressed_sniffers[datatype_class].append(compressed_datatype_instance)
# Processing the new datatype elem is now complete, so make sure the element defining it is retained by appending
# the new datatype to the in-memory list of datatype elems to enable persistence.
self.datatype_elems.append(elem)
else:
if extension is not None:
if dtype is not None or type_extension is not None:
@@ -488,8 +387,6 @@ class Registry:
# since sniffers are not tightly coupled with datatypes.
self.load_datatype_sniffers(
root,
deactivate=deactivate,
handling_proprietary_datatypes=handling_proprietary_datatypes,
override=override,
compressed_sniffers=compressed_sniffers,
)
@@ -570,16 +467,10 @@ class Registry:
def get_display_sites(self, site_type):
return self.display_sites.get(site_type, [])
def load_datatype_sniffers(
self, root, deactivate=False, handling_proprietary_datatypes=False, override=False, compressed_sniffers=None
):
def load_datatype_sniffers(self, root, override=False, compressed_sniffers=None):
"""
Process the sniffers element from a parsed a datatypes XML file located at root_dir/config (if processing the Galaxy
distributed config) or contained within an installed Tool Shed repository. If deactivate is True, an installed Tool
Shed repository that includes custom sniffers is being deactivated or uninstalled, so appropriate loaded sniffers will
be removed from the registry. The value of override will be False when a Tool Shed repository is being installed.
Since installation is occurring after the datatypes registry has been initialized at server startup, its contents
cannot be overridden by newly introduced conflicting sniffers.
distributed config) or contained within an installed Tool Shed repository.
"""
sniffer_elem_classes = [e.attrib["type"] for e in self.sniffer_elems]
sniffers = root.find("sniffers")
@@ -598,12 +489,6 @@ class Registry:
self.log.exception("Error determining datatype class or module for dtype %s", str(dtype))
ok = False
if ok:
if handling_proprietary_datatypes:
# See if one of the imported modules contains the datatype class name.
for imported_module in self.imported_modules:
if hasattr(imported_module, datatype_class_name):
module = imported_module
break
if module is None:
try:
# The datatype class name must be contained in one of the datatype modules in the Galaxy distribution.
@@ -622,47 +507,27 @@ class Registry:
)
ok = False
if ok:
if deactivate:
# We are deactivating or uninstalling an installed Tool Shed repository, so eliminate the appropriate sniffers.
sniffer_class = elem.get("type", None)
if sniffer_class is not None:
for index, s_e_c in enumerate(sniffer_elem_classes):
if sniffer_class == s_e_c:
del self.sniffer_elems[index]
sniffer_elem_classes = [
elem.attrib["type"] for elem in self.sniffer_elems
]
self.log.debug(f"Removed sniffer element for datatype '{str(dtype)}'")
break
for sniffer_class in self.sniff_order:
if sniffer_class.__class__ == aclass.__class__:
self.sniff_order.remove(sniffer_class)
self.log.debug(
f"Removed sniffer class for datatype '{str(dtype)}' from sniff order"
)
break
else:
# We are loading new sniffer, so see if we have a conflicting sniffer already loaded.
conflict = False
for conflict_loc, sniffer_class in enumerate(self.sniff_order):
if sniffer_class.__class__ == aclass.__class__:
# We have a conflicting sniffer, so replace the one previously loaded.
conflict = True
if override:
del self.sniff_order[conflict_loc]
self.log.debug(f"Removed conflicting sniffer for datatype '{dtype}'")
break
if not conflict or override:
if compressed_sniffers and aclass.__class__ in compressed_sniffers:
for compressed_sniffer in compressed_sniffers[aclass.__class__]:
self.sniff_order.append(compressed_sniffer)
self.sniff_order.append(aclass)
self.log.debug(f"Loaded sniffer for datatype '{dtype}'")
# Processing the new sniffer elem is now complete, so make sure the element defining it is loaded if necessary.
sniffer_class = elem.get("type", None)
if sniffer_class is not None:
if sniffer_class not in sniffer_elem_classes:
self.sniffer_elems.append(elem)
# We are loading new sniffer, so see if we have a conflicting sniffer already loaded.
conflict = False
for conflict_loc, sniffer_class in enumerate(self.sniff_order):
if sniffer_class.__class__ == aclass.__class__:
# We have a conflicting sniffer, so replace the one previously loaded.
conflict = True
if override:
del self.sniff_order[conflict_loc]
self.log.debug(f"Removed conflicting sniffer for datatype '{dtype}'")
break
if not conflict or override:
if compressed_sniffers and aclass.__class__ in compressed_sniffers:
for compressed_sniffer in compressed_sniffers[aclass.__class__]:
self.sniff_order.append(compressed_sniffer)
self.sniff_order.append(aclass)
self.log.debug(f"Loaded sniffer for datatype '{dtype}'")
# Processing the new sniffer elem is now complete, so make sure the element defining it is loaded if necessary.
sniffer_class = elem.get("type", None)
if sniffer_class is not None:
if sniffer_class not in sniffer_elem_classes:
self.sniffer_elems.append(elem)
def get_datatype_from_filename(self, name):
max_extension_parts = 3
@@ -734,149 +599,64 @@ class Registry:
data.init_meta(copy_from=data)
return data
def load_datatype_converters(self, toolbox, installed_repository_dict=None, deactivate=False, use_cached=False):
def load_datatype_converters(self, toolbox, use_cached=False):
"""
If deactivate is False, add datatype converters from self.converters or self.proprietary_converters
to the calling app's toolbox. If deactivate is True, eliminates relevant converters from the calling
app's toolbox.
Add datatype converters from self.converters to the calling app's toolbox.
"""
if installed_repository_dict:
# Load converters defined by datatypes_conf.xml included in installed tool shed repository.
converters = self.proprietary_converters
else:
# Load converters defined by local datatypes_conf.xml.
converters = self.converters
# Load converters defined by local datatypes_conf.xml.
converters = self.converters
for elem in converters:
tool_config = elem[0]
source_datatype = elem[1]
target_datatype = elem[2]
if installed_repository_dict:
converter_path = installed_repository_dict["converter_path"]
else:
converter_path = self.converters_path
converter_path = self.converters_path
try:
config_path = os.path.join(converter_path, tool_config)
converter = toolbox.load_tool(config_path, use_cached=use_cached)
self.converter_tools.add(converter)
if installed_repository_dict:
# If the converter is included in an installed tool shed repository, set the tool
# shed related tool attributes.
converter.tool_shed = installed_repository_dict["tool_shed"]
converter.repository_name = installed_repository_dict["repository_name"]
converter.repository_owner = installed_repository_dict["repository_owner"]
converter.installed_changeset_revision = installed_repository_dict["installed_changeset_revision"]
converter.old_id = converter.id
# The converter should be included in the list of tools defined in tool_dicts.
tool_dicts = installed_repository_dict["tool_dicts"]
for tool_dict in tool_dicts:
if tool_dict["id"] == converter.id:
converter.guid = tool_dict["guid"]
converter.id = tool_dict["guid"]
break
if deactivate:
toolbox.remove_tool_by_id(converter.id, remove_from_panel=False)
if source_datatype in self.datatype_converters:
if target_datatype in self.datatype_converters[source_datatype]:
del self.datatype_converters[source_datatype][target_datatype]
self.log.debug("Deactivated converter: %s", converter.id)
else:
toolbox.register_tool(converter)
if source_datatype not in self.datatype_converters:
self.datatype_converters[source_datatype] = {}
self.datatype_converters[source_datatype][target_datatype] = converter
if not hasattr(toolbox.app, "tool_cache") or converter.id in toolbox.app.tool_cache._new_tool_ids:
self.log.debug("Loaded converter: %s", converter.id)
toolbox.register_tool(converter)
if source_datatype not in self.datatype_converters:
self.datatype_converters[source_datatype] = {}
self.datatype_converters[source_datatype][target_datatype] = converter
if not hasattr(toolbox.app, "tool_cache") or converter.id in toolbox.app.tool_cache._new_tool_ids:
self.log.debug("Loaded converter: %s", converter.id)
except Exception:
if deactivate:
self.log.exception(f"Error deactivating converter from ({converter_path})")
else:
self.log.exception(f"Error loading converter ({converter_path})")
self.log.exception(f"Error loading converter ({converter_path})")
def load_display_applications(self, app, installed_repository_dict=None, deactivate=False):
def load_display_applications(self, app):
"""
If deactivate is False, add display applications from self.display_app_containers or
self.proprietary_display_app_containers to appropriate datatypes. If deactivate is
True, eliminates relevant display applications from appropriate datatypes.
Add display applications from self.display_app_containers or to appropriate datatypes.
"""
if installed_repository_dict:
# Load display applications defined by datatypes_conf.xml included in installed tool shed repository.
datatype_elems = self.proprietary_display_app_containers
else:
# Load display applications defined by local datatypes_conf.xml.
datatype_elems = self.display_app_containers
# Load display applications defined by local datatypes_conf.xml.
datatype_elems = self.display_app_containers
for elem in datatype_elems:
extension = self.get_extension(elem)
for display_app in elem.findall("display"):
display_file = display_app.get("file", None)
if installed_repository_dict:
display_path = installed_repository_dict["display_path"]
display_file_head, display_file_tail = os.path.split(display_file)
config_path = os.path.join(display_path, display_file_tail)
else:
config_path = os.path.join(self.display_applications_path, display_file)
config_path = os.path.join(self.display_applications_path, display_file)
try:
inherit = galaxy.util.string_as_bool(display_app.get("inherit", "False"))
display_app = DisplayApplication.from_file(config_path, app)
if display_app:
if display_app.id in self.display_applications:
if deactivate:
del self.display_applications[display_app.id]
else:
# If we already loaded this display application, we'll use the first one loaded.
display_app = self.display_applications[display_app.id]
elif installed_repository_dict:
# If the display application is included in an installed tool shed repository,
# set the tool shed related tool attributes.
display_app.tool_shed = installed_repository_dict["tool_shed"]
display_app.repository_name = installed_repository_dict["repository_name"]
display_app.repository_owner = installed_repository_dict["repository_owner"]
display_app.installed_changeset_revision = installed_repository_dict[
"installed_changeset_revision"
]
display_app.old_id = display_app.id
# The display application should be included in the list of tools defined in tool_dicts.
tool_dicts = installed_repository_dict["tool_dicts"]
for tool_dict in tool_dicts:
if tool_dict["id"] == display_app.id:
display_app.guid = tool_dict["guid"]
display_app.id = tool_dict["guid"]
break
if deactivate:
if display_app.id in self.display_applications:
del self.display_applications[display_app.id]
if extension in self.datatypes_by_extension:
if display_app.id in self.datatypes_by_extension[extension].display_applications:
del self.datatypes_by_extension[extension].display_applications[display_app.id]
if (
inherit
and (self.datatypes_by_extension[extension], display_app)
in self.inherit_display_application_by_class
):
self.inherit_display_application_by_class.remove(
(self.datatypes_by_extension[extension], display_app)
)
self.log.debug(
f"Deactivated display application '{display_app.id}' for datatype '{extension}'."
)
else:
self.display_applications[display_app.id] = display_app
self.datatypes_by_extension[extension].add_display_application(display_app)
if (
inherit
and (self.datatypes_by_extension[extension], display_app)
not in self.inherit_display_application_by_class
):
self.inherit_display_application_by_class.append(
(self.datatypes_by_extension[extension], display_app)
)
self.log.debug(
f"Loaded display application '{display_app.id}' for datatype '{extension}', inherit={inherit}."
# If we already loaded this display application, we'll use the first one loaded.
display_app = self.display_applications[display_app.id]
self.display_applications[display_app.id] = display_app
self.datatypes_by_extension[extension].add_display_application(display_app)
if (
inherit
and (self.datatypes_by_extension[extension], display_app)
not in self.inherit_display_application_by_class
):
self.inherit_display_application_by_class.append(
(self.datatypes_by_extension[extension], display_app)
)
self.log.debug(
f"Loaded display application '{display_app.id}' for datatype '{extension}', inherit={inherit}."
)
except Exception:
if deactivate:
self.log.exception(f"Error deactivating display application ({config_path})")
else:
self.log.exception(f"Error loading display application ({config_path})")
self.log.exception(f"Error loading display application ({config_path})")
# Handle display_application subclass inheritance.
for extension, d_type1 in self.datatypes_by_extension.items():
for d_type2, display_app in self.inherit_display_application_by_class:
@@ -1,182 +0,0 @@
import logging
import os
from galaxy.tool_shed.util.basic_util import strip_path
from galaxy.tool_shed.util.hg_util import get_config_from_disk
from galaxy.tool_shed.util.shed_util_common import DATATYPES_CONFIG_FILENAME
from galaxy.util.tool_shed.xml_util import parse_xml
log = logging.getLogger(__name__)
class CustomDatatypeLoader:
def __init__(self, app):
self.app = app
def _alter_config_and_load_proprietary_datatypes(
self, datatypes_config, relative_install_dir, deactivate=False, override=True
):
"""
Parse a custom datatypes config (a datatypes_conf.xml file included in an installed
tool shed repository) and add information to appropriate element attributes that will
enable custom datatype class modules, datatypes converters and display applications
to be discovered and properly imported by the datatypes registry. The value of override
will be False when a tool shed repository is being installed. Since installation is
occurring after the datatypes registry has been initialized, the registry's contents
cannot be overridden by conflicting data types.
"""
tree, error_message = parse_xml(datatypes_config)
if tree is None:
return None, None
datatypes_config_root = tree.getroot()
registration = datatypes_config_root.find("registration")
if registration is None:
# We have valid XML, but not a valid custom datatypes definition.
return None, None
converter_path, display_path = self._get_converter_and_display_paths(registration, relative_install_dir)
if converter_path:
# Path to datatype converters
registration.attrib["proprietary_converter_path"] = converter_path
if display_path:
# Path to datatype display applications
registration.attrib["proprietary_display_path"] = display_path
relative_path_to_datatype_file_name = None
datatype_files = datatypes_config_root.find("datatype_files")
datatype_class_modules = []
if datatype_files is not None:
# The <datatype_files> tag set contains any number of <datatype_file> tags.
# <datatype_files>
# <datatype_file name="gmap.py"/>
# <datatype_file name="metagenomics.py"/>
# </datatype_files>
# We'll add attributes to the datatype tag sets so that the modules can be properly imported
# by the datatypes registry.
for elem in datatype_files.findall("datatype_file"):
datatype_file_name = elem.get("name", None)
if datatype_file_name:
# Find the file in the installed repository.
for root, _dirs, files in os.walk(relative_install_dir):
if root.find(".hg") < 0:
for name in files:
if name == datatype_file_name:
datatype_class_modules.append(os.path.join(root, name))
break
break
if datatype_class_modules:
for relative_path_to_datatype_file_name in datatype_class_modules:
datatype_file_name_path, datatype_file_name = os.path.split(relative_path_to_datatype_file_name)
for elem in registration.findall("datatype"):
# Handle 'type' attribute which should be something like one of the following:
# type="gmap:GmapDB"
# type="galaxy.datatypes.gmap:GmapDB"
dtype = elem.get("type", None)
if dtype:
fields = dtype.split(":")
proprietary_datatype_module = fields[0]
if proprietary_datatype_module.find(".") >= 0:
# Handle the case where datatype_module is "galaxy.datatypes.gmap".
proprietary_datatype_module = proprietary_datatype_module.split(".")[-1]
# The value of proprietary_path must be an absolute path due to job_working_directory.
elem.attrib["proprietary_path"] = os.path.abspath(datatype_file_name_path)
elem.attrib["proprietary_datatype_module"] = proprietary_datatype_module
# Load custom datatypes
self.app.datatypes_registry.load_datatypes(
root_dir=self.app.config.root, config=datatypes_config_root, deactivate=deactivate, override=override
)
return converter_path, display_path
def _create_repository_dict_for_proprietary_datatypes(
self, tool_shed, name, owner, installed_changeset_revision, tool_dicts, converter_path=None, display_path=None
):
return dict(
tool_shed=tool_shed,
repository_name=name,
repository_owner=owner,
installed_changeset_revision=installed_changeset_revision,
tool_dicts=tool_dicts,
converter_path=converter_path,
display_path=display_path,
)
def _get_converter_and_display_paths(self, registration_elem, relative_install_dir):
"""
Find the relative path to data type converters and display applications included
in installed tool shed repositories.
"""
converter_path = None
display_path = None
for elem in registration_elem.findall("datatype"):
if not converter_path:
# If any of the <datatype> tag sets contain <converter> tags, set the converter_path
# if it is not already set. This requires developers to place all converters in the
# same subdirectory within the repository hierarchy.
for converter in elem.findall("converter"):
converter_config = converter.get("file", None)
if converter_config:
converter_config_file_name = strip_path(converter_config)
for root, _dirs, files in os.walk(relative_install_dir):
if root.find(".hg") < 0:
for name in files:
if name == converter_config_file_name:
# The value of converter_path must be absolute due to job_working_directory.
converter_path = os.path.abspath(root)
break
if converter_path:
break
if not display_path:
# If any of the <datatype> tag sets contain <display> tags, set the display_path
# if it is not already set. This requires developers to place all display acpplications
# in the same subdirectory within the repository hierarchy.
for display_app in elem.findall("display"):
display_config = display_app.get("file", None)
if display_config:
display_config_file_name = strip_path(display_config)
for root, _dirs, files in os.walk(relative_install_dir):
if root.find(".hg") < 0:
for name in files:
if name == display_config_file_name:
# The value of display_path must be absolute due to job_working_directory.
display_path = os.path.abspath(root)
break
if display_path:
break
if converter_path and display_path:
break
return converter_path, display_path
def load_installed_datatype_converters(self, installed_repository_dict, deactivate=False):
"""Load or deactivate proprietary datatype converters."""
self.app.datatypes_registry.load_datatype_converters(
self.app.toolbox, installed_repository_dict=installed_repository_dict, deactivate=deactivate
)
def load_installed_datatypes(self, repository, relative_install_dir, deactivate=False):
"""
Load proprietary datatypes and return information needed for loading custom
datatypes converters and display applications later.
"""
metadata = repository.metadata_
repository_dict = None
datatypes_config = get_config_from_disk(DATATYPES_CONFIG_FILENAME, relative_install_dir)
if datatypes_config:
converter_path, display_path = self._alter_config_and_load_proprietary_datatypes(
datatypes_config, relative_install_dir, deactivate=deactivate
)
if converter_path or display_path:
# Create a dictionary of tool shed repository related information.
repository_dict = self._create_repository_dict_for_proprietary_datatypes(
tool_shed=repository.tool_shed,
name=repository.name,
owner=repository.owner,
installed_changeset_revision=repository.installed_changeset_revision,
tool_dicts=metadata.get("tools", []),
converter_path=converter_path,
display_path=display_path,
)
return repository_dict
def load_installed_display_applications(self, installed_repository_dict, deactivate=False):
"""Load or deactivate custom datatype display applications."""
self.app.datatypes_registry.load_display_applications(
self.app, installed_repository_dict=installed_repository_dict, deactivate=deactivate
)
@@ -6,14 +6,7 @@ import logging
import os
import shutil
from sqlalchemy import (
and_,
false,
true,
)
from galaxy import util
from galaxy.tool_shed.galaxy_install.datatypes import custom_datatype_manager
from galaxy.tool_shed.galaxy_install.metadata.installed_repository_metadata_manager import (
InstalledRepositoryMetadataManager,
)
@@ -619,32 +612,6 @@ class InstalledRepositoryManager:
deleted_tool_dependency_names.append(original_dependency_val_dict["name"])
return updated_tool_dependency_names, deleted_tool_dependency_names
def load_proprietary_datatypes(self):
cdl = custom_datatype_manager.CustomDatatypeLoader(self.app)
for tool_shed_repository in (
self.context.query(self.install_model.ToolShedRepository)
.filter(
and_(
self.install_model.ToolShedRepository.table.c.includes_datatypes == true(),
self.install_model.ToolShedRepository.table.c.deleted == false(),
)
)
.order_by(self.install_model.ToolShedRepository.table.c.id)
):
relative_install_dir = self.get_repository_install_dir(tool_shed_repository)
if relative_install_dir:
installed_repository_dict = cdl.load_installed_datatypes(tool_shed_repository, relative_install_dir)
if installed_repository_dict:
self.installed_repository_dicts.append(installed_repository_dict)
def load_proprietary_converters_and_display_applications(self, deactivate=False):
cdl = custom_datatype_manager.CustomDatatypeLoader(self.app)
for installed_repository_dict in self.installed_repository_dicts:
if installed_repository_dict["converter_path"]:
cdl.load_installed_datatype_converters(installed_repository_dict, deactivate=deactivate)
if installed_repository_dict["display_path"]:
cdl.load_installed_display_applications(installed_repository_dict, deactivate=deactivate)
def uninstall_repository(self, repository, remove_from_disk=True):
errors = ""
shed_tool_conf, tool_path, relative_install_dir = suc.get_tool_panel_config_tool_path_install_dir(