view env/lib/python3.9/site-packages/planemo/conda.py @ 0:4f3585e2f14b draft default tip

"planemo upload commit 60cee0fc7c0cda8592644e1aad72851dec82c959"

"""Planemo specific utilities for dealing with conda.

The extend galaxy-tool-util's features with planemo specific idioms.
"""

from __future__ import absolute_import

import collections
import os
import threading

from galaxy.tool_util.deps import conda_util
from galaxy.util import unicodify

from planemo.exit_codes import EXIT_CODE_FAILED_DEPENDENCIES, ExitCodeException
from planemo.io import error, shell
from planemo.tools import yield_tool_sources_on_paths

MESSAGE_ERROR_FAILED_INSTALL = "Attempted to install conda and failed."
MESSAGE_ERROR_CANNOT_INSTALL = "Cannot install Conda - perhaps due to a failed installation or permission problems."
MESSAGE_ERROR_NOT_INSTALLING = "Conda not configured - run ``planemo conda_init`` or pass ``--conda_auto_init`` to continue."

BEST_PRACTICE_CHANNELS = ["conda-forge", "bioconda", "defaults"]


def build_conda_context(ctx, **kwds):
    """Build a galaxy-tool-util CondaContext tailored to planemo use.

    Using planemo's common command-line/global config options.
    """
    condarc_override_default = os.path.join(ctx.workspace, "condarc")
    conda_prefix = kwds.get("conda_prefix", None)
    use_planemo_shell = kwds.get("use_planemo_shell_exec", True)
    ensure_channels = kwds.get("conda_ensure_channels", "")
    condarc_override = kwds.get("condarc", condarc_override_default)
    use_local = kwds.get("conda_use_local", False)
    shell_exec = shell if use_planemo_shell else None
    conda_context = conda_util.CondaContext(conda_prefix=conda_prefix,
                                            ensure_channels=ensure_channels,
                                            condarc_override=condarc_override,
                                            use_local=use_local,
                                            shell_exec=shell_exec)
    handle_auto_init = kwds.get("handle_auto_init", False)
    if handle_auto_init and not conda_context.is_installed():
        auto_init = kwds.get("conda_auto_init", True)
        failed = True
        if auto_init:
            if conda_context.can_install_conda():
                if conda_util.install_conda(conda_context):
                    error(MESSAGE_ERROR_FAILED_INSTALL)
                else:
                    failed = False
            else:
                error(MESSAGE_ERROR_CANNOT_INSTALL)
        else:
            error(MESSAGE_ERROR_NOT_INSTALLING)

        if failed:
            raise ExitCodeException(EXIT_CODE_FAILED_DEPENDENCIES)
    if handle_auto_init:
        conda_context.ensure_conda_build_installed_if_needed()
    return conda_context


def collect_conda_targets(ctx, paths, recursive=False, found_tool_callback=None):
    """Load CondaTarget objects from supplied artifact sources.

    If a tool contains more than one requirement, the requirements will each
    appear once in the output.
    """
    conda_targets = set([])
    real_paths = []
    for path in paths:
        if not os.path.exists(path):
            targets = target_str_to_targets(path)
            [conda_targets.add(_) for _ in targets]
        else:
            real_paths.append(path)

    for (tool_path, tool_source) in yield_tool_sources_on_paths(ctx, real_paths, recursive=recursive, exclude_deprecated=True):
        if found_tool_callback:
            found_tool_callback(tool_path)
        for target in tool_source_conda_targets(tool_source):
            conda_targets.add(target)
    return conda_targets


# Copied and modified from mulled stuff - need to syncronize these concepts.
def target_str_to_targets(targets_raw):
    def parse_target(target_str):
        if "=" in target_str:
            package_name, version = target_str.split("=", 1)
        else:
            package_name = target_str
            version = None
        target = conda_util.CondaTarget(package_name, version)
        return target

    targets = [parse_target(_) for _ in targets_raw.split(",")]
    return targets


def collect_conda_target_lists(ctx, paths, recursive=False, found_tool_callback=None):
    """Load CondaTarget lists from supplied artifact sources.

    If a tool contains more than one requirement, the requirements will all
    appear together as one list element of the output list.
    """
    conda_target_lists, _ = collect_conda_target_lists_and_tool_paths(ctx, paths, recursive=recursive, found_tool_callback=found_tool_callback)
    return conda_target_lists


def collect_conda_target_lists_and_tool_paths(ctx, paths, recursive=False, found_tool_callback=None):
    """Load CondaTarget lists from supplied artifact sources.

    If a tool contains more than one requirement, the requirements will all
    appear together as one list element of the output list.
    """
    conda_target_lists = set([])
    tool_paths = collections.defaultdict(list)
    for (tool_path, tool_source) in yield_tool_sources_on_paths(ctx, paths, recursive=recursive, yield_load_errors=False):
        try:
            if found_tool_callback:
                found_tool_callback(tool_path)
            targets = frozenset(tool_source_conda_targets(tool_source))
            conda_target_lists.add(targets)
            tool_paths[targets].append(tool_path)
        except Exception as e:
            ctx.log(f"Error while collecting list of conda targets for '{tool_path}': {unicodify(e)}")

    # Turn them into lists so the order matches before returning...
    conda_target_lists = list(conda_target_lists)
    conda_target_tool_paths = [tool_paths[c] for c in conda_target_lists]

    return conda_target_lists, conda_target_tool_paths


def tool_source_conda_targets(tool_source):
    """Load CondaTarget object from supplied abstract tool source."""
    requirements, _ = tool_source.parse_requirements_and_containers()
    return conda_util.requirements_to_conda_targets(requirements)


best_practice_search_first = threading.local()


def best_practice_search(conda_target, conda_context=None, platform=None):
    # Call it in offline mode after the first time.
    try:
        best_practice_search_first.previously_called
        # TODO: Undo this...
        offline = False
    except AttributeError:
        best_practice_search_first.previously_called = True
        offline = False

    if not conda_context:
        conda_context = conda_util.CondaContext()
    return conda_util.best_search_result(
        conda_target,
        conda_context=conda_context,
        channels_override=BEST_PRACTICE_CHANNELS,
        offline=offline,
        platform=platform,
    )


__all__ = (
    "BEST_PRACTICE_CHANNELS",
    "best_practice_search",
    "build_conda_context",
    "collect_conda_targets",
    "collect_conda_target_lists",
    "collect_conda_target_lists_and_tool_paths",
    "tool_source_conda_targets",
)