libmambapy add_constraint adds constraint packages to install

[minrk]

Troubleshooting docs

• My problem is not solved in the Troubleshooting docs

Anaconda default channels

• I do NOT use the Anaconda default channels (pkgs/* etc.)

How did you install Mamba?

Mambaforge or latest Miniforge

Search tried in issue tracker

searched for add_constraint, could find no mentions of issues

Latest version of Mamba

• My problem is not solved with the latest version

Tried in Conda?

I do not have this problem with Conda, just with Mamba

Describe your issue

libmambapy 1.5.8

attempting to add constraints via solver.add_constraint(["package version"]) results in constraints being installed, instead of merely constraining the install.

test case

import json
from collections import OrderedDict

import libmambapy


def get_index(
    channel_urls=(),
    platform=None,
    use_local=False,
    use_cache=False,
    unknown=None,
    prefix=None,
    repodata_fn="repodata.json",
    *,
    package_cache,
):
    if isinstance(platform, str):
        platform = [platform, "noarch"]

    all_channels = []
    if use_local:
        all_channels.append("local")
    all_channels.extend(channel_urls)
    # check_allowlist(all_channels)

    # Remove duplicates but retain order
    all_channels = list(OrderedDict.fromkeys(all_channels))

    dlist = libmambapy.DownloadTargetList()

    index = []

    def fixup_channel_spec(spec):
        at_count = spec.count("@")
        if at_count > 1:
            first_at = spec.find("@")
            spec = (
                spec[:first_at]
                + urllib.parse.quote(spec[first_at])
                + spec[first_at + 1 :]
            )
        if platform:
            spec = spec + "[" + ",".join(platform) + "]"
        return spec

    all_channels = list(map(fixup_channel_spec, all_channels))
    libmambapy.create_cache_dir(package_cache.first_writable_path)

    for channel in libmambapy.get_channels(all_channels):
        for channel_platform, url in channel.platform_urls(with_credentials=True):
            full_url = url

            sd = libmambapy.SubdirData(
                channel, channel_platform, full_url, package_cache, repodata_fn
            )

            needs_finalising = sd.download_and_check_targets(dlist)
            index.append(
                (
                    sd,
                    {
                        "platform": channel_platform,
                        "url": url,
                        "channel": channel,
                        "needs_finalising": needs_finalising,
                    },
                )
            )

    for sd, info in index:
        if info["needs_finalising"]:
            sd.finalize_checks()
        dlist.add(sd)

    is_downloaded = dlist.download(libmambapy.MAMBA_DOWNLOAD_FAILFAST)

    if not is_downloaded:
        raise RuntimeError("Error downloading repodata.")

    return index


class MambaSolver:
    def __init__(self, prefix="/tmp/mamba-test", channels=("conda-forge",), platform="osx-arm64"):

        api_ctx = libmambapy.Context()
        api_ctx.prefix_params.conda_prefix = prefix
        self.package_cache = libmambapy.MultiPackageCache([str(Path(prefix) / "conda/pkgs")])


        self.channels = channels
        self.platform = platform
        self.index = get_index(channels, platform=platform, package_cache=self.package_cache)
        self.local_index = []
        self.pool = libmambapy.Pool()
        self.repos = []

        start_prio = len(channels)
        priority = start_prio
        subpriority = 0  # wrong! :)
        for subdir, channel in self.index:
            repo = libmambapy.Repo(
                self.pool,
                str(channel),
                subdir.cache_path(),
                channel["url"],
            )
            repo.set_priority(start_prio, subpriority)
            start_prio -= 1
            self.repos.append(repo)

        self.local_repos = {}

    def solve(self, specs, constraints=None):
        """Solve given a set of specs.
        Parameters
        ----------
        specs : list of str
            A list of package specs. You can use `conda.models.match_spec.MatchSpec`
            to get them to the right form by calling
            `MatchSpec(mypec).conda_build_form()`
        Returns
        -------
        solvable : bool
            True if the set of specs has a solution, False otherwise.
        """
        solver_options = [(libmambapy.SOLVER_FLAG_ALLOW_DOWNGRADE, 1)]
        api_solver = libmambapy.Solver(self.pool, solver_options)
        _specs = specs
        for constraint in constraints or []:
            api_solver.add_constraint(constraint)

        api_solver.add_jobs(_specs, libmambapy.SOLVER_INSTALL)
        success = api_solver.try_solve()

        if not success:
            error_string = "Mamba failed to solve:\n"
            for s in _specs:
                error_string += f" - {s}\n"
            error_string += "\nwith channels:\n"
            for c in self.channels:
                error_string += f" - {c}\n"
            pstring = api_solver.problems_to_str()
            pstring = "\n".join(["   " + l for l in pstring.split("\n")])
            error_string += f"\nThe reported errors are:\n⇟{pstring}"
            print(error_string)
            exit(1)

        t = libmambapy.Transaction(self.pool, api_solver, self.package_cache)
        _, to_link, _ = t.to_conda()
        pkgs = []
        for channel, fname, jdata in to_link:
            pkg_info = json.loads(jdata)
            pkgs.append("{name} {version} {build}".format(**pkg_info))
        return pkgs

With the above MambaSolver implementation, the following:

solver = MambaSolver("/tmp/mamba-test", channels=("conda-forge",), platform="osx-arm64")

solver.solve(['zlib'], constraints=["libzlib 1.2.*"])
# ['libzlib 1.2.13 h53f4e23_5', 'zlib 1.2.13 h53f4e23_5'] - correct
solver.solve(['libzlib'], constraints=["libsodium 1.*"])
# ['libsodium 1.0.18 h27ca646_1', 'libzlib 1.3.1 h0d3ecfb_0'] incorrect, includes libsodium

I don't know if this is a libmambapy bug or a libmamba bug, or if I'm just using libmambapy wrong, but I don't see any tests of constraints actually being applied in the solver, nor any examples suggesting how it should be used.

mamba info / micromamba info

mamba version : 1.5.8
     active environment : cf-scripts
    active env location : /Users/minrk/conda/envs/cf-scripts
            shell level : 1
       user config file : /Users/minrk/.condarc
 populated config files : /Users/minrk/conda/.condarc
                          /Users/minrk/.condarc
          conda version : 24.3.0
    conda-build version : 24.3.0
         python version : 3.10.13.final.0
                 solver : libmamba (default)
       virtual packages : __archspec=1=m1
                          __conda=24.3.0=0
                          __osx=14.4.1=0
                          __unix=0=0
       base environment : /Users/minrk/conda  (writable)
      conda av data dir : /Users/minrk/conda/etc/conda
  conda av metadata url : None
           channel URLs : https://conda.anaconda.org/conda-forge/osx-arm64
                          https://conda.anaconda.org/conda-forge/noarch
          package cache : /Users/minrk/conda/pkgs
                          /Users/minrk/.conda/pkgs
       envs directories : /Users/minrk/conda/envs
                          /Users/minrk/.conda/envs
               platform : osx-arm64
             user-agent : conda/24.3.0 requests/2.31.0 CPython/3.10.13 Darwin/23.4.0 OSX/14.4.1 solver/libmamba conda-libmamba-solver/23.11.1 libmambapy/1.5.8
                UID:GID : 501:20
             netrc file : /Users/minrk/.netrc
           offline mode : False

Logs

No response

environment.yml

No response

~/.condarc

No response

[minrk]

FWIW, in regro/conda-forge-feedstock-check-solvable#31 it worked when I switched to add_pin instead of add_constraint, though I had to call repo.set_installed() to avoid an error, which I don't understand, but it worked.