sxs 2023.3.1__py3-none-any.whl → 2024.0.0__py3-none-any.whl

sxs/simulations/simulation.py

@@ -0,0 +1,521 @@
+from pathlib import Path
+from warnings import warn
+from .. import doi_url, Metadata
+from ..utilities import (
+    sxs_id_and_version, lev_number, sxs_path_to_system_path,
+    download_file, sxs_directory, read_config
+)
+
+
+def Simulation(location, *args, **kwargs):
+    """Construct a Simulation object from a location string
+
+    The location string should be an SXS ID with an optional version
+    number, and possibly a Lev specification, as in
+    "SXS:BBH:1234v2.0/Lev5".  The default version number is "v2.0",
+    and the default Lev number is the highest available.  Note that
+    the version number must be in the form "v2.0", not "2.0", and it
+    must exist, or no files will be found to load the data from.
+
+    The returned object will be either a `Simulation_v1` or
+    `Simulation_v2` object, depending on the version number.
+    Hopefully, most details of the two versions will be hidden from
+    the user, so that the interface is identical.
+
+    Note that some simulations are deprecated and/or superseded by
+    other simulations.  By default, this function will raise an error
+    if you try to load a deprecated or superseded simulation.  There
+    are several ways to override this behavior:
+
+    1. Pass `ignore_deprecation=True` to completely bypass even
+        checking for deprecation or supersession.  No warnings or
+        errors will be issued.
+    2. Include an explicit version number in the `location`
+        string, as in "SXS:BBH:0001v2.0".  A warning will be issued
+        that the simulation is deprecated, but it will be loaded
+        anyway.
+    3. Pass `auto_supersede=True` to automatically load the
+        superseding simulation, if there is only one.  Because no
+        superseding simulation can be *precisely* the same as the
+        deprecated one, there may be multiple superseding simulations
+        that have very similar parameters, in which case an error will
+        be raised and you must explicitly choose one.  If there is
+        only one, a warning will be issued, but the superseding
+        simulation will be loaded.
+
+    Otherwise, a `ValueError` will be raised, with an explanation and
+    suggestions on what you might want to do.
+
+    Parameters
+    ----------
+    location : str
+        The location string for the simulation.  This must include the
+        SXS ID, but can also include the version and Lev number, as
+        described above.
+
+    Other parameters
+    ----------------
+    ignore_deprecation : bool
+        If `True`, completely bypass checking for deprecation or
+        supersession.  No warnings or errors will be issued.
+    auto_supersede : bool
+        If `True`, automatically load the superseding simulation, if
+        there is only one.  If there are multiple superseding
+        simulations, an error will be raised, and you must explicitly
+        choose one.  If there is only one, a warning will still be
+        issued, but the superseding simulation will be loaded.  Note
+        that this can also be set in the configuration file with
+        `sxs.write_config(auto_supersede=True)`.
+    extrapolation : str
+        The extrapolation order to use for the strain and Psi4 data.
+        This is only relevant for versions 1 and 2 of the data format,
+        both of which default to "N2".  Other options include "N3",
+        "N4", and "Outer".  "Nx" refers to extrapolation by
+        polynomials in 1/r with degree `x`, while "Outer" refers to
+        data extracted at the outermost extraction radius but
+        corrected for time-dilation and areal-radius effects.
+
+    Returns
+    -------
+    simulation : SimulationBase
+        A `Simulation_v1` or `Simulation_v2` object, depending on the
+        version of the simulation data.
+
+    Note that all remaining arguments (including keyword arguments)
+    are passed on to the `SimulationBase`, `Simulation_v1`, and/or
+    `Simulation_v2` constructors, though none currently recognize any
+    arguments other than those listed above.
+    
+    """
+    from .. import load
+
+    # Extract the simulation ID, version, and Lev from the location string
+    simulation_id, input_version = sxs_id_and_version(location)
+    if not simulation_id:
+        raise ValueError(f"Invalid SXS ID in '{simulation_id}'")
+    input_lev_number = lev_number(location)  # Will be `None` if not present
+
+    # Load the simulation catalog and check if simulation ID exists in the catalog
+    simulations = load("simulations")
+    if simulation_id not in simulations:
+        raise ValueError(f"Simulation '{simulation_id}' not found in simulation catalog")
+
+    # Attach metadata to this object
+    metadata = Metadata(simulations[simulation_id])
+    series = simulations.dataframe.loc[simulation_id]
+
+    # Check if the specified version exists in the simulation catalog
+    if input_version not in metadata.DOI_versions:
+        raise ValueError(f"Version '{input_version}' not found in simulation catalog for '{simulation_id}'")
+
+    # Set various pieces of information about the simulation
+    version = input_version or max(metadata.DOI_versions)
+    if not version.startswith("v"):
+        raise ValueError(f"Invalid version string '{version}'")
+    sxs_id_stem = simulation_id
+    sxs_id = f"{sxs_id_stem}{version}"
+    url = f"{doi_url}{sxs_id}"
+
+    # Deal with "superseded_by" field, or "deprecated" keyword in the metadata
+    if not kwargs.get("ignore_deprecation", False):
+        auto_supersede = kwargs.get("auto_supersede", read_config("auto_supersede", False))
+        if (
+            input_version
+            and not auto_supersede
+            and ("deprecated" in metadata.get("keywords", []) or metadata.get("superseded_by", False))
+        ):
+            message = ("\n"
+                + f"Simulation '{sxs_id_stem}' is deprecated and/or superseded.\n"
+                + "Normally, this simulation should no longer be used, but you\n"
+                + f"explicitly requested version '{input_version}', so it is being used.\n"
+            )
+            warn(message, DeprecationWarning)
+        else:
+            if "superseded_by" in metadata:
+                superseded_by = metadata["superseded_by"]
+                if auto_supersede and isinstance(superseded_by, list):
+                    raise ValueError(
+                        f"`auto_supersede` is enabled, but simulation '{sxs_id}' is\n"
+                        + "superseded by multiple simulations.  You must choose one\n"
+                        + "explicitly from the list:\n"
+                        + "\n".join(f"  {s}" for s in superseded_by)
+                        + "\nAlternatively, you could pass `ignore_deprecation=True` or\n"
+                        + "specify a version to load this waveform anyway."
+                    )
+                elif auto_supersede and isinstance(superseded_by, str):
+                    message = f"Simulation '{sxs_id}' is being automatically superseded by '{superseded_by}'."
+                    warn(message, DeprecationWarning)
+                    new_location = f"{superseded_by}{input_version}"
+                    if input_lev_number:
+                        new_location += f"/Lev{input_lev_number}"
+                    return Simulation(new_location, *args, **kwargs)
+                elif isinstance(superseded_by, list):
+                    raise ValueError(
+                        f"Simulation '{sxs_id}' is superseded by multiple simulations.\n"
+                        + "Even if you enable `auto_supersede`, with multiple options, you\n"
+                        + "must choose one explicitly from the list:\n"
+                        + "\n".join(f"  {s}" for s in superseded_by)
+                        + "\nAlternatively, you could pass `ignore_deprecation=True` or\n"
+                        + "specify a version to load this waveform anyway."
+                    )
+                elif isinstance(superseded_by, str):
+                    raise ValueError(
+                        f"Simulation '{sxs_id}' is superseded by '{superseded_by}'.\n"
+                        + "Note that you could enable `auto_supersede` to automatically\n"
+                        + "load the superseding simulation.  Alternatively, you could\n"
+                        + "pass `ignore_deprecation=True` or specify a version to load\n"
+                        + "this waveform anyway."
+                    )
+                else:
+                    raise ValueError(
+                        f"Simulation '{sxs_id}' is superseded by '{superseded_by}'.\n"
+                        + "Note that you could pass `ignore_deprecation=True` or\n"
+                        + "specify a version to load this waveform anyway."
+                    )
+            if "deprecated" in metadata.get("keywords", []):
+                raise ValueError(
+                    f"Simulation '{sxs_id}' is deprecated but has no superseding simulation.\n"
+                    + "Note that you could pass `ignore_deprecation=True` or specify a version\n"
+                    + "to  to load this waveform anyway."
+                )
+
+    # We want to do this *after* deprecation checking, to avoid possibly unnecessary web requests
+    files = get_file_info(metadata, sxs_id)
+
+    # If Lev is given as part of `location`, use it; otherwise, use the highest available
+    lev_numbers = sorted({lev for f in files if (lev:=lev_number(f))})
+    output_lev_number = input_lev_number or max(lev_numbers)
+    location = f"{sxs_id_stem}{version}/Lev{output_lev_number}"
+
+    # Finally, figure out which version of the simulation to load and dispatch
+    version_number = float(version[1:])
+    if 1 <= version_number < 2.0:
+        return Simulation_v1(
+            metadata, series, version, sxs_id_stem, sxs_id, url, files, lev_numbers, output_lev_number, location, *args, **kwargs
+        )
+    elif 2 <= version_number < 3.0:
+        return Simulation_v2(
+            metadata, series, version, sxs_id_stem, sxs_id, url, files, lev_numbers, output_lev_number, location, *args, **kwargs
+        )
+    else:
+        raise ValueError(f"Version '{version}' not yet supported")
+
+
+class SimulationBase:
+    """Base class for Simulation objects
+
+    Note that users almost certainly never need to call this function;
+    see the `Simulation` function or `sxs.load` function instead.
+    
+    Attributes
+    ----------
+    metadata : Metadata
+        Metadata object for the simulation
+    series : pandas.Series
+        The metadata, as extracted from the `simulations.dataframe`,
+        meaning that it has columns consistent with other simulations,
+        even when the underlying Metadata objects do not.  Note that
+        `metadata` is an alias for this attribute, just based on the
+        use of that name for `simulations`, but technically `pandas`
+        distinguishes a single row like this as a `Series` object.
+    version : str
+        Version number of the simulation
+    sxs_id_stem : str
+        SXS ID without the version number or Lev
+    sxs_id : str
+        SXS ID with the version number
+    location : str
+        Location string for the simulation, including the SXS ID,
+        version number, and Lev number.
+    url : str
+        URL for the DOI of the simulation
+    files : dict
+        Dictionary of file information for the simulation.  The keys
+        are of the form "Lev5:Horizons.h5", and the values are
+        dictionaries with keys "checksum", "size", and "link".
+    lev_numbers : list
+        List of available Lev numbers for the simulation.
+    lev_number : int
+        Chosen Lev number for the simulation.
+    horizons : Horizons
+        Horizons object for the simulation
+    strain : Waveform
+        Strain Waveform object for the simulation.  Note that `h` is
+        an alias for this attribute, both of which are case
+        insensitive: `Strain` and `H` are also acceptable.
+    psi4 : Waveform
+        Psi4 Waveform object for the simulation.  Note that this
+        attribute is also case insensitive: `Psi4` is also acceptable.
+    psi3 : Waveform
+        Psi3 Waveform object for the simulation.  Note that this
+        attribute is also case insensitive: `Psi3` is also acceptable.
+        In versions 1 and 2, this attribute will raise an error
+        because the data is not available.
+    psi2 : Waveform
+        Psi2 Waveform object for the simulation.  Note that this
+        attribute is also case insensitive: `Psi2` is also acceptable.
+        In versions 1 and 2, this attribute will raise an error
+        because the data is not available.
+    psi1 : Waveform
+        Psi1 Waveform object for the simulation.  Note that this
+        attribute is also case insensitive: `Psi1` is also acceptable.
+        In versions 1 and 2, this attribute will raise an error
+        because the data is not available.
+    psi0 : Waveform
+        Psi0 Waveform object for the simulation.  Note that this
+        attribute is also case insensitive: `Psi0` is also acceptable.
+        In versions 1 and 2, this attribute will raise an error
+        because the data is not available.
+    """
+    def __init__(self,
+        metadata, series, version, sxs_id_stem, sxs_id, url, files, lev_numbers, lev_number, location,
+        *args, **kwargs
+    ):
+        self.metadata = metadata
+        self.series = series
+        self.version = version
+        self.sxs_id_stem = sxs_id_stem
+        self.sxs_id = sxs_id
+        self.url = url
+        self.files = files
+        self.lev_numbers = lev_numbers
+        self.lev_number = lev_number
+        self.location = location
+
+    def __repr__(self):
+        return f"""{type(self).__qualname__}("{self.sxs_id}")"""
+
+    def __str__(self):
+        return repr(self)
+    
+    @property
+    def dataframe(self):
+        return self.series
+
+    @property
+    def versions(self):
+        return self.metadata.DOI_versions
+
+    @property
+    def lev(self):
+        return f"Lev{self.lev_number}"
+
+    @property
+    def Lev(self):
+        return self.lev
+
+    def load_horizons(self):
+        from .. import load
+        sxs_id_path = Path(sxs_path_to_system_path(self.sxs_id))
+        horizons_path = self.horizons_path
+        horizons_location = self.files.get(horizons_path)["link"]
+        horizons_truepath = sxs_id_path / sxs_path_to_system_path(horizons_path)
+        return load(horizons_location, truepath=horizons_truepath)
+
+    @property
+    def horizons(self):
+        if not hasattr(self, "_horizons"):
+            self._horizons = self.load_horizons()
+        return self._horizons
+
+    @property
+    def strain(self):
+        if not hasattr(self, "_strain"):
+            self._strain = self.load_waveform(*self.strain_path)
+        return self._strain
+    Strain = strain
+    h = strain
+    H = strain
+
+    # I'm not entirely sure about the conjugations and factors of 2 in
+    # shear and news in our conventions.  These will have to wait for
+    # later.
+    #
+    # @property
+    # def shear(self):
+    #     if not hasattr(self, "_shear"):
+    #         self._shear = self.strain.bar / 2
+    #     return self._shear
+    # sigma = shear
+    # σ = shear
+    # Shear = shear
+    # Sigma = shear
+    # Σ = shear
+    #
+    # @property
+    # def news(self):
+    #     if not hasattr(self, "_news"):
+    #         self._news = self.strain.dot
+    #     return self._news
+    # News = news
+
+    @property
+    def psi4(self):
+        if not hasattr(self, "_psi4"):
+            self._psi4 = self.load_waveform(*self.psi4_path)
+        return self._psi4
+    Psi4 = psi4
+
+    @property
+    def psi3(self):
+        raise AttributeError(f"Psi3 is not available for version {self.version} of the data")
+    Psi3 = psi3
+
+    @property
+    def psi2(self):
+        raise AttributeError(f"Psi2 is not available for version {self.version} of the data")
+    Psi2 = psi2
+
+    @property
+    def psi1(self):
+        raise AttributeError(f"Psi1 is not available for version {self.version} of the data")
+    Psi1 = psi1
+
+    @property
+    def psi0(self):
+        raise AttributeError(f"Psi0 is not available for version {self.version} of the data")
+    Psi0 = psi0
+
+
+class Simulation_v1(SimulationBase):
+    """Simulation object for version 1 of the data format
+    
+    Note that users almost certainly never need to call this function;
+    see the `Simulation` function or `sxs.load` function instead.  See
+    also `SimulationBase` for the base class that this class inherits
+    from.
+    """
+    # We have to deal with the fact that some early file paths on
+    # Zenodo included the SXS ID as a prefix, while others did not.
+    # This means that we have to check for both possibilities in
+    # `load_horizons` and `load_waveform`.
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.extrapolation = kwargs.get("extrapolation", "N2")
+
+    @property
+    def horizons_path(self):
+        return f"{self.lev}/Horizons.h5"
+    
+    def load_horizons(self):
+        from .. import load
+        sxs_id_path = Path(sxs_path_to_system_path(self.sxs_id))
+        horizons_path = self.horizons_path
+        if horizons_path in self.files:
+            horizons_location = self.files.get(horizons_path)["link"]
+        else:
+            if (extended_horizons_path := f"{self.sxs_id_stem}/{horizons_path}") in self.files:
+                horizons_location = self.files.get(extended_horizons_path)["link"]
+            else:
+                raise ValueError(f"File '{horizons_path}' not found in simulation files")
+        horizons_truepath = sxs_id_path / sxs_path_to_system_path(horizons_path)
+        return load(horizons_location, truepath=horizons_truepath)
+
+    @property
+    def strain_path(self):
+        extrapolation = (
+            f"Extrapolated_{self.extrapolation}.dir"
+            if self.extrapolation != "Outer"
+            else "OutermostExtraction.dir"
+        )
+        return (
+            f"{self.lev}/rhOverM_Asymptotic_GeometricUnits_CoM.h5",
+            extrapolation
+        )
+
+    @property
+    def psi4_path(self):
+        extrapolation = (
+            f"Extrapolated_{self.extrapolation}.dir"
+            if self.extrapolation != "Outer"
+            else "OutermostExtraction.dir"
+        )
+        return (
+            f"{self.lev}/rMPsi4_Asymptotic_GeometricUnits_CoM.h5",
+            extrapolation
+        )
+
+    def load_waveform(self, file_name, group):
+        from .. import load
+        if file_name in self.files:
+            location = self.files.get(file_name)["link"]
+        else:
+            if (extended_file_name := f"{self.sxs_id_stem}/{file_name}") in self.files:
+                location = self.files.get(extended_file_name)["link"]
+            else:
+                raise ValueError(f"File '{file_name}' not found in simulation files")
+        sxs_id_path = Path(sxs_path_to_system_path(self.sxs_id))
+        truepath = sxs_id_path / sxs_path_to_system_path(file_name)
+        w = load(location, truepath=truepath, extrapolation_order=group)
+        w.metadata = self.metadata
+        return w
+
+
+class Simulation_v2(SimulationBase):
+    """Simulation object for version 2 of the data format
+        
+    Note that users almost certainly never need to call this function;
+    see the `Simulation` function or `sxs.load` function instead.  See
+    also `SimulationBase` for the base class that this class inherits
+    from.
+    """
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.extrapolation = kwargs.get("extrapolation", "N2")
+
+    @property
+    def horizons_path(self):
+        return f"{self.lev}:Horizons.h5"
+
+    @property
+    def strain_path(self):
+        return (
+            f"{self.lev}:Strain_{self.extrapolation}",
+            "/"
+        )
+
+    @property
+    def psi4_path(self):
+        extrapolation = (
+            f"Extrapolated_{self.extrapolation}.dir"
+            if self.extrapolation != "Outer"
+            else "OutermostExtraction.dir"
+        )
+        return (
+            f"{self.lev}:ExtraWaveforms",
+            f"/rMPsi4_Asymptotic_GeometricUnits_CoM_Mem/{extrapolation}"
+        )
+
+    def load_waveform(self, file_name, group):
+        from .. import load
+        # Note that `name` should not have the file ending on input,
+        # but we will strip it regardless with `.stem`.
+        file_name = Path(file_name).stem
+        sxs_id_path = Path(sxs_path_to_system_path(self.sxs_id))
+        h5_path = f"{file_name}.h5"
+        json_path = f"{file_name}.json"
+        h5_location = self.files.get(h5_path)["link"]
+        json_location = self.files.get(json_path)["link"]
+        h5_truepath = sxs_id_path / sxs_path_to_system_path(h5_path)
+        json_truepath = sxs_id_path / sxs_path_to_system_path(json_path)
+        if not json_truepath.exists():
+            download_file(json_location, sxs_directory("cache") / json_truepath)
+        return load(h5_location, truepath=h5_truepath, group=group, metadata=self.metadata)
+
+
+def get_file_info(metadata, sxs_id):
+    from .. import load_via_sxs_id
+    if "files" in metadata:
+        return metadata["files"]
+    truepath = Path(sxs_path_to_system_path(sxs_id)) / "zenodo_metadata.json"
+    record = load_via_sxs_id(sxs_id, "export/json", truepath=truepath)
+    entries = record["files"]["entries"]
+    return {
+        str(filename): {
+            "checksum": entry["checksum"],
+            "size": entry["size"],
+            "link": entry["links"]["content"],
+        }
+        for filename, entry in entries.items()
+    }