Simplify CLI with hydra instantiate (#230)

* update poetry

* more error checking in draw_blends for MR

* bug in this line when using basic measure function and MR

* catch error related to meas_band_num in metrics

* use hydra instantiate and call rather than available_*

* update documentation

* test MR on main, remove error checking now done by hydra

* Revert "update poetry"

This reverts commit aad74ea.

btk/draw_blends.py

@@ -3,6 +3,7 @@
 import os
 from abc import ABC
 from abc import abstractmethod
+from collections.abc import Iterable
 from itertools import chain
 
 import galsim
@@ -169,14 +170,20 @@ def __init__(
         if isinstance(surveys, Survey):
             self.surveys = [surveys]
             self.check_compatibility(surveys)
-        elif isinstance(surveys, list):
+        elif isinstance(surveys, Iterable):
             for s in surveys:
                 if not isinstance(s, Survey):
-                    raise TypeError("surveys must be a Survey object or a list of Survey objects.")
+                    raise TypeError(
+                        f"surveys must be a Survey object or an Iterable of Survey objects, but"
+                        f"Iterable contained object of type {type(s)}"
+                    )
                 self.check_compatibility(s)
             self.surveys = surveys
         else:
-            raise TypeError("surveys must be a Survey object or a list of Survey objects.")
+            raise TypeError(
+                f"surveys must be a Survey object or an Iterable of Survey objects,"
+                f"but surveys is type {type(surveys)}"
+            )
         self.is_multiresolution = len(self.surveys) > 1
 
         self.stamp_size = stamp_size
@@ -643,10 +650,3 @@ def render_single(self, entry, filt, psf, survey, extra_data):
             nx=pix_stamp_size, ny=pix_stamp_size, scale=survey.pixel_scale, dtype=np.float64
         )
         return galaxy_image
-
-
-available_draw_blends = {
-    "CatsimGenerator": CatsimGenerator,
-    "CosmosGenerator": CosmosGenerator,
-    "GalsimHubGenerator": GalsimHubGenerator,
-}

btk/main.py

@@ -1,44 +1,21 @@
 """Implements main function to run BTK end-to-end."""
 from collections.abc import Iterable
 
+from hydra.utils import instantiate
 from omegaconf import OmegaConf
 
-from btk.catalog import available_catalogs
-from btk.draw_blends import available_draw_blends
 from btk.measure import available_measure_functions
 from btk.measure import MeasureGenerator
 from btk.metrics import MetricsGenerator
-from btk.sampling_functions import available_sampling_functions
 from btk.survey import _get_survey_from_cfg
 
 
 def main(cfg: OmegaConf):
     """Run BTK from end-to-end using a hydra configuration object."""
-    # get catalog
-    if cfg.catalog.name not in available_catalogs:
-        raise ValueError(f"Catalog '{cfg.catalog.name}' is not implemented in BTK.")
-    catalog = available_catalogs[cfg.catalog.name].from_file(cfg.catalog.catalog_files)
-
-    # get sampling function
-    if cfg.sampling.name not in available_sampling_functions:
-        raise ValueError(f"Sampling function '{cfg.sampling.name}' is not implemented in BTK.")
-    sampling_function = available_sampling_functions[cfg.sampling.name](**cfg.sampling.kwargs)
-
-    # get survey(s) to be used.
-    if not isinstance(cfg.surveys, Iterable):
-        cfg.surveys = [cfg.surveys]
-
-    surveys = []
-    for survey_name in cfg.surveys:
-        survey = _get_survey_from_cfg(cfg.surveys[survey_name])
-        surveys.append(survey)
+    surveys = [_get_survey_from_cfg(cfg.surveys[survey_name]) for survey_name in cfg.surveys]
 
     # get draw blends generator.
-    if cfg.draw_blends.name not in available_draw_blends:
-        raise ValueError("DrawBlendGenerator specified is not implemented in BTK.")
-    draw_blend_generator = available_draw_blends[cfg.draw_blends.name](
-        catalog, sampling_function, surveys, **cfg.draw_blends.kwargs
-    )
+    draw_blend_generator = instantiate(cfg.draw_blends, surveys=surveys)
 
     # get measure_functions.
     measure_functions = []

btk/measure.py

@@ -111,7 +111,6 @@ def basic_measure(
     if is_multiresolution:
         if surveys is None:
             raise ValueError("surveys are required in order to use the MR feature.")
-        surveys = kwargs.get("surveys", None)
         survey_name = surveys[0].name
         coadd = np.mean(batch["blend_images"][survey_name][idx], axis=channel_indx)
         wcs = batch["wcs"][survey_name]

btk/metrics.py

@@ -53,6 +53,7 @@
 
 """
 import os
+from collections.abc import Iterable
 
 import astropy.table
 import galsim
@@ -730,20 +731,31 @@ def __next__(self):
         """Returns metric results calculated on one batch."""
         blend_results, measure_results = next(self.measure_generator)
         surveys = self.measure_generator.draw_blend_generator.surveys
+        meas_band_num = self.meas_band_num
 
         metrics_results = {}
         for meas_func in measure_results["catalog"].keys():
             if self.is_multiresolution:
+                if not isinstance(meas_band_num, Iterable) and hasattr(meas_band_num, "__len__"):
+                    raise ValueError(
+                        f"meas_band_num is required to be an finite length iterable,"
+                        f"instead type was {type(meas_band_num)}"
+                    )
+                if not len(meas_band_num) == len(surveys):
+                    raise ValueError(
+                        f"meas_band_num should be an iterable of exactly the same len as the"
+                        f"number of surveys: {len(surveys)}, instead len is {len(meas_band_num)}"
+                    )
                 metrics_results_f = {}
                 for i, surv in enumerate(blend_results["isolated_images"].keys()):
                     additional_params = {
                         "psf": blend_results["psf"][surv],
                         "pixel_scale": surveys[i].pixel_scale,
-                        "meas_band_num": self.meas_band_num[i],
+                        "meas_band_num": meas_band_num[i],
                         "verbose": self.verbose,
                     }
                     noise_threshold = self.noise_threshold_factor * np.sqrt(
-                        get_mean_sky_level(surveys[i], surveys[i].filters[self.meas_band_num[i]])
+                        get_mean_sky_level(surveys[i], surveys[i].filters[meas_band_num[i]])
                     )
                     target_meas = {}
                     for k in self.target_meas.keys():
@@ -756,7 +768,7 @@ def __next__(self):
                         measure_results["segmentation"][meas_func][surv],
                         measure_results["deblended_images"][meas_func][surv],
                         noise_threshold,
-                        self.meas_band_num[i],
+                        meas_band_num[i],
                         target_meas,
                         channels_last=self.measure_generator.channels_last,
                         save_path=os.path.join(self.save_path, meas_func, surv)
@@ -771,11 +783,11 @@ def __next__(self):
                 additional_params = {
                     "psf": blend_results["psf"],
                     "pixel_scale": surveys[0].pixel_scale,
-                    "meas_band_num": self.meas_band_num,
+                    "meas_band_num": meas_band_num,
                     "verbose": self.verbose,
                 }
                 noise_threshold = self.noise_threshold_factor * np.sqrt(
-                    get_mean_sky_level(surveys[0], surveys[0].filters[self.meas_band_num])
+                    get_mean_sky_level(surveys[0], surveys[0].filters[meas_band_num])
                 )
                 target_meas = {}
                 for k in self.target_meas.keys():
@@ -788,7 +800,7 @@ def __next__(self):
                     measure_results["segmentation"][meas_func],
                     measure_results["deblended_images"][meas_func],
                     noise_threshold,
-                    self.meas_band_num,
+                    meas_band_num,
                     target_meas,
                     channels_last=self.measure_generator.channels_last,
                     save_path=os.path.join(self.save_path, meas_func, surveys[0].name)

btk/sampling_functions.py

@@ -437,11 +437,3 @@ def __call__(self, table, **kwargs):
         )
         assert len(no_boundary) <= self.max_number, message
         return no_boundary
-
-
-available_sampling_functions = {
-    "DefaultSampling": DefaultSampling,
-    "DefaultSamplingGalsimHub": DefaultSamplingGalsimHub,
-    "GroupSampling": GroupSampling,
-    "GroupSamplingNumbered": GroupSamplingNumbered,
-}

conf/catalog/catsim.yaml

@@ -0,0 +1,2 @@
+_target_: btk.catalog.CatsimCatalog.from_file
+catalog_files: ${paths.data}/sample_input_catalog.fits

conf/catalog/cosmos.yaml

@@ -0,0 +1,4 @@
+name: btk.catalog.CosmosCatalog.from_file
+catalog_files:
+  - ${paths.data}/cosmos/real_galaxy_catalog_23.5_example.fits
+  - ${paths.data}/cosmos/real_galaxy_catalog_23.5_example_fits.fits

conf/config.yaml

@@ -1,15 +1,12 @@
 defaults:
   - _self_
+  - catalog: catsim
   - sampling: default
   - draw_blends: catsim
   - surveys:
       - Rubin
   - override hydra/help: btk_help
 
-catalog:
-  name: catsim
-  catalog_files: ${paths.data}/sample_input_catalog.fits
-
 ### all arguments ###
 
 # system

conf/draw_blends/catsim.yaml

@@ -1,11 +1,13 @@
-name: "CatsimGenerator"
-kwargs:
-  batch_size: ${batch_size}
-  stamp_size: ${stamp_size}
-  cpus: ${cpus}
-  verbose: ${verbose}
-  add_noise: ${add_noise}
-  shifts: Null
-  indexes: Null
-  channels_last: ${channels_last}
-  save_path: ${save_path}
+_target_: btk.draw_blends.CatsimGenerator
+catalog: ${catalog}
+sampling_function: ${sampling}
+surveys: Null
+batch_size: ${batch_size}
+stamp_size: ${stamp_size}
+cpus: ${cpus}
+verbose: ${verbose}
+add_noise: ${add_noise}
+shifts: Null
+indexes: Null
+channels_last: ${channels_last}
+save_path: ${save_path}

conf/draw_blends/cosmos.yaml

@@ -1,11 +1,13 @@
-name: "CosmosGenerator"
-kwargs:
-  batch_size: ${batch_size}
-  stamp_size: ${stamp_size}
-  cpus: ${cpus}
-  verbose: ${verbose}
-  add_noise: ${add_noise}
-  shifts: Null
-  indexes: Null
-  channels_last: ${channels_last}
-  save_path: ${save_path}
+_target_: btk.draw_blends.CosmosGenerator
+catalog: ${catalog}
+sampling_function: ${sampling}
+surveys: Null
+batch_size: ${batch_size}
+stamp_size: ${stamp_size}
+cpus: ${cpus}
+verbose: ${verbose}
+add_noise: ${add_noise}
+shifts: Null
+indexes: Null
+channels_last: ${channels_last}
+save_path: ${save_path}

conf/draw_blends/galsim_hub.yaml

@@ -1,15 +1,15 @@
-name: "GalsimHubGenerator"
-kwargs:
-  catalog: ${catalog}
-  surveys: Null
-  batch_size: ${batch_size}
-  stamp_size: ${stamp_size}
-  cpus: ${cpus}
-  verbose: ${verbose}
-  add_noise: ${add_noise}
-  shifts: Null
-  indexes: Null
-  channels_last: ${channels_last}
-  galsim_hub_model: ${galsim_hub_model}
-  param_names: ${param_names}
-  save_path: ${save_path}
+_target_: btk.draw_blends.GalsimHubGenerator
+catalog: ${catalog}
+sampling: ${sampling}
+surveys: Null
+batch_size: ${batch_size}
+stamp_size: ${stamp_size}
+cpus: ${cpus}
+verbose: ${verbose}
+add_noise: ${add_noise}
+shifts: Null
+indexes: Null
+channels_last: ${channels_last}
+galsim_hub_model: ${galsim_hub_model}
+param_names: ${param_names}
+save_path: ${save_path}

conf/hydra/help/btk_help.yaml

@@ -27,7 +27,7 @@ template: |-
   Assuming that BTK has been pip installed, you can run btk from the command line like e.g.
 
   btk sampling=default draw_blends=catsim max_number=3 save_path=/directory/to/save/results cpus=1
-    verbose=False surveys=[Rubin, HST] surveys.Rubin.airmass=1.1
+    verbose=False surveys=[Rubin,HST] surveys.Rubin.airmass=1.1
     sampling=default catalog.name=catsim use_metrics=['detection', 'segmentation'] (...)
 
   You need to create the directory to save results yourself (preferably an empty directory) and specify its absolute path when you run the CLI via the `save_path` parameter.
@@ -39,17 +39,17 @@ template: |-
       {default, galsim_hub, group_sampling, group_sampling_numbered}
 
   * catalog: Attribute group consisting of two sub-attributes.
-      * catalog.name: Name of the BTK catalog class, options: {catsim, cosmos}
+      * catalog: Name of the BTK catalog class, options: {catsim, cosmos}
       * catalog.catalog_files: Path to files containing catalog information. The 'catsim'
-          catalog requires one path, while the `cosmos` type requires two paths specified as
+          catalog requires one path, while the 'cosmos' type requires two paths specified as
           a list. (see documentation for more details).
 
     * surveys: Name of the survey(s) you want to use, options are
       {Rubin, HST, HSC, DES, CFHT, Euclid} and correspond to each of the config files available
       in conf/surveys. You can pass in a list of surveys for multi-resolution
       studies too. For example:
 
-      btk surveys=[Rubin, HST] (...)
+      btk surveys=[Rubin,HST] (...)
 
       Assuming that you want to use e.g. the Rubin survey default parameters but with a couple
       of changes, you can modify individual parameters of a given survey directly from the
@@ -83,14 +83,14 @@ template: |-
     * galsim_hub: Attribute group consisting of options:
         * model: Which galsim_hub model to use (default: 'hub:Lanusse2020')
         * param_names: list of the parameters with which the generation is parametrized; this
-          is unique to each model (default: `['flux_radious, 'mag_auto', 'zphot']`).
+          is unique to each model (default: `['flux_radius', 'mag_auto', 'zphot']`).
 
-        NOTE: This flag is only used when `draw_blends=galsim_hub`
+        NOTE: This flag has no effect unless `draw_blends=galsim_hub`
 
     * measure_kwargs: Dictionary or list of dictionaries containing the keyword arguments to be
         passed in to each measure_function.
 
-    * measure_functions: List of measure_functions to be ran, options {'basic', 'sep'}.
+    * measure_functions: List of measure_functions to be ran, options {basic, sep}.
 
     * use_metrics: List of metrics to return, options are:
       {'detection', 'segmentation', 'reconstruction'}

conf/sampling/default.yaml

@@ -1,5 +1,4 @@
-name: "DefaultSampling"
-kwargs:
-  max_number: ${max_number}
-  stamp_size: ${stamp_size}
-  maxshift: ${max_shift}
+_target_: btk.sampling_functions.DefaultSampling
+max_number: ${max_number}
+stamp_size: ${stamp_size}
+maxshift: ${max_shift}

conf/sampling/galsim_hub.yaml

@@ -1,5 +1,4 @@
-name: "DefaultSamplingGalsimHub"
-kwargs:
-  max_number: ${max_number}
-  stamp_size: ${stamp_size}
-  maxshift: ${max_shift}
+_target_: btk.sampling_functions.DefaultSamplingGalsimHub
+max_number: ${max_number}
+stamp_size: ${stamp_size}
+maxshift: ${max_shift}

conf/sampling/group_sampling.yaml

@@ -1,7 +1,6 @@
-name: "GroupSamplingFunction"
-kwargs:
-  max_number: ${max_number}
-  stamp_size: ${stamp_size}
-  maxshift: ${max_shift}
-  wld_catalog_name: ${paths.data}/sample_group_catalog.fits
-  shift: Null
+_target_: btk.sampling_functions.GroupSampling
+max_number: ${max_number}
+stamp_size: ${stamp_size}
+maxshift: ${max_shift}
+wld_catalog_name: ${paths.data}/sample_group_catalog.fits
+shift: Null

conf/sampling/group_sampling_numbered.yaml

@@ -1,7 +1,6 @@
-name: "GroupSamplingFunctionNumbered"
-kwargs:
-  max_number: ${max_number}
-  stamp_size: ${stamp_size}
-  maxshift: ${max_shift}
-  wld_catalog_name: ${paths.data}/sample_group_catalog.fits
-  shift: Null
+_target_: btk.sampling_functions.GroupSamplingNumbered
+max_number: ${max_number}
+stamp_size: ${stamp_size}
+maxshift: ${max_shift}
+wld_catalog_name: ${paths.data}/sample_group_catalog.fits
+shift: Null