Coverage for python/lsst/meas/algorithms/convertReferenceCatalog.py: 32%
171 statements
« prev ^ index » next coverage.py v6.5.0, created at 2022-10-27 02:15 -0700
« prev ^ index » next coverage.py v6.5.0, created at 2022-10-27 02:15 -0700
1# This file is part of meas_algorithms.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (https://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This program is free software: you can redistribute it and/or modify
10# it under the terms of the GNU General Public License as published by
11# the Free Software Foundation, either version 3 of the License, or
12# (at your option) any later version.
13#
14# This program is distributed in the hope that it will be useful,
15# but WITHOUT ANY WARRANTY; without even the implied warranty of
16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17# GNU General Public License for more details.
18#
19# You should have received a copy of the GNU General Public License
20# along with this program. If not, see <https://www.gnu.org/licenses/>.
22"""
23Convert an external reference catalog into the hierarchical triangular mesh
24(HTM) sharded LSST-style format, to be ingested into the butler.
25"""
27__all__ = ["ConvertReferenceCatalogTask", "ConvertReferenceCatalogConfig", "DatasetConfig"]
29import argparse
30import glob
31import os
32import pathlib
33import logging
35import astropy
37import lsst.afw.table
38import lsst.pipe.base
39import lsst.pex.config as pexConfig
40from lsst.daf.base import PropertyList
42from .indexerRegistry import IndexerRegistry
43from .readTextCatalogTask import ReadTextCatalogTask
44from . import convertRefcatManager
45from . import ReferenceObjectLoader
47# The most recent Indexed Reference Catalog on-disk format version.
48LATEST_FORMAT_VERSION = 1
51def addRefCatMetadata(catalog):
52 """Add metadata to a new (not yet populated) reference catalog.
54 Parameters
55 ----------
56 catalog : `lsst.afw.table.SimpleCatalog`
57 Catalog to which metadata should be attached. Will be modified
58 in-place.
59 """
60 md = catalog.getMetadata()
61 if md is None:
62 md = PropertyList()
63 md.set("REFCAT_FORMAT_VERSION", LATEST_FORMAT_VERSION)
64 catalog.setMetadata(md)
67class DatasetConfig(pexConfig.Config):
68 """Description of the on-disk storage format for the converted reference
69 catalog.
70 """
71 format_version = pexConfig.Field(
72 dtype=int,
73 doc="Version number of the persisted on-disk storage format."
74 "\nVersion 0 had Jy as flux units (default 0 for unversioned catalogs)."
75 "\nVersion 1 had nJy as flux units.",
76 default=0 # This needs to always be 0, so that unversioned catalogs are interpreted as version 0.
77 )
78 ref_dataset_name = pexConfig.Field(
79 dtype=str,
80 doc="Name of this reference catalog; this should match the name used during butler ingest.",
81 )
82 indexer = IndexerRegistry.makeField(
83 default='HTM',
84 doc='Name of indexer algoritm to use. Default is HTM',
85 )
88class ConvertReferenceCatalogConfig(pexConfig.Config):
89 dataset_config = pexConfig.ConfigField(
90 dtype=DatasetConfig,
91 doc="Configuration for reading the ingested data",
92 )
93 n_processes = pexConfig.Field(
94 dtype=int,
95 doc=("Number of python processes to use when ingesting."),
96 default=1
97 )
98 manager = pexConfig.ConfigurableField(
99 target=convertRefcatManager.ConvertRefcatManager,
100 doc="Multiprocessing manager to perform the actual conversion of values, file-by-file."
101 )
102 file_reader = pexConfig.ConfigurableField(
103 target=ReadTextCatalogTask,
104 doc='Task to use to read the files. Default is to expect text files.'
105 )
106 ra_name = pexConfig.Field(
107 dtype=str,
108 doc="Name of RA column (values in decimal degrees)",
109 )
110 dec_name = pexConfig.Field(
111 dtype=str,
112 doc="Name of Dec column (values in decimal degrees)",
113 )
114 ra_err_name = pexConfig.Field(
115 dtype=str,
116 doc="Name of RA error column",
117 optional=True,
118 )
119 dec_err_name = pexConfig.Field(
120 dtype=str,
121 doc="Name of Dec error column",
122 optional=True,
123 )
124 coord_err_unit = pexConfig.Field(
125 dtype=str,
126 doc="Unit of RA/Dec error fields (astropy.unit.Unit compatible)",
127 optional=True
128 )
129 mag_column_list = pexConfig.ListField(
130 dtype=str,
131 doc="The values in the reference catalog are assumed to be in AB magnitudes. "
132 "List of column names to use for photometric information. At least one entry is required."
133 )
134 mag_err_column_map = pexConfig.DictField(
135 keytype=str,
136 itemtype=str,
137 default={},
138 doc="A map of magnitude column name (key) to magnitude error column (value)."
139 )
140 is_photometric_name = pexConfig.Field(
141 dtype=str,
142 optional=True,
143 doc='Name of column stating if satisfactory for photometric calibration (optional).'
144 )
145 is_resolved_name = pexConfig.Field(
146 dtype=str,
147 optional=True,
148 doc='Name of column stating if the object is resolved (optional).'
149 )
150 is_variable_name = pexConfig.Field(
151 dtype=str,
152 optional=True,
153 doc='Name of column stating if the object is measured to be variable (optional).'
154 )
155 id_name = pexConfig.Field(
156 dtype=str,
157 optional=True,
158 doc='Name of column to use as an identifier (optional).'
159 )
160 pm_ra_name = pexConfig.Field(
161 dtype=str,
162 doc="Name of proper motion RA column",
163 optional=True,
164 )
165 pm_dec_name = pexConfig.Field(
166 dtype=str,
167 doc="Name of proper motion Dec column",
168 optional=True,
169 )
170 pm_ra_err_name = pexConfig.Field(
171 dtype=str,
172 doc="Name of proper motion RA error column",
173 optional=True,
174 )
175 pm_dec_err_name = pexConfig.Field(
176 dtype=str,
177 doc="Name of proper motion Dec error column",
178 optional=True,
179 )
180 pm_scale = pexConfig.Field(
181 dtype=float,
182 doc="Scale factor by which to multiply proper motion values to obtain units of milliarcsec/year",
183 default=1.0,
184 )
185 parallax_name = pexConfig.Field(
186 dtype=str,
187 doc="Name of parallax column",
188 optional=True,
189 )
190 parallax_err_name = pexConfig.Field(
191 dtype=str,
192 doc="Name of parallax error column",
193 optional=True,
194 )
195 parallax_scale = pexConfig.Field(
196 dtype=float,
197 doc="Scale factor by which to multiply parallax values to obtain units of milliarcsec",
198 default=1.0,
199 )
200 epoch_name = pexConfig.Field(
201 dtype=str,
202 doc="Name of epoch column",
203 optional=True,
204 )
205 epoch_format = pexConfig.Field(
206 dtype=str,
207 doc="Format of epoch column: any value accepted by astropy.time.Time, e.g. 'iso' or 'unix'",
208 optional=True,
209 )
210 epoch_scale = pexConfig.Field(
211 dtype=str,
212 doc="Scale of epoch column: any value accepted by astropy.time.Time, e.g. 'utc'",
213 optional=True,
214 )
215 extra_col_names = pexConfig.ListField(
216 dtype=str,
217 default=[],
218 doc='Extra columns to add to the reference catalog.'
219 )
221 def setDefaults(self):
222 # Newly ingested reference catalogs always have the latest format_version.
223 self.dataset_config.format_version = LATEST_FORMAT_VERSION
224 # gen3 refcats are all depth=7
225 self.dataset_config.indexer['HTM'].depth = 7
227 def validate(self):
228 super().validate()
230 def assertAllOrNone(*names):
231 """Raise ValueError unless all the named fields are set or are
232 all none (or blank).
233 """
234 setNames = [name for name in names if bool(getattr(self, name))]
235 if len(setNames) in (len(names), 0):
236 return
237 prefix = "Both or neither" if len(names) == 2 else "All or none"
238 raise ValueError("{} of {} must be set, but only {} are set".format(
239 prefix, ", ".join(names), ", ".join(setNames)))
241 if not (self.ra_name and self.dec_name and self.mag_column_list):
242 raise ValueError(
243 "ra_name and dec_name and at least one entry in mag_column_list must be supplied.")
244 if self.mag_err_column_map and set(self.mag_column_list) != set(self.mag_err_column_map.keys()):
245 raise ValueError(
246 "mag_err_column_map specified, but keys do not match mag_column_list: {} != {}".format(
247 sorted(self.mag_err_column_map.keys()), sorted(self.mag_column_list)))
248 assertAllOrNone("ra_err_name", "dec_err_name", "coord_err_unit")
249 if self.coord_err_unit is not None:
250 result = astropy.units.Unit(self.coord_err_unit, parse_strict='silent')
251 if isinstance(result, astropy.units.UnrecognizedUnit):
252 msg = f"{self.coord_err_unit} is not a valid astropy unit string."
253 raise pexConfig.FieldValidationError(ConvertReferenceCatalogConfig.coord_err_unit, self, msg)
255 assertAllOrNone("epoch_name", "epoch_format", "epoch_scale")
256 assertAllOrNone("pm_ra_name", "pm_dec_name")
257 assertAllOrNone("pm_ra_err_name", "pm_dec_err_name")
258 assertAllOrNone("parallax_name", "parallax_err_name")
259 if self.pm_ra_err_name and not self.pm_ra_name:
260 raise ValueError('"pm_ra/dec_name" must be specified if "pm_ra/dec_err_name" are specified')
261 if (self.pm_ra_name or self.parallax_name) and not self.epoch_name:
262 raise ValueError(
263 '"epoch_name" must be specified if "pm_ra/dec_name" or "parallax_name" are specified')
266class ConvertReferenceCatalogTask(lsst.pipe.base.Task):
267 """Class for producing HTM-indexed reference catalogs from external
268 catalog data.
270 This implements an indexing scheme based on hierarchical triangular
271 mesh (HTM). The term index really means breaking the catalog into
272 localized chunks called shards. In this case each shard contains
273 the entries from the catalog in a single HTM trixel
275 For producing catalogs this task makes the following assumptions
276 about the input catalogs:
278 - RA, Dec are in decimal degrees.
279 - Epoch is available in a column, in a format supported by astropy.time.Time.
280 - There are no off-diagonal covariance terms, such as covariance
281 between RA and Dec, or between PM RA and PM Dec. Support for such
282 covariance would have to be added to to the config, including consideration
283 of the units in the input catalog.
285 Parameters
286 ----------
287 output_dir : `str`
288 The path to write the output files to, in a subdirectory defined by
289 ``DatasetConfig.ref_dataset_name``.
290 """
291 canMultiprocess = False
292 ConfigClass = ConvertReferenceCatalogConfig
293 _DefaultName = 'ConvertReferenceCatalogTask'
295 def __init__(self, *, output_dir=None, **kwargs):
296 super().__init__(**kwargs)
297 if output_dir is None:
298 raise RuntimeError("Must specify output_dir.")
299 self.base_dir = output_dir
300 self.output_dir = os.path.join(output_dir, self.config.dataset_config.ref_dataset_name)
301 self.ingest_table_file = os.path.join(self.base_dir, "filename_to_htm.ecsv")
302 self.indexer = IndexerRegistry[self.config.dataset_config.indexer.name](
303 self.config.dataset_config.indexer.active)
304 self.makeSubtask('file_reader')
306 def run(self, inputFiles):
307 """Index a set of files comprising a reference catalog.
309 Outputs are persisted in the butler repository.
311 Parameters
312 ----------
313 inputFiles : `list`
314 A list of file paths to read.
315 """
316 # Create the output path, if it doesn't exist; fail if the path exists:
317 # we don't want to accidentally append to existing files.
318 pathlib.Path(self.output_dir).mkdir(exist_ok=False)
320 schema, key_map = self._writeMasterSchema(inputFiles[0])
321 # create an HTM we can interrogate about pixel ids
322 htm = lsst.sphgeom.HtmPixelization(self.indexer.htm.get_depth())
323 filenames = self._getOutputFilenames(htm)
324 worker = self.config.manager.target(filenames,
325 self.config,
326 self.file_reader,
327 self.indexer,
328 schema,
329 key_map,
330 htm.universe()[0],
331 addRefCatMetadata,
332 self.log)
333 result = worker.run(inputFiles)
335 self._writeConfig()
336 self._writeIngestHelperFile(result)
338 def _writeIngestHelperFile(self, result):
339 """Write the astropy table containing the htm->filename relationship,
340 used for the ``butler ingest-files`` command after this task completes.
341 """
342 dimension = f"htm{self.config.dataset_config.indexer.active.depth}"
343 table = astropy.table.Table(names=("filename", dimension), dtype=('str', 'int'))
344 for key in result:
345 table.add_row((result[key], key))
346 table.write(self.ingest_table_file)
348 def _writeConfig(self):
349 """Write the config that was used to generate the refcat."""
350 filename = os.path.join(self.output_dir, "config.py")
351 with open(filename, 'w') as file:
352 self.config.dataset_config.saveToStream(file)
354 def _getOutputFilenames(self, htm):
355 """Get filenames from the butler for each output htm pixel.
357 Parameters
358 ----------
359 htm : `lsst.sphgeom.HtmPixelization`
360 The HTM pixelization scheme to be used to build filenames.
362 Returns
363 -------
364 filenames : `list [str]`
365 List of filenames to write each HTM pixel to.
366 """
367 filenames = {}
368 start, end = htm.universe()[0]
369 path = os.path.join(self.output_dir, f"{self.indexer.htm}.fits")
370 base = os.path.join(os.path.dirname(path), "%d"+os.path.splitext(path)[1])
371 for pixelId in range(start, end):
372 filenames[pixelId] = base % pixelId
374 return filenames
376 def makeSchema(self, dtype):
377 """Make the schema to use in constructing the persisted catalogs.
379 Parameters
380 ----------
381 dtype : `numpy.dtype`
382 Data type describing each entry in ``config.extra_col_names``
383 for the catalogs being ingested.
385 Returns
386 -------
387 schemaAndKeyMap : `tuple` of (`lsst.afw.table.Schema`, `dict`)
388 A tuple containing two items:
389 - The schema for the output source catalog.
390 - A map of catalog keys to use in filling the record
391 """
392 # make a schema with the standard fields
393 schema = ReferenceObjectLoader.makeMinimalSchema(
394 filterNameList=self.config.mag_column_list,
395 addCentroid=False,
396 addIsPhotometric=bool(self.config.is_photometric_name),
397 addIsResolved=bool(self.config.is_resolved_name),
398 addIsVariable=bool(self.config.is_variable_name),
399 coordErrDim=2 if bool(self.config.ra_err_name) else 0,
400 addProperMotion=2 if bool(self.config.pm_ra_name) else 0,
401 properMotionErrDim=2 if bool(self.config.pm_ra_err_name) else 0,
402 addParallax=bool(self.config.parallax_name),
403 )
404 keysToSkip = set(("id", "centroid_x", "centroid_y", "hasCentroid"))
405 key_map = {fieldName: schema[fieldName].asKey() for fieldName in schema.getOrderedNames()
406 if fieldName not in keysToSkip}
408 def addField(name):
409 if dtype[name].kind == 'U':
410 # dealing with a string like thing. Need to get type and size.
411 at_size = dtype[name].itemsize
412 return schema.addField(name, type=str, size=at_size)
413 else:
414 at_type = dtype[name].type
415 return schema.addField(name, at_type)
417 for col in self.config.extra_col_names:
418 key_map[col] = addField(col)
419 return schema, key_map
421 def _writeMasterSchema(self, inputfile):
422 """Generate and save the master catalog schema.
424 Parameters
425 ----------
426 inputfile : `str`
427 An input file to read to get the input dtype.
428 """
429 arr = self.file_reader.run(inputfile)
430 schema, key_map = self.makeSchema(arr.dtype)
432 catalog = lsst.afw.table.SimpleCatalog(schema)
433 addRefCatMetadata(catalog)
434 outputfile = os.path.join(self.output_dir, "master_schema.fits")
435 catalog.writeFits(outputfile)
436 return schema, key_map
438 def _reduce_kwargs(self):
439 # Need to be able to pickle this class to use the multiprocess manager.
440 kwargs = super()._reduce_kwargs()
441 kwargs['output_dir'] = self.base_dir
442 return kwargs
445def build_argparser():
446 """Construct an argument parser for the ``convertReferenceCatalog`` script.
448 Returns
449 -------
450 argparser : `argparse.ArgumentParser`
451 The argument parser that defines the ``convertReferenceCatalog``
452 command-line interface.
453 """
454 parser = argparse.ArgumentParser(
455 description=__doc__,
456 formatter_class=argparse.RawDescriptionHelpFormatter,
457 epilog='More information is available at https://pipelines.lsst.io.'
458 )
459 parser.add_argument("outputDir",
460 help="Path to write the output shard files, configs, and `ingest-files` table to.")
461 parser.add_argument("configFile",
462 help="File containing the ConvertReferenceCatalogConfig fields.")
463 # Use a "+"-list here, so we can produce a more useful error if the user
464 # uses an unquoted glob that gets shell expanded.
465 parser.add_argument("fileglob", nargs="+",
466 help="Quoted glob for the files to be read in and converted."
467 " Example (note required quotes to prevent shell expansion):"
468 ' "gaia_source/csv/GaiaSource*"')
469 return parser
472def run_convert(outputDir, configFile, fileglob):
473 """Run `ConvertReferenceCatalogTask` on the input arguments.
475 Parameters
476 ----------
477 outputDir : `str`
478 Path to write the output files to.
479 configFile : `str`
480 File specifying the ``ConvertReferenceCatalogConfig`` fields.
481 fileglob : `str`
482 Quoted glob for the files to be read in and converted.
483 """
484 # We have to initialize the logger manually when running from the commandline.
485 logging.basicConfig(level=logging.INFO, format="{name} {levelname}: {message}", style="{")
487 config = ConvertReferenceCatalogTask.ConfigClass()
488 config.load(configFile)
489 converter = ConvertReferenceCatalogTask(output_dir=outputDir, config=config)
490 files = glob.glob(fileglob)
491 converter.run(files)
492 with open(os.path.join(outputDir, "convertReferenceCatalogConfig.py"), "w") as outfile:
493 converter.config.saveToStream(outfile)
494 msg = ("Completed refcat conversion.\n\n"
495 "Ingest the resulting files with the following commands, substituting the path\n"
496 "to your butler repo for `REPO`, and the ticket number you are tracking this\n"
497 "ingest on for `DM-NNNNN`:\n"
498 f"\n butler register-dataset-type REPO {config.dataset_config.ref_dataset_name} "
499 "SimpleCatalog htm7"
500 "\n butler ingest-files -t direct REPO gaia_dr2 refcats/DM-NNNNN "
501 f"{converter.ingest_table_file}"
502 "\n butler collection-chain REPO --mode extend refcats refcats/DM-NNNNN")
503 print(msg)
506def main():
507 args = build_argparser().parse_args()
508 if len(args.fileglob) > 1:
509 raise RuntimeError("Final argument must be a quoted file glob, not a shell-expanded list of files.")
510 # Fileglob comes out as a length=1 list, so we can test it above.
511 run_convert(args.outputDir, args.configFile, args.fileglob[0])