Coverage for python / lsst / images / cells / _provenance.py: 25%
140 statements
« prev ^ index » next coverage.py v7.14.0, created at 2026-05-15 08:42 +0000
1# This file is part of lsst-images.
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# Use of this source code is governed by a 3-clause BSD-style
10# license that can be found in the LICENSE file.
12from __future__ import annotations
14__all__ = ("CoaddProvenance", "CoaddProvenanceSerializationModel")
16from collections.abc import Iterable
17from typing import TYPE_CHECKING, Any, ClassVar
19import astropy.table
20import astropy.units as u
21import numpy as np
22import pydantic
24from .._cell_grid import CellIJ
25from .._polygon import Polygon
26from ..serialization import ArchiveTree, InputArchive, OutputArchive, TableModel
28if TYPE_CHECKING:
29 try:
30 from lsst.cell_coadds import CoaddInputs as LegacyCoaddInputs
31 from lsst.cell_coadds import MultipleCellCoadd, ObservationIdentifiers
32 from lsst.skymap import Index2D
33 except ImportError:
34 type Index2D = Any # type: ignore[no-redef]
35 type LegacyCoaddInputs = Any # type: ignore[no-redef]
36 type MultipleCellCoadd = Any # type: ignore[no-redef]
37 type ObservationIdentifiers = Any # type: ignore[no-redef]
40class CoaddProvenance:
41 """A pair of tables that record the inputs to a cell-based coadd.
43 Parameters
44 ----------
45 inputs
46 A table of {visit, detector} combinations that contribute to any cell
47 in the coadd.
48 contributions
49 A table of {visit, detector, cell} combinations that describe how an
50 observation contributed to a cell.
52 Notes
53 -----
54 This object can represent the provenance of a whole patch, a single cell,
55 or anything in between. In the single-cell case, the ``inputs`` and
56 ``contributions`` tables have the same number of rows (but may not be
57 ordered the same way!).
58 """
60 def __init__(self, inputs: astropy.table.Table, contributions: astropy.table.Table):
61 self._inputs = inputs
62 self._contributions = contributions
64 _INPUT_TABLE_COLUMNS: ClassVar[list[tuple[str, type, str]]] = [
65 ("instrument", np.object_, "Name of the instrument."),
66 ("visit", np.uint64, "ID of the visit."),
67 ("detector", np.uint16, "ID of the detector."),
68 ("physical_filter", np.object_, "Full name of the bandpass filter."),
69 ("day_obs", np.uint32, "Observation night as a YYYYMMDD integer."),
70 (
71 "polygon",
72 np.object_,
73 (
74 "Polygon that approximates the overlap of the observation and the coadd patch, "
75 "in coadd coordinates."
76 ),
77 ),
78 ]
80 _CONTRIBUTION_TABLE_COLUMNS: ClassVar[list[tuple[str, type, str, u.UnitBase | None]]] = [
81 ("cell_i", np.uint16, "Y-axis index of the cell within the patch.", None),
82 ("cell_j", np.uint16, "X-axis index of the cell within the patch.", None),
83 ("instrument", np.object_, "Name of the instrument.", None),
84 ("visit", np.uint64, "ID of the visit.", None),
85 ("detector", np.uint16, "ID of the detector.", None),
86 ("overlaps_center", np.bool_, "Whether a this observation overlaps the center of the cell.", None),
87 ("overlap_fraction", np.float64, "Fraction of the cell that is covered by the overlap region.", None),
88 ("weight", np.float64, "Weight to be used for this input in this cell.", None),
89 ("psf_shape_xx", np.float64, "Second order moments of the PSF.", u.pix**2),
90 ("psf_shape_yy", np.float64, "Second order moments of the PSF.", u.pix**2),
91 ("psf_shape_xy", np.float64, "Second order moments of the PSF.", u.pix**2),
92 (
93 "psf_shape_flag",
94 np.bool_,
95 "Flag indicating whether the PSF shape measurement was successful.",
96 None,
97 ),
98 ]
100 @classmethod
101 def make_empty_input_table(cls, n_rows: int) -> astropy.table.Table:
102 """Make an empty `inputs` table with a set number of rows."""
103 return astropy.table.Table(
104 [
105 astropy.table.Column(name=name, length=n_rows, dtype=dtype, description=description)
106 for name, dtype, description in cls._INPUT_TABLE_COLUMNS
107 ]
108 )
110 @classmethod
111 def make_empty_contribution_table(cls, n_rows: int) -> astropy.table.Table:
112 """Make an empty `contributions` table with a set number of rows."""
113 return astropy.table.Table(
114 [
115 astropy.table.Column(
116 name=name, length=n_rows, dtype=dtype, description=description, unit=unit
117 )
118 for name, dtype, description, unit in cls._CONTRIBUTION_TABLE_COLUMNS
119 ]
120 )
122 @property
123 def inputs(self) -> astropy.table.Table:
124 """A table of {visit, detector} combinations that contribute to any
125 cell in the coadd.
126 """
127 return self._inputs
129 @property
130 def contributions(self) -> astropy.table.Table:
131 """A table of {visit, detector, cell} combinations that describe how an
132 observation contributed to a cell.
133 """
134 return self._contributions
136 def __getitem__(self, cell: CellIJ) -> CoaddProvenance:
137 return self.subset([cell])
139 def subset(self, cells: Iterable[CellIJ]) -> CoaddProvenance:
140 """Return a new provenance object with just the given cells."""
141 cells_to_keep = astropy.table.Table(
142 rows=[(index.i, index.j) for index in cells],
143 names=["cell_i", "cell_j"],
144 dtype=[np.uint16, np.uint16],
145 )
146 contributions = astropy.table.join(self._contributions, cells_to_keep)
147 assert contributions.columns.keys() == {name for name, _, _, _ in self._CONTRIBUTION_TABLE_COLUMNS}
148 inputs = astropy.table.join(contributions["instrument", "visit", "detector"], self._inputs)
149 assert inputs.columns.keys() == {name for name, _, _ in self._INPUT_TABLE_COLUMNS}
150 return CoaddProvenance(inputs=inputs, contributions=contributions)
152 def serialize(self, archive: OutputArchive[Any]) -> CoaddProvenanceSerializationModel:
153 """Serialize the provenance to an output archive.
155 Parameters
156 ----------
157 archive
158 Archive to write to.
159 """
160 inputs = self._inputs.copy(copy_data=False)
161 contributions = self._contributions.copy(copy_data=False)
162 instrument = CoaddProvenanceSerializationModel._fix_str_for_serialization(
163 "instrument", inputs, contributions
164 )
165 physical_filter = CoaddProvenanceSerializationModel._fix_str_for_serialization(
166 "physical_filter", inputs
167 )
168 CoaddProvenanceSerializationModel._fix_polygon_for_serialization(inputs)
169 inputs_model = archive.add_table(inputs, name="inputs")
170 contributions_model = archive.add_table(contributions, name="contributions")
171 return CoaddProvenanceSerializationModel(
172 instrument=instrument,
173 physical_filter=physical_filter,
174 inputs=inputs_model,
175 contributions=contributions_model,
176 )
178 @staticmethod
179 def from_legacy(legacy_cell_coadd: MultipleCellCoadd) -> CoaddProvenance:
180 """Extract provenance from a legacy
181 `lsst.cell_coadds.MultipleCellCoadd` object.
182 """
183 inputs = CoaddProvenance.make_empty_input_table(len(legacy_cell_coadd.common.visit_polygons))
184 for n, (legacy_identifiers, legacy_polygon) in enumerate(
185 legacy_cell_coadd.common.visit_polygons.items()
186 ):
187 inputs["instrument"][n] = legacy_identifiers.instrument
188 inputs["visit"][n] = legacy_identifiers.visit
189 inputs["detector"][n] = legacy_identifiers.detector
190 inputs["physical_filter"][n] = legacy_identifiers.physical_filter
191 inputs["day_obs"][n] = legacy_identifiers.day_obs
192 inputs["polygon"][n] = Polygon.from_legacy(legacy_polygon)
193 n_contributions = 0
194 for legacy_cell in legacy_cell_coadd.cells.values():
195 n_contributions += len(legacy_cell.inputs)
196 contributions = CoaddProvenance.make_empty_contribution_table(n_contributions)
197 n = 0
198 for legacy_cell in legacy_cell_coadd.cells.values():
199 for legacy_identifiers, legacy_inputs in legacy_cell.inputs.items():
200 contributions["cell_i"][n] = legacy_cell.identifiers.cell.y
201 contributions["cell_j"][n] = legacy_cell.identifiers.cell.x
202 contributions["instrument"][n] = legacy_identifiers.instrument
203 contributions["visit"][n] = legacy_identifiers.visit
204 contributions["detector"][n] = legacy_identifiers.detector
205 contributions["overlaps_center"][n] = legacy_inputs.overlaps_center
206 contributions["overlap_fraction"][n] = legacy_inputs.overlap_fraction
207 contributions["weight"][n] = legacy_inputs.weight
208 contributions["psf_shape_xx"][n] = legacy_inputs.psf_shape.getIxx()
209 contributions["psf_shape_yy"][n] = legacy_inputs.psf_shape.getIyy()
210 contributions["psf_shape_xy"][n] = legacy_inputs.psf_shape.getIxy()
211 contributions["psf_shape_flag"][n] = legacy_inputs.psf_shape_flag
212 n += 1
213 return CoaddProvenance(inputs=inputs, contributions=contributions)
216class CoaddProvenanceSerializationModel(ArchiveTree):
217 """A Pydantic model used to represent a serialized `CoaddProvenance`.
219 Notes
220 -----
221 We can't rewrite the Astropy tables directly into the archive (e.g. as
222 FITS binary tables for a FITS archive), because:
224 - `str` columns are a huge pain in both Numpy and FITS;
225 - the polygon columns need to be rewritten as array-valued columns.
227 To deal with the string columns (``instrument`` and ``physical_filter``)
228 we do dictionary compression: we map each distinct value of those columns
229 to an integer, and then we save that mapping to the model while saving
230 an integer version of that column in the table. But if there is actually
231 only one value in that column (the most common case by far) we just drop
232 the column and store that value directly in the model.
233 """
235 instrument: str | dict[str, int] = pydantic.Field(
236 description=(
237 "Instrument name for all inputs to this coadd, or a mapping from "
238 "instrument name to the integer used in its place in the tables."
239 )
240 )
241 physical_filter: str | dict[str, int] = pydantic.Field(
242 description="Physical filter name for all inputs to this coadd."
243 )
244 inputs: TableModel = pydantic.Field(description="Table of all inputs to the coadd.")
245 contributions: TableModel = pydantic.Field(description="Table of per-cell contributions to the coadd.")
247 def deserialize(self, archive: InputArchive[Any]) -> CoaddProvenance:
248 """Deserialize a provenance from an input archive.
250 Parameters
251 ----------
252 archive
253 Archive to read from.
255 Notes
256 -----
257 While `CoaddProvenance.subset` can be used to filter provenance
258 information down to just certain cells, there is no advantage to be
259 had from doing this during deserialization (the table data is not
260 ordered by cell, and hence there's read-slicing we can do).
261 """
262 inputs = archive.get_table(self.inputs)
263 contributions = archive.get_table(self.contributions)
264 CoaddProvenanceSerializationModel._fix_str_for_deserialization(
265 "instrument", self.instrument, inputs, contributions
266 )
267 CoaddProvenanceSerializationModel._fix_str_for_deserialization(
268 "physical_filter", self.physical_filter, inputs
269 )
270 CoaddProvenanceSerializationModel._fix_polygon_for_deserialization(inputs)
271 for name, _, description in CoaddProvenance._INPUT_TABLE_COLUMNS:
272 inputs.columns[name].description = description
273 for name, _, description, unit in CoaddProvenance._CONTRIBUTION_TABLE_COLUMNS:
274 contributions.columns[name].description = description
275 contributions.columns[name].unit = unit
276 return CoaddProvenance(inputs=inputs, contributions=contributions)
278 @staticmethod
279 def _fix_str_for_serialization(column: str, *tables: astropy.table.Table) -> str | dict[str, int]:
280 """Rewrite a string column as an integer column or drop it.
282 Parameters
283 ----------
284 column
285 Name of the column to rewrite.
286 *tables
287 One or more astropy tables to rewrite. The first table is assumed
288 to have all values for this column that might appear in any other
289 tables.
291 Returns
292 -------
293 `str` | `dict` [`str`, `int`]
294 If there is only one unique value for this column in the first
295 table, that value (and the column will have been dropped from
296 all givne tables). If the tables are empty, the column is
297 dropped and an empty `dict` is returned. In all other cases the
298 given column is replaced with an integer column in all given
299 tables and the mapping from strings to integers is returned.
300 """
301 result: str | dict[str, int] = {name: n for n, name in enumerate(sorted(set(tables[0][column])))}
302 match len(result):
303 case 0:
304 pass
305 case 1:
306 (result,) = result.keys() # type: ignore[union-attr]
307 case _:
308 for table in tables:
309 table.columns[column] = astropy.table.Column(
310 data=[result[k] for k in table.columns[column]],
311 name=column,
312 dtype=np.uint8,
313 description=f"Integer mapped to {column} name.",
314 )
315 return result
316 # If we didn't remap to an integer (case 0 and 1 above), delete the
317 # column.
318 for table in tables:
319 del table.columns[column]
320 return result
322 @staticmethod
323 def _fix_str_for_deserialization(
324 column: str, value: str | dict[str, int], *tables: astropy.table.Table
325 ) -> None:
326 """Rewrite an integer column back to a string one.
328 Parameters
329 ----------
330 column
331 Name of the column to rewrite.
332 value
333 Value or mapping of values returned by
334 `_fix_str_for_serialization`.
335 tables
336 Tables to rewrite this column in.
337 """
338 match value:
339 case str():
340 for table in tables:
341 table.columns[column] = astropy.table.Column([value] * len(table), dtype=object)
342 case dict():
343 mapping = {v: k for k, v in value.items()}
344 for table in tables:
345 table.columns[column] = astropy.table.Column(
346 [mapping[k] for k in table[column]], dtype=object
347 )
349 @staticmethod
350 def _fix_polygon_for_serialization(inputs: astropy.table.Table) -> None:
351 """Rewrite a polygon `object` column as a pair of array-valued columns
352 and an array-size column.
354 Parameters
355 ----------
356 inputs
357 A copy of the in-memory coadd inputs table to modify in-place into
358 its serialization form.
359 """
360 max_n_vertices = max(p.n_vertices for p in inputs["polygon"])
361 inputs["n_vertices"] = astropy.table.Column(
362 [p.n_vertices for p in inputs["polygon"]],
363 name="n_vertices",
364 dtype=np.uint8,
365 description="Number of polygon vertices.",
366 )
367 inputs["x_vertices"] = astropy.table.Column(
368 name="x_vertices",
369 dtype=np.float64,
370 length=len(inputs),
371 shape=(max_n_vertices,),
372 description="X coordinates of polygon vertices, in tract coordinates.",
373 )
374 inputs["x_vertices"][:, :] = np.nan
375 inputs["y_vertices"] = astropy.table.Column(
376 name="y_vertices",
377 dtype=np.float64,
378 length=len(inputs),
379 shape=(max_n_vertices,),
380 description="Y coordinates of polygon vertices, in tract coordinates.",
381 )
382 inputs["y_vertices"][:, :] = np.nan
383 for i, polygon in enumerate(inputs["polygon"]):
384 inputs["n_vertices"][i] = polygon.n_vertices
385 inputs["x_vertices"][i][: polygon.n_vertices] = polygon.x_vertices
386 inputs["y_vertices"][i][: polygon.n_vertices] = polygon.y_vertices
387 del inputs["polygon"]
389 @staticmethod
390 def _fix_polygon_for_deserialization(inputs: astropy.table.Table) -> None:
391 """Rewrite a a pair of array-valued columns and an array-size column
392 into a polygon `object` column.
394 Parameters
395 ----------
396 inputs
397 The serialized version of the coadd inputs table, to be modified
398 in-place into its in-memory form.
399 """
400 polygons = [
401 Polygon(x_vertices=x_vertices[:n_vertices], y_vertices=y_vertices[:n_vertices])
402 for n_vertices, x_vertices, y_vertices in zip(
403 inputs["n_vertices"], inputs["x_vertices"], inputs["y_vertices"]
404 )
405 ]
406 del inputs["n_vertices"]
407 del inputs["x_vertices"]
408 del inputs["y_vertices"]
409 inputs["polygon"] = astropy.table.Column(polygons, name="polygon", dtype=np.object_)