Ingest parallaxes from Ultracool sheet

scripts/ingests/ultracool_sheet/Ingest_Parallax.py

@@ -0,0 +1,101 @@
+from astrodb_utils import (
+    load_astrodb,
+    find_source_in_db,
+    AstroDBError,
+)
+import sys
+
+sys.path.append(".")
+import logging
+from astropy.io import ascii
+from simple.schema import Photometry
+from simple.schema import REFERENCE_TABLES
+from math import isnan
+import sqlalchemy.exc
+from simple.utils.astrometry import ingest_parallax
+from scripts.ingests.ultracool_sheet.references import uc_ref_to_simple_ref
+
+logger = logging.getLogger("AstroDB")
+logger.setLevel(logging.DEBUG)
+
+DB_SAVE = False
+RECREATE_DB = True
+db = load_astrodb(
+    "SIMPLE.sqlite", recreatedb=RECREATE_DB, reference_tables=REFERENCE_TABLES
+)
+
+
+# Load Ultracool sheet
+doc_id = "1i98ft8g5mzPp2DNno0kcz4B9nzMxdpyz5UquAVhz-U8"
+sheet_id = "361525788"
+link = (
+    f"https://docs.google.com/spreadsheets/d/{doc_id}/export?format=csv&gid={sheet_id}"
+)
+
+# read the csv data into an astropy table
+uc_sheet_table = ascii.read(
+    link,
+    format="csv",
+    data_start=1,
+    header_start=0,
+    guess=False,
+    fast_reader=False,
+    delimiter=",",
+)
+
+no_sources = 0
+multiple_sources = 0
+ingested = 0
+already_exists = 0
+
+# Ingest loop
+for source in uc_sheet_table:
+    if isnan(source["plx_lit"]):  # skip if no data
+        continue
+    uc_sheet_name = source["name"]
+    match = find_source_in_db(
+        db,
+        uc_sheet_name,
+        ra=source["ra_j2000_formula"],
+        dec=source["dec_j2000_formula"],
+    )
+
+    if len(match) == 1:
+        # 1 Match found. INGEST!
+        simple_source = match[0]
+        logger.info(f"Match found for {uc_sheet_name}: {simple_source}")
+
+        try:
+            references = source["ref_plx_lit"].split(";")
+            comment = None
+            if len(references) > 1:
+                comment = f"other references: {uc_ref_to_simple_ref(db, references[1])}"
+            ingest_parallax(
+                db,
+                simple_source,
+                source["plx_lit"],
+                source["plxerr_lit"],
+                uc_ref_to_simple_ref(db, references[0]),
+                comment,
+            )
+            ingested += 1
+        except AstroDBError as e:
+            msg = "ingest failed with error: " + str(e)
+            if "Duplicate measurement exists" not in str(e):
+                logger.warning(msg)
+                raise AstroDBError(msg) from e
+            already_exists += 1
+    elif len(match) == 0:
+        no_sources += 1
+    else:
+        multiple_sources += 1
+
+
+# 1108 data points in UC sheet in total
+logger.info(f"ingested:{ingested}")  # 899 ingested
+logger.info(f"already exists:{already_exists}")  # skipped 463 due to preexisting data
+logger.info(f"no sources:{no_sources}")  # skipped 129 due to 0 matches
+logger.info(f"multiple sources:{multiple_sources}")  # skipped 0 due to multiple matches
+logger.info(f"total:{ingested+already_exists+no_sources+multiple_sources}")  # 1491
+if DB_SAVE:
+    db.save_database(directory="data/")

scripts/ingests/ultracool_sheet/Ingest_spitzer_photometry.py

@@ -17,7 +17,7 @@
 logger = logging.getLogger("AstroDB")
 logger.setLevel(logging.INFO)
 
-DB_SAVE = True
+DB_SAVE = False
 RECREATE_DB = True
 db = load_astrodb(
     "SIMPLE.sqlite", recreatedb=RECREATE_DB, reference_tables=REFERENCE_TABLES

simple/utils/astrometry.py

@@ -2,6 +2,7 @@
 from typing import Optional, Union
 from sqlalchemy import and_
 import sqlalchemy.exc
+from simple.schema import Parallaxes
 from astropy.units import Quantity
 from astropy.table import Table
 from astrodbkit2.astrodb import Database
@@ -181,6 +182,103 @@ def ingest_parallaxes(db, sources, plxs, plx_errs, plx_refs, comments=None):
     return
 
 
+def ingest_parallax(
+    db,
+    source: str = None,
+    parallax: float = None,
+    plx_error: float = None,
+    reference: str = None,
+    comment: str = None,
+):
+    # Search for existing parallax data and determine if this is the best
+    # If no previous measurement exists, set the new one to the Adopted measurement
+    adopted = None
+    source_plx_data: Table = (
+        db.query(db.Parallaxes).filter(db.Parallaxes.c.source == source).table()
+    )
+
+    if source_plx_data is None or len(source_plx_data) == 0:
+        # if there's no other measurements in the database,
+        # set new data Adopted = True
+        adopted = True
+        # old_adopted = None  # not used
+        logger.debug("No other measurement")
+    elif len(source_plx_data) > 0:  # Parallax data already exists
+        # check for duplicate measurement
+        dupe_ind = source_plx_data["reference"] == reference
+        if sum(dupe_ind):
+            logger.debug(f"Duplicate measurement\n, {source_plx_data[dupe_ind]}")
+            msg = "Duplicate measurement exists with same reference"
+            raise AstroDBError(msg)
+        else:
+            logger.debug("!!! Another parallax measurement exists,")
+            if logger.level == 10:
+                source_plx_data.pprint_all()
+
+        # check for previous adopted measurement and find new adopted
+        adopted_ind = source_plx_data["adopted"] == 1
+        if sum(adopted_ind):
+            old_adopted = source_plx_data[adopted_ind]
+            # if errors of new data are less than other measurements,
+            # set Adopted = True.
+            if plx_error < min(source_plx_data["parallax_error"]):
+                adopted = True
+
+                # unset old adopted
+                if old_adopted:
+                    with db.engine.connect() as conn:
+                        conn.execute(
+                            db.Parallaxes.update()
+                            .where(
+                                and_(
+                                    db.Parallaxes.c.source == old_adopted["source"][0],
+                                    db.Parallaxes.c.reference
+                                    == old_adopted["reference"][0],
+                                )
+                            )
+                            .values(adopted=False)
+                        )
+                        conn.commit()
+                    # check that adopted flag is successfully changed
+                    old_adopted_data = (
+                        db.query(db.Parallaxes)
+                        .filter(
+                            and_(
+                                db.Parallaxes.c.source == old_adopted["source"][0],
+                                db.Parallaxes.c.reference
+                                == old_adopted["reference"][0],
+                            )
+                        )
+                        .table()
+                    )
+                    logger.debug("Old adopted measurement unset")
+                    if logger.level == 10:
+                        old_adopted_data.pprint_all()
+            else:
+                adopted = False
+            logger.debug(f"The new measurement's adopted flag is:, {adopted}")
+    else:
+        msg = "Unexpected state"
+        logger.error(msg)
+        raise RuntimeError(msg)
+
+    # Construct data to be added
+    parallax_data = {
+        "source": source,
+        "parallax": parallax,
+        "parallax_error": plx_error,
+        "reference": reference,
+        "adopted": adopted,
+        "comments": comment,
+    }
+
+    plx_obj = Parallaxes(**parallax_data)
+    with db.session as session:
+        session.add(plx_obj)
+        session.commit()
+    logger.info(f" Photometry added to database: {parallax_data}\n")
+
+
 # PROPER MOTIONS
 def ingest_proper_motions(
     db, sources, pm_ras, pm_ra_errs, pm_decs, pm_dec_errs, pm_references

tests/test_astrometry.py

@@ -2,7 +2,7 @@
 from astropy.table import Table
 from astrodb_utils import AstroDBError
 from simple.utils.astrometry import (
-    ingest_parallaxes,
+    ingest_parallax,
     ingest_proper_motions,
     ingest_radial_velocity,
 )
@@ -15,6 +15,7 @@ def t_plx():
         [
             {"source": "Fake 1", "plx": 113.0, "plx_err": 0.3, "plx_ref": "Ref 1"},
             {"source": "Fake 2", "plx": 145.0, "plx_err": 0.5, "plx_ref": "Ref 1"},
+            {"source": "Fake 1", "plx": 113.0, "plx_err": 0.2, "plx_ref": "Ref 2"},
             {"source": "Fake 3", "plx": 155.0, "plx_err": 0.6, "plx_ref": "Ref 2"},
         ]
     )
@@ -68,25 +69,33 @@ def t_rv():
 
 def test_ingest_parallaxes(temp_db, t_plx):
     # Test ingest of parallax data
-    ingest_parallaxes(
-        temp_db, t_plx["source"], t_plx["plx"], t_plx["plx_err"], t_plx["plx_ref"]
-    )
+
+    for row in t_plx:
+        ingest_parallax(
+            temp_db,
+            row["source"],
+            row["plx"],
+            row["plx_err"],
+            row["plx_ref"],
+        )
 
     results = (
         temp_db.query(temp_db.Parallaxes)
         .filter(temp_db.Parallaxes.c.reference == "Ref 1")
         .table()
     )
     assert len(results) == 2
+    assert not results["adopted"][0]
     results = (
         temp_db.query(temp_db.Parallaxes)
         .filter(temp_db.Parallaxes.c.reference == "Ref 2")
         .table()
     )
-    assert len(results) == 1
-    assert results["source"][0] == "Fake 3"
-    assert results["parallax"][0] == 155
-    assert results["parallax_error"][0] == 0.6
+    assert len(results) == 2
+    assert results["source"][1] == "Fake 3"
+    assert results["parallax"][1] == 155
+    assert results["parallax_error"][1] == 0.6
+    assert results["adopted"][0]
 
 
 def test_ingest_proper_motions(temp_db, t_pm):