Merge pull request #24 from tnakazato/refactor-filler-field

refactor: filler for FIELD table

src/nro45data/psw/ms2/filler/__init__.py

@@ -8,7 +8,7 @@
 from .antenna import fill_antenna
 from .data_description import fill_data_description
 from .feed import fill_feed
-from .field import _fill_field_columns, _get_field_columns
+from .field import fill_field
 from .observation import _fill_observation_columns, _get_observation_columns
 from .polarization import _fill_polarization_columns, _get_polarization_columns
 from .main import fill_main
@@ -26,11 +26,6 @@
 LOG = logging.getLogger(__name__)
 
 
-def fill_field(msfile: str, hdu: BinTableHDU):
-    columns = _get_field_columns(hdu)
-    _fill_field_columns(msfile, columns)
-
-
 def fill_observation(msfile: str, hdu: BinTableHDU):
     columns = _get_observation_columns(hdu)
     _fill_observation_columns(msfile, columns)

src/nro45data/psw/ms2/filler/field.py

@@ -1,19 +1,32 @@
+from __future__ import annotations
+
 import logging
-from typing import TYPE_CHECKING
+from typing import Generator, TYPE_CHECKING
 
 import numpy as np
 
-from .._casa import open_table
+from .._casa import datestr2mjd
 from .._casa import convert_str_angle_to_rad
-from .utils import fix_nrow_to
+from .utils import fill_ms_table
 
 if TYPE_CHECKING:
     from astropy.io.fits.hdu.BinTableHDU import BinTableHDU
 
 LOG = logging.getLogger(__name__)
 
 
-def _get_field_columns(hdu: "BinTableHDU") -> dict:
+def _get_field_row(hdu: BinTableHDU) -> Generator[dict, None, dict]:
+    """Provide field row information.
+
+    Args:
+        hdu: NRO45m psw data in the form of BinTableHDU object.
+
+    Yields:
+        Dictionary containing field row information.
+
+    Returns:
+        Dictionary containing data-dependent column keywords.
+    """
     # NAME
     field_name = hdu.header["OBJECT"].strip()
     LOG.debug("field_name: %s", field_name)
@@ -25,8 +38,11 @@ def _get_field_columns(hdu: "BinTableHDU") -> dict:
     # use start time of the observation
     history_cards = hdu.header["HISTORY"]
     start_time_card = [x for x in history_cards if x.startswith("NEWSTAR START-TIME")]
-    field_time = float(start_time_card[0].split("=")[-1].strip(" '"))
-    LOG.debug("field_time: %s", field_time)
+    sstr = start_time_card[0].split("=")[-1].strip(" '")
+    LOG.debug("sstr: %s", sstr)
+    datestr = sstr[0:4] + "/" + sstr[4:6] + "/" + sstr[6:8] + " " + sstr[8:10] + ":" + sstr[10:12] + ":" + sstr[12:14]
+    LOG.debug("formatted sstr: %s", datestr)
+    field_time = datestr2mjd(datestr) - 9 * 3600
 
     # NUM_POLY
     num_poly = 0
@@ -62,37 +78,34 @@ def _get_field_columns(hdu: "BinTableHDU") -> dict:
     # FLAG_ROW
     flag_row = False
 
-    columns = {
+    row = {
         "NAME": field_name,
         "CODE": field_code,
         "TIME": field_time,
         "NUM_POLY": num_poly,
-        "FIELD_EPOCH": field_epoch,
         "DELAY_DIR": delay_dir,
         "PHASE_DIR": phase_dir,
         "REFERENCE_DIR": reference_dir,
         "SOURCE_ID": source_id,
         "FLAG_ROW": flag_row,
     }
 
-    return columns
-
-
-def _fill_field_columns(msfile: str, columns: dict):
-    with open_table(msfile + "/FIELD", read_only=False) as tb:
-        fix_nrow_to(1, tb)
-
-        tb.putcell("NAME", 0, columns["NAME"])
-        tb.putcell("CODE", 0, columns["CODE"])
-        tb.putcell("TIME", 0, columns["TIME"])
-        tb.putcell("NUM_POLY", 0, columns["NUM_POLY"])
-        tb.putcell("DELAY_DIR", 0, columns["DELAY_DIR"])
-        colkeywords = tb.getcolkeywords("DELAY_DIR")
-        colkeywords["MEASINFO"]["Ref"] = columns["FIELD_EPOCH"]
-        tb.putcolkeywords("DELAY_DIR", colkeywords)
-        tb.putcell("PHASE_DIR", 0, columns["PHASE_DIR"])
-        tb.putcolkeywords("PHASE_DIR", colkeywords)
-        tb.putcell("REFERENCE_DIR", 0, columns["REFERENCE_DIR"])
-        tb.putcolkeywords("REFERENCE_DIR", colkeywords)
-        tb.putcell("SOURCE_ID", 0, columns["SOURCE_ID"])
-        tb.putcell("FLAG_ROW", 0, columns["FLAG_ROW"])
+    yield row
+
+    column_keywords = {
+        "DELAY_DIR": {"MEASINFO": {"Ref": field_epoch}},
+        "PHASE_DIR": {"MEASINFO": {"Ref": field_epoch}},
+        "REFERENCE_DIR": {"MEASINFO": {"Ref": field_epoch}}
+    }
+
+    return column_keywords  # noqa
+
+
+def fill_field(msfile: str, hdu: BinTableHDU):
+    """Fill MS FIELD table.
+
+    Args:
+        msfile: Name of MS file.
+        hdu: NRO45m psw data in the form of BinTableHDU object.
+    """
+    fill_ms_table(msfile, hdu, "FIELD", _get_field_row)

src/nro45data/psw/ms2/filler/utils.py

@@ -2,7 +2,7 @@
 
 import logging
 import pprint
-from typing import Any, Callable, Optional, TYPE_CHECKING
+from typing import Callable, TYPE_CHECKING
 
 import numpy as np
 
@@ -281,8 +281,7 @@ def fill_ms_table(
         msfile: str,
         hdu: BinTableHDU,
         table_name: str,
-        row_generator: Callable,
-        column_keywords: Optional[dict[str, dict[str, Any]]] = None
+        row_generator: Callable
 ):
     """Fill MS table.
 
@@ -291,36 +290,42 @@ def fill_ms_table(
         hdu: NRO45m psw data in the form of BinTableHDU object.
         table_name: Name of subtable or "MAIN" for MAIN table.
         row_generator: Generator to yield table row.
-        column_keywords: Optional information to update column
-            keywords. Defaults to None.
-            Key is column name and value is a dictionary of
-            column keywords to be updated.
     """
     if table_name.upper() == "MAIN":
         table_path = msfile
     else:
         table_path = f"{msfile}/{table_name.upper()}"
 
     with open_table(table_path, read_only=False) as tb:
-        # update table column keywords
-        if column_keywords is None:
-            column_keywords = {}
-
-        for colname, colkeywords_new in column_keywords.items():
-            colkeywords = tb.getcolkeywords(colname)
-            for key, value_new in colkeywords_new.items():
-                if key in colkeywords and isinstance(value_new, dict):
-                    colkeywords[key].update(value_new)
-                else:
-                    colkeywords[key] = value_new
-            tb.putcolkeywords(colname, colkeywords)
-
-        # fill table rows
-        for row_id, row in enumerate(row_generator(hdu)):
-            if tb.nrows() <= row_id:
-                tb.addrows(tb.nrows() - row_id + 1)
-
-            for key, value in row.items():
-                LOG.debug("row %d key %s", row_id, key)
-                tb.putcell(key, row_id, value)
-            LOG.debug("%s table %d row %s", table_name, row_id, row)
+        # iterator should provide row dictionary in order,
+        # and then, if necessary, column keywords dictionary
+        # should be returned additionally
+        iterator = enumerate(row_generator(hdu))
+        try:
+            # fill table rows
+            while True:
+                row_id, row = next(iterator)
+                if tb.nrows() <= row_id:
+                    tb.addrows(tb.nrows() - row_id + 1)
+
+                for key, value in row.items():
+                    LOG.debug("row %d key %s", row_id, key)
+                    tb.putcell(key, row_id, value)
+                LOG.debug("%s table %d row %s", table_name, row_id, row)
+
+        except StopIteration as s:
+            # update column keywords if necessary
+            column_keywords = s.value
+            LOG.debug("return value: %s", s.value)
+            if isinstance(column_keywords, dict):
+                LOG.debug("column_keywords: %s", column_keywords)
+                for colname, colkeywords_new in column_keywords.items():
+                    colkeywords = tb.getcolkeywords(colname)
+                    for key, value_new in colkeywords_new.items():
+                        if key in colkeywords and isinstance(value_new, dict):
+                            LOG.debug("updating column keyword %s: %s, %s", colname, key, value_new)
+                            colkeywords[key].update(value_new)
+                        else:
+                            LOG.debug("adding column keyword %s: %s, %s", colname, key, value_new)
+                            colkeywords[key] = value_new
+                    tb.putcolkeywords(colname, colkeywords)

tests/ms2_filler/test_forest.py

@@ -104,6 +104,31 @@ def test_forest_ms2_structure(msfile):
             assert receptor_angle.shape == (2,)
             assert np.all(receptor_angle == 0.0)
 
+    with open_table(os.path.join(msfile, "FIELD")) as tb:
+        assert tb.nrows() == 1
+        assert tb.getcell("NAME", 0) == "NML-Tau"
+        assert tb.getcell("CODE", 0) == ""
+        # start time: 2024/09/30 17:49:19
+        field_time = mjd2datetime(tb.getcell("TIME", 0))
+        time_expected = datetime.datetime(2024, 9, 30, 17, 49, 19, tzinfo=datetime.timezone.utc)
+        assert abs((field_time - time_expected).total_seconds()) < 1e-3
+        assert tb.getcell("NUM_POLY", 0) == 0
+        # RA  = 03:53:28.860
+        # DEC = +11:24:22.40
+        # Epoch = J2000
+        ra_expected = 1.0187530477122202  # rad
+        dec_expected = 2.9861419948788317  # rad
+        for col in ["DELAY_DIR", "PHASE_DIR", "REFERENCE_DIR"]:
+            meas_info = tb.getcolkeyword(col, "MEASINFO")
+            assert "Ref" in meas_info
+            assert meas_info["Ref"] == "J2000"
+            direction = tb.getcell(col, 0)
+            assert direction.shape == (1, 2)
+            assert abs(direction[0, 0] - ra_expected) / ra_expected < 1e-6
+            assert abs(direction[0, 1] - dec_expected) / dec_expected < 1e-6
+        assert tb.getcell("SOURCE_ID", 0) == 0
+        assert tb.getcell("FLAG_ROW", 0) is False
+
     with open_table(os.path.join(msfile, "STATE")) as tb:
         intents_map = dict((i, v) for i, v in enumerate(tb.getcol("OBS_MODE")))
 

tests/ms2_filler/test_h40.py

@@ -106,6 +106,31 @@ def test_h40_ms2_structure(msfile):
             assert receptor_angle.shape == (2,)
             assert np.all(receptor_angle == 0.0)
 
+    with open_table(os.path.join(msfile, "FIELD")) as tb:
+        assert tb.nrows() == 1
+        assert tb.getcell("NAME", 0) == "NML-Tau"
+        assert tb.getcell("CODE", 0) == ""
+        # start time: 2024/09/25 15:58:54
+        field_time = mjd2datetime(tb.getcell("TIME", 0))
+        time_expected = datetime.datetime(2024, 9, 25, 15, 58, 54, tzinfo=datetime.timezone.utc)
+        assert abs((field_time - time_expected).total_seconds()) < 1e-3
+        assert tb.getcell("NUM_POLY", 0) == 0
+        # RA  = 03:53:28.860
+        # DEC = +11:24:22.40
+        # Epoch = J2000
+        ra_expected = 1.0187530477122202  # rad
+        dec_expected = 2.9861419948788317  # rad
+        for col in ["DELAY_DIR", "PHASE_DIR", "REFERENCE_DIR"]:
+            meas_info = tb.getcolkeyword(col, "MEASINFO")
+            assert "Ref" in meas_info
+            assert meas_info["Ref"] == "J2000"
+            direction = tb.getcell(col, 0)
+            assert direction.shape == (1, 2)
+            assert abs(direction[0, 0] - ra_expected) / ra_expected < 1e-6
+            assert abs(direction[0, 1] - dec_expected) / dec_expected < 1e-6
+        assert tb.getcell("SOURCE_ID", 0) == 0
+        assert tb.getcell("FLAG_ROW", 0) is False
+
     with open_table(os.path.join(msfile, "STATE")) as tb:
         intents_map = dict((i, v) for i, v in enumerate(tb.getcol("OBS_MODE")))
 

tests/ms2_filler/test_z45.py

@@ -105,6 +105,31 @@ def test_z45_ms2_structure(msfile):
             assert receptor_angle.shape == (2,)
             assert np.all(receptor_angle == 0.0)
 
+    with open_table(os.path.join(msfile, "FIELD")) as tb:
+        assert tb.nrows() == 1
+        assert tb.getcell("NAME", 0) == "NML-Tau"
+        assert tb.getcell("CODE", 0) == ""
+        # start time: 2024/09/30 18:00:32
+        field_time = mjd2datetime(tb.getcell("TIME", 0))
+        time_expected = datetime.datetime(2024, 9, 30, 18, 0, 32, tzinfo=datetime.timezone.utc)
+        assert abs((field_time - time_expected).total_seconds()) < 1e-3
+        assert tb.getcell("NUM_POLY", 0) == 0
+        # RA  = 03:53:28.860
+        # DEC = +11:24:22.40
+        # Epoch = J2000
+        ra_expected = 1.0187530477122202  # rad
+        dec_expected = 2.9861419948788317  # rad
+        for col in ["DELAY_DIR", "PHASE_DIR", "REFERENCE_DIR"]:
+            meas_info = tb.getcolkeyword(col, "MEASINFO")
+            assert "Ref" in meas_info
+            assert meas_info["Ref"] == "J2000"
+            direction = tb.getcell(col, 0)
+            assert direction.shape == (1, 2)
+            assert abs(direction[0, 0] - ra_expected) / ra_expected < 1e-6
+            assert abs(direction[0, 1] - dec_expected) / dec_expected < 1e-6
+        assert tb.getcell("SOURCE_ID", 0) == 0
+        assert tb.getcell("FLAG_ROW", 0) is False
+
     with open_table(os.path.join(msfile, "STATE")) as tb:
         intents_map = dict((i, v) for i, v in enumerate(tb.getcol("OBS_MODE")))