move to stable polars, pin numpy below 2

.github/workflows/tests.yml

@@ -7,7 +7,7 @@ jobs:
     strategy:
       matrix:
         os: ["ubuntu-latest"]
-        python-version: ["3.9", "3.10"]
+        python-version: ["3.9", "3.10", "3.11", "3.12"]
     steps:
       - uses: actions/checkout@v2
       - name: Cache conda
@@ -25,7 +25,7 @@ jobs:
           environment-file: tests/environment.yml
           python-version: ${{ matrix.python-version }}
           channel-priority: strict
-          use-only-tar-bz2: true # IMPORTANT: This needs to be set for caching to work properly!
+          use-only-tar-bz2: false # IMPORTANT: This needs to be set for caching to work properly!
       - name: Conda info
         shell: bash -l {0}
         run: conda info; conda list
@@ -45,4 +45,4 @@ jobs:
         shell: bash -l {0}
         run: which python; pytest --cov --cov-report xml
       - name: Upload coverage to Codecov
-        uses: codecov/codecov-action@v3
+        uses: codecov/codecov-action@v3

environment.yml

@@ -9,9 +9,9 @@ dependencies:
   - dask
   - netcdf4
   - gsw
+  - polars>=1.1
   - scipy
   - bitstring
   - pooch
   - pip:
     - dbdreader
-    - polars

pyglider/ncprocess.py

@@ -70,14 +70,14 @@ def extract_timeseries_profiles(inname, outdir, deploymentyaml, force=False):
                     dss['v'] = dss.water_velocity_northward.mean()
                     dss['v'].attrs = profile_meta['v']
                 elif 'u' in profile_meta:
-                    dss['u'] = profile_meta['u'].get('_FillValue', np.NaN)
+                    dss['u'] = profile_meta['u'].get('_FillValue', np.nan)
                     dss['u'].attrs = profile_meta['u']
 
-                    dss['v'] = profile_meta['v'].get('_FillValue', np.NaN)
+                    dss['v'] = profile_meta['v'].get('_FillValue', np.nan)
                     dss['v'].attrs = profile_meta['v']
                 else:
-                    dss['u'] = np.NaN
-                    dss['v'] = np.NaN
+                    dss['u'] = np.nan
+                    dss['v'] = np.nan
 
 
                 dss['profile_id'] = np.int32(p)
@@ -117,11 +117,11 @@ def extract_timeseries_profiles(inname, outdir, deploymentyaml, force=False):
                     dss['platform'].attrs['_FillValue'] = -1
 
 
-                dss['lat_uv'] = np.NaN
+                dss['lat_uv'] = np.nan
                 dss['lat_uv'].attrs = profile_meta['lat_uv']
-                dss['lon_uv'] = np.NaN
+                dss['lon_uv'] = np.nan
                 dss['lon_uv'].attrs = profile_meta['lon_uv']
-                dss['time_uv'] = np.NaN
+                dss['time_uv'] = np.nan
                 dss['time_uv'].attrs = profile_meta['time_uv']
 
                 dss['instrument_ctd'] = np.int32(1.0)
@@ -328,7 +328,7 @@ def make_gridfiles(inname, outdir, deploymentyaml, *, fnamesuffix='', dz=1, star
     dsout.to_netcdf(
         outname,
         encoding={'time': {'units': 'seconds since 1970-01-01T00:00:00Z',
-                           '_FillValue': np.NaN,
+                           '_FillValue': np.nan,
                            'calendar': 'gregorian',
                            'dtype': 'float64'}})
     _log.info('Done gridding')

pyglider/slocum.py

@@ -345,8 +345,8 @@ def dbd_to_dict(dinkum_file, cachedir, keys=None):
     _log.debug('Diagnostic check passed.  Endian is %s', endian)
 
     nsensors = int(meta['sensors_per_cycle'])
-    currentValues = np.zeros(int(meta['sensors_per_cycle'])) + np.NaN
-    data = np.zeros((DINKUMCHUNKSIZE, nsensors)) + np.NaN
+    currentValues = np.zeros(int(meta['sensors_per_cycle'])) + np.nan
+    data = np.zeros((DINKUMCHUNKSIZE, nsensors)) + np.nan
     # Then there's a data cycle with every sensor marked as updated, giving
     # us our initial values.
     # 01 means updated with 'same value', 10 means updated with a new value,
@@ -370,7 +370,7 @@ def dbd_to_dict(dinkum_file, cachedir, keys=None):
         binaryData.bytealign()
         for i, code in enumerate(updatedCode):
             if code == '00':  # No new value
-                currentValues[i] = np.NaN
+                currentValues[i] = np.nan
             elif code == '01':  # Same value as before.
                 continue
             elif code == '10':  # New value.
@@ -404,7 +404,7 @@ def dbd_to_dict(dinkum_file, cachedir, keys=None):
             if ndata % DINKUMCHUNKSIZE == 0:
                 # need to allocate more data!
                 data = np.concatenate(
-                    (data, np.NaN + np.zeros((DINKUMCHUNKSIZE, nsensors))),
+                    (data, np.nan + np.zeros((DINKUMCHUNKSIZE, nsensors))),
                     axis=0)
         elif d == 'X':
             # End of file cycle tag. We made it through.
@@ -496,7 +496,7 @@ def add_times_flight_sci(fdata, sdata=None):
             sdata['m_present_time_sci'] = np.interp(
                 sdata['sci_m_present_time'], tf, pt, np.nan, np.nan)
         else:
-            sdata['m_present_time_sci'] = np.NaN * sdata['sci_m_present_time']
+            sdata['m_present_time_sci'] = np.nan * sdata['sci_m_present_time']
 
     return fdata, sdata
 
@@ -731,7 +731,7 @@ def raw_to_timeseries(indir, outdir, deploymentyaml, *,
             _log.debug('EBD sensorname %s', sensorname)
             val = ebd[sensorname]
             val = utils._zero_screen(val)
-    #        val[val==0] = np.NaN
+    #        val[val==0] = np.nan
             val = convert(val)
         else:
             _log.debug('DBD sensorname %s', sensorname)
@@ -922,7 +922,7 @@ def binary_to_timeseries(indir, cachedir, outdir, deploymentyaml, *,
     ds = utils.get_derived_eos_raw(ds)
 
     # screen out-of-range times; these won't convert:
-    ds['time'] = ds.time.where((ds.time>0) & (ds.time<6.4e9), np.NaN)
+    ds['time'] = ds.time.where((ds.time>0) & (ds.time<6.4e9), np.nan)
     # convert time to datetime64:
     ds['time'] = (ds.time*1e9).astype('datetime64[ns]')
     ds['time'].attrs = attr
@@ -954,7 +954,7 @@ def binary_to_timeseries(indir, cachedir, outdir, deploymentyaml, *,
     # as a unit:
     ds.to_netcdf(outname, 'w',
                  encoding={'time': {'units': 'seconds since 1970-01-01T00:00:00Z',
-                                    '_FillValue': np.NaN,
+                                    '_FillValue': np.nan,
                                     'dtype': 'float64'}})
 
     return outname
@@ -1076,9 +1076,9 @@ def parse_logfiles(files):
     # now parse them
     out = xr.Dataset(
         coords={'time': ('surfacing', np.zeros(ntimes, dtype='datetime64[ns]'))})
-    out['ampH'] = ('surfacing', np.zeros(ntimes) * np.NaN)
-    out['lon'] = ('surfacing', np.zeros(ntimes) * np.NaN)
-    out['lat'] = ('surfacing', np.zeros(ntimes) * np.NaN)
+    out['ampH'] = ('surfacing', np.zeros(ntimes) * np.nan)
+    out['lon'] = ('surfacing', np.zeros(ntimes) * np.nan)
+    out['lat'] = ('surfacing', np.zeros(ntimes) * np.nan)
 
     for i in range(ntimes):
         timestring = times[i][11:-13]
@@ -1151,10 +1151,10 @@ def parse_logfiles_maybe(files):
 
     # now parse them
     out = xr.Dataset(coords={'time': ('surfacing', np.zeros(ntimes, dtype='datetime64[ns]'))})
-    out['ampH'] = ('surfacing', np.zeros(ntimes) * np.NaN)
-    out['lon'] = ('surfacing', np.zeros(ntimes) * np.NaN)
-    out['lat'] = ('surfacing', np.zeros(ntimes) * np.NaN)
-    out['missionnum'] = ('surfacing', np.zeros(ntimes) * np.NaN)
+    out['ampH'] = ('surfacing', np.zeros(ntimes) * np.nan)
+    out['lon'] = ('surfacing', np.zeros(ntimes) * np.nan)
+    out['lat'] = ('surfacing', np.zeros(ntimes) * np.nan)
+    out['missionnum'] = ('surfacing', np.zeros(ntimes) * np.nan)
     out.attrs['surfacereason'] = surfacereason
     # ABORT HISTORY: last abort segment: hal_1002-2024-183-0-0 (0171.0000)
     out.attrs['abortsegment'] = float(abortsegment[-11:-2])

pyglider/utils.py

@@ -92,8 +92,8 @@ def get_profiles(ds, min_dp=10.0, inversion=3., filt_length=7,
     make two variables: profile_direction and profile_index; this version
     is good for lots of data.  Less good for sparse data
     """
-    profile = ds.pressure.values * np.NaN
-    direction = ds.pressure.values * np.NaN
+    profile = ds.pressure.values * np.nan
+    direction = ds.pressure.values * np.nan
     pronum = 1
     lastpronum = 0
 
@@ -500,7 +500,7 @@ def fill_metadata(ds, metadata, sensor_data):
 
 
 def _zero_screen(val):
-    val[val == 0] = np.NaN
+    val[val == 0] = np.nan
     return val
 
 

requirements.txt

@@ -7,4 +7,4 @@ numpy
 pooch
 scipy
 xarray
-polars>0.16
+polars>=1.1

tests/environment.yml

@@ -11,6 +11,7 @@ dependencies:
   - gsw
   - scipy
   - bitstring
+  - polars>=1.1
   - pytest
   - pytest-cov
   - pooch
@@ -19,4 +20,3 @@ dependencies:
   - cc-plugin-glider
   - pip:
     - dbdreader
-    - polars