From ee606ebb6445d443f368c12d3b931893ecbc7d16 Mon Sep 17 00:00:00 2001
From: Nicholas Shavers <nshavers@contractor.usgs.gov>
Date: Mon, 9 Dec 2024 13:10:09 -0800
Subject: [PATCH] general clean up
---
geomagio/ImagCDFFactory.py | 56 ++++++++++++++++++--------------------
1 file changed, 26 insertions(+), 30 deletions(-)
diff --git a/geomagio/ImagCDFFactory.py b/geomagio/ImagCDFFactory.py
index 6ba44f3e..3d8ba136 100644
--- a/geomagio/ImagCDFFactory.py
+++ b/geomagio/ImagCDFFactory.py
@@ -31,6 +31,8 @@ from . import TimeseriesUtility
from . import Util
import cdflib
+from cdflib.cdfwrite import CDF as CDFWriter
+from cdflib.cdfread import CDF as CDFReader
import tempfile
@@ -48,7 +50,7 @@ class IMCDFPublicationLevel:
4: Definitive data with no further changes expected.
Reference:
- - ImagCDF Documentation Section 4.2: Attributes that Uniquely Identify the Data
+ - ImagCDF Technical Documentation: Attributes that Uniquely Identify the Data
"""
class PublicationLevel:
@@ -137,13 +139,13 @@ class ImagCDFFactory(TimeseriesFactory):
# Initialize the CDF writer
cdf_spec = {
'Compressed': 9, # Enable compression (0-9)
- # 'Majority': 'row_major', # Data layout - gets set automatically
- # 'Encoding': 'ibmpc', # Corrupts CDF - gets set automatically
+ 'Majority': CDFWriter.ROW_MAJOR, # Data layout - gets set automatically
+ 'Encoding': CDFWriter.IBMPC_ENCODING, # gets set automatically
'Checksum': True, # Disable checksum for faster writes (optional)
'rDim_sizes': [], # Applicable only if using rVariables - CDF protocol recommends only using zVariables.
}
- cdf_writer = cdflib.cdfwrite.CDF(path=tmp_file_path, cdf_spec=cdf_spec, delete=True)
+ cdf_writer = CDFWriter(path=tmp_file_path, cdf_spec=cdf_spec, delete=True)
# Write global attributes
global_attrs = self._create_global_attributes(timeseries, channels)
@@ -155,13 +157,13 @@ class ImagCDFFactory(TimeseriesFactory):
# Define time variable specification
var_spec = {
'Variable': ts_name,
- 'Data_Type': 33, # CDF_TIME_TT2000
+ 'Data_Type': CDFWriter.CDF_TIME_TT2000, # CDF_TIME_TT2000
'Num_Elements': 1,
'Rec_Vary': True,
'Var_Type': 'zVariable',
'Dim_Sizes': [],
'Sparse': 'no_sparse',
- 'Compress': 6,
+ 'Compress': 9,
'Pad': None,
}
# Define time variable attributes
@@ -182,9 +184,7 @@ class ImagCDFFactory(TimeseriesFactory):
var_name = f"GeomagneticField{channel}"
data_type = self._get_cdf_data_type(trace)
num_elements = 1
- CDF_CHAR = 51
- CDF_UCHAR = 52
- if data_type in [CDF_CHAR, CDF_UCHAR]: # Handle string types
+ if data_type in [CDFWriter.CDF_CHAR, CDFWriter.CDF_UCHAR]: # Handle string types
num_elements = len(trace.data[0]) if len(trace.data) > 0 else 1
var_spec = {
@@ -195,7 +195,7 @@ class ImagCDFFactory(TimeseriesFactory):
'Var_Type': 'zVariable',
'Dim_Sizes': [],
'Sparse': 'no_sparse',
- 'Compress': 6,
+ 'Compress': 9,
'Pad': None,
}
var_attrs = self._create_var_attrs(trace, temperature_index, self.isUniqueTimes)
@@ -294,7 +294,7 @@ class ImagCDFFactory(TimeseriesFactory):
if os.path.isfile(url_file):
try:
# Read existing data to merge with new data
- existing_cdf = cdflib.cdfread.CDF(url_file)
+ existing_cdf = CDFReader(url_file)
existing_stream = self._read_cdf(existing_cdf)
# existing_cdf.close() #no close method?
existing_data = existing_stream
@@ -371,7 +371,7 @@ class ImagCDFFactory(TimeseriesFactory):
try:
# Read CDF data and merge
- cdf = cdflib.cdfread.CDF(url_file)
+ cdf = CDFReader(url_file)
file_stream = self._read_cdf(cdf)
# Attempt to select only requested channels
selected = Stream()
@@ -418,7 +418,7 @@ class ImagCDFFactory(TimeseriesFactory):
descriptions.
References:
- - ImagCDF Documentation Section 4: ImagCDF Global Attributes
+ - ImagCDF Technical Documentation: ImagCDF Global Attributes
"""
stats = timeseries[0].stats if len(timeseries) > 0 else None
@@ -458,10 +458,10 @@ class ImagCDFFactory(TimeseriesFactory):
# 'UniqueIdentifier': {0: ''},
# 'ParentIdentifiers': {0: ''},
# 'ReferenceLinks': {0: ''}, #links to /ws, plots, USGS.gov
- 'SensorSamplingRate': {0: sensor_sampling_rate},
- 'DataType': {0: data_type},
- 'Comments': {0: comments},
- 'DeclinationBase': {0: declination_base},
+ 'SensorSamplingRate': {0: sensor_sampling_rate}, #Optional
+ 'DataType': {0: data_type},#Optional
+ 'Comments': {0: comments}, #Optional
+ 'DeclinationBase': {0: declination_base}, #Optional
}
return global_attrs
@@ -470,7 +470,6 @@ class ImagCDFFactory(TimeseriesFactory):
scalar_times = None
temperature_times = {}
temperature_index = 1
- print(timeseries[0].stats)
for trace in timeseries:
channel = trace.stats.channel
@@ -615,9 +614,9 @@ class ImagCDFFactory(TimeseriesFactory):
Create a dictionary of time variable attributes.
These attributes provide metadata for time variables.
- Note: None of these attributes are required for the time stamp variables GeomagneticVectorTimes and GeomagneticScalarTimes.
+ Note: None of these attributes are required for the time stamp variables.
Reference:
- - ImagCDF Documentation Section 3: ImagCDF Data
+ - ImagCDF Technical Documentation: ImagCDF Data
"""
# var_attrs = {
# 'UNITS': 'TT2000',
@@ -639,21 +638,18 @@ class ImagCDFFactory(TimeseriesFactory):
- CDF_INT4 (41) for integer data.
Reference:
- - CDF Data Types: http://cdf.gsfc.nasa.gov/html/cdfdatatypes.html
+ - See CDF for more data types
"""
- # CDF data type constants
- CDF_DOUBLE = 45 # CDF_DOUBLE corresponds to 64-bit float
- CDF_INT4 = 41 # CDF_INT4 corresponds to 32-bit int
if trace.data.dtype in [np.float32, np.float64]:
- return CDF_DOUBLE
+ return CDFWriter.CDF_DOUBLE
elif trace.data.dtype in [np.int32, np.int64]:
- return CDF_INT4
+ return CDFWriter.CDF_INT4
else:
# Default to double precision float
- return CDF_DOUBLE
+ return CDFWriter.CDF_DOUBLE
- def _read_cdf(self, cdf: cdflib.cdfread.CDF) -> Stream:
+ def _read_cdf(self, cdf: CDFReader) -> Stream:
"""
Read CDF data into an ObsPy Stream.
@@ -677,7 +673,7 @@ class ImagCDFFactory(TimeseriesFactory):
institution = global_attrs.get('Institution', [''])[0]
latitude = global_attrs.get('Latitude', [0.0])[0]
longitude = global_attrs.get('Longitude', [0.0])[0]
- elevation = global_attrs.get('Elevation', [99_999.0])[0]
+ elevation = global_attrs.get('Elevation', [99_999.0])[0] #default to 99_999 per technical documents.
sensor_sampling_rate = global_attrs.get('SensorSamplingRate', [0.0])[0]
sensor_orientation = global_attrs.get('VectorSensOrient', [''])[0]
data_type = global_attrs.get('DataType', ['variation'])[0]
@@ -816,7 +812,7 @@ class ImagCDFFactory(TimeseriesFactory):
- The formatted file URL or path.
Reference:
- - ImagCDF Documentation Section 5: ImagCDF File Names
+ - ImagCDF Technical Documentation: ImagCDF File Names
"""
# Get the publication level for the type
publication_level = IMCDFPublicationLevel(data_type=type).to_string()
--
GitLab