from __future__ import annotations
import logging
import gzip
import json
import pandas as pd
from pathlib import Path
from typing import Any, Literal, TYPE_CHECKING
from addict import Dict as aDict
if TYPE_CHECKING:
from openghg.dataobjects import ObsData
logger = logging.getLogger("openghg.util")
logger.setLevel(logging.DEBUG) # Have to set level for logger as well as handler
[docs]
def to_dashboard(
data: ObsData | list[ObsData],
selected_vars: list,
downsample_n: int = 3,
filename: str | None = None,
) -> dict | None:
"""Takes a Dataset produced by OpenGHG and outputs it into a JSON
format readable by the OpenGHG dashboard or a related project.
This also exports a separate file with the locations of the sites
for use with map selector component.
Note - this function does not currently support export of data from multiple
inlets.
Args:
data: Dictionary of retrieved data
selected_vars: The variables to want to export
downsample_n: Take every nth value from the data
filename: filename to write output to
Returns:
None
"""
to_export = aDict()
if not isinstance(selected_vars, list):
selected_vars = [selected_vars]
selected_vars = [str(c).lower() for c in selected_vars]
if not isinstance(data, list):
data = [data]
for obs in data:
measurement_data = obs.data
attributes = measurement_data.attrs
metadata = obs.metadata
df = measurement_data.to_dataframe()
rename_lower = {c: str(c).lower() for c in df.columns}
df = df.rename(columns=rename_lower)
# We just want the selected variables
to_extract = [c for c in df.columns if c in selected_vars]
if not to_extract:
continue
df = df[to_extract]
# Downsample the data
if downsample_n > 1:
df = df.iloc[::downsample_n]
network = metadata["network"]
instrument = metadata["instrument"]
try:
latitude = attributes["latitude"]
except KeyError:
latitude = metadata["latitude"]
try:
longitude = attributes["longitude"]
except KeyError:
longitude = metadata["longitude"]
# TODO - remove this if we add site location to standard metadata
location = {
"latitude": latitude,
"longitude": longitude,
}
metadata.update(location)
json_data = json.loads(df.to_json())
species = metadata["species"]
site = metadata["site"]
inlet = metadata["inlet"]
to_export[species][network][site][inlet][instrument] = {
"data": json_data,
"metadata": metadata,
}
if filename is not None:
with open(filename, "w") as f:
json.dump(obj=to_export, fp=f)
return None
else:
# TODO - remove this once addict is stubbed
export_dict: dict = to_export.to_dict()
return export_dict
[docs]
def to_dashboard_mobile(data: dict, filename: str | Path | None = None) -> dict | None:
"""Export the Glasgow LICOR data to JSON for the dashboard
Args:
data: Data dictionary
filename: Filename for export of JSON
Returns:
dict or None: Dictonary if no filename given
"""
to_export = aDict()
for species, species_data in data.items():
spec_data = species_data["data"]
metadata = species_data["metadata"]
latitude = spec_data["latitude"].values.tolist()
longitude = spec_data["longitude"].values.tolist()
ch4 = spec_data["ch4"].values.tolist()
to_export[species]["data"] = {"lat": latitude, "lon": longitude, "z": ch4}
to_export[species]["metadata"] = metadata
if filename is not None:
with open(filename, "w") as f:
json.dump(to_export, f)
return None
else:
to_return: dict = to_export.to_dict()
return to_return
def to_dashboard_agage(
data: ObsData | list[ObsData],
export_folder: str | Path,
downsample_n: int = 3,
compress_json: bool = False,
float_to_int: bool = False,
selection_level: Literal["site", "inlet"] = "inlet",
mock_inlet: bool = False,
drop_na: bool = True,
default_site: str | None = None,
default_species: str | None = None,
default_inlet: str | None = None,
) -> None:
"""Takes ObsData objects produced by OpenGHG and outputs them to JSON
files. Files are named using the following convention:
TODO - add compression of metadata file
if selection_level == "site":
export_filename = f"{species}_{network}_{site}.json"
elif selection_level == "inlet":
export_filename = f"{species}_{network}_{site}_{inlet}_{instrument}.json"
A separate metadata file, metadata_complete.json is created in the same
export folder containing the metadata for each site and the filename for
that specific data set. This chunking allows a larger amount of data to be used
by the dashboard due to the separation into separate files.
Args:
data: ObsData object or list of ObsData objects
export_folder: Folder path to write files
selection_level: Do we want the user to select by site or inlet in the dashboard
downsample_n: Take every nth value from the data
compress_json: compress JSON using gzip
float_to_int: Convert floats to ints by multiplying by 100
mock_inlet: Use a mock "888m" inlet for the the dashboard as it doesn't currently support
selection only by site.
drop_na: Drop any NaNs from the datasets
default_site: Set a default site for the dashboard
default_species: Set a default species for the dashboard
default_inlet: Set a default inlet for the dashboard
Returns:
None
"""
allowed_selection_levels = ("site", "inlet")
if selection_level not in allowed_selection_levels:
raise ValueError(f"Invalid selection level, please select one of {allowed_selection_levels}")
if selection_level == "site":
raise NotImplementedError("Selection by site is not currently supported")
if mock_inlet:
logger.warning("Assuming multiple inlet height data and setting inlet = 'single_inlet'")
export_folder = Path(export_folder)
if not export_folder.exists():
logger.info(f"Creating export folder at {export_folder}")
export_folder.mkdir()
# Here we'll store the metadata that can be used to populate the interface
# it'll also hold the filenames for the retrieval of data
metadata_complete_filepath = export_folder.joinpath("metadata_complete.json")
dashboard_config_filepath = export_folder.joinpath("dashboard_config.json")
# Create the data directory
data_foldername = "measurements"
data_dir = export_folder.joinpath(data_foldername)
data_dir.mkdir(exist_ok=True)
if not isinstance(data, list):
data = [data]
# Hold a list of all the files we export so we can check the size of the exported data
file_sizes_bytes = 0
one_MB = 1024 * 1024
# We'll use this to convert floats to ints
float_to_int_multiplier = 1000
dashboard_config: dict[str, Any] = {}
dashboard_config["selection_level"] = selection_level
dashboard_config["float_to_int"] = float_to_int
dashboard_config["compressed_json"] = compress_json
if default_site is not None:
dashboard_config["default_site"] = default_site
if default_species is not None:
dashboard_config["default_species"] = default_species
if default_inlet is not None:
dashboard_config["default_inlet"] = default_inlet
if float_to_int:
dashboard_config["float_to_int_multiplier"] = float_to_int_multiplier
# We'll store the filename information and source metadata here
metadata_complete = aDict()
# We'll record the inlets for each site
# so we can warn the user if they're exporting multiple inlets
# for the same site
site_inlets = aDict()
for obs in data:
measurement_data = obs.data
attributes = measurement_data.attrs
metadata = obs.metadata
df: pd.DataFrame = measurement_data.to_dataframe()
rename_lower = {c: str(c).lower() for c in df.columns}
df = df.rename(columns=rename_lower)
species_name = obs.metadata["species"]
# Some of the AGAGE data variables are named differently from the species in the metadata
try:
df = df[[species_name]]
except KeyError:
species_label = obs.metadata["species_label"]
df = df[[species_label]]
# Drop any NaNs
if drop_na:
df = df.dropna()
if float_to_int:
key = next(iter(df))
df[key] = df[key] * float_to_int_multiplier
df = df.astype(int)
# Downsample the data
if downsample_n > 1:
df = df.iloc[::downsample_n]
try:
station_latitude = attributes["station_latitude"]
except KeyError:
try:
station_latitude = obs.metadata["station_latitude"]
except KeyError:
station_latitude = obs.metadata["inlet_latitude"]
try:
station_longitude = attributes["station_longitude"]
except KeyError:
try:
station_longitude = obs.metadata["station_longitude"]
except KeyError:
station_longitude = obs.metadata["inlet_longitude"]
species = metadata["species"]
site = metadata["site"]
network = obs.metadata["network"]
# TODO - remove this as we won't want to select by instrument
# use a mock instrument name for now
instrument = "instrument_key"
# This is all the metadata we need for the dashboard itself
source_metadata = {
"station_latitude": station_latitude,
"station_longitude": station_longitude,
"species": species,
"site": site,
"network": network,
"instrument": instrument,
"units": obs.metadata["units"],
"station_long_name": obs.metadata["station_long_name"],
}
# TODO - remove this once we've updated the dashboard to support selection by site or inlet
if mock_inlet:
inlet = "single_inlet"
else:
try:
inlet = obs.metadata["inlet"]
except ValueError:
inlet = str(int(float(obs.metadata["inlet"])))
source_metadata["inlet"] = inlet
file_extension = ".json"
if compress_json:
file_extension += ".gz"
if selection_level == "site":
export_filename = f"{species}_{network}_{site}{file_extension}"
else:
export_filename = f"{species}_{network}_{site}_{inlet}_{instrument}{file_extension}"
export_filepath = data_dir.joinpath(export_filename)
file_data = {
"metadata": source_metadata,
"filepath": f"{data_foldername}/{export_filename.lower()}",
}
if selection_level == "site":
metadata_complete[species][network][site] = file_data
site_inlets[species][network][site] = (
site_inlets[species][network][site].get(site, []).append(inlet)
)
else:
metadata_complete[species][network][site][inlet][instrument] = file_data
# TODO - Check if this hoop jumping is required, I can't remember exactly why
# I did it
data_dict = json.loads(df.to_json())
# Let's trim the species name as we don't need that
key = next(iter(data_dict))
data_dict = data_dict[key]
for_export_str = json.dumps(data_dict)
if compress_json:
for_export_bytes = gzip.compress(for_export_str.encode())
export_filepath.write_bytes(for_export_bytes)
else:
export_filepath.write_text(for_export_str)
logger.info(f"Writing dashboard data to: {export_filename}")
file_size = export_filepath.stat().st_size
file_sizes_bytes += file_size
if file_size > one_MB:
logger.warn(
msg=f"The file {export_filename} is larger than 1 MB, consider ways to reduce its size."
)
if selection_level == "site":
for species, networkData in site_inlets.items():
for network, siteData in networkData.items():
for site, inlets in siteData.items():
if len(inlets) > 1:
logger.warn(
msg=f"Site {site} has multiple inlets: {inlets}. "
"You've set selection_level == 'site' meaning only data for the last inlet will be kept."
"Please make sure are more specific in your selection of data or select"
"selection_level == 'inlet'"
)
# Add in the config
dashboard_config_filepath.write_text(json.dumps(dashboard_config))
metadata_complete_filepath.write_text(json.dumps(metadata_complete))
file_sizes_bytes += metadata_complete_filepath.stat().st_size
file_sizes_bytes += dashboard_config_filepath.stat().st_size
logger.info(f"\n\nComplete metadata file written to: {metadata_complete_filepath}")
logger.info(f"Dashboard configuration file written to: {metadata_complete_filepath}")
logger.info(f"\nTotal size of exported data package: {file_sizes_bytes / one_MB:.2f} MB")
