ArgoFloat spec and online/offline implementations

this is a refactoring

argopy/stores/float/argo_float.py

@@ -23,5 +23,5 @@ class ArgoFloat(FloatStore):
     """
     Main docstring for ArgoFloat
     """
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)

argopy/stores/float/implementations/argo_float_offline.py

@@ -1,7 +1,9 @@
-from errors import InvalidOption
+import pandas as pd
+import numpy as np
 from pathlib import Path
 import logging
 
+from ....errors import InvalidOption
 from ..spec import ArgoFloatProto
 
 
@@ -13,29 +15,63 @@ class ArgoFloatOffline(ArgoFloatProto):
 
     def __init__(
         self,
+        *args,
         **kwargs,
     ):
-        super().__init__(**kwargs)
+        super().__init__(*args, **kwargs)
 
         if self.host_protocol != "file":
-            raise InvalidOption("Trying to work with the offline store using a remote host !")
+            raise InvalidOption(
+                "Trying to work with the offline store using a remote host !"
+            )
 
         # Load some data (in a perfect world, this should be done asynchronously):
         # self.load_index()
+        self.load_dac()  # must come before metadata
         self.load_metadata()
-        self.load_dac()
 
     def load_metadata(self):
         """Method to load float meta-data"""
         data = {}
-        data.update({"deployment": {"launchDate": "?"}})
-        data.update({"cycles": []})
-        data.update({"networks": ['?']})
-        data.update({"platform": {'type': '?'}})
-        data.update({"maker": '?'})
+        # data.update({"cycles": []})
+
+        ds = self.open_dataset("meta")
+        data.update(
+            {
+                "deployment": {
+                    "launchDate": pd.to_datetime(ds["LAUNCH_DATE"].values, utc=True)
+                }
+            }
+        )
+        data.update(
+            {"platform": {"type": ds["PLATFORM_TYPE"].values[np.newaxis][0].strip()}}
+        )
+        data.update(
+            {"maker": ds["PLATFORM_MAKER"].values[np.newaxis][0].strip()}
+        )
+
+        def infer_network(this_ds):
+            if this_ds['PLATFORM_FAMILY'].values[np.newaxis][0].strip() == 'FLOAT_DEEP':
+                network = ['DEEP']
+                if len(this_ds['SENSOR'].values) > 4:
+                    network.append('BGC')
+
+            elif this_ds['PLATFORM_FAMILY'].values[np.newaxis][0].strip() == 'FLOAT':
+                if len(this_ds['SENSOR'].values) > 4:
+                    network = ['BGC']
+                else:
+                    network = ['CORE']
+
+            else:
+                network = ['?']
+
+            return network
+
+        data.update({"networks": infer_network(ds)})
+
+        data.update({"cycles": np.unique(self.open_dataset("prof")['CYCLE_NUMBER'])})
 
         self._metadata = data
-        return self._metadata
 
     def load_dac(self):
         """Load the DAC short name for this float"""

argopy/stores/float/implementations/argo_float_online.py

@@ -16,12 +16,15 @@
 
 class ArgoFloatOnline(ArgoFloatProto):
     _online = True
+    _eafleetmonitoring_server = "https://fleetmonitoring.euro-argo.eu"
+    _technicaldata = None
 
     def __init__(
         self,
+        *args,
         **kwargs,
     ):
-        super().__init__(**kwargs)
+        super().__init__(*args, **kwargs)
 
         if self.host_protocol == "s3":
             self.host = self.host.replace(
@@ -30,14 +33,13 @@ def __init__(
 
         # Load some data (in a perfect world, this should be done asynchronously):
         # self.load_index()
-        self.load_metadata()
+        self.load_metadata() # must come before load_dac
         self.load_dac()
 
     def load_metadata(self):
         """Load float meta data from Euro-Argo fleet-monitoring API"""
-        api_server = "https://fleetmonitoring.euro-argo.eu"
-        # api_point = f"{api_server}/floats/basic/{self.WMO}"
-        api_point = f"{api_server}/floats/{self.WMO}"
+        # api_point = f"{self._eafleetmonitoring_server}/floats/basic/{self.WMO}"
+        api_point = f"{self._eafleetmonitoring_server}/floats/{self.WMO}"
         self._metadata = httpstore(
             cache=self.cache, cachedir=self.cachedir
         ).open_json(api_point)
@@ -47,6 +49,22 @@ def load_metadata(self):
             self._metadata["deployment"]["launchDate"]
         )
 
+    def load_technicaldata(self):
+        """Load float meta data from Euro-Argo fleet-monitoring API"""
+        # api_point = f"{self._eafleetmonitoring_server}/technical-data/basic/{self.WMO}"
+        api_point = f"{self._eafleetmonitoring_server}/technical-data/{self.WMO}"
+        self._technicaldata = httpstore(
+            cache=self.cache, cachedir=self.cachedir
+        ).open_json(api_point)
+
+    @property
+    def technicaldata(self) -> dict:
+        """A dictionary holding float technical-data, based on the EA fleet-monitoring API json schema
+        """
+        if self._technicaldata is None:
+            self.load_technicaldata()
+        return self._technicaldata
+
     def load_dac(self):
         """Load the DAC short name for this float"""
         try:
@@ -59,29 +77,3 @@ def load_dac(self):
 
         # For the record, another method to get the DAC name, based on the profile index
         # self._dac = self.idx.search_wmo(self.WMO).read_dac_wmo()[0][0] # Get DAC from Argo index
-
-    def __repr__(self):
-        summary = ["<argofloat.%i.%s>" % (self.WMO, self.host_protocol)]
-        status = "online ✅" if isconnected(self.path, maxtry=1) else "offline 🚫"
-        summary.append("GDAC host: %s [%s]" % (self.host, status))
-        summary.append("DAC name: %s" % self.dac)
-        summary.append("Network(s): %s" % self.metadata['networks'])
-
-        launchDate = self.metadata["deployment"]["launchDate"]
-        today = pd.to_datetime('now', utc=True)
-        summary.append(
-            "Deployment date: %s [%s ago]"
-            % (launchDate, (today-launchDate).days)
-        )
-        summary.append(
-            "Float type and manufacturer: %s [%s]"
-            % (
-                self.metadata["platform"]["type"],
-                self.metadata["maker"],
-            )
-        )
-        summary.append("Number of cycles: %s" % self.N_CYCLES)
-        summary.append("Dashboard: %s" % dashboard(wmo=self.WMO, url_only=True))
-        summary.append("Netcdf dataset available: %s" % list(self.avail_dataset().keys()))
-
-        return "\n".join(summary)

argopy/stores/float/spec.py

@@ -28,6 +28,7 @@ def __init__(
         cache: bool = False,
         cachedir: str = "",
         timeout: int = 0,
+        **kwargs,
     ):
 
         self.WMO = check_wmo(wmo)[0]
@@ -51,6 +52,10 @@ def __init__(
             raise InvalidOption("Trying to work with remote host '%s' without a web connection. Check your connection parameters or try to work with a local GDAC path." % self.host)
 
 
+    def load_index(self):
+        """Load the Argo full index in memory and trigger search for this float"""
+        self.idx.load().search_wmo(self.WMO)
+
     @property
     def metadata(self) -> dict:
         """A dictionary holding float meta-data, based on the EA fleet-monitoring API json schema
@@ -59,7 +64,7 @@ def metadata(self) -> dict:
         self._metadata["deployment"]["launchDate"]  # pd.Datetime
         self._metadata['cycles']  # list
         self.metadata['networks']  # list of str
-        self.metadata["platform"]["type"]  #
+        self.metadata["platform"]["type"]  # from Reference table 23
         self.metadata["maker"]  #
         """
         if self._metadata is None:
@@ -71,15 +76,16 @@ def load_metadata(self):
         """Method to load float meta-data"""
         raise NotImplementedError("Not implemented")
 
-    def load_index(self):
-        """Load the Argo full index in memory and trigger search for this float"""
-        self.idx.load().search_wmo(self.WMO)
-
     @abstractmethod
     def load_dac(self):
         """Load the DAC short name for this float"""
         raise NotImplementedError("Not implemented")
 
+    @property
+    def dac(self) -> str:
+        """Name of the DAC responsible for this float"""
+        return self._dac
+
     @property
     def host_protocol(self) -> str:
         """Protocol of the GDAC host"""
@@ -90,11 +96,6 @@ def host_sep(self) -> str:
         """Host path separator"""
         return self.fs.fs.sep
 
-    @property
-    def dac(self) -> str:
-        """Name of the DAC responsible for this float"""
-        return self._dac
-
     @property
     def path(self) -> str:
         """Return float path
@@ -144,7 +145,7 @@ def N_CYCLES(self) -> int:
 
         If the float is still active, this is the current value.
         """
-        return len(self.fleetmonitoring_metadata['cycles'])
+        return len(self.metadata['cycles'])
 
     def describe_profiles(self) -> pd.DataFrame:
         """Return a :class:`pd.DataFrame` describing profile files"""
@@ -171,16 +172,17 @@ def describe_profiles(self) -> pd.DataFrame:
 
     def __repr__(self):
         summary = ["<argofloat.%i.%s>" % (self.WMO, self.host_protocol)]
-        status = "online ✅" if isconnected(self.path, maxtry=1) else "offline 🚫"
-        summary.append("GDAC host: %s [%s]" % (self.host, status))
+        # status = "online ✅" if isconnected(self.path, maxtry=1) else "offline 🚫"
+        # summary.append("GDAC host: %s [%s]" % (self.host, status))
+        summary.append("GDAC host: %s" % self.host)
         summary.append("DAC name: %s" % self.dac)
         summary.append("Network(s): %s" % self.metadata['networks'])
 
         launchDate = self.metadata["deployment"]["launchDate"]
         today = pd.to_datetime('now', utc=True)
         summary.append(
-            "Deployment date: %s [%s ago]"
-            % (launchDate, (today-launchDate).days)
+            "Deployment date: %s [%s days ago]"
+            % (launchDate.strftime("%Y-%m-%d %H:%M"), (today-launchDate).days)
         )
         summary.append(
             "Float type and manufacturer: %s [%s]"
@@ -190,7 +192,8 @@ def __repr__(self):
             )
         )
         summary.append("Number of cycles: %s" % self.N_CYCLES)
-        summary.append("Dashboard: %s" % dashboard(wmo=self.WMO, url_only=True))
+        if self._online:
+            summary.append("Dashboard: %s" % dashboard(wmo=self.WMO, url_only=True))
         summary.append("Netcdf dataset available: %s" % list(self.avail_dataset().keys()))
 
         return "\n".join(summary)