removed the usage of parameter in parsing (#1214)

echopype/convert/api.py

@@ -411,17 +411,11 @@ def open_raw(
     # Check file extension and existence
     file_chk, xml_chk = _check_file(raw_file, sonar_model, xml_path, storage_options)
 
-    # TODO: the if-else below only works for the AZFP vs EK contrast,
-    #  but is brittle since it is abusing params by using it implicitly
-    if SONAR_MODELS[sonar_model]["xml"]:
-        params = xml_chk
-    else:
-        params = "ALL"  # reserved to control if only wants to parse a certain type of datagram
-
     # Parse raw file and organize data into groups
     parser = SONAR_MODELS[sonar_model]["parser"](
         file_chk,
-        params=params,
+        # Currently used only for AZFP XML File
+        file_meta=xml_chk,
         storage_options=storage_options,
         sonar_model=sonar_model,
     )

echopype/convert/parse_ad2cp.py

@@ -219,7 +219,9 @@ class NoMorePackets(Exception):
 
 
 class ParseAd2cp(ParseBase):
-    def __init__(self, file, params, storage_options={}, dgram_zarr_vars={}, sonar_model="AD2CP"):
+    def __init__(
+        self, file, file_meta, storage_options={}, dgram_zarr_vars={}, sonar_model="AD2CP"
+    ):
         super().__init__(file, storage_options, sonar_model)
         self.config = None
         self.packets: List[Ad2cpDataPacket] = []

echopype/convert/parse_azfp.py

@@ -73,12 +73,12 @@ class ParseAZFP(ParseBase):
     HEADER_FORMAT = ">HHHHIHHHHHHHHHHHHHHHHHHHHHHHHHHHHHBBBBHBBBBBBBBHHHHHHHHHHHHHHHHHHHH"
     FILE_TYPE = 64770
 
-    def __init__(self, file, params, storage_options={}, dgram_zarr_vars={}, sonar_model="AZFP"):
+    def __init__(self, file, file_meta, storage_options={}, dgram_zarr_vars={}, sonar_model="AZFP"):
         super().__init__(file, storage_options, sonar_model)
         # Parent class attributes
         #  regex pattern used to grab datetime embedded in filename
         self.timestamp_pattern = FILENAME_DATETIME_AZFP
-        self.xml_path = params
+        self.xml_path = file_meta
 
         # Class attributes
         self.parameters = defaultdict(list)

echopype/convert/parse_base.py

@@ -43,7 +43,7 @@ def _print_status(self):
 class ParseEK(ParseBase):
     """Class for converting data from Simrad echosounders."""
 
-    def __init__(self, file, params, storage_options, sonar_model):
+    def __init__(self, file, storage_options, sonar_model):
         super().__init__(file, storage_options, sonar_model)
 
         # Parent class attributes
@@ -59,7 +59,6 @@ def __init__(self, file, params, storage_options, sonar_model):
         self.ch_ids = defaultdict(
             list
         )  # Stores the channel ids for each data type (power, angle, complex)
-        self.data_type = self._select_datagrams(params)
 
         self.nmea = defaultdict(list)  # Dictionary to store NMEA data(timestamp and string)
         self.mru = defaultdict(list)  # Dictionary to store MRU data (heading, pitch, roll, heave)
@@ -371,19 +370,8 @@ def parse_raw(self):
                         # and in the form of floats separated by semicolons
                         v["pulse_duration"] = [float(x) for x in v["pulse_length"].split(";")]
 
-            # If exporting to XML file (EK80/EA640 only), print a message
-            if "print_export_msg" in self.data_type:
-                if "ENV" in self.data_type:
-                    xml_type = "environment"
-                elif "CONFIG" in self.data_type:
-                    xml_type = "configuration"
-                logger.info(f"exporting {xml_type} XML file")
-                # Don't parse anything else if only the config xml is required.
-                if "CONFIG" in self.data_type:
-                    return
-            # If not exporting to XML, print the usual converting message
-            else:
-                self._print_status()
+            # print the usual converting message
+            self._print_status()
 
             # Check if reading an ME70 file with a CON1 datagram.
             next_datagram = fid.peek()
@@ -398,11 +386,10 @@ def parse_raw(self):
             # Read the rest of datagrams
             self._read_datagrams(fid)
 
-        if "ALL" in self.data_type:
-            # Convert ping time to 1D numpy array, stored in dict indexed by channel,
-            #  this will help merge data from all channels into a cube
-            for ch, val in self.ping_time.items():
-                self.ping_time[ch] = np.array(val, dtype="datetime64[ns]")
+        # Convert ping time to 1D numpy array, stored in dict indexed by channel,
+        #  this will help merge data from all channels into a cube
+        for ch, val in self.ping_time.items():
+            self.ping_time[ch] = np.array(val, dtype="datetime64[ns]")
 
     def _read_datagrams(self, fid):
         """Read all datagrams.
@@ -494,25 +481,13 @@ def _read_datagrams(self, fid):
 
             num_datagrams_parsed += 1
 
-            # Skip any datagram that the user does not want to save
-            if (
-                not any(new_datagram["type"].startswith(dgram) for dgram in self.data_type)
-                and "ALL" not in self.data_type
-            ):
-                continue
-
             # XML datagrams store environment or instrument parameters for EK80
             if new_datagram["type"].startswith("XML"):
-                if new_datagram["subtype"] == "environment" and (
-                    "ENV" in self.data_type or "ALL" in self.data_type
-                ):
+                if new_datagram["subtype"] == "environment":
                     self.environment = new_datagram["environment"]
                     self.environment["xml"] = new_datagram["xml"]
                     self.environment["timestamp"] = new_datagram["timestamp"]
-                    # Don't parse anything else if only the environment xml is required.
-                    if "ENV" in self.data_type:
-                        break
-                elif new_datagram["subtype"] == "parameter" and ("ALL" in self.data_type):
+                elif new_datagram["subtype"] == "parameter":
                     current_parameters = new_datagram["parameter"]
 
             # RAW0 datagrams store raw acoustic data for a channel for EK60
@@ -661,17 +636,3 @@ def pad_shorter_ping(data_list) -> np.ndarray:
         else:
             out_array = np.array(data_list)
         return out_array
-
-    def _select_datagrams(self, params):
-        """Translates user input into specific datagrams or ALL
-
-        Valid use cases:
-        # get GPS info only (EK60, EK80)
-        # ec.to_netcdf(data_type='GPS')
-
-        # get configuration XML only (EK80)
-        # ec.to_netcdf(data_type='CONFIG')
-
-        # get environment XML only (EK80)
-        # ec.to_netcdf(data_type='ENV')
-        """

echopype/convert/parse_ek60.py

@@ -4,14 +4,5 @@
 class ParseEK60(ParseEK):
     """Class for converting data from Simrad EK60 echosounders."""
 
-    def __init__(self, file, params, storage_options={}, sonar_model="EK60"):
-        super().__init__(file, params, storage_options, sonar_model)
-
-    def _select_datagrams(self, params):
-        # Translates user input into specific datagrams or ALL
-        if params == "ALL":
-            return ["ALL"]
-        elif params == "GPS":
-            return ["NME"]
-        else:
-            raise ValueError("Unknown data type", params)
+    def __init__(self, file, file_meta, storage_options={}, sonar_model="EK60"):
+        super().__init__(file, storage_options, sonar_model)

echopype/convert/parse_ek80.py

@@ -4,44 +4,6 @@
 class ParseEK80(ParseEK):
     """Class for converting data from Simrad EK80 echosounders."""
 
-    def __init__(self, file, params, storage_options={}, sonar_model="EK80"):
-        super().__init__(file, params, storage_options, sonar_model)
+    def __init__(self, file, file_meta, storage_options={}, sonar_model="EK80"):
+        super().__init__(file, storage_options, sonar_model)
         self.environment = {}  # dictionary to store environment data
-
-    def _select_datagrams(self, params):
-        """Translates user input into specific datagrams or ALL"""
-
-        def translate_to_dgram(s):
-            if s == "ALL":
-                return ["ALL"]
-            # The GPS flag indicates that only the NME and MRU datagrams are parsed.
-            # It is kept in the list because it is used in SetGroups
-            # to flag that only the platform group is saved.
-            elif s == "GPS":
-                return ["NME", "MRU"]
-            # CONFIG flag indicates that only the configuration XML is parsed.
-            # The XML flag is not needed because the configuration is always
-            # the first datagram parsed.
-            elif s == "CONFIG":
-                return ["CONFIG"]
-            # XML flag indicates that XML0 datagrams should be read.
-            # ENV flag indicates that of the XML datagrams,
-            # only keep the environment datagrams
-            elif s == "ENV":
-                return ["XML", "ENV"]
-            # EXPORT_XML flag passed in only by the to_xml function
-            # Used to print the export message when writing to an xml file
-            elif s == "EXPORT_XML":
-                return ["print_export_msg"]
-            else:
-                raise ValueError("Unknown data type", params)
-
-        # Params is a string when user sets data_type in to_netcdf/to_zarr
-        if isinstance(params, str):
-            dgrams = translate_to_dgram(params)
-        # Params is a list when the parse classes are called by to_xml
-        else:
-            dgrams = []
-            for p in params:
-                dgrams += translate_to_dgram(p)
-        return dgrams

echopype/tests/convert/test_parse_base.py

@@ -42,14 +42,12 @@ def test__print_status(self):
 class TestParseEK:
     file = "./my_file.raw"
     storage_options = {}
-    params = "ALL"
     data_types = ["power", "angle", "complex"]
     raw_types = ["receive", "transmit"]
 
     def _get_parser(self, sonar_model, ping_data_dict):
         parser = ParseEK(
             file=self.file,
-            params=self.params,
             storage_options=self.storage_options,
             sonar_model=sonar_model,
         )
@@ -61,7 +59,6 @@ def _get_parser(self, sonar_model, ping_data_dict):
     def test_constructor(self, sonar_model):
         parser = ParseEK(
             file=self.file,
-            params=self.params,
             storage_options=self.storage_options,
             sonar_model=sonar_model,
         )