Begin refactoring RAOB parser

lib/xmet/raob.py

@@ -60,39 +60,19 @@ class RAOBSounding():
 
         return sounding
 
-    def sample(self, pressure: float):
+    def record(self, sample: SoundingSample):
-        if pressure in self.samples:
+        dest = SoundingSample()
-            return self.samples[pressure]
+        dest.pressure_qa = ' '
+        dest.height_qa   = ' '
+        dest.temp_qa     = ' '
+        dest.pressure    = sample.pressure
+        dest.height      = sample.height
+        dest.temp        = sample.temp
+        dest.dewpoint    = sample.dewpoint
+        dest.wind_speed  = sample.wind_speed
+        dest.wind_dir    = sample.wind_dir
 
-        sample = SoundingSample()
+        self.samples[sample.pressure] = dest
-        sample.pressure    = pressure
-        sample.pressure_qa = ' '
-        sample.height_qa   = ' '
-        sample.temp_qa     = ' '
-
-        self.samples[pressure] = sample
-
-        return sample
-
-    def record_height(self, pressure: float, height: float):
-        sample = self.sample(pressure)
-        sample.height = height
-
-    def record_temp_dewpoint(self,
-                             pressure: float,
-                             temp: float,
-                             dewpoint: float):
-        sample = self.sample(pressure)
-        sample.temp     = temp
-        sample.dewpoint = dewpoint
-
-    def record_wind_speed_dir(self,
-                              pressure: float,
-                              wind_speed: float,
-                              wind_dir: float):
-        sample = self.sample(pressure)
-        sample.wind_speed = wind_speed
-        sample.wind_dir   = wind_dir
 
 class RAOBObs():
     DATA_SOURCE = 'UCAR'
@@ -312,7 +292,7 @@ class RAOBObs():
 
         return sample
 
-    def parse_ttaa(self) -> Sounding:
+    def parse_ttaa(self) -> dict:
         #
         # Return None if there is no height data up to 100mb.
         #
@@ -330,16 +310,9 @@ class RAOBObs():
         if sample is None:
             return
 
-        timestamp = self.parse_timestamp(self.tokens[0])
+        station   = self.tokens[1]
-
+        timestamp = self.tokens[0]
-        sounding = Sounding()
+        samples   = list()
-        sounding.samples = [sample]
-        sounding.station = self.tokens[1]
-
-        sounding.data_source_pressure = self.DATA_SOURCE
-        sounding.data_source_other    = self.DATA_SOURCE
-        sounding.timestamp_observed   = timestamp
-        sounding.timestamp_released   = timestamp - datetime.timedelta(minutes=45)
 
         for i in range(5, len(self.tokens), 3):
             if len(self.tokens) < i+3 or self.tokens[i][-1] == '=':
@@ -356,9 +329,13 @@ class RAOBObs():
             if sample is None:
                 continue
 
-            sounding.samples.append(sample)
+            samples.append(sample)
 
-        return sounding
+        return {
+            'station':   station,
+            'timestamp': timestamp,
+            'samples':   samples
+        }
 
 class RAOBChunk():
     def __init__(self,
@@ -390,15 +367,15 @@ class RAOBChunk():
         if obs is not None:
             yield obs
 
-    def each_sounding(self):
+    def each_data(self):
         for obs in self.each_obs():
             if obs.kind == 'TTAA':
-                sounding = obs.parse_ttaa()
+                data = obs.parse_ttaa()
 
-                if sounding is None or len(sounding.samples) == 0:
+                if data is None or len(data['samples']) == 0:
                     continue
 
-                yield sounding
+                yield data
 
 class RAOBReader():
     """
@@ -414,6 +391,18 @@ class RAOBReader():
         self.fh        = fh
         self.soundings = dict()
 
+    def sounding(self, station: str, timestamp: str) -> RAOBSounding:
+        key = station + '.' + timestamp
+
+        if key in self.soundings:
+            return self.soundings[key]
+
+        sounding = RAOBSounding(station, timestamp)
+
+        self.soundings[key] = sounding
+
+        return sounding
+
     def parse_chunk(self, text: str) -> RAOBChunk:
         meta = {
             'wfo':     None, # NWS forecast office
@@ -474,26 +463,34 @@ class RAOBReader():
             tokens.extend(re.split(r'\s+', line))
 
         return RAOBChunk(meta['wfo'],
-                           meta['product'],
+                         meta['product'],
-                           tokens)
+                         tokens)
 
     def each_chunk(self):
         for text in each_chunk(self.fh, CHUNK_SEP, CHUNK_STRIP_CHARS):
             yield self.parse_chunk(text)
 
-    def each_obs(self):
+    def load_soundings(self):
         for chunk in self.each_chunk():
-            yield from chunk.each_obs()
+            for data in chunk.each_data():
+                station   = data['station']
+                timestamp = data['timestamp']
+                samples   = data['samples']
 
-    def each_sounding(self):
+                sounding = self.sounding(station, timestamp)
-        for chunk in self.each_chunk():
+
-            yield from chunk.each_sounding()
+                for sample in samples:
+                    sounding.record(sample)
+
+        for key in self.soundings:
+            yield self.soundings[key].finish()
 
     @staticmethod
     def each_sounding_from_fh(fh: io.TextIOBase):
         reader = RAOBReader(fh)
 
-        yield from reader.each_sounding()
+        for sounding in reader.load_soundings():
+            yield sounding
 
     @staticmethod
     def each_sounding_from_file(path: str):