Tweak SPC outlook parser to ingest all archival products

lib/xmet/spc.py

@@ -25,9 +25,9 @@ RE_HEADER = re.compile(r'''
     \s+ (?P<day>\d+)
     \s+ CONVECTIVE
     \s+ OUTLOOK
-''', re.X)
+''', re.IGNORECASE | re.X)
 
-RE_OFFICE = re.compile(r'.* STORM PREDICTION CENTER .*')
+RE_OFFICE = re.compile(r'.*STORM PREDICTION CENTER.*')
 
 RE_ISSUANCE = re.compile(r'''
        ^(?P<hour>\d{2})
@@ -38,7 +38,7 @@ RE_ISSUANCE = re.compile(r'''
     \s+ (?P<month>[A-Z]{3})
     \s+ (?P<day>\d{2})
     \s+ (?P<year>\d{4})$
-''', re.X)
+''', re.IGNORECASE | re.X)
 
 RE_VALIDITY = re.compile(r'''
     ^VALID \s+ TIME
@@ -49,7 +49,7 @@ RE_VALIDITY = re.compile(r'''
     \s+ (?P<day_end>\d{2})
         (?P<hour_end>\d{2})
         (?P<minute_end>\d{2})Z$
-''', re.X)
+''', re.IGNORECASE | re.X)
 
 RE_AREA_TYPE = re.compile(r'^(?P<type>[A-Z ]+) OUTLOOK POINTS DAY .*')
 
@@ -57,14 +57,14 @@ RE_HAZARD = re.compile(r'''
     ^(?:\.\.\.)
     \s+ (?P<type>[A-Z ]+)
     \s+ (?:\.\.\.)$
-''', re.X)
+''', re.IGNORECASE | re.X)
 
 RE_POINTS_START = re.compile(r'''
     ^(?P<category>[A-Z0-9\.]+)
      (?P<rest>(?:\s+\d{8}){1,6})
-''', re.X)
+''', re.IGNORECASE | re.X)
 
-RE_POINTS = re.compile(r'^(?:\s+\d{8}){1,6}$')
+RE_POINTS = re.compile(r'^(?:\s+\d{8}){1,6}$', re.IGNORECASE)
 
 CITIES = {
     'CA': (
@@ -174,7 +174,12 @@ def each_poly(parts: list[str]):
     #
     # 3. Convert segments into what they are
     #
+    try:
         polygons, interiors, linestrings = convert_segments(segments)
+    except shapely.errors.TopologicalError:
+        return
+    except shapely.errors.GEOSException:
+        return
 
     #
     # We do our winding logic now
@@ -203,6 +208,10 @@ def each_poly(parts: list[str]):
         polygons[i] = polygon.buffer(0)
 
     for polygon in polygons:
+        if polygon.geom_type == 'MultiPolygon':
+            for poly in polygon.geoms:
+                yield poly
+        else:
             yield polygon
 
 class SPCOutlookArea(DatabaseTable):
@@ -408,7 +417,7 @@ class SPCOutlookParser():
         day_start   = int(match['day_start'])
         day_end     = int(match['day_end'])
 
-        if day_start > day_end:
+        if day_start > day_end and month_end == 12:
             month_end = (month_end + 1) % 12
 
         if month_start > month_end:
@@ -446,11 +455,16 @@ class SPCOutlookParser():
                 if self.category == 'SIGN':
                     area.probability = None
                     area.sig         = True
+
+                    self.outlook.probabilities.append(area)
                 else:
+                    try:
                         area.probability = float(self.category)
                         area.sig         = False
 
                         self.outlook.probabilities.append(area)
+                    except ValueError:
+                        pass
             elif self.area_type == 'CATEGORICAL':
                 area = SPCOutlookCategoryArea()
                 area.category = self.category
@@ -540,7 +554,11 @@ class SPCOutlookParser():
             elif self.state is SPCOutlookParserState.ISSUANCE:
                 self.parse_issuance(line)
             elif self.state is SPCOutlookParserState.VALIDITY:
+                try:
                     self.parse_validity(line)
+                except SPCOutlookParserException:
+                    self.outlook.timestamp_start = self.outlook.timestamp_issued
+                    self.outlook.timestamp_end   = self.outlook.timestamp_issued + datetime.timedelta(days = 1)
             elif self.state is SPCOutlookParserState.AREA_THREAT:
                 self.parse_area_hazard(line)
             elif self.state is SPCOutlookParserState.BODY: