Hook up fallback parsing

PyRoute/Inputs/BaseTransformer.py

@@ -390,9 +390,21 @@ def trim_raw_string(self, tree):
                             self.raw = self.raw[index:]
 
     def _square_up_parsed_zero(self, rawstring, parsed):
+        from PyRoute.Inputs.ParseStarInput import ParseStarInput
         bitz = [item for item in rawstring.split(' ') if '' != item]
         if 3 == len(bitz) and bitz[0] == parsed['nobles'] and bitz[1] == parsed['base'] and bitz[2] == parsed['zone']:
             return parsed
+        if 2 == len(bitz) and "" == parsed['zone']:
+            if 2 < len(bitz[0]):  # bitz[0] can only possibly be nobles, so return
+                return parsed
+            if 1 < len(bitz[1]):  # if bitz[1] is more than one char, it can't be a trade zone, so return
+                return parsed
+            non_noble = [item for item in bitz[0] if item not in ParseStarInput.valid_nobles]
+            if 0 < len(non_noble):  # If one or more chars in bitz[0] is not a valid noble call, then we have a base code and trade zone
+                parsed['zone'] = parsed['base']
+                parsed['base'] = parsed['nobles']
+                parsed['nobles'] = ''
+                return parsed
         if 3 == len(bitz):
             parsed['nobles'] = bitz[0]
             parsed['base'] = bitz[1]

PyRoute/Inputs/ParseStarInput.py

@@ -21,9 +21,9 @@ class ParseStarInput:
     regex = """
 ^([0-3]\d[0-4]\d) +
 (.{15,}) +
-(\w\w\w\w\w\w\w-\w|\?\?\?\?\?\?\?-\?|[\w\?]{7,7}-[\w\?]) +
+([A-HXYa-hxy][0-9A-Fa-f]\w\w[0-9A-Fa-f][0-9A-Xa-x][0-9A-Ja-j]-\w|\?\?\?\?\?\?\?-\?|[A-HXYa-hxy\?][0-9A-Fa-f\?][\w\?]{2,2}[0-9A-Fa-f\?][0-9A-Xa-x\?][0-9A-Ja-j\?]-[\w\?]) +
 (.{15,}) +
-((\{ *[+-]?[0-6] ?\}|-) +(\([0-9A-Za-z]{3}[+-]\d\)|- ) +(\[[0-9A-Za-z]{4}\]|- )|( ) ( ) ( )) +
+((\{ *[ +-]?[0-6] ?\}|-) +(\([0-9A-Za-z]{3}[+-]\d\)|- ) +(\[[0-9A-Za-z]{4}\]|- )|( ) ( ) ( )) +
 ([BcCDeEfFGH]{1,5}|-| ) +
 ([A-Z]{1,3}|-|\*) +
 ([ARUFGBarufgb]|-| ) +
@@ -50,6 +50,8 @@ class ParseStarInput:
     station_parser = None
     station_transformer = None
     deep_space = {}
+    valid_zone = 'arufgbARUFGB-'
+    valid_nobles = 'BCcDEeFfGH-'
 
     @staticmethod
     def parse_line_into_star_core(star, line, sector, pop_code, ru_calc, fix_pop=False):
@@ -84,7 +86,7 @@ def parse_line_into_star_core(star, line, sector, pop_code, ru_calc, fix_pop=Fal
         if ('' == star.baseCode) or ('-' != star.baseCode and 1 == len(star.baseCode) and not star.baseCode.isalpha()):
             star.baseCode = '-'
         star.zone = data[13].strip()
-        if not star.zone or star.zone not in 'arufgbARUFGB-':
+        if not star.zone or star.zone not in ParseStarInput.valid_zone:
             star.zone = '-'
         star.zone = star.zone.upper()
         star.ggCount = 0 if (len(data[14]) < 3 or not data[14][2] or data[14][2] in 'X?') else int(data[14][2], 16)
@@ -117,7 +119,10 @@ def parse_line_into_star_core(star, line, sector, pop_code, ru_calc, fix_pop=Fal
         star.tradeCode.check_world_codes(star, fix_pop=fix_pop)
 
         if data[5] and data[5].startswith('{'):
-            imp = int(data[5][1:-1].strip())
+            raw_imp = data[5][1:-1].strip()
+            imp = 0
+            if '' != raw_imp:
+                imp = int(raw_imp)
             star.calculate_importance()
             if imp != star.importance:
                 star.logger.warning(
@@ -197,11 +202,15 @@ def _unpack_starline(star, line, sector):
             else:
                 result, line = ParseStarInput.parser.parse(line)
         except UnexpectedCharacters:
-            star.logger.error("Unmatched line: {}".format(line))
-            return None, None
+            result = ParseStarInput._unpack_starline_fallback(line)
+            if result is None:
+                star.logger.error("Unmatched line: {}".format(line))
+                return None, None
         except UnexpectedEOF:
-            star.logger.error("Unmatched line: {}".format(line))
-            return None, None
+            result = ParseStarInput._unpack_starline_fallback(line)
+            if result is None:
+                star.logger.error("Unmatched line: {}".format(line))
+                return None, None
         if ParseStarInput.transformer is None:
             ParseStarInput.transformer = StarlineTransformer(raw=line)
         else:
@@ -213,13 +222,47 @@ def _unpack_starline(star, line, sector):
             ParseStarInput.station_transformer.raw = line
             ParseStarInput.station_transformer.crankshaft = False
 
-        if is_station:
-            transformed = ParseStarInput.station_transformer.transform(result)
+        if not isinstance(result, list):
+            if is_station:
+                transformed = ParseStarInput.station_transformer.transform(result)
+            else:
+                transformed = ParseStarInput.transformer.transform(result)
         else:
-            transformed = ParseStarInput.transformer.transform(result)
+            transformed = result
 
         return transformed, is_station
 
+    @staticmethod
+    def _unpack_starline_fallback(line):
+        matches = ParseStarInput.starline.match(line)
+        if matches is None:
+            return
+        data = list(matches.groups())
+        parsed = {'position': data[0], 'name': data[1], 'uwp': data[2], 'trade': data[3]}
+        raw_extensions = data[4].replace('  ', ' ').replace('{ ', '{').replace(' }', '}')
+        if 2 <= raw_extensions.count(' '):
+            bitz = raw_extensions.split(' ')
+            parsed['ix'] = bitz[0]
+            parsed['ex'] = bitz[1]
+            parsed['cx'] = bitz[2]
+        else:
+            parsed['ix'] = ' ' if data[8] is None else data[8]
+            parsed['ex'] = ' ' if data[9] is None else data[9]
+            parsed['cx'] = ' ' if data[10] is None else data[10]
+        parsed['nobles'] = data[11]
+        parsed['base'] = data[12]
+        parsed['zone'] = data[13]
+        parsed['pbg'] = data[14]
+        parsed['worlds'] = data[15]
+        parsed['allegiance'] = data[16]
+        parsed['residual'] = data[17]
+
+        extensions = parsed['ix'] + ' ' + parsed['ex'] + ' ' + parsed['cx']
+
+        spacer = ' '
+        data = [parsed['position'], parsed['name'], parsed['uwp'], parsed['trade'], extensions, parsed['ix'], parsed['ex'], parsed['cx'], spacer, spacer, spacer, parsed['nobles'], parsed['base'], parsed['zone'].upper(), parsed['pbg'], parsed['worlds'], parsed['allegiance'], parsed['residual']]
+        return data
+
     @staticmethod
     def check_tl(star, fullmsg=None):
         if '???-' in str(star.uwp) or star.tl_unknown:

PyRoute/Inputs/StarlineParser.py

@@ -20,7 +20,7 @@ class StarlineParser:
 
     position: /^([0-3]\d[0-4]\d)/
 
-    starname: /(.{15,}) ([\w\?]{7,7}-[\w\?]) /
+    starname: /(.{15,}) ([A-HXYa-hxy\?][0-9A-Fa-f\?][\w\?]{2,2}[0-9A-Fa-f\?][0-9A-Xa-x\?][0-9A-Ja-j\?]-[\w\?]) /
 
     trade: TRADECODE*
     TRADECODE: MINOR_DIEBACK | BINARY | POPCODE | MINOR_SOPHONT | OWNED_COLONY | MAJOR_SOPHONT | RESIDUAL | SINGLETON
@@ -35,7 +35,7 @@ class StarlineParser:
 
     extensions: ix ex cx | /( ) ( ) ( )/
 
-    ix: /\{ *[+-]?[0-6] ?\}|-/
+    ix: /\{ *[ +-]?[0-6] ?\}|-/
     ex: /\([0-9A-Za-z]{3}[+-]\d\)|-/
     cx: /(\[[0-9A-Za-z]{4}[\]\}]|-)/
 

PyRoute/Inputs/StarlineStationParser.py

@@ -20,7 +20,7 @@ class StarlineStationParser:
 
     position: /^([0-3]\d[0-4]\d)/
 
-    starname: /(.{15,}) ([\w\?]{7,7}-[\w\?]) /
+    starname: /(.{15,}) ([A-HXYa-hxy\?][0-9A-Fa-f\?][\w\?]{2,2}[0-9A-Fa-f\?][0-9A-Xa-x\?][0-9A-Ja-j\?]-[\w\?]) /
 
     trade: TRADECODE*
     TRADECODE: MINOR_DIEBACK | BINARY | POPCODE | MINOR_SOPHONT | OWNED_COLONY | MAJOR_SOPHONT | RESIDUAL | SINGLETON
@@ -35,7 +35,7 @@ class StarlineStationParser:
 
     extensions: ix ex cx | /( ) ( ) ( )/
 
-    ix: /\{ *[+-]?[0-6] ?\}/
+    ix: /\{ *[ +-]?[0-6] ?\}/
     ex: /\([0-9A-Za-z]{3}[+-]\d\)|-/
     cx: /(\[[0-9A-Za-z]{4}[\]\}]|-)/
 

PyRoute/SystemData/UWP.py

@@ -13,7 +13,7 @@
 
 class UWP(object):
     # Port code, size, atmo, hydro, pop, gov, law, the all-important hyphen, then TL
-    match_string = '^([A-HXYa-hxy\?])([0-9A-Fa-f\?])([0-9A-Za-z\?])([0-9A-Za-z\?])([0-9A-Fa-f\?])([0-9A-Za-z\?])([0-9A-Ja-j\?])-([0-9A-Za-z\?])'
+    match_string = '^([A-HXYa-hxy\?])([0-9A-Fa-f\?])([0-9A-Za-z\?])([0-9A-Za-z\?])([0-9A-Fa-f\?])([0-9A-Xa-x\?])([0-9A-Ja-j\?])-([0-9A-Za-z\?])'
 
     match = re.compile(match_string)
 

Tests/Hypothesis/Inputs/testHypothesisStarlineParser.py

@@ -81,7 +81,8 @@ def comparison_line(draw):
         '0000 000000000000000 ???????-? (0 0000000000)A       - - A 000    ?0',
         '2732 sa0)OkeWHEk4Mmb ???????-? (8PvzPS]UbZypt6 - (8Pv-2) [GZks] f  * g 88C   E7',
         '0000 000000000000000 ???????-? [00000000000]0]       - - A 000    ?0',
-        '0000 000000000000000 ???????-? (00000000000000 - (000-0) [0000] BB     - A 000   ?0'
+        '0000 000000000000000 ???????-? (00000000000000 - (000-0) [0000] BB     - A 000   ?0',
+        '0000 000000000000000 ???????-? (00000000000)0)       - - A 000   00?'
     ]
 
     candidate = draw(from_regex(regex=ParseStarInput.starline, alphabet='0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWYXZ -{}()[]?\'+*'))
@@ -320,6 +321,7 @@ class testHypothesisStarlineParser(unittest.TestCase):
     @example('0001 000000000000000 ???????-? 000000000000000       - - A 000   --0', False)
     @example('0111 -2Z4ig11RbxW010 0000001-1 wJED9E(E(T (HN6 -  (113-0) -  - - B  114       00y', False)
     @example('0000 000000000000000 ???????-? [0000000000]00]       - - A 000    ?0', False)
+    @example('0000 000000000000000 ???????-? (00000000000)0)       - - A 000   00?', False)
     # Weird parsing cases
     @example('0000 000000000000000 ???????-? (00000000000000       - - 0 000   00?)', 'weird')
     @example('0000 000000000000000 ???????-? [00000000000000       - - 0 000   00?]', 'weird')

Tests/Hypothesis/testStar.py

@@ -116,6 +116,14 @@ def setUp(self) -> None:
     @example('0101 000000000000000 A0000y0-0 000000000000000       - - 0 000   0000D')
     @example('0110 000000000000000 ???????-? 000000000000000       H B A 000 0 00')
     @example('0110 000000000000000 ???????-? 000000000000000       e - A 000   00')
+    @example('0000 000000000000000 0000000-0 0000000000000 B       - A A 000   00')
+    @example('0000 000000000000000 0000000-0 [00000000000000 - -  [0000] - - A 000   00')
+    @example('0110 000000000000000 ???????-? 0000000000 (000 - (000-0) [0000] - - A 000 0 00')
+    @example('0101 000000000000000 0000000-0 000000000000000       BB A      A 000 0 00')
+    @example('0101 000000000000000 ???????-? 0000000000000 0 0       B - A 000   00')
+    @example('0101 000000000000000 ???????-? 000000000000000 {-0} (000-0)     -    * A 000   00')
+    @example('0101 000000000000000 ???????-? 000000000000000 {-0} (000-0) -  - - A 000   00')
+    @example("0101 000000000000000  ?000000-0                 {3 }  (7TG-1) [2IBt]   EHEG  -    569 9 C5'")
     def test_parse_line_to_star(self, s):
         hyp_line = "Hypothesis input: " + s
         sector = Sector('# Core', '# 0, 0')